Research in Economic History Volume 26

Citation preview

RESEARCH IN ECONOMIC HISTORY Series Editor: Alexander J. Field

RESEARCH IN ECONOMIC HISTORY

VOLUME 26

RESEARCH IN ECONOMIC HISTORY EDITED BY

ALEXANDER J. FIELD Department of Economics, Santa Clara University, USA CO-EDITED BY

GREGORY CLARK Department of Economics, University of California, Davis, USA

WILLIAM A. SUNDSTROM Department of Economics, Santa Clara University, USA

United Kingdom – North America – Japan India – Malaysia – China

JAI Press is an imprint of Emerald Group Publishing Limited Howard House, Wagon Lane, Bingley BD16 1WA, UK First edition 2008 Copyright r 2008 Emerald Group Publishing Limited Reprints and permission service Contact: [email protected] No part of this book may be reproduced, stored in a retrieval system, transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without either the prior written permission of the publisher or a licence permitting restricted copying issued in the UK by The Copyright Licensing Agency and in the USA by The Copyright Clearance Center. No responsibility is accepted for the accuracy of information contained in the text, illustrations or advertisements. The opinions expressed in these chapters are not necessarily those of the Editor or the publisher. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN: 978-1-84855-336-1 ISSN: 0363-3268 (Series)

Awarded in recognition of Emerald’s production department’s adherence to quality systems and processes when preparing scholarly journals for print

LIST OF CONTRIBUTORS James L. Butkiewicz

Department of Economics, University of Delaware, Newark, DE, USA

Carlo Ciccarelli

Faculty of Economics, University of Rome ‘‘Tor Vergata’’, Rome, Italy

Stefano Fenoaltea

Faculty of Economics, University of Rome ‘‘Tor Vergata’’, Rome, Italy

Bernard Harris

Division of Sociology and Social Policy, School of Social Sciences, University of Southampton, Southampton, UK

Mark T. Kanazawa

Department of Economics, Carleton College, Northfield, MN, USA

Peter G. Klein

Department of Economics, University of Missouri, Columbia, MO, USA

Markus Lampe

Institute for Economic and Social History, University of Mu¨nster, Mu¨nster, Westfalen, Germany

Jason E. Taylor

Department of Economics, Central Michigan University, Mount Pleasant, MI, USA

vii

EDITOR’S INTRODUCTION Volume 26 of Research in Economic History includes six papers, evenly divided between European and North American topics. Beginning with the European topics, we have two data intensive papers and a survey/synthesis. Stefano Fenoaltea and Carlo Ciccarelli provide new estimates of social overhead investment in the different regions of Italy. This is followed by Markus Lampe’s paper reporting data on bilateral trade flows in Europe between 1857 and 1875. The third contribution on a European topic is Bernard Harris’s survey of the literature on gender, health, and welfare in England and Wales since industrialization. Turning west, Mark Kanazawa studies conflicts between ranchers and miners over who should bear the burden of taxation in nineteenth century California. The second paper on a US topic is Jason Taylor and Peter Klein’s study of cartel behavior under the National Industrial Recovery Act. Finally, James Butkiewicz mines archival material to provide a new perspective on and some rehabilitation of Eugene Meyer’s role as Governor of the Federal Reserve Board between 1930 and 1933. Research in Economic History welcomes innovative contributions to economic history in any area. We continue to have more flexibility than other outlets in publishing data-rich papers running to somewhat greater length. Potential authors may submit their work in hard copy or (this is preferred) as attachments in an email addressed to the editor (afield@ scu.edu). We are flexible on matters of style and formatting for the first round of submission. If it looks as if the paper has a reasonable likelihood of acceptance, we will ask that you prepare it according to JAI/Emerald guidelines, and are happy at any stage to send you a copy of these guidelines as well as a recently prepared manuscript for use as a template. Authors may wish to consult a recent volume of Research in Economic History for examples of house style. Alexander J. Field Series Editor

ix

SOCIAL-OVERHEAD CONSTRUCTION IN ITALY’S REGIONS, 1861–1913 Carlo Ciccarelli and Stefano Fenoaltea ABSTRACT This article presents estimates of social-overhead construction in Italy’s regions. The new-construction series point to a largely common cycle in non-railway work, and largely idiosyncratic bursts of railway building. Maintenance doubles as an index of the underlying stock, which cannot be calculated from the flows alone; one finds limited convergence, and only in railway infrastructure. Industrial and overall growth are increasingly correlated both with the initial stock, and with its increment. Direct measures of welfare improvements are uncertain, but the relative increases in draftees’ mean heights correlate in particular with socialoverhead investment.

1. INTRODUCTION The failure of Southern Italy to match the North’s industrial growth in the decades that followed Unification (1861) has attracted endless analysis and interpretation – and, until comparatively recently, little effort to establish the facts of the case. The first systematic cross-section estimates of regional Research in Economic History, Volume 26, 1–80 Copyright r 2008 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(08)26001-5

1

2

CARLO CICCARELLI AND STEFANO FENOALTEA

industrial production, for 1911 alone, were pioneered by Vera Zamagni in the 1970s. A second benchmark (c. 1891), by Alfredo Esposto, appeared over a decade later; and after still another decade these were joined by the first diachronic estimates, limited however to the four census years (1871, 1881, 1901, 1911). Since then, however, the systematic reconstruction of regional production series has gotten under way, and the first results are already in the public domain.1 This paper presents regional time-series estimates of social-overhead construction. These distinguish total new construction and total maintenance, and the corresponding subtotals for railways (and tramways), other civil projects, and military or naval projects; they cover work on all structures, privately or publicly financed, other than buildings.2 As components of broader aggregates, they measure production by the corresponding value added at constant (1911) prices; but they serve as more than that.3 The new-construction series double directly as measures of (1911-price) investment, as in the three broad groups identified here the (1911-price) ratio of value added to value was virtually identical. These series point to a largely common (Kuznets) cycle in non-railway investment, with a sharp upswing in the pre-war boom. Railway investment appears much more idiosyncratic, but this may reflect no more than scale, and does not in fact imply that railway and non-railway investment responded to different cyclical influences. Mean levels of new construction, standardized for regional size, depend on whether size is related to surface area or to population; they tend to vary less on the latter measure, but on either basis the greatest flows appear to have been captured by Latium, with the national capital, and even more, unexpectedly, by Liguria. Maintenance work, which shows little cyclical variation, outweighed new construction when the latter was particularly depressed. The maintenance figures serve here as indices of the social-overhead capital stock; the latter cannot be calculated from the flows alone, as the relevant capital goods were extremely long-lived. In general, these stock indices point to a measure of convergence in the endowment of railways, but not in other civil infrastructure. The investigation of the relationship between social-overhead capital and growth is constrained by the available evidence. The reconstructed benchmarks suggest that industrial and overall growth did not always go hand in hand – the overall leaders include Apulia and Sicily, where industry grew little – but that on either measure growth was increasingly associated both with initial infrastructure, and with current additions to infrastructure.

Social-Overhead Construction in Italy’s Regions, 1861–1913

3

The increasing role of the initial stock may reflect the transition from artisanal production to more infrastructure-sensitive factory production; the increasing association with current investment may reflect a transition from remedial investment in the less-endowed regions to adaptive investment in the regions where strong growth strained the available infrastructure. The relationship between social-overhead capital and other measures of welfare is even harder to assess, not least because the measures themselves are highly tentative; the present aggregate estimates yield very weak results, but these may mask lower-level associations.

2. THE NEW ESTIMATES The annual regional estimates for railway, other civil, military, and total new construction (Appendix A) and maintenance (Appendix B) are, for the period at hand, the first of their kind. In the extant literature, the regional series are limited to State spending for new civil construction (Picci, 2002); local-government spending has been estimated only for aggregates of regions, and for a handful of benchmarks (De Nardo, 1957); and privately funded (non-rail) infrastructure has been neglected altogether.4 In general, the new series replicate, at the regional level, the corresponding national figures; like these, they are built up from a broad spectrum of real and financial indicators of new construction on the one hand and maintenance on the other. The estimates of construction for railways and private utilities are typically real measures, combined with 1911-price value added weights; the others are derived from public budget data (and estimates of complementary private spending), duly deflated and scaled by the (1911-price) ratio of value added to value. State-spending data are typically ex post, annual, and highly detailed. Much of the regional evidence used here is taken from existing retrospectives; only a minor residual needs to be calculated by drawing directly on the public budgets, and only part of that (concentrated, however, in military works) lacks a regional tag in the source.5 Municipal and provincial budget data are all regionally tagged, but ex ante, and, especially after 1900, available only for sample years; the estimates of complementary private spending are typically crude, and crudely allocated among the regions. Systematic public budget data are not available for 1861, or for Venetia and Latium before their incorporation in the Kingdom; the missing regional figures are estimated ad hoc, and the 1861 estimates typically disaggregate national figures with shares of a succeeding year.

4

CARLO CICCARELLI AND STEFANO FENOALTEA

The railway-construction estimates are built up from five components. The first refers to the new construction for network extensions of railways strictly defined. It disaggregates the corresponding national series in proportion to the regional shares of annual equivalent kilometers built; the latter are themselves calculated from data on the length of the new lines opened each year, allowing for the different construction period, and unit cost, of major lines and minor lines. The second refers to the maintenance of railways strictly defined. It disaggregates the corresponding national series in proportion to the regional shares of the length of the net, weighted by approximate estimates of regional traffic per unit length. The third refers to railway improvements (new construction on existing lines); it includes some direct evidence on the doubling of track, but in the main it disaggregates the national series on the basis of benchmark figures on the main stations’ traffic. The fourth refers to the new construction of tramways. It is estimated analogously to the construction of new railway lines, distinguishing between urban and suburban tramways. The fifth refers to the maintenance of tramway lines. It disaggregates the national series in direct proportion to cumulative (urban, suburban) equivalent kilometers built. Other civil construction covers a multitude of components, almost all of them in a double set to separate new construction from maintenance. The time series are variously derived from public budget data and real measures of construction. A first group refers to road, river-control, land-reclamation, and harbor construction. These estimates are anchored by region-specific State, municipal, and provincial public-works budget figures. The roadrelated figures are complemented by the national estimates of private expenditure to maintain rural-access roads, allocated in proportion to the regional shares of total cultivated land. The river-control estimates ignore the minor expenditure in other State budgets, and include only small, uniform corrections to allow for private expenditure. The land-reclamation estimates include the figures in other public budgets, and estimates of private spending, disaggregated with ancillary project-specific evidence. A second group refers to State-financed construction, in various budgets, for specific local projects. These include disaster-related works, regionally allocated by identifying the specific earthquake, volcanic eruption, etc., concerned; the Apulian aqueduct; and the Cavour canal system in Piedmont. A third group refers to the construction undertaken by private utilities (including those that were municipally owned, but legally separate). Private (net) spending on the Apulian aqueduct is estimated from company records,

Social-Overhead Construction in Italy’s Regions, 1861–1913

5

and allocated using direct evidence of the location of its various parts. Construction to build and maintain power-generating systems (dams and power lines), water-distribution systems (aqueducts and local nets), and gasdistribution nets is estimated, element by element, from the regional series derived to estimate the utilities’ production (Ciccarelli & Fenoaltea, 2008a). Military construction also covers a multitude of (double) components. Expenditure by the Department of War is typically dominated by general categories, such as ‘‘fortifications,’’ without direct regional tags; these are crudely allocated along Italy’s land border, with the allocation suggested by ancillary evidence (most to the Northwest, with a shift to the Northeast only at the end of the period at hand; the Swiss border remained unfortified). The early estimates include hypothetical figures for Austrian spending in Venetia, and Papal spending in Latium. Naval expenditure data are almost a mirror image of Army expenditure data: all major projects (naval bases) are separately identified, and only minor items are crudely allocated to their likely destinations. The final estimates are all referred to value added at 1911 prices. The real series are simply multiplied through by an estimate of 1911-price value added per real unit. The expenditure series, at current prices, are instead deflated with (national) price indices for maintenance work on the one hand, and new construction on the other; the resulting constant-price series are then scaled by ratios of value added to value at 1911 prices, again in maintenance on the one hand and in new construction on the other.6 The maintenance ratio is somewhat higher of the two, but the 1911-price value of new construction can be taken as simply twice the value added estimates presented here. By construction, as it were, the sums of the present regional series return the corresponding national series.7

3. NEW CONSTRUCTION: REGIONAL PATHS At the national level, as can be seen from Fig. 1, public-works new construction displayed an early burst ended by the crisis of 1866, and then the classic Kuznets-cycle sequence (tied to British capital exports: Fenoaltea, 1988) with peaks in the early 1870s, the late 1880s, and the last years before the War; and in its broad features this pattern was common to railway construction (illustrated as the difference between the total and non-railway construction), other civil construction, and military construction.

6

CARLO CICCARELLI AND STEFANO FENOALTEA 300

250

____ Total ______ Non-railway (total) _ _ _ _ _ _ Non-railway (civil)

200

150

100

50

0 65

Fig. 1.

70

75

80

85

90

95

00

05

10

Social-Overhead New Construction: National Estimates (Million Lire of Value Added at 1911 Prices).

The time paths of the regional new-construction estimates are illustrated, in index form and on a common scale, in Fig. 2. To avoid clutter, the rarely significant military component is not separated out; the indices of total and non-railway construction for each region are the 1911-price series divided by the same scalar – the region’s average total new construction at 1911 prices – so the railway component is again illustrated as the difference between the two. At the regional level different patterns emerge. On the one hand, the regions are similar in that most of the major movements were tied to bursts of railway construction; other construction was much less volatile, especially over the later nineteenth century. This is confirmed by the average coefficient of variation, which equals 1.07 for the regional railwaynew-construction series and .55 for the other-new-construction series (and .59 for the totals). On the other hand, the paths of railway construction seem to differ more, across regions, than the paths of the residual; this is confirmed by the average off-diagonal correlation coefficient, which equals just .13 for the railway-construction series, against .61 for the otherconstruction series (and .23 for the totals). As far as non-railway construction is concerned, most regions display a mild long swing, with a late 1880s peak, decline and recovery over the 1890s,

7

Social-Overhead Construction in Italy’s Regions, 1861–1913 PIEDMONT

LIGURIA

LOMBARDY

VENETIA

400

400

400

400

300

300

300

300

200

200

200

200

100

100

100

100

0

0

0

70

80

90

00

10

70

EMILIA

80

90

00

10

0 70

TUSCANY

80

90

00

10

70

MARCHES

400

400

400

400

300

300

300

300

200

200

200

200

100

100

100

100

0

0

0

70

80

90

00

10

70

LATIUM

80

90

00

10

300 200 100 0 80

90

00

80

90

00

10

70

CAMPANIA 400

400

300

300

300

200

200

200

100

100

100

10

BASILICATA

80

90

00

70

10

CALABRIA

80

90

00

10

70

SICILY

400

400

400

300

300

300

300

200

200

200

200

100

100

100

100

0

0

0

80

90

00

10

70

80

______

90

00

Total

10

80

90

00

10

80

90

00

10

SARDINIA

400

70

10

0

0 70

00

APULIA

400

0 70

90

0 70

ABRUZZI

400

80

UMBRIA

0 70

____

80

90

00

10

70

80

90

00

10

Non-railway

Fig. 2. Social-Overhead New Construction: Indices of Value Added at 1911 Prices (For Each Region, Average Total New Construction ¼ 100).

and faster growth in the pre-war boom. A much stronger cycle, with a massive peak in the 1880s, appears only in three regions: Liguria, where almost half was due to defense work (mainly the new naval base at La Spezia); Latium, where a fortune was spent on the city of Rome; and Lombardy, where it was due entirely to municipal spending on urban improvement in the city of Milan. The post-Unification cycle (with a minimum in the later 1860s) was also relatively localized. In most regions,

8

CARLO CICCARELLI AND STEFANO FENOALTEA

construction appears to have grown, or at worst remained relatively constant, through the 1860s and 1870s. A possible exception is Venetia, assuming, as the present figures do, heavy military spending by the Austrian authorities in the run-up to the war of 1866; but the cycle was certainly strongest in Piedmont, where it was driven by the construction of the Cavour canal and related irrigation projects, and also (as again in the 1880s) by military work along the new French border. The final construction boom, in contrast, was very widely diffused; the acceleration was however particularly sharp in Piedmont (thanks to major hydroelectric construction), in Venetia (where military construction surged as it was finally realized that the next European war need not be on Italy’s western front), in Basilicata and Apulia (reflecting early work on the Apulian aqueduct), and again, as in the 1880s, in Liguria and Latium. The only truly individual cycle appears in Umbria, with an extra peak in the early 1900s, which can be traced to hydroelectric construction for chemical works. The paths of railway construction are instead not surprisingly a mixture of common and idiosyncratic elements. The latter are tied most obviously to the initial construction of the national trunks. The quick advance of the railhead from Bologna to and down the Adriatic coast and along the boot’s instep to its toe can be traced in the early peak that appears first in Emilia and the Marches, then in the Abruzzi, and later still in Apulia, Basilicata, and Calabria; the early linking of the Tuscan and Roman lines along the West coast and through Umbria is similarly apparent, as is the slightly later construction of the Ligurian coast line and the island trunks, and, later still, in the 1890s, the final link-up of Naples and Reggio Calabria along the West coast, and of Naples and Taranto (in Apulia) across Basilicata. The common elements are essentially two. One is the wave of minor-line construction in the 1880s and early 1890s, quite apparent in most regions, but with significant exceptions in the continental South and in Sardinia; the second is the final wave of improvements, even more concentrated in the North and Center.

4. INTERPRETING IDIOSYNCRATIC CYCLES: OF SCALE AND GEOGRAPHIC HIERARCHY The regional railway- and non-railway-construction cycles were very different. On the face of it, they were shaped by different mechanisms: non-railway work apparently responded to largely common, supra-regional

Social-Overhead Construction in Italy’s Regions, 1861–1913

9

influences, while the determinants of railway work seem idiosyncratic, more strictly regional. That is the message of Fig. 2, and it is, at its own level, undeniable. Undeniable, but perhaps in some sense superficial. In the case of nonrailway construction, the regional paths in Fig. 2 and the national paths in Fig. 1 dovetail nicely: the largely common Kuznets cycles of the regional series add up to the evident Kuznets cycle of the corresponding national aggregate. In the case of railway construction, on the other hand, the different cycles of the various regions do not add up to a series that is merely noisy: at the national level, the Kuznets cycle is again clearly apparent. These different national and regional patterns beg to be reconciled, but alternative interpretations are possible. The Kuznets-cycle story ties construction in Italy to the loosening and tightening of financial constraints as the supply of foreign capital rose and fell (with the Kuznets-cycle variations in capital exports from the British core to the entire periphery: Fenoaltea, 1988). An obvious possibility is to argue that the regional evidence tends to confirm that story for non-railway construction alone. The non-railway Kuznets cycle remains visible with the increase in magnification from the national to the regional level, the railway Kuznets cycle dissolves into noise: the national non-railway Kuznets cycle appears as a real signal built on signals, the national railway Kuznets cycle as a false signal generated by the cumulation of noise. This view would be consistent with Figs. 1 and 2; but it may be worth digging deeper. Regional railway-construction cycles appear idiosyncratic, as noted, because railway construction is lumpy and long-lived: a line is built from here to there, and once built, in the main, it stays built. But this is equally true of other social-overhead capital: once a canal, a road, a hydroelectric dam or a military fort has been built, it too, in the main, stays built. If our observations were further magnified, to ever narrower categories and smaller geographic units, all our cycles would become idiosyncratic, for any particular local work would show up as a sudden and short-lived burst of activity at some particular point in time. As the national railway cycle dissolves if one moves to the regional level, so the common regional cycle in other construction would presumably dissolve if one moved to a finer grid of functional and geographic categories. A common cycle requires sufficient aggregation; and this suggests that the apparent difference between railway and non-railway construction reflects not a difference in the forces that generated their cyclical paths, but simple scale. Railway projects are few and large relative to the region: regional

10

CARLO CICCARELLI AND STEFANO FENOALTEA

construction is perforce idiosyncratic, and the overall cycle can be measured only at a supra-regional level. Non-railway projects are few and large if considered singly, at the local level, but many and small if considered together at the regional level, so the overall cycle emerges already at that level. In this alternative and arguably deeper perspective, Figs. 1 and 2 remain consistent with the national Kuznets-cycle story for railway construction as for other construction. The Kuznets cycle appears where it can, and not where the unit of observation is too small to allow it: for railway construction in Fig. 1 but not Fig. 2, for non-railway construction in Figs. 1 and 2 but not in further figures devoted for example to military construction in a particular mountain valley.

5. NEW CONSTRUCTION: REGIONAL LEVELS New construction varies across regions as well as over time. Perhaps the least meaningful source of interregional variation is simply a scale effect, as some regions were much larger than others; but the appropriate standardization is moot. The traditional dimensions are measures of population or of surface area (Nitti, 1900); the former seems the more appropriate for comparisons of regional product, the latter for investigations of the growth-enhancing effects of large-scale infrastructure. Both are presented here: demographic densities varied significantly, and the pictures they yield are very different. Table 1 presents, in panel A, the overall annual averages of the newconstruction series (columns 1–4) and alternative measures of regional scale (columns 5–6). The demographic variable, as in related work, is here the male population of working age (15 plus); since (for present purposes) its distribution varied comparatively little over time, that for 1901 is selected as reasonably representative. The regions’ areas are the (almost) unequivocal alternative; the difference between the two is illustrated by their ratio (column 7, in essence a density measure), which varies over the larger part of a full order of magnitude.8 These densities are also converted into percentages of the national mean (column 8), and, in that form, illustrated in Fig. 3. Panels B–E report, in columns 1 and 2, the average level of annual new construction per demographic and geographic unit. Columns 3–8 report, in index form, annual average new construction by decade (or near decade), with the full-period average of each component in each region (again) set

A. Average new construction and regional characteristics (1) (2) (3) (4) Average Construction, 1861–1913 (Million Lire) Railways

Other civil

Military

Total

(5) Males Aged 15þ in 1901 (Millions)

Piedmont Liguria Lombardy Venetia

8.18 3.52 7.51 7.33

8.54 5.33 8.31 6.12

1.62 2.01 .00 1.26

18.34 10.86 15.81 14.71

1.082 .370 1.397 .982

Emilia Tuscany Marches Umbria

4.76 5.29 2.00 1.82

5.56 5.18 1.92 2.31

.00 .06 .09 .00

10.32 10.52 4.01 4.13

Latium Abruzzi Campania Apulia

4.08 3.75 4.86 4.86

7.63 4.37 8.28 3.44

.42 .00 .09 .41

Basilicata Calabria Sicily Sardinia

1.43 3.79 6.20 1.87

2.03 4.05 8.49 1.98

71.24

83.54

Italy

(6) Area (Million Square Kilometers)

(7)

(8) Density

Ratio (column 5/ column 6)

Index (Italy ¼ 100)

.0294 .0053 .0241 .0245

36.8 69.8 58.0 40.1

101 191 158 110

.817 .853 .334 .229

.0207 .0241 .0097 .0097

39.5 35.4 34.4 23.6

108 97 94 64

12.13 8.12 13.23 8.70

.425 .436 .988 .625

.0121 .0165 .0163 .0191

35.1 26.4 60.6 32.7

96 72 166 89

.00 .08 .20 .14

3.47 7.92 14.88 3.99

.144 .390 1.139 .272

.0100 .0151 .0257 .0241

14.4 25.8 44.3 11.3

39 70 121 31

6.38

161.14

10.483

.2864

36.6

100

Social-Overhead Construction in Italy’s Regions, 1861–1913

Social-Overhead New Construction: Annual Averages (Value Added at 1911 Prices).

11

12

(Continued). B. Railway new construction: Scaled estimates and indices (3)

(4) (5) (6) (7) Indices of Average Annual Construction

(8)

Lire/Male 15þ

Lire/square kilometer

1861–1870

1871–1880

1881–1890

1891–1900

1901–1913

1861–1913

Piedmont Liguria Lombardy Venetia

7.56 9.52 5.37 7.46

279 668 312 299

65 98 114 23

63 117 88 111

198 108 174 182

39 59 45 15

127 114 84 153

100 100 100 100

Emilia Tuscany Marches Umbria

5.83 6.20 6.00 7.95

230 219 206 187

74 183 194 325

31 32 17 101

226 109 154 86

48 50 112 3

116 120 40 11

100 100 100 100

Latium Abruzzi Campania Apulia

9.60 8.60 4.92 7.77

338 227 298 254

111 93 114 284

25 99 55 8

211 174 181 132

112 138 119 67

54 21 46 30

100 100 100 100

Basilicata Calabria Sicily Sardinia

9.96 9.72 5.44 6.86

144 251 241 77

60 130 115 103

234 149 159 347

70 55 136 53

166 173 96 19

0 17 19 6

100 100 100 100

Italy

6.80

249

118

88

155

72

75

100

CARLO CICCARELLI AND STEFANO FENOALTEA

(1) (2) Average Construction, 1861–1913

(1) (2) Average Construction, 1861–1913

(3)

(4) (5) (6) (7) Indices of Average Annual Construction

(8)

Lire/Male 15þ

Lire/square kilometer

1861–1870

1871–1880

1881–1890

1891–1900

1901–1913

1861–1913

Piedmont Liguria Lombardy Venetia

7.90 14.40 5.95 6.23

291 1,009 345 249

122 45 43 65

64 60 50 66

70 135 128 96

66 79 70 83

160 162 184 170

100 100 100 100

Emilia Tuscany Marches Umbria

6.80 6.08 5.75 10.10

268 215 198 238

50 107 67 31

77 90 75 72

87 83 109 69

84 69 89 126

179 139 146 178

100 100 100 100

Latium Abruzzi Campania Apulia

17.96 10.02 8.38 5.50

632 264 508 180

34 60 49 57

68 90 97 106

198 120 112 95

72 92 76 62

122 130 151 162

100 100 100 100

Basilicata Calabria Sicily Sardinia

14.12 10.38 7.45 7.27

204 269 330 82

30 29 53 94

116 101 88 129

121 103 98 94

71 90 91 61

148 160 154 117

100 100 100 100

7.97

292

60

79

110

78

156

100

Italy

Social-Overhead Construction in Italy’s Regions, 1861–1913

C. Other civil new construction: Scaled estimates and indices

13

14

(Continued). D. Military new construction: Scaled estimates and indices (3)

(4) (5) (6) (7) Indices of Average Annual Construction

(8)

Lire/Male 15þ

Lire/square kilometer

1861–1870

1871–1880

1881–1890

1891–1900

1901–1913

1861–1913

1.50 5.43 .00 1.29

55 380 0 52

172 186 – 174

49 74 – 56

196 207 – 51

57 45 – 19

44 14 – 177

100 100 100 100

Emilia Tuscany Marches Umbria

.00 .07 .26 .00

0 2 9 0

– 0 14 –

– 18 0 –

– 255 156 –

– 115 101 –

– 109 200 –

100 100 100 100

Latium Abruzzi Campania Apulia

.99 .00 .09 .65

35 0 5 21

173 – 0 0

0 – 12 0

304 – 294 262

43 – 124 136

8 – 77 101

100 100 100 100

Basilicata Calabria Sicily Sardinia

.00 .22 .17 .52

0 6 8 6

– 0 0 0

– 0 5 0

– 314 337 404

– 125 138 115

– 70 39 9

100 100 100 100

Italy

.61

22

148

47

194

56

65

100

Piedmont Liguria Lombardy Venetia

CARLO CICCARELLI AND STEFANO FENOALTEA

(1) (2) Average Construction, 1861–1913

(1) (2) Average Construction, 1861–1913

(3)

(4) (5) (6) (7) Indices of Average Annual Construction.

(8)

Lire/Male 15þ

Lire/square kilometer

1861–1870

1871–1880

1881–1890

1891–1900

1901–1913

1861–1913

Piedmont Liguria Lombardy Venetia

16.95 29.35 11.32 14.98

625 2,057 657 599

101 88 76 53

62 81 68 88

138 140 150 135

53 67 58 44

135 119 136 162

100 100 100 100

Emilia Tuscany Marches Umbria

12.63 12.34 12.00 18.05

498 437 413 426

61 145 130 160

56 60 44 85

151 97 133 76

67 60 101 72

150 130 94 105

100 100 100 100

Latium Abruzzi Campania Apulia

28.54 18.62 13.39 13.92

1,004 491 812 455

65 75 73 181

51 94 81 46

206 145 138 123

84 113 92 68

95 79 112 85

100 100 100 100

Basilicata Calabria Sicily Sardinia

24.08 20.31 13.07 14.65

348 526 578 165

42 77 78 95

165 123 116 226

100 82 117 86

110 130 93 43

87 91 96 61

100 100 100 100

Italy

15.37

563

89

82

133

75

116

100

Social-Overhead Construction in Italy’s Regions, 1861–1913

E. Total new construction: Scaled estimates and indices

15

16

(Continued). F. Indices of average scaled new construction, 1861–1913 (Italy ¼ 100) (1)

(2) (3) Lire/Male 15þ

(4)

(5)

(6) (7) Lire/Square Kilometer

(8)

Other civil

Military

Total

Railways

Other civil

Military

Total

Piedmont Liguria Lombardy Venetia

111 140 79 110

99 181 75 78

246 893 0 212

110 191 74 97

112 268 125 120

100 346 118 85

248 1,710 0 232

111 366 117 107

Emilia Tuscany Marches Umbria

86 91 88 117

85 76 72 127

0 11 42 0

82 80 78 117

92 88 83 75

92 74 68 82

0 10 40 0

89 78 73 76

Latium Abruzzi Campania Apulia

141 127 72 114

225 126 105 69

164 0 14 107

186 121 87 91

136 91 120 102

217 91 174 62

157 0 24 95

178 87 144 81

Basilicata Calabria Sicily Sardinia

147 143 80 101

177 130 94 91

0 35 29 86

157 132 85 95

58 101 97 31

70 92 113 28

0 25 35 26

62 93 103 29

Italy

100

100

100

100

100

100

100

100

CARLO CICCARELLI AND STEFANO FENOALTEA

Railways

Social-Overhead Construction in Italy’s Regions, 1861–1913

Fig. 3.

17

Indices of Demographic Density (Males 15 and Over Per Square Kilometer in 1901; Italy ¼ 100).

equal to 100. Any such index value, multiplied by column 1 or column 2 (and divided by 100), returns the actual average level of construction per (demographic or geographic) unit in the relevant region and period. These actual (quasi) decadal and overall averages are illustrated in Figs. 4–9; to keep the number of maps within reasonable bounds, the rarely significant military component is not separated out, and all non-railway new construction is considered together. As the maps recognize only six ranges, some borderline cases are inevitable. Panel F of Table 1 returns to the long-run averages per male 15 years of age and over and per square kilometer reported in columns 1 and 2 of the preceding panels, and presents them as column-specific indices, with the national means equal to 100.9 These highlight the differences in relative levels of construction across categories with a given metric, and across metrics within given categories. The latter comparison reflects of course

18

CARLO CICCARELLI AND STEFANO FENOALTEA 1861-1870

1871-1880

1881-1890

1891-1900

1901-1913

1861-1913

Fig. 4.

27.50 and over

22.50 to 27.49

17.50 to 22.49

12.50 to 17.49

7.50 to 12.49

up to 7.49

Railway New Construction: Annual Average Value Added at 1911 Prices Per Male 15 and Over in 1901 (Lire).

the different densities in panel A, columns 7 and 8. Regions (like Piedmont) with a density close to the national average look much the same; the outliers look very different, with the high-density regions (like Liguria) scoring much higher per unit area, and the low-density regions (like Basilicata) per head. These particular differences are illustrated by a comparison of the corresponding maps of Figs. 4 and 5, or Figs. 6 and 7, or Figs. 8 and 9. Considering only average total new construction, the results may be summarized as follows. Within the near-average-density group, with either

19

Social-Overhead Construction in Italy’s Regions, 1861–1913 1861-1870

1871-1880

1881-1890

1891-1900

1901-1913

1861-1913

Fig. 5.

600 and over

480 to 599

360 to 479

240 to 359

120 to 239

up to 119

Railway New Construction: Annual Average Value Added at 1911 Prices Per Square Kilometer (Lire).

measure Latium appears far above the national average, Piedmont slightly above it, Venetia near it, and Emilia, Tuscany, the Marches, and Apulia rather below it. Within the low-density group, where all decline as one moves from the demographic base to the geographic, Sardinia goes from average to dead last, Umbria, the Abruzzi, and Calabria from well above average to below it; Basilicata drops all the way from third highest to second lowest. Within the high-density group, conversely, Sicily moves up from below-average to average, Lombardy and Campania from last and sixth-last

20

CARLO CICCARELLI AND STEFANO FENOALTEA 1861-1870

1871-1880

1881-1890

1891-1900

1901-1913

1861-1913

Fig. 6.

27.50 and over

22.50 to 27.49

17.50 to 22.49

12.50 to 17.49

7.50 to 12.49

up to 7.49

Other Social-Overhead New Construction: Annual Average Value Added at 1911 Prices Per Male 15 and Over in 1901 (Lire).

to fourth and third, respectively. Liguria is first on either score: per man by a whisker, per unit area by a country mile. Considering the regions together in the light of panel F, the differences among them are altogether less extreme per man than per unit area: for total new construction, in particular, the former (column 4) yields for example a range of some 117 index points, a maximum/minimum ratio near 2.6, and a coefficient of variation of .34, against 337, 12.6, and .67,

21

Social-Overhead Construction in Italy’s Regions, 1861–1913 1861-1870

1871-1880

1881-1890

1891-1900

1901-1913

1861-1913

Fig. 7.

1,500 and over

1,200 to 1,499

900 to 1,199

600 to 899

300 to 599

up to 299

Other Social-Overhead New Construction: Annual Average Value Added at 1911 Prices Per Square Kilometer (Lire).

respectively, for the latter (column 8). If one mixes the two measures, as the weight of the second increases these statistics grow monotonically, essentially because the rise of Liguria’s winning score overwhelms the convergence of the other regions with unusually high or low density. Measures that are not as sensitive to outliers yield different results: the interquartile range, for example, is lower per-area (37) than per man (43), and lower still if the two measures are combined (26, for example, with equal weights).

22

CARLO CICCARELLI AND STEFANO FENOALTEA 1861-1870

1871-1880

1881-1890

1891-1900

1901-1913

1861-1913

Fig. 8.

55.00 and over

45.00 to 54.99

35.00 to 44.99

25.00 to 34.99

15.00 to 24.99

up to 14.99

Aggregate Social-Overhead New Construction: Annual Average Value Added at 1911 Prices Per Male 15 and Over in 1901 (Lire).

6. MAINTENANCE AND THE CONVERGENCE OF REGIONAL ENDOWMENTS The national estimates of 1911-price value added in maintenance are illustrated in Fig. 10; the railway component is illustrated as a difference (as in Fig. 1), and the small military component is not separated out (as in Fig. 2). The aggregate new-construction series (from Fig. 1) is included to

23

Social-Overhead Construction in Italy’s Regions, 1861–1913 1861-1870

1871-1880

1881-1890

1891-1900

1901-1913

1861-1913

Fig. 9.

2,100 and over

1,680 to 2,099

1,260 to 1,679

840 to 1,259

420 to 839

up to 419

Aggregate Social-Overhead New Construction: Annual Average Value Added at 1911 Prices Per Square Kilometer (Lire).

provide perspective; briefly, in the depths of the depression of the 1890s, maintenance outweighed new construction. Maintenance is stock-related, and the regional maintenance series too are generally trend-dominated (with minor co-movements that derive from the common deflator of the underlying expenditure series). These estimates accordingly have a negligible effect on measured cyclical movements; they are of interest here precisely as indices of the underlying stock.

24

CARLO CICCARELLI AND STEFANO FENOALTEA 300 ____ ______ 250

Total new construction Total maintenance Non-railway maintenance

200 150 100 50 0 65

Fig. 10.

70

75

80

85

90

95

00

05

10

Social-Overhead Maintenance: National Estimates (Million Lire of Value Added at 1911 Prices).

In the sources, the stock of social-overhead capital is often reported in physical units (‘‘kilometers of roads’’), but such measures are inevitably partial. In the secondary literature, comprehensive measures of the real capital stock have been obtained through the ‘‘perpetual inventory’’ approach, which estimates capital by cumulating, and depreciating, the investment flow alone.10 An early example is provided by the national estimates described in Vitali (1969); the capital-stock series begins in 1881, and with structure lives of a century or more the investment series had to be extrapolated back to the later 1700s. More recently, ambitious regional estimates have appeared in Bonaglia and Picci (2000); but these refer only to the later twentieth century, and even assuming much shorter structure lives (under half a century) they required the backward extrapolation of the available investment data. In the case at hand, the permanent-inventory approach seems to lead nowhere: the present series are themselves just half a century long, and even if one assumed structure lives no longer than that one could obtain no more than a point estimate of the capital stock at the very end of the period at hand. Empirically, however, the evidence does not support the assumption that by 1913 the stock inherited in 1861 was all but irrelevant. As is evident from Fig. 10, at the national level the railways’ share of total maintenance increased over time: from trivial levels in the early years to about a third

Social-Overhead Construction in Italy’s Regions, 1861–1913

25

at the end. Railway new construction was just over half the total before falling off after 1895, and remained close to half over the full period at hand (44 percent, from Table 1, panel A, columns 1 and 4). The disproportion between a third (or less) and close to half implies either that railways required relatively little maintenance, or, more plausibly, that the initial non-railway stock was both large and extremely long-lived.11 The conclusion that regional capital stocks cannot be calculated from the present new-construction series alone seems inescapable; nor of course can those series be used to tease annual capital-stock estimates out of an independent early benchmark, as at the time of writing such a benchmark simply does not exist.12 In the circumstances, the present maintenance series serve as useful indices of the underlying stock. They are, by construction, comprehensive. They are obtained in the main directly from expenditure data in public budgets, and are correspondingly independent of the new-construction series: to that extent they incorporate neither the errors of the latter, nor assumed asset lives. Moreover, they adjust automatically for the actual usefulness of the infrastructure under consideration, to the extent that roadway repairs, for example, vary directly with traffic. On the other hand, the unweighted sum of maintenance figures will overstate the relative stock of regions disproportionately endowed with maintenance-intensive capital, and correspondingly understate that of regions with, for example, a concentration of hydroelectric dams.13 Table 2 summarizes the regional maintenance estimates for four quinquennia centered on the census years 1871, 1881, 1901, and 1911; only civil maintenance is considered here, on the presumption that military infrastructure produced national, rather than local, public goods. Like the aggregates in Table 1, the estimates in Table 2 are standardized both per male of working age (at each census year) and per unit area. Fig. 11 illustrates the standardized figures for 1869–1873 and 1909–1913, rescaled as column-specific indices with the corresponding national means equal to 100. The regions are ordered on the horizontal axis, from Piedmont to Sardinia, as in Table 2; the left-hand graphs correspond to columns 4–6, the right-hand graphs to columns 7–9. As with the analogous newconstruction indices in Table 1, panel F, the differences between the two standardizations mirror the relative densities illustrated above in Fig. 3. In general, the lines in Fig. 11 move downward from left to right, pointing to progressively smaller average endowments as one moves from the NorthWest down to the boot’s toe and finally to the major islands; they appear to become somewhat flatter over time on a per-man basis (left column),

26

CARLO CICCARELLI AND STEFANO FENOALTEA

Social-Overhead Civil Maintenance: Annual Averages (Value Added at 1911 Prices). A. 1869–1873 (1)

(2) (3) Million Lire

Railways Other

(4) (5) (6) Lire/Male 15þ in 1871

(7) (8) (9) Lire/Square Kilometer

Total

Railways

Other

Total

Railways

Other

Total

Piedmont Liguria Lombardy Venetia

.92 .39 1.25 .42

4.77 1.38 6.37 7.17

5.68 1.77 7.62 7.60

.94 1.39 1.05 .48

4.91 4.90 5.37 8.06

5.85 6.29 6.42 8.54

31 74 52 17

162 262 264 292

193 336 316 310

Emilia Tuscany Marches Umbria

.72 1.05 .28 .28

6.53 5.19 1.41 .90

7.26 6.25 1.69 1.19

.98 1.42 .91 1.47

8.91 7.01 4.66 4.67

9.90 8.43 5.57 6.15

35 44 28 29

316 215 145 93

351 259 174 122

Latium Abruzzi Campania Apulia

.34 .17 .36 .40

1.99 1.39 3.67 2.41

2.33 1.56 4.03 2.81

1.08 .41 .39 .87

6.26 3.33 3.93 5.18

7.34 3.74 4.31 6.05

28 10 22 21

165 84 225 126

193 95 247 147

Basilicata Calabria Sicily Sardinia

.03 .14 .16 .01

.63 1.32 4.85 1.59

0.65 1.46 5.00 1.60

.16 .35 .19 .06

3.78 3.39 5.76 7.25

3.94 3.75 5.94 7.32

3 9 6 1

63 88 188 66

66 97 194 66

6.94

51.57

58.51

.77

5.70

6.46

24

180

204

Italy

but not, railways apart, per unit area (right column). The same information yields the maps in Figs. 12–14. In general, these illustrations suggest significant overall convergence in railways, but only weak convergence, if any, in other or total civil infrastructure, especially per unit area. A variety of summary statistics confirm this impression: from 1869–1873 to 1909–1913 the coefficient of variation of railway maintenance drops, per man (column 4), from .62 to .33, and, per unit area (column 7), from .76 to .65; those of other and total civil maintenance drop, per man (columns 5 and 6), from .30 to .28, and .29 to .24, respectively, but rise, per unit area (columns 8 and 9), from .48 to .51, and .49 to .54, respectively. The simple correlation between the initial level and the mean annual growth rate between the initial and final level (which if negative points to convergence, absent significant rank reversals) equals, for the per-man measures, .86 for railway maintenance,

27

Social-Overhead Construction in Italy’s Regions, 1861–1913

(Continued). B. 1879–1883 (1)

(2) (3) Million Lire

Railways Other

(4) (5) (6) Lire/Male 15þ in 1881

(7) (8) (9) Lire/Square Kilometer

Total

Railways

Other

Total

Railways

Other

Total

Piedmont Liguria Lombardy Venetia

1.58 .91 2.37 1.00

5.20 1.70 7.77 9.59

6.77 2.61 10.14 10.59

1.55 3.03 1.89 1.07

5.10 5.68 6.20 10.27

6.65 8.71 8.10 11.34

54 172 98 41

177 323 323 391

231 495 421 432

Emilia Tuscany Marches Umbria

1.04 1.48 .37 .48

6.85 4.50 1.56 .87

7.89 5.97 1.93 1.35

1.36 1.92 1.17 2.36

8.93 5.84 4.88 4.22

10.29 7.76 6.04 6.59

50 61 38 50

331 187 160 89

381 248 199 139

Latium Abruzzi Campania Apulia

.49 .42 .60 .54

2.71 1.60 3.96 2.62

3.20 2.02 4.55 3.16

1.40 .99 .61 1.02

7.79 3.76 4.04 5.00

9.19 4.76 4.65 6.02

40 26 37 28

224 97 243 137

265 122 279 165

Basilicata Calabria Sicily Sardinia

.16 .42 .49 .09

.85 1.51 4.94 2.13

1.01 1.93 5.43 2.22

.98 1.04 .52 .38

5.23 3.76 5.15 8.95

6.21 4.80 5.67 9.33

16 28 19 4

86 100 192 88

102 128 211 92

12.43

58.34

70.77

1.29

6.08

7.37

43

204

247

Italy

.47 for other civil maintenance, and .58 for aggregate civil maintenance; for the per-unit-area measures it equals .70 for railway maintenance, but just .14 for other civil maintenance, and .12 for aggregate civil maintenance. The corresponding scatter diagrams are illustrated in Fig. 15. Further evidence on convergence can be obtained by correlating the initial levels reported in Table 2 with the mean flows reported in Table 1. Adjusting the latter to exclude the initial decade, per man the correlation between the initial level of maintenance (Table 2, panel A, column 6), again interpreted as a stock index, and the subsequent mean flow to 1913 (Table 1, panel B, column 1 plus panel C, column 1, suitably adjusted) was negative, but low (.27); per unit area (Table 2, panel A, column 9 and Table 1, panel B, column 2 plus panel C, column 2, suitably adjusted), it is positive and significant (.55): again, there appears to have been limited convergence per man, and not even that per unit area.

28

CARLO CICCARELLI AND STEFANO FENOALTEA

(Continued). C. 1899–1903 (1)

(2) (3) Million Lire

Railways Other

(4) (5) (6) Lire/Male 15þ in 1901

(7) (8) (9) Lire/Square Kilometer

Total

Railways

Other

Total

Railways

Other

Total

Piedmont Liguria Lombardy Venetia

3.20 1.35 4.97 2.04

6.32 2.49 9.39 9.27

9.52 3.84 14.36 11.30

2.96 3.65 3.56 2.08

5.84 6.74 6.72 9.44

8.80 10.39 10.28 11.51

109 256 206 83

215 472 390 377

324 728 596 461

Emilia Tuscany Marches Umbria

2.96 2.43 .75 .81

8.22 5.22 2.00 1.08

11.17 7.64 2.75 1.90

3.62 2.84 2.26 3.55

10.06 6.11 5.98 4.73

13.68 8.96 8.24 8.29

143 101 78 84

397 216 206 112

540 317 283 195

Latium Abruzzi Campania Apulia

1.25 1.14 1.46 1.05

3.62 1.82 4.73 3.47

4.87 2.96 6.19 4.51

2.95 2.61 1.48 1.67

8.51 4.18 4.78 5.55

11.46 6.80 6.26 7.22

104 69 90 55

299 110 290 181

403 179 380 236

Basilicata Calabria Sicily Sardinia

.33 .82 1.20 .30

1.03 2.00 5.60 2.40

1.37 2.81 6.79 2.70

2.32 2.09 1.05 1.11

7.18 5.12 4.91 8.81

9.50 7.21 5.96 9.92

34 54 47 13

104 132 217 99

137 187 264 112

26.06

68.65

94.71

2.49

6.55

9.03

91

240

331

Italy

7. PUBLIC WORKS AND GROWTH: PRELIMINARY RESULTS The relation between infrastructure investment and growth is at the heart of the post-war debate on the economic development of post-Unification Italy, and much current literature besides. The principals in that debate – Alexander Gerschenkron and Rosario Romeo – differed as to whether infrastructure investment should precede, or accompany, economic development; that economic development meant industrial development went without saying (Romeo, 1959; Gerschenkron, 1962, 1968). In the case at hand, regional gross product series are not yet available; but preliminary census-year estimates of regional industrial product have been compiled, and overall regional growth can be proxied by the growth of the male population (in essence, labor force) from census to census.

29

Social-Overhead Construction in Italy’s Regions, 1861–1913

(Continued). D. 1909–1913 (1)

(2) (3) Million Lire

Railways Other

(4) (5) (6) Lire/Male 15þ in 1911

(7) (8) (9) Lire/Square Kilometer

Total

Railways

Other

Total

Railways

Other

Total

Piedmont Liguria Lombardy Venetia

4.68 1.94 7.35 3.18

6.96 2.80 10.98 9.83

11.63 4.73 18.34 13.01

4.11 4.60 4.80 3.11

6.11 6.64 7.17 9.61

10.21 11.24 11.97 12.72

159 367 305 130

237 530 456 400

396 897 761 530

Emilia Tuscany Marches Umbria

4.43 3.50 1.20 1.12

9.85 5.88 2.30 1.21

14.28 9.38 3.50 2.32

4.98 3.91 3.59 4.87

11.08 6.56 6.90 5.28

16.06 10.46 10.49 10.15

214 145 124 115

476 244 237 124

690 389 361 239

Latium Abruzzi Campania Apulia

1.83 1.68 2.25 1.71

4.65 2.25 5.13 3.72

6.48 3.93 7.38 5.43

4.09 4.09 2.19 2.55

10.38 5.49 4.98 5.56

14.47 9.58 7.17 8.11

152 102 138 89

385 136 315 195

537 238 453 284

Basilicata Calabria Sicily Sardinia

.45 1.12 1.75 .41

1.28 2.14 6.01 2.42

1.73 3.25 7.76 2.83

3.27 2.86 1.45 1.45

9.19 5.46 4.99 8.57

12.46 8.32 6.44 10.03

46 74 68 17

128 142 233 100

174 216 301 117

77.40 116.00

3.50

7.01

10.51

135

270

405

Italy

38.60

The manufacturing-product estimates and the demographic data for 1871, 1881, 1901, and 1911 are transcribed in Table 3, panel A; panel B reports the corresponding relative increments over each intercensal period and again from 1871 to 1911.14 Fig. 16 plots the regional ratios of the male labor force in 1911 to that in 1871, and the ratios of manufacturing output in 1911 to that in 1871, scaled by the corresponding national ratios (Table 3, panel B, columns 4 and 8); the regions are arranged, in the usual order, on the horizontal axis. The surprising lack of congruence between industrial and overall (male labor force) growth is readily apparent: while industrial growth generally declines from the Northwest to the South and islands, overall growth does not, and Latium, Apulia, and Sicily outstrip all but Liguria.15 Table 3, panel C presents measures of investment in civil infrastructure over the corresponding intercensal periods. Investment in railway and other

30

CARLO CICCARELLI AND STEFANO FENOALTEA RAILWAYS / MAN

RAILWAYS / SQ. KM

350

350

300

300

250

250

200

200

150

150

100

100

50

50

0

0 PI LI LO VE EM TO MA UM LA AB CM PU BA CL SI SA

PI LI LO VE EM TO MA UM LA AB CM PU BA CL SI SA

OTHER CIVIL / MAN

OTHER CIVIL / SQ. KM.

350

350

300

300

250

250

200

200

150

150

100

100

50

50 0

0 PI LI LO VE EM TO MA UM LA AB CM PU BA CL SI SA

PI LI LO VE EM TO MA UM LA AB CM PU BA CL SI SA

TOTAL CIVIL / MAN

TOTAL CIVIL / SQ. KM.

350

350

300

300

250

250

200

200

150

150

100

100

50

50

0

0 PI LI LO VE EM TO MA UM LA AB CM PU BA CL SI SA

1869-1873

Fig. 11.

PI LI LO VE EM TO MA UM LA AB CM PU BA CL SI SA

1909-1913

Indices of Civil Maintenance (For Each Series, Italy ¼ 100).

31

Social-Overhead Construction in Italy’s Regions, 1861–1913 Per male 15 and over:

1909-1913

150 and over

120 to 149.99

90 to 119.99

60 to 89.99

30 to 59.99

up to 29.99

Per square kilometer:

Fig. 12.

1869-1873

1869-1873

1909-1913

300 and over

240 to 299.99

180 to 239.99

120 to 179.99

60 to 119.99

up to 59.99

Railway Maintenance: Indices of Annual Average Value Added at 1911 Prices (in Each Period, Italy ¼ 100).

civil projects, again proxied by new-construction value added, is reported per unit area; as noted above, in investigating effects on growth the unscaled and per-man figures seem altogether less meaningful, and they are not reported here.

32

CARLO CICCARELLI AND STEFANO FENOALTEA

Per male 15 and over:

1909-1913

150 and over

120 to 149.99

90 to 119.99

60 to 89.99

30 to 59.99

up to 29.99

Per square kilometer:

Fig. 13.

1869-1873

1869-1873

1909-1913

300 and over

240 to 299.99

180 to 239.99

120 to 179.99

60 to 119.99

up to 59.99

Other Civil Social-Overhead Maintenance: Indices of Annual Average Value Added at 1911 Prices (in Each Period, Italy ¼ 100).

Panel D reports the simple correlations between growth and the capital stock calculated over each intercensal period, and for the full span from 1871 to 1911. Growth is measured both in industrial terms (columns 1–3) and in demographic terms (columns 4–6), using the relatives in panel B. Capital is measured separately for railways, other civil projects, and their

33

Social-Overhead Construction in Italy’s Regions, 1861–1913 Per male 15 and over:

1869-1873

1909-1913

150 and over

120 to 149.99

90 to 119.99

60 to 89.99

30 to 59.99

up to 29.99

Per square kilometer:

1869-1873

1909-1913

300 and over

240 to 299.99

180 to 239.99

120 to 179.99

60 to 119.99

up to 59.99

Fig. 14. Aggregate Civil Social-Overhead Maintenance: Indices of Annual Average Value Added at 1911 Prices (in Each Period, Italy ¼ 100).

sum (total civil capital); growth is correlated in lines 1–4 with the initial level of capital (proxied by the maintenance figures in Table 2, panels A–C, columns 7–9), and in lines 5–8 with its mean increments (panel C). Figs. 17–20 present the corresponding scatter diagrams for the two components of capital, and the three intercensal sub-periods.

34

CARLO CICCARELLI AND STEFANO FENOALTEA SA BA

6

7

AB SI CL VE CM

4

EM PILO MALA

MEAN GROWTH RATE

MEAN GROWTH RATE

8

LIUM TO

PU

SA BA SI AB

6

CL 5

VE CM EM LA PI PUMA UM

4

0

0.5

1

1.5

2

2.5

0

INITIAL RAILWAYS / MAN

20

40

80

100

LA

BA 2

LA

AB CL MA LI LO CM PI UM PU

1

0

MEAN GROWTH RATE

MEAN GROWTH RATE

60

INITIAL RAILWAYS / SQ. KM.

2

EM SA VE

SI

TO

BA

LI

1.5

LO MA AB CL PU SA PI

1

EM CM

UM

VE

SI

0.5

TO 0

-1 2

4 6 8 INITIAL OTHER CIVIL / MAN

3

10

0

BA

CL

CM

MA LO PILI UM

LA EM VE

1

200

PU SA TO

AB LO CL

2

MAPI UMPU 1.5

VE

1

Fig. 15.

4 6 8 10 INITIAL TOTAL CIVIL / MAN

12

0

EM CM

SA SI

0 2

400

LI

BA

SI 0

300

LA

2.5

AB 2

100

INITIAL OTHER CIVIL / SQ. KM.

MEAN GROWTH RATE

MEAN GROWTH RATE

LI

TO

3

2

LO

100

200

TO 300

400

INITIAL TOTAL CIVIL / SQ. KM.

Civil Maintenance: Initial Levels (1869–73) and Average Annual Growth Rates (1869–1873 to 1909–1913).

In general, these correlations are positive.16 Negative correlations appear only in the first decade, and specifically between railway initial capital or current railway investment and the growth of the male population of working age; this result is driven by Apulia and Sicily, where very high

35

Social-Overhead Construction in Italy’s Regions, 1861–1913

Social-Overhead Civil Capital and Intercensal Growth. A. Census-year levels (1) (2) (3) (4) Value Added in Manufacturing (Million 1911 Lire)

(5) (6) (7) (8) Male Population Over Age 15 (Millions)

1871

1881

1901

1911

1871

1881

1901

1911

Piedmont Liguria Lombardy Venetia

167 47 255 137

210 69 311 157

339 122 568 230

522 224 918 343

.971 .282 1.187 .890

1.018 .300 1.252 .934

1.082 .370 1.397 .982

1.139 .421 1.532 1.023

Emilia Tuscany Marches Umbria

96 115 39 20

110 140 45 22

157 214 61 40

274 332 82 59

.733 .741 .303 .193

.767 .770 .319 .205

.817 .853 .334 .229

.889 .897 .334 .229

Latium Abruzzi Campania Apulia

44 37 141 57

56 44 182 66

85 55 239 99

122 71 339 152

.318 .418 .934 .465

.348 .425 .978 .524

.425 .436 .988 .625

.448 .410 1.029 .669

Basilicata Calabria Sicily Sardinia

17 42 119 17

19 50 139 23

20 62 194 32

24 82 253 51

.166 .389 .842 .219

.163 .401 .958 .238

.144 .390 1.139 .272

.139 .391 1.205 .282

1,351

1,644

2,518

3,847

9.051

9.600

10.483

11.037

Italy

demographic growth was registered with only a poor-to-middling initial stock or investment flow.17 The correlations with the initial level of capital (lines 1–4) all rise from one intercensal period to the next; the related scatter diagrams (Figs. 17 and 18) suggest that these increases reflect a widespread convergence to a common relationship rather than the movement of a few outliers. This progressive tightening of the apparent link between industrial growth or overall growth and social-overhead capital is intriguing; one might associate it with the progressive transformation of the economy, and the rise of new activities more closely dependent on infrastructure than the traditional activities they displaced. Artisans tend to locate where their customers congregate, factory industry where it is well served by infrastructure; as the latter displaced the former, the growth-inducing role of infrastructure naturally gathered strength.18

36

CARLO CICCARELLI AND STEFANO FENOALTEA

(Continued). B. Intercensal relatives (1) (2) (3) (4) Value Added in Manufacturing: Ratio

(5) (6) (7) (8) Male Population Over Age 15: Ratio

1881/ 1871

1901/ 1881

1911/ 1901

1911/ 1871

1881/ 1871

1901/ 1881

1911/ 1901

1911/ 1871

Piedmont Liguria Lombardy Venetia

1.26 1.47 1.22 1.14

1.61 1.78 1.83 1.47

1.54 1.83 1.62 1.49

3.12 4.78 3.59 2.50

1.05 1.06 1.05 1.05

1.06 1.23 1.12 1.05

1.05 1.14 1.10 1.04

1.17 1.49 1.29 1.15

Emilia Tuscany Marches Umbria

1.15 1.22 1.14 1.10

1.42 1.53 1.36 1.80

1.74 1.55 1.36 1.45

2.85 2.90 2.11 2.87

1.05 1.04 1.05 1.06

1.07 1.11 1.05 1.12

1.09 1.05 1.00 1.00

1.21 1.21 1.10 1.19

Latium Abruzzi Campania Apulia

1.27 1.19 1.29 1.16

1.52 1.25 1.31 1.50

1.44 1.29 1.42 1.55

2.77 1.92 2.40 2.69

1.09 1.02 1.05 1.13

1.22 1.03 1.01 1.19

1.05 .94 1.04 1.07

1.41 .98 1.10 1.44

Basilicata Calabria Sicily Sardinia

1.11 1.20 1.18 1.31

1.06 1.23 1.39 1.38

1.19 1.33 1.30 1.60

1.39 1.95 2.13 2.89

.98 1.03 1.14 1.09

.88 .97 1.19 1.14

.97 1.00 1.06 1.04

.84 1.01 1.43 1.29

Italy

1.22

1.53

1.53

2.85

1.06

1.09

1.05

1.22

The correlations with the current additions to capital (lines 5–8) are more ambiguous. Substantively, they pick up a bidirectional relationship, as within a decade (or two) investment can induce growth, and growth investment.19 Formally, the monotonic increase in the correlations themselves reappears with the demographic measure (columns 4–6), but not the industrial measures (columns 1–3); and as suggested by the related scatter diagrams (Figs. 19 and 20) the results are sensitive to a few outliers. These two formal complaints tend in part to offset each other. The relatively high manufacturing–growth correlations over the initial decade (row 5, columns 1–3) are due largely to the Ligurian outlier; suppressing that one observation, the correlations in columns 1–3 fall to .05, .37, and .20, respectively. Over the second decade, conversely, the relatively low correlation with railway investment (row 6, column 1) is tied to the exceptional rise of manufacturing in Umbria, which reflects the opening of

37

Social-Overhead Construction in Italy’s Regions, 1861–1913

(Continued). C. Intercensal new construction (average 1911-price value added) (1) (2) (3) (4) Railways (Lire/Square kilometer)

(5) (6) (7) (8) Other Civil (Lire/Square kilometer)

1871– 1880

1881– 1900

1901– 1910

1871– 1910

1871– 1880

1881– 1900

1901– 1910

1871– 1910

Piedmont Liguria Lombardy Venetia

175 779 275 332

330 559 342 293

354 606 227 328

297 626 296 312

187 605 173 164

198 1,083 342 223

333 1,453 607 379

229 1,056 366 247

Emilia Tuscany Marches Umbria

71 71 35 190

316 174 274 83

254 214 51 18

239 158 159 94

208 192 148 172

229 163 196 232

407 272 245 356

268 198 196 248

Latium Abruzzi Campania Apulia

83 224 165 20

546 353 448 253

168 49 123 48

336 245 296 143

432 238 494 190

851 280 476 142

593 329 689 193

682 281 534 167

Basilicata Calabria Sicily Sardinia

337 375 382 269

170 287 279 28

0 8 31 2

169 239 243 82

237 272 289 105

196 258 311 64

235 389 455 83

216 294 342 79

Italy

218

282

157

235

231

275

390

293

the Terni steel works; without that one observation, the correlation rises to .40. With a judicious pruning of the sample, therefore, the pattern of rising correlations can be recovered. But the removal of outliers cuts both ways. If Liguria and Latium are pruned from the sample, for example, the correlations in rows 5–7 fall successively to .26, .01 and .03 in column 2, to .13, .21, and .53 in column 4, and to .07, .20, and .30 in column 5. The prudent conclusion suggested by the scatter diagrams is that the correlation between growth and current investment in railway and other civil infrastructure was generally higher in the last intercensal period than in earlier times. In the case of railway construction, this change is readily associated with the shift from remedial investment to adaptive investment: the former aimed at the construction of railways where these were not already in place, the latter at the expansion of existing facilities strained by heavy traffic. A similar mechanism may have been at work for other civil

38

CARLO CICCARELLI AND STEFANO FENOALTEA

(Continued). D. Simple correlations: initial stock/investment per square kilometer and growth (1) (2) (3) Relative Manufacturing Product

(4) (5) (6) Relative Male Population Aged 15þ

Railways

Other

Total

Railways

Other

Total

Initial stock: 1. 1881/1871 2. 1901/1881 3. 1911/1901 4. 1911/1871

.51 .70 .71 .86

.23 .42 .68 .52

.30 .55 .72 .62

.02 .41 .71 .47

.10 .27 .79 .38

.08 .34 .79 .42

Investment: 5. 1881/1871 6. 1901/1881 7. 1911/1901 8. 1911/1871

.54 .17 .73 .61

.68 .35 .48 .60

.69 .30 .60 .62

.11 .26 .70 .35

.04 .46 .60 .44

.05 .41 .67 .42

1.8 Manufacturing product Male labor force

1.6

1.4

1.2

1

0.8

0.6

0.4 PI

Fig. 16.

LI

LO VE EM TO MA UM LA AB CM PU BA CL

SI

SA

Manufacturing Growth and Male Labor-Force Growth (Regional Ratios, 1911 Over 1871, Scaled by the Corresponding National Ratio).

39

MALE LABOR FORCE, 1881/1871

MANUFACTURING PRODUCT, 1881/1871

Social-Overhead Construction in Italy’s Regions, 1861–1913

1.5 LI 1.4 1.3 SA CL AB SI

1.2

1.1

CM

BA

LA PI TO LO

PU VE MA EM UM

0

20

40

60

80

MALE LABOR FORCE, 1901/1881

MANUFACTURING PRODUCT, 1901/1881

LO

UM

LI

PI TO PU LA VE EM 1.4 SA SI MA CM AB CL 1.2 1.6

BA 1.0 0

50

100

150

1.6

LO

SA PU

TOPI VE UM LA CM CL MA SI AB

1.4

1.2 0

BA 50

100

150

200

250

AVERAGE MAINTENANCE / SQ. KM., 1899-1903

Fig. 17.

SA

LA UM VE CM MAPIEM

TO

CL AB

LI

LO

1.00 BA 0.95

1.30 LI

LA SIPU

1.20 SA 1.10

1.00

0.90

UMTO EM MA VE PI AB CM CL

LO

BA 0

50

100

150

200

AVERAGE MAINTENANCE / SQ. KM., 1879-1883

MALE LABOR FORCE, 1911/1901

MANUFACTURING PRODUCT, 1911/1901

LI EM

PU

1.05

200

AVERAGE MAINTENANCE / SQ. KM., 1879-1883

1.8

1.10

SI

0 20 40 60 80 AVERAGE MAINTENANCE / SQ. KM., 1869-1873

AVERAGE MAINTENANCE / SQ. KM., 1869-1873

1.8

1.15

1.15

LI

1.10

EM PU SI

1.05

SA

LO

TO LA PI VE CM

CL MA UM

1.00 BA 0.95

AB 0

50

100

150

200

250

AVERAGE MAINTENANCE / SQ. KM., 1899-1903

Railway Capital and Growth: Initial Levels.

construction, too; and that would account for the higher correlations with overall labor-force growth than with manufacturing growth alone.20 None of this closes the Gerschenkron–Romeo debate, and those glorious old warriors can fight on in Valhalla. The relatively high correlations of manufacturing growth with the initial stock of railway capital (column 1,

1.5

LI

1.4 SA

1.3

TO

CL AB

1.2

LO

SI

PU MA 1.1

CM

LA PI

VE EM

BA UM 50

100

150

200

250

MALE LABOR FORCE, 1881/1871

CARLO CICCARELLI AND STEFANO FENOALTEA

1.6 PU 1.4 SA 1.2

PI TO LA

MA SI

AB CL

EM

VE

CM

BA 1.0 100

200

300

400

AVERAGE MAINTENANCE / SQ. KM.,1879-1883

LI

1.8 EM 1.6 SA

LO PU TO PI

UM 1.4 CL AB

LA CM

VE

MA SI

1.2 BA 100 200 300 400 500 AVERAGE MAINTENANCE / SQ. KM.,1899-1903

Fig. 18.

SI

PU 1.10

LA

SA UM

1.05

MAPI

LI LO VE EM

CM TO

CL AB 1.00 BA 0.95 50

100

150

200

250

300

AVERAGE MAINTENANCE / SQ. KM.,1869-1873

MALE LABOR FORCE, 1901/1881

LO LI

1.8 UM

1.15

300

AVERAGE MAINTENANCE / SQ. KM.,1869-1873

1.30

1.10

PU SA UM AB

1.00 0.90

LI

LA

1.20

SI LO

TO

EM

MAPI

VE

CM

CL BA 100

200

300

400

AVERAGE MAINTENANCE / SQ. KM.,1879-1883

MALE LABOR FORCE, 1911/1901

MANUFACTURING PRODUCT, 1911/1901

MANUFACTURING PRODUCT, 1901/1881

MANUFACTURING PRODUCT, 1881/1871

40

1.15

LI

1.10 1.05 1.00

LO EM SA UMCL

PU SI TO PI

LA CM

VE

MA

BA 0.95

AB 100 200 300 400 500 AVERAGE MAINTENANCE / SQ. KM.,1899-1903

Other Civil Capital and Growth: Initial Levels.

rows 1–4) would seem to support Romeo’s emphasis on the importance of the transportation system, but do not speak to his specific claims that the critical (‘‘prerequisite’’) lines were the national trunks built in the 1860s and 1870s, and that that system had to be ‘‘completed’’ before industry could develop; the similarly high correlation with current investment in the 1900s

41

1.5 LI 1.4

1.3 LA TO

1.2

SA

CM PI

LO AB

PUEM MA 1.1

UM 0

CL SI VE BA

200

400

600

MALE LABOR FORCE, 1881/1871

MANUFACTURING PRODUCT, 1881/1871

Social-Overhead Construction in Italy’s Regions, 1861–1913

SA

PU VE EM SI MA

LA

CM

CL AB

1.2 BA 1.0

LI EM LO

1.6 SA

TO

PU UM 1.4

PI VE

CMLA

MA CL SI AB

1.2 BA 0 200 400 600 AVERAGE NEW CONSTRUCTION/SQ. KM.,1901-1910

Fig. 19.

SA

UM MA EM CM PI LO VE TO CL AB

LI

1.00 BA 0.95 0

200

400

600

800

AVERAGE NEW CONSTRUCTION/SQ. KM.,1871-1880

1.30

1.15

0 200 400 600 AVERAGE NEW CONSTRUCTION/SQ. KM.,1881-1900

1.8

1.05

MALE LABOR FORCE, 1901/1881

TO 1.4

LI

PI

1.6

LA

MALE LABOR FORCE, 1911/1901

MANUFACTURING PRODUCT, 1901/1881 MANUFACTURING PRODUCT, 1911/1901

LO

UM

SI

PU 1.10

800

AVERAGE NEW CONSTRUCTION/SQ. KM.,1871-1880

1.8

1.15

LI LA

1.20

PUSI SA

1.10

UM

LO

TO

EM MA VEPI AB

1.00

CM

CL

0.90

BA 0 200 400 600 AVERAGE NEW CONSTRUCTION/SQ. KM.,1881-1900

LI LO EM

1.10 1.05

PU SI LA TO SA CM

PI VE

UMMA 1.00 CL BA 0.95

AB

0 200 400 600 AVERAGE NEW CONSTRUCTION/SQ. KM.,1901-1910

Railway Capital and Growth: Current Additions.

(row 7, column 1) might suggest otherwise, but in Romeo’s scheme it could be dismissed as a case of reverse causation. Gerschenkron’s argument is not touched at all: the latter turns on the demand-pull effects of railway construction on industrial growth, but the market for rails and rolling stock is a national one, and no connection between the consumption and the production of railway equipment is implied at the regional level.

CARLO CICCARELLI AND STEFANO FENOALTEA

1.5 LI 1.4 1.3 SA LA

PI LO TO

ABCLSI PU EM MA VE UM BA 1.1 100 200 300 1.2

400

CM

500

MALE LABOR FORCE, 1881/1871

MANUFACTURING PRODUCT, 1881/1871

42

LO

LI

PI TO PU VE EM SA MA SI

1.6 1.4

LA

CM

AB CL

1.2

BA 1.0 0

200

400

600

800

LO

SA

1.6

PUTOPI VE UM 1.4

LACM

MA

CL AB SI

1.2

BA 0

500

1000

1500

AVERAGE NEW CONSTRUCTION/SQ. KM.,1901-1910

Fig. 20.

LI

CM

1.00 BA 0.95 200

300

400

500

600

1.30 LI

LA

1.20

PU SA

1.10

SI

TOUM LO EM PI MA VE AB

1.00

CM

CL

0.90 0

BA 200

400

600

800

1000

AVERAGE NEW CONSTRUCTION/SQ. KM.,1881-1900

MALE LABOR FORCE, 1911/1901

MANUFACTURING PRODUCT, 1911/1901

LI

UM MA VE LO PIEM TO CL AB

1.05

1000

EM

LA

SA

100

AVERAGE NEW CONSTRUCTION/SQ. KM.,1881-1900

1.8

1.10

AVERAGE NEW CONSTRUCTION/SQ. KM.,1871-1880

MALE LABOR FORCE, 1901/1881

MANUFACTURING PRODUCT, 1901/1881

UM

SI

PU

600

AVERAGE NEW CONSTRUCTION/SQ. KM.,1871-1880

1.8

1.15

1.15

LI LO EM PU SI TOPI LA SA VE CM

1.10 1.05

MAUMCL

1.00

BA 0.95

AB 0

500

1000

1500

AVERAGE NEW CONSTRUCTION/SQ. KM.,1901-1910

Other Civil Capital and Growth: Current Additions.

8. PUBLIC WORKS AND WELFARE IMPROVEMENTS: PRELIMINARY NON-RESULTS Table 4 is analogous to Table 3, but is concerned with welfare improvements rather (or other) than growth. Panel A transcribes estimates of three

43

Social-Overhead Construction in Italy’s Regions, 1861–1913

Social-Overhead Civil Capital and Long-Term Welfare Improvements. A. Welfare indicators (1) (2) (3) Late Fetal Infant Mortality (Per Thousand)

(4) (5) (6) Life Expectancy at One Year of Age (Years)

(7) (8) Mean Height of Army Draftees, by Birth Year (centimeters)

c.1871

c.1891

c.1911

1871

1891

1911

1874

1918

Piedmont Liguria Lombardy Venetia

239 230 255 253

206 197 230 209

163 159 195 163

46 43 42 46

51 48 49 53

55 52 51 55

164.6 165.1 164.7 166.7

168.1 168.8 167.4 168.6

Emilia Tuscany Marches Umbria

294 242 271 255

252 208 238 219

181 152 165 168

44 38 45 47

50 49 51 48

55 54 56 56

165.3 164.9 163.8 163.9

167.8 168.1 165.6 166.0

Latium Abruzzi Campania Apulia

260 232 241 249

202 217 224 223

160 175 181 192

37 38 39 40

46 43 43 44

51 45 53 47

164.0 162.5 162.9 162.2

166.4 164.5 165.0 163.5

Basilicata Calabria Sicily Sardinia

273 228 227 226

245 215 225 178

191 170 199 150

38 37 44 39

44 43 44 43

50 51 47 49

159.3 161.1 161.7 159.9

162.0 163.7 164.2 161.9

Italy

248

220

177

42

47

51

163.8

166.2

indicators near the beginning, middle, and end of the period at hand. The first refers to late fetal/infant mortality, calculated as the number of stillbirths plus deaths within one year of birth, divided by total births, in 1870–1872, 1890–1892, and 1910–1912 (columns 1–3).21 The second refers to life expectancies at one year of age (to exclude the otherwise dominant effect of infant mortality) in 1871, 1891, and 1911 (columns 4–6). These were calculated from life expectancies at birth generously provided by Leandro Conte, using the corresponding year-specific data on live births and infant deaths.22 The third refers to the mean height of army draftees born in 1874 and in 1918 (columns 7–8). These data are taken directly from Costanzo (1948, p. 80); no other regional figures are available for the period at hand.

44

CARLO CICCARELLI AND STEFANO FENOALTEA

(Continued). B. Welfare improvements (1) (2) (3) Late Fetal/Infant Mortality (Per Thousand)

(4) (5) (6) Life Expectancy at One Year of Age (Years)

(7) Mean Height of Army Draftees, by Birth Year (centimeters)

c.1891/ c.1871

c.1911/ c.1891

c.1911/ c.1871

1891/ 1871

1911/ 1891

1911/ 1871

1918/ 1874

Piedmont Liguria Lombardy Venetia

.86 .86 .90 .83

.79 .81 .85 .78

.68 .69 .76 .64

1.11 1.12 1.17 1.15

1.08 1.08 1.04 1.04

1.20 1.21 1.21 1.20

1.021 1.022 1.016 1.011

Emilia Tuscany Marches Umbria

.86 .86 .88 .86

.72 .73 .69 .77

.62 .63 .61 .66

1.14 1.29 1.13 1.02

1.10 1.10 1.10 1.17

1.25 1.42 1.24 1.19

1.015 1.019 1.011 1.013

Latium Abruzzi Campania Apulia

.78 .94 .93 .90

.79 .81 .81 .86

.62 .75 .75 .77

1.24 1.13 1.10 1.10

1.11 1.05 1.23 1.07

1.38 1.18 1.36 1.18

1.015 1.012 1.013 1.008

Basilicata Calabria Sicily Sardinia

.90 .94 .99 .79

.78 .79 .88 .84

.70 .75 .88 .66

1.16 1.16 1.00 1.10

1.14 1.19 1.07 1.14

1.32 1.38 1.07 1.26

1.017 1.016 1.015 1.013

Italy

.89

.80

.71

1.12

1.09

1.21

1.015

These estimates are what can be taken from the current literature; but they must be considered tentative. There are apparent incongruities in the infant-mortality cross-sections (columns 1–3), which suggest that the data are not in fact homogeneous across space (and perhaps over time).23 The life expectancies at birth that underlie the present estimates at one year of age contain, in 1871 and 1891, a single estimate for the entire continental Mezzogiorno; the differences in the present regional estimates from the Abruzzi to Calabria (columns 4–5) thus stem entirely from differences in infant mortality (which may well also be distorted).24 Costanzo’s figures would appear to be the best of this poor lot, not least because Italy’s internal migrations did not become massive until much later.

45

Social-Overhead Construction in Italy’s Regions, 1861–1913

(Continued). C. Simple correlations: initial stock/investment per square kilometer and welfare improvements (1) (2) (3) (4) (5) (6) Late Fetal/Infant Life Expectancy at One Mortality (Per Thousand) Year of Age (Years)

Railways Other Total

Railways Other Total

(7) (8) (9) Mean Height of Army Draftees, by Birth Year (centimeters)a Railways Other Total

Initial stock: 1. 1891/1871 2. 1911/1891 3. 1911/1871

.22 .08 .27

.10 .07 .14

.13 .08 .17

.27 .35 .03

.19 .34 .07

.22 .36 .06

.51

.22

.29

Investment: 4. 1891/1871 5. 1911/1891 6. 1911/1871

.03 .02 .06

.14 .13 .04

.11 .09 .05

.01 .14 .06

.18 .01 .08

.11 .06 .03

.54

.46

.50

a

Calculated as 1918/1874.

These caveats are reinforced if the indicators are considered together. One would expect a strong negative correlation between infant mortality on the one hand and life expectancy at one year of age on the other; the late observations (columns 3 and 6) yield a weak correlation of the correct sign (.42), the early observations not even that (.22 in 1871 and .12 in 1891). This may be due to a differential improvement of the life-expectancy estimates (initially weakened by the uniform life-expectancy-at-birth figure for the continental South), or to a differential improvement of the infant-mortality figures; in either case, the region-specific improvements would also be distorted. Panel B reports the welfare improvements yielded by these three indicators (measured simply as the ratios of the corresponding levels). These improvements too are not strongly correlated. The signs of the correlation coefficients are the expected ones, but their absolute values are disappointing: .40, .16, and .44 for infant-mortality and life-expectancy improvements (respectively columns 1 and 4, 2 and 5, and 3 and 6), .08 for infant-mortality and height improvements (columns 3 and 7), .22 for lifeexpectancy and height improvements (columns 6 and 7). Over the long term, the greatest declines in infant mortality were concentrated in central Italy, from Venetia to Latium; the gains in life expectancy were often close to the

46

CARLO CICCARELLI AND STEFANO FENOALTEA

national norm, with some high gains associated with low initial levels (Tuscany, Latium, Campania, Basilicata, and Calabria) and vice-versa (Sicily); the gains in height were well above average in Tuscany and especially Piedmont and Liguria, and well below average in Venetia, the Marches, and especially Apulia. Panel C reports the simple correlations between, on the one hand, the measured welfare improvements, and, on the other, per-unit-area socialoverhead capital, measured both by the initial stocks (calculated in, or as in, Table 2) and by the total flows of new construction (calculated in, or as in, Table 3). In general, these correlations are disappointing. The correlations between the initial stocks and the changes in infant mortality (rows 1–3, columns 1–3) are all of the expected sign, but the absolute values are low or very low. The correlations between the initial stocks and the changes in life expectancies (rows 1–3, columns 4–6) are even worse, and under half are so much as of the expected sign. The corresponding correlations with the current investment flows (rows 4–6, columns 1–6) are no better: all are near zero, many of the wrong sign. The correlations between the initial stocks, or current investment flows, and the changes in heights (rows 3 and 6, columns 7–9) are, perhaps not coincidentally, the best of this poor lot: the signs are correct, the absolute values respectable. Not unreasonably, height gains appear more closely tied to current investment than to the prior stock; the former correlations (row 6, columns 7–9) are the highest of those obtained here. But even these results are of small comfort, for they are driven essentially by the Ligurian outlier. That region displayed the highest relative height gain, and received an enormous amount of investment per unit area (Table 1): if it is dropped from the sample, these correlations drop to .25 (railways), .09 (other), and .16 (total).25 On balance, these results are encouraging. They certainly encourage the pursuit of better results, as may be expected with improved estimates of the welfare indicators – and with disaggregated measures of social-overhead capital, for the disease environment is surely more affected by proper aqueducts and sewers than by proper harbors or street lighting.

9. CONCLUSION This article presents time series estimates of social-overhead construction in Italy’s regions; they cover private as well as public projects, and maintenance as well as new construction.

Social-Overhead Construction in Italy’s Regions, 1861–1913

47

The new-construction series point to a largely common cycle in nonrailway work, and largely idiosyncratic bursts of railway construction; but there may be less there than meets the eye, as the relative uniformity of the former may reflect no more than the aggregation of a much larger number of equally idiosyncratic series. Average social-overhead investment was more nearly uniform per man than per square kilometer, but on either measure Liguria did best, with Latium second. Maintenance, which at times outweighed new construction, serves here as an index of the underlying stock; given the extremely long lives of social-overhead capital goods, a perpetual-inventory measure is simply not feasible. There is limited evidence of convergence in railway infrastructure, and not even that in other civil infrastructure. Both the initial endowments and the mean additions to infrastructure are increasingly correlated with long-term industrial and overall growth. The modernization of the economy may have strengthened the growthinducing role of the social-overhead capital stock; conversely, current investment may have been shaped initially by geo-political considerations, and later by the need to loosen the infrastructure capacity constraints induced by growth itself. The new estimates shed less light on the relation between social-overhead capital and other welfare indicators, in part because the latter measures are still rudimentary. The strongest correlations are observed between the investment flows, or the initial stocks, and the gain in draftees’ mean heights; the correlations with the improvements in infant mortality or life expectancies are very weak, but these data are poor, and in any case the influence of social-overhead capital is surely more fruitfully investigated with infrastructure statistics far more detailed than the aggregates presented here.

NOTES 1. The early regional benchmark estimates appeared in Zamagni (1978), Esposto (1992), and Fenoaltea (2003b); for refined versions see Felice (2005, 2007). In principle, the new regional series aim to replicate, at the regional level, the richly documented estimates compiled for the country as a whole (for example, Fenoaltea, 2003a). In practice, the detail that can be captured is often less: at the national level, for example, the data on international trade (in intermediate goods) are used to separate the time paths of successive stages of production, but interregional trade was not recorded. The new regional production series for the textile, mining, chemical, utilities, and metalmaking industries appear in Fenoaltea (2004), Fenoaltea and Ciccarelli (2006), and Ciccarelli and Fenoaltea (2007, 2008a, 2008b).

48

CARLO CICCARELLI AND STEFANO FENOALTEA

2. These regional series disaggregate the national estimates presented in Fenoaltea (1984, 1986). Those national series included the new construction of public buildings (separately estimated), and their maintenance; the present estimates do not. The distinction between civil and military (non-railway, non-building) construction is new. 3. ‘‘Real’’ value added is measured, in construction as in other industries, by summing over 1911-price-value-added-weighted physical series (or, equivalently, own-price-deflated, and scaled, current expenditure series) only as a practical expedient. The desired measure is a deflation of current-price value added by a common price index (Fenoaltea, 1976; Fua`, 1993), but the latter measure requires far more information, and cannot yet be calculated. 4. For a broad review of infrastructure development in Italy from Unification to the present day, and the related literature, see Cannari and Chiri (2002). 5. The familiar retrospectives are Ministero dei Lavori pubblici, Ragioneria centrale (1900) and De’ Stefani (1925); both provide a regional breakdown of new construction by the Ministero dei lavori pubblici, but only the earlier similarly breaks down the corresponding maintenance expenditure. Neither provides a systematic regional breakdown of expenditure by other Departments (notably War and Navy). 6. The deflators are relatively crude. The basic building blocks are price series for high-grade construction materials, and a construction-wage series. The former are combined into a high-grade-materials cost index; the latter serves as a cost index for labor, and for low-grade materials (which presumably lacked scarcity value, and were worth little beyond the labor they embodied). These two cost indices are in turn combined, with different weights that reflect their share of total cost at 1911 prices, into separate cost indices for maintenance on the one hand and new construction on the other. The cost shares, and the ratio of value added to value, were calculated from the cost figures for different types of construction in contemporary civilengineering manuals. For new construction, in particular, the cyclical variations in the undeflated and deflated series far exceed those of the deflator. 7. A detailed summary of the sources and methods underlying the national estimates appears in Fenoaltea (1984, 1986). Since the regional estimates essentially subaggregate the local-level components of those national figures, that summary provides further details on the sources and methods underlying the regional series as well. A full account of the derivation of the national and regional series is available on request. 8. One might wish to exclude uninhabitable areas such as high mountains or swamps, but such corrections are not pursued here. In the interest of simplicity, similarly, the male population of working age is identified directly with the corresponding labor force. 9. These are computed from the unrounded figures underlying those reported in panels B–E. 10. The use of investment series and a depreciation rate to annualize an extraneous initial estimate of the capital stock is at times also subsumed under the ‘‘permanent-inventory’’ approach (for example, Gleed & Rees, 1979); but this terminology masks the characteristic feature of that approach, which was invented precisely to obtain stock estimates from the flow data themselves, and those alone.

Social-Overhead Construction in Italy’s Regions, 1861–1913

49

11. To hazard an estimate, one may note that cumulated (1911-price value added in) new construction from 1861 to 1913 equaled some 3.8 billion lire for railways, and 4.8 billion for other social-overhead structures; in 1913, the corresponding maintenance burdens were 41.7 and 85.5 million lire (again in 1911-price value added terms). If the railways’ maintenance bill in 1913 is associated with the corresponding cumulative investment since 1861, one obtains a ratio equal to (41.7/3,800); and if one sets it equal to the corresponding ratio for other structures, or (85.5/(4,800þx)), where x is the capital stock in 1913 put in place before 1861, one obtains an estimate of x equal to some 40% of the total non-railway stock in 1913. Even that is a lower bound, as it assumes that railways were no more maintenance-intensive than other structures; were they assumed to require 20% more maintenance, the estimate of x would rise to nearer 50% of the non-rail total extant in 1913. 12. Nor does it seems possible, appearances perhaps to the contrary, to tease capital-stock estimates from the present new-construction and maintenance estimates together; see below, note 13. 13. This weakness is not trivial. On the simple assumption of infinite lives, an initial estimate of the capital stock could be obtained from the maintenance and newconstruction estimates together. Cumulative new construction measures the absolute increment in the capital stock, dK; the relative increment in the capital stock dK/Ko is approximated by the relative increment in maintenance, dm/mo; and Ko is estimated directly as dK(mo/dm). For non-railway civil capital, the national estimates in Table 2 for 1869–1873 and 1909–1913 yield mo ¼ 51.6 and dm ¼ 25.8 million lire (column 2); from 1871 to 1910 dK ¼ 3,354.8 million lire (from Appendix A, summing over regions), whence K1871 ¼ 6,699.5 (and K1911 ¼ 10,054.3) million lire. The implied capital/maintenance ratio (K/m), constant by assumption, equals 130. Repeating the calculation with the regional figures, one obtains (K/m) ratios ranging from 67 to 315. These are the inverse of the maintenance-intensity of the new capital; the measure of the latter is particularly low where much was spent on new construction, and maintenance expenditure grew little if at all. The source of this variation may reflect anything from composition effects, as noted in the text, to the incidence of abandoned projects; the heroic assumption is the extrapolation of these local effects, measured for the period at hand, to the capital in place at the start, and thus to the regions’ prior history. In the circumstances, a correction of the maintenance figures to reflect the calculated (K/m) ratios seems imprudent, and the maintenance figures are here cautiously examined simply as they are. 14. The initial census-year estimates of industrial production (Fenoaltea, 2003b, pp. 1061–1068) simply allocated the national 1911-price value added for each of 15 sectors in proportion to the regional distribution of its labor force. The present figures include the updated (time-series) estimates for textiles, clothing, metalmaking, non-metallic mineral products, and chemicals, and the original estimates for the other sectors (foodstuffs, tobacco, leather products, wood products, engineering, paper and publishing, and sundry manufacturing). Table 3 considers manufacturing output alone; it is less sensitive than aggregate industrial production to short-run swings in construction, and thus to the specific years in the sample. The differences in male labor-force growth rates are in turn tied to differential migration rates, rather

50

CARLO CICCARELLI AND STEFANO FENOALTEA

than to differential rates of natural growth; they fail to track changes in relative products to the extent that relative labor-intensities varied over time. 15. While three of the four overall leaders were not industrial leaders, among the also-rans manufacturing and overall growth were closely associated: for the 12 other regions, the correlation between the two 1911/1871 relatives (panel B, columns 4 and 8) equals .93. 16. With the per-man measures of initial capital, the correlations typically remain positive, but tend to be lower than those with the per-area measure; with the per-man measures of investment, they are negative as often as not. Negative correlations are also frequently obtained if the growth measures are correlated directly with the relative maintenance burdens calculated from the successive panels of Table 2. The underlying problem seems to be that already highlighted in Note 13 above: the ratio of the change in maintenance dm to the increment in the capital stock dK varies not only across regions, but from period to period (in Table 2, from panel A to panel B, Tuscany actually registers a decline in columns 2 and 3). The maintenance measures of the capital stock perform better than the maintenance measures of the changes in the capital stock simply because the index of the capital stock (moþdm) is less affected by dm than the index of the change in capital represented by dm itself. 17. Excluding Apulia and Sicily the correlation in column 4 rises from .02 to .27 in row 1, and from .11 to .17 in row 4. 18. So interpreted, the pattern of correlations dovetails with the regional relocation of industry described in Fenoaltea (2003b, pp. 1073–1074). 19. The investment flow may depend on overall growth (which generates a demand for additional infrastructure) and indeed on sectoral growth (whence high investment in Liguria and Latium, where industry, trade, and government services boosted utilities- and transport-intensive urban growth, and comparatively low investment in Apulia and Sicily, where rapid overall growth seems to have derived overwhelmingly from agriculture); it may even depend negatively on growth, to the extent that the national government pursues regional policies and (as in the later twentieth century) pours infrastructure into the laggards. 20. Manufacturing growth may have become increasingly sensitive to new infrastructure (for example, to generate hydroelectric power), but this supply-side effect appears to have been secondary. 21. These are calculated from the data in Istituto centrale di statistica (1965, pp. 687–757); the early figure for Latium refers to 1872 alone. 22. Life expectancies at birth were calculated for 1871, 1891, and 1911 as elements of the Human Development Index presented in Conte, Della Torre, and Vasta (2007). 23. The surprisingly low levels obtained for some heavily rural regions may point to underreporting, especially of stillbirths. 24. The middle benchmark may further be distorted, with respect to trend levels, because the early 1890s were marked by widespread hardship. 25. These correlations are all, as noted, with the per-area measures of capital. With the per-man measures of initial capital, the correlations with the change in infant mortality are much strengthened (partly because Venetia, Emilia, and Tuscany lead on both scores), but those with the change in life expectancy change little, and those with the change in heights sharply fall. Similar results obtain with the per-man investment flows: the infant-mortality correlations are much improved

Social-Overhead Construction in Italy’s Regions, 1861–1913

51

(albeit at times only in the sense that a wrong result becomes no result at all), the life-expectancy correlations also improve, at least for non-railway (and total) investment, and the height-change correlations again deteriorate.

ACKNOWLEDGMENTS The regional estimates were produced as part of the research project on ‘‘Unita` d’Italia e sviluppo disuguale: la struttura creditizia e la crescita industriale per regioni dal 1861 al 1913’’ of the Ufficio Ricerche Storiche of the Bank of Italy. The material presented here represents the views of the authors alone, and cannot be attributed to the Bank. The anonymous referee, whose prompt and perceptive suggestions much improved the paper, is gratefully thanked.

REFERENCES Bonaglia, F., & Picci, L. (2000). Lo stock di capitale nelle regioni italiane. Working Paper no 374, Bologna: Dipartimento di scienze economiche, Universita` di Bologna. Cannari, L., & Chiri, S. (2002). Le infrastrutture economiche dall’Unita`. In: P. Ciocca & G. Toniolo (Eds), Storia economica d’Italia. 3. Industrie, mercati, istituzioni. 1. Le strutture dell’economia (pp. 225–297). Milan-Rome-Bari: IntesaBci and Laterza. Ciccarelli, C., & Fenoaltea, S. (2007). The chemical, coal and petroleum products, and rubber industries in Italy’s regions, 1861–1913: Time series estimates. Quaderni dell’Ufficio Ricerche Storiche, no 16, Rome: Banca d’Italia. Ciccarelli, C., & Fenoaltea, S. (2008a). The growth of the utilities industries in Italy’s regions, 1861–1913. Rivista di storia economica, 24, 175–205. Ciccarelli, C., & Fenoaltea, S. (2008b). Metalmaking in Italy, 1861–1913: National and regional time series. Unpublished ms. Rome: Universita` di Roma ‘‘Tor Vergata’’. Conte, L., Della Torre, G., & Vasta, M. (2007). The Human Development Index in historical perspective: Italy from political Unification to the present day. Quaderni del Dipartimento di Economia politica, no 491, Siena: Universita` degli Studi di Siena. Costanzo, A. (1948). La statura degli italiani ventenni nati dal 1854 al 1920. In: Istituto centrale di statistica, Annali di statistica, ser. VIII, vol. 2: Studi di statistica metodologica e demografica (pp. 59–123). Rome: A.B.E.T.E. De Nardo, V. (1957). Le spese per opere pubbliche dei comuni e delle province dal 1871 al 1955. Informazioni Svimez, 35–40. De’ Stefani, A. (1925). L’azione dello Stato italiano per le opere pubbliche (1862–1924). Rome: Libreria dello Stato. Esposto, A. G. (1992). Italian industrialization and the Gerschenkronian ‘great spurt’: A regional analysis. Journal of Economic History, 52, 353–362. Felice, E. (2005). Il valore aggiunto regionale. Una stima per il 1891 e per il 1911 e alcune elaborazioni di lungo periodo (1891–1971). Rivista di storia economica, 21, 273–314.

52

CARLO CICCARELLI AND STEFANO FENOALTEA

Felice, E. (2007). Divari regionali e intervento pubblico. Per una rilettura dello sviluppo in Italia. Bologna: Il Mulino. Fenoaltea, S. (1976). Real value added and the measurement of industrial production. Annals of Economic and Social Measurement, 5, 113–139. Fenoaltea, S. (1984). Railway construction in Italy, 1861–1913. Rivista di storia economica, 1(International issue), 27–58. Fenoaltea, S. (1986). Public works construction in Italy, 1861–1913. Rivista di storia economica, 3(International issue), 1–33. Fenoaltea, S. (1988). International resource flows and construction movements in the Atlantic economy: The Kuznets cycle in Italy, 1861–1913. Journal of Economic History, 48, 605–638. Fenoaltea, S. (2003a). Notes on the rate of industrial growth in Italy, 1861–1913. Journal of Economic History, 63, 695–735. Fenoaltea, S. (2003b). Peeking backward: Regional aspects of industrial growth in postUnification Italy. Journal of Economic History, 63, 1059–1102. Fenoaltea, S. (2004). Textile production in Italy’s regions, 1861–1913. Rivista di storia economica, 20, 145–174. Fenoaltea, S., & Ciccarelli, C. (2006). Mining production in Italy, 1861–1913: National and regional time series. Rivista di storia economica, 22, 141–208. Fua`, G. (1993). Crescita economica. Le insidie delle cifre. Bologna: Il Mulino. Gerschenkron, A. (1962). Rosario Romeo and the original accumulation of capital. In: A. Gerschenkron (Ed.), Economic Backwardness in Historical Perspective. Cambridge, MA: Harvard University Press. Gerschenkron, A. (1968). The industrial development of Italy: A debate with Rosario Romeo. In: A. Gerschenkron (Ed.), Continuity in history and other essays. Cambridge, MA: Harvard University Press. Gleed, R. H., & Rees, R. D. (1979). The derivation of regional capital stock estimates for UK manufacturing industries, 1951–1973. Journal of the Royal Statistical Society, Series A (General), 142, 330–346. Istituto centrale di statistica. (1965). Annali di statistica, ser. VIII, vol. 17: Sviluppo della popolazione italiana dal 1861 al 1961. Rome: Istituto centrale di statistica. Ministero dei Lavori pubblici, Ragioneria centrale. (1900). I pagamenti fatti dallo Stato per opere pubbliche negli esercizi finanziari dal 1862 al 1897–98: Relazione compilata in occasione della Esposizione universale di Parigi dell’anno 1900, 3 vols, Rome: Tipografia dell’Unione Cooperativa Editrice. Nitti, F. S. (1900). Nord e Sud. Prime linee di una inchiesta sulla ripartizione territoriale delle entrate e delle spese dello Stato in Italia. Turin: Roux e Viarengo. Picci, L. (2002). Le opere pubbliche dall’Unita` d’Italia: l’informazione statistica. Rivista di Politica Economica, 92, 29–80. Romeo, R. (1959). Risorgimento e capitalismo. Bari: Laterza. Vitali, O. (1969). La stima degli investimenti e dello stock di capitale. In: G. Fua` (Ed.), Lo sviluppo economico in Italia (vol. 3, pp. 478–537). Milan: Franco Angeli. Zamagni, V. (1978). Industrializzazione e squilibri regionali in Italia. Bilancio dell’eta` giolittiana. Bologna: Il Mulino.

53

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX A. SOCIAL-OVERHEAD NEW CONSTRUCTION: VALUE ADDED AT 1911 PRICES (MILLION LIRE) (1)

(2) (3) Piedmont

(4)

(5)

(6) Liguria

(7)

(8)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

3.04 4.95 6.75 6.97

2.82 3.51 19.75 25.66

7.68 7.25 3.94 2.47

13.54 15.71 30.45 35.10

.95 1.31 1.44 2.01

1.62 1.79 1.82 1.86

4.02 3.79 3.00 1.93

6.58 6.89 6.25 5.79

1865 1866 1867 1868 1869

3.40 2.04 4.40 6.15 7.21

26.69 9.75 2.92 5.99 3.24

2.10 2.85 1.03 .12 .23

32.19 14.65 8.35 12.26 10.68

2.45 3.73 4.74 5.75 4.42

3.49 2.32 1.75 3.15 2.64

4.97 5.78 4.98 4.52 2.30

10.91 11.83 11.46 13.42 9.35

1870 1871 1872 1873 1874

8.64 6.35 5.85 8.49 7.34

3.82 5.76 8.42 6.95 4.76

.10 .79 .24 .16 .15

12.56 12.91 14.51 15.60 12.24

7.61 10.76 10.51 6.59 5.47

3.59 2.48 2.31 2.13 3.10

1.96 2.01 1.65 1.07 1.07

13.16 15.25 14.47 9.79 9.64

1875 1876 1877 1878 1879

2.06 2.07 1.89 2.66 5.08

4.04 3.90 4.59 5.17 5.47

.06 .39 1.08 2.98 1.17

6.16 6.36 7.56 10.81 11.72

1.19 1.60 1.46 1.35 .96

3.37 2.77 2.93 4.32 3.89

1.13 1.60 1.76 1.91 1.32

5.68 5.97 6.15 7.57 6.17

1880 1881 1882 1883 1884

9.56 14.30 19.15 21.89 21.47

5.76 5.30 4.97 5.14 6.51

.90 1.50 3.22 3.78 3.46

16.22 21.10 27.33 30.81 31.44

1.25 2.49 3.24 3.72 2.98

4.62 4.08 6.80 5.24 7.58

1.35 1.52 1.57 2.90 2.24

7.22 8.09 11.60 11.86 12.80

1885 1886 1887 1888 1889

20.30 18.94 15.20 11.00 9.97

6.67 5.60 5.54 6.67 6.88

4.72 3.47 2.23 2.43 3.47

31.69 28.01 22.97 20.09 20.33

3.08 3.52 4.76 6.11 4.97

8.96 7.49 6.61 7.90 9.29

2.85 3.82 5.15 7.12 8.19

14.89 14.83 16.52 21.13 22.45

54

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX A. (Continued ) (1)

(2) (3) Piedmont

(4)

(5)

(6) Liguria

(7)

(8)

Railways Other civil Military Total Railways Other civil Military Total 1890 1891 1892 1893 1894

9.69 7.92 7.43 4.57 .76

6.22 5.36 5.09 5.06 4.69

3.39 2.17 1.54 1.04 .89

19.30 15.45 14.06 10.67 6.34

3.23 2.10 2.09 2.72 3.62

8.16 6.69 5.63 4.73 3.83

6.15 3.60 1.89 .99 .85

17.54 12.39 9.61 8.44 8.30

1895 1896 1897 1898 1899

.13 .45 1.28 3.79 2.97

5.08 5.61 5.72 5.67 6.71

.75 .49 .51 .68 .64

5.96 6.55 7.50 10.14 10.32

.94 1.52 1.74 1.44 2.20

3.36 3.11 3.00 3.26 3.91

.47 .26 .24 .29 .26

4.76 4.89 4.98 4.99 6.38

1900 1901 1902 1903 1904

2.58 4.33 9.76 15.91 19.11

7.65 7.15 7.20 6.36 6.11

.52 .69 1.27 1.32 .83

10.75 12.16 18.23 23.59 26.05

2.54 1.76 2.62 2.35 2.34

4.74 4.80 5.40 6.04 5.81

.25 .21 .25 .24 .39

7.54 6.77 8.27 8.63 8.54

1905 1906 1907 1908 1909

17.49 9.75 6.93 5.63 6.35

7.87 9.74 9.73 9.14 14.38

.87 .87 .61 .36 .16

26.24 20.36 17.28 15.12 20.89

2.73 3.56 3.83 3.75 3.94

6.27 7.71 8.86 9.40 10.28

.39 .15 .14 .18 .47

9.39 11.42 12.82 13.33 14.69

1910 1911 1912 1913

8.73 11.95 11.51 7.53

20.20 22.41 26.51 30.86

.34 .61 .64 .60

29.27 34.97 38.66 38.99

5.12 5.74 6.89 7.58

12.11 12.46 11.92 10.96

.57 .33 .20 .15

17.79 18.53 19.01 18.69

(12)

(13)

(14) (15) Venetia

(9)

(10) (11) Lombardy

(16)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

14.63 14.06 11.69 9.28

2.83 3.40 3.75 4.32

.00 .00 .00 .00

17.46 17.46 15.44 13.60

.00 1.34 2.69 3.87

3.56 3.36 4.08 4.09

4.08 4.03 3.99 3.91

7.64 8.72 10.76 11.86

55

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX A. (Continued ) (9)

(10) (11) Lombardy

(12)

(13)

(14) (15) Venetia

(16)

Railways Other civil Military Total Railways Other civil Military Total 1865 1866 1867 1868 1869

11.04 8.94 3.38 2.83 4.21

7.35 2.29 2.30 2.71 3.34

.00 .00 .00 .00 .00

18.39 11.23 5.68 5.54 7.55

5.12 4.01 .00 .00 .00

3.43 3.87 4.59 3.99 3.88

3.84 1.89 .01 .01 .06

12.39 9.77 4.60 4.00 3.94

1870 1871 1872 1873 1874

5.25 4.54 6.68 8.19 6.26

3.15 2.48 2.74 6.44 5.43

.00 .00 .00 .00 .00

8.40 7.02 9.42 14.63 11.69

.00 .50 2.71 8.69 14.40

4.65 3.50 3.81 5.03 4.54

.16 .17 .60 .59 1.04

4.80 4.17 7.12 14.30 19.98

1875 1876 1877 1878 1879

4.15 4.63 4.25 6.45 8.47

3.73 3.45 3.81 4.02 4.64

.00 .00 .00 .00 .00

7.88 8.08 8.06 10.47 13.11

19.41 18.38 10.32 2.39 1.91

4.00 3.45 3.74 3.49 3.88

.86 .92 .89 1.08 .53

24.27 22.76 14.95 6.96 6.32

1880 1881 1882 1883 1884

12.54 17.02 16.60 14.68 15.82

4.81 3.34 3.27 3.16 3.72

.00 .00 .00 .00 .00

17.35 20.36 19.87 17.84 19.54

2.74 5.74 12.63 18.55 23.85

4.76 4.62 5.06 6.23 6.75

.40 .54 .48 .55 .60

7.90 10.90 18.17 25.32 31.20

1885 1886 1887 1888 1889

12.56 13.05 10.83 11.74 8.48

3.51 14.31 26.40 25.71 15.80

.00 .00 .00 .00 .00

16.07 27.36 37.23 37.45 24.28

25.92 18.69 10.93 8.95 4.53

6.91 5.94 5.56 5.47 5.36

.74 .53 .46 .66 .97

33.56 25.16 16.95 15.08 10.86

1890 1891 1892 1893 1894

9.85 7.32 6.63 5.04 3.10

7.47 6.59 5.66 5.58 5.46

.00 .00 .00 .00 .00

17.32 13.91 12.29 10.62 8.56

3.26 1.76 1.50 1.23 .71

6.89 7.50 5.87 4.84 4.70

.95 .68 .40 .17 .17

11.10 9.94 7.77 6.25 5.57

1895 1896 1897 1898 1899

.79 1.54 2.53 1.42 1.92

5.10 6.39 6.93 5.66 4.74

.00 .00 .00 .00 .00

5.89 7.93 9.46 7.08 6.66

.54 .81 1.01 1.51 .78

5.40 5.57 5.25 4.49 3.47

.19 .14 .13 .14 .15

6.13 6.52 6.39 6.14 4.39

56

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX A. (Continued ) (9)

(10) (11) Lombardy

(12)

(13)

(14) (15) Venetia

(16)

Railways Other civil Military Total Railways Other civil Military Total 1900 1901 1902 1903 1904

3.64 4.03 3.69 3.10 3.46

6.00 7.26 9.29 11.36 13.22

.00 .00 .00 .00 .00

9.64 11.29 12.98 14.46 16.68

1.05 1.05 1.42 1.33 2.86

3.76 5.67 6.55 6.47 6.43

.17 .15 .19 .19 .11

4.98 6.86 8.16 7.99 9.40

1905 1906 1907 1908 1909

3.52 5.50 7.09 7.07 7.92

15.04 16.33 14.26 16.36 20.66

.00 .00 .00 .00 .00

18.56 21.83 21.35 23.43 28.58

4.82 8.90 11.50 12.27 15.86

8.14 9.33 11.21 14.32 13.66

.14 .39 .97 1.82 2.43

13.09 18.61 23.68 28.40 31.94

1910 1911 1912 1913

9.39 8.56 8.54 9.87

22.33 19.74 16.47 16.13

.00 .00 .00 .00

31.72 28.30 25.01 26.00

20.58 23.21 22.64 19.49

11.35 13.58 14.94 13.36

4.32 6.29 6.33 5.81

36.25 43.08 43.91 38.66

(17)

(18) (19) Emilia

(20)

(21)

(22) (23) Tuscany

(24)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

17.39 10.14 4.09 .00

2.59 2.70 3.81 2.64

.00 .00 .00 .00

19.98 12.84 7.90 2.64

19.84 23.52 21.70 12.00

3.58 4.71 4.58 3.42

.00 .00 .00 .00

23.42 28.23 26.28 15.42

1865 1866 1867 1868 1869

.00 .00 .00 .61 1.16

2.75 2.21 2.38 2.62 2.49

.00 .00 .00 .00 .00

2.75 2.21 2.38 3.23 3.65

6.92 4.81 .77 1.35 2.32

5.11 6.84 6.97 7.20 6.18

.00 .00 .00 .00 .00

12.03 11.65 7.74 8.55 8.50

1870 1871 1872 1873 1874

1.68 2.25 1.75 .34 .59

3.50 3.28 4.09 5.02 5.26

.00 .00 .00 .00 .00

5.18 5.53 5.84 5.36 5.85

3.53 4.45 3.32 1.35 1.37

6.98 5.61 6.45 6.66 5.71

.00 .00 .00 .00 .00

10.51 10.06 9.77 8.01 7.08

57

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX A. (Continued ) (17)

(18) (19) Emilia

(20)

(21)

(22) (23) Tuscany

(24)

Railways Other civil Military Total Railways Other civil Military Total 1875 1876 1877 1878 1879

.96 1.30 1.13 1.01 1.70

4.55 4.21 4.27 4.09 3.77

.00 .00 .00 .00 .00

5.51 5.51 5.40 5.10 5.47

.98 1.25 1.04 .93 1.41

4.41 3.86 3.69 3.67 3.12

.00 .00 .00 .00 .02

5.39 5.11 4.73 4.60 4.55

1880 1881 1882 1883 1884

3.77 7.65 10.93 13.30 7.84

4.51 4.14 4.59 4.96 4.98

.00 .00 .00 .00 .00

8.28 11.79 15.52 18.26 12.82

.91 2.05 2.58 3.32 3.65

3.22 3.03 2.93 3.78 3.90

.08 .07 .04 .07 .07

4.21 5.16 5.55 7.17 7.62

1885 1886 1887 1888 1889

10.08 14.06 15.59 14.58 10.49

4.95 4.91 4.64 4.47 5.07

.00 .00 .00 .00 .00

15.03 18.97 20.23 19.05 15.56

3.78 5.52 8.35 9.75 9.24

3.83 4.18 4.76 5.68 5.56

.10 .11 .16 .22 .29

7.71 9.81 13.27 15.65 15.08

1890 1891 1892 1893 1894

3.20 2.91 4.33 6.04 3.00

5.36 5.15 4.68 4.81 5.20

.00 .00 .00 .00 .00

8.56 8.06 9.01 10.85 8.20

9.19 5.07 5.29 2.98 2.28

5.17 5.09 4.37 3.79 3.44

.29 .23 .13 .04 .04

14.65 10.38 9.79 6.81 5.75

1895 1896 1897 1898 1899

.09 .48 .96 .67 1.39

5.35 5.15 3.99 3.51 4.28

.00 .00 .00 .00 .00

5.44 5.63 4.95 4.18 5.67

1.77 1.63 2.04 2.45 1.33

3.45 3.12 2.85 3.03 3.28

.04 .03 .03 .03 .03

5.26 4.78 4.92 5.51 4.64

1900 1901 1902 1903 1904

3.17 3.74 3.33 3.28 3.96

4.47 4.37 5.07 5.78 6.23

.00 .00 .00 .00 .00

7.64 8.11 8.40 9.06 10.19

1.48 1.56 1.48 1.02 1.29

3.55 3.94 4.51 4.83 4.95

.04 .03 .03 .02 .01

5.07 5.53 6.02 5.87 6.26

58

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX A. (Continued ) (17)

(18) (19) Emilia

(20)

(21)

(22) (23) Tuscany

(24)

Railways Other civil Military Total Railways Other civil Military Total 1905 1906 1907 1908 1909

4.25 5.38 4.99 5.51 8.41

6.72 10.01 11.27 11.87 11.00

.00 .00 .00 .00 .00

10.97 15.39 16.26 17.38 19.41

1.33 3.51 7.56 8.86 11.63

5.63 6.84 7.97 8.34 9.24

.01 .02 .03 .05 .13

6.98 10.37 15.56 17.24 21.00

1910 1911 1912 1913

9.64 6.34 6.67 6.15

11.97 13.48 14.72 16.65

.00 .00 .00 .00

21.61 19.82 21.39 22.80

13.40 12.34 10.16 8.45

9.41 9.16 9.46 9.69

.17 .13 .09 .07

22.98 21.62 19.71 18.21

(25)

(26) (27) Marches

(28)

(29)

(30) (31) Umbria

(32)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

9.29 8.32 7.99 3.75

1.61 1.67 1.34 1.05

.00 .00 .00 .04

10.90 9.99 9.33 4.85

1.60 5.78 9.29 13.21

.17 .22 .26 .33

.00 .00 .00 .00

1.77 6.00 9.55 13.54

1865 1866 1867 1868 1869

4.96 3.88 .00 .00 .25

1.79 .89 .95 1.19 1.14

.02 .02 .02 .02 .00

6.76 4.80 .97 1.21 1.40

16.62 12.03 .00 .00 .00

.60 .69 .96 1.24 1.25

.00 .00 .00 .00 .00

17.22 12.72 .96 1.24 1.25

1870 1871 1872 1873 1874

.50 .73 .97 .80 .03

1.22 1.00 1.24 1.42 1.29

.00 .00 .00 .00 .00

1.72 1.73 2.21 2.22 1.32

.63 1.79 2.91 4.11 3.98

1.42 1.09 1.47 1.55 1.63

.00 .00 .00 .00 .00

2.05 2.88 4.38 5.66 5.61

1875 1876 1877 1878 1879

.05 .07 .06 .06 .04

1.13 1.33 1.63 1.70 1.75

.00 .00 .00 .00 .00

1.18 1.40 1.69 1.76 1.79

1.55 .00 .00 .00 1.40

1.69 1.76 2.06 1.97 1.78

.00 .00 .00 .00 .00

3.24 1.76 2.06 1.97 3.18

59

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX A. (Continued ) (25)

(26) (27) Marches

(28)

(29)

(30) (31) Umbria

(32)

Railways Other civil Military Total Railways Other civil Military Total 1880 1881 1882 1883 1884

.63 1.47 3.21 5.08 6.01

1.89 1.89 1.83 1.82 2.02

.00 .01 .02 .06 .07

2.52 3.36 5.06 6.96 8.10

2.69 4.10 5.78 4.67 .13

1.74 1.54 1.49 1.58 1.55

.00 .00 .00 .00 .00

4.43 5.64 7.27 6.25 1.68

1885 1886 1887 1888 1889

5.36 4.71 2.62 2.04 .20

1.89 1.96 2.41 2.59 2.44

.10 .11 .16 .22 .29

7.35 6.78 5.19 4.84 2.93

.45 .32 .00 .03 .05

1.38 1.42 1.62 1.72 1.70

.00 .00 .00 .00 .00

1.83 1.74 1.62 1.75 1.75

1890 1891 1892 1893 1894

.16 .64 1.24 1.87 4.74

2.16 1.85 1.77 1.65 1.54

.29 .23 .13 .04 .06

2.62 2.72 3.14 3.56 6.34

.06 .02 .01 .00 .02

1.93 2.09 1.83 1.60 1.40

.00 .00 .00 .00 .00

1.99 2.11 1.84 1.60 1.42

1895 1896 1897 1898 1899

3.95 2.88 3.85 3.03 .09

1.52 1.39 1.50 1.89 2.11

.09 .06 .06 .05 .06

5.56 4.33 5.40 4.97 2.26

.01 .02 .03 .12 .20

1.68 2.25 2.90 3.89 5.38

.00 .00 .00 .00 .00

1.69 2.27 2.93 4.01 5.58

1900 1901 1902 1903 1904

.12 .12 .25 .31 .41

1.84 1.74 1.89 1.95 1.96

.08 .06 .05 .05 .03

2.04 1.92 2.19 2.31 2.40

.18 .35 .10 .07 .08

6.15 4.13 2.29 2.08 1.37

.00 .00 .00 .00 .00

6.33 4.48 2.39 2.15 1.45

1905 1906 1907 1908 1909

.53 1.12 .92 .54 .43

2.30 2.44 2.44 2.60 3.00

.03 .06 .08 .13 .37

2.86 3.62 3.44 3.28 3.80

.10 .13 .16 .23 .33

2.49 4.01 4.88 4.39 4.03

.00 .00 .00 .00 .00

2.59 4.14 5.04 4.62 4.36

1910 1911 1912 1913

.29 .43 1.80 3.38

3.50 3.74 4.50 4.38

.49 .38 .28 .21

4.29 4.55 6.57 7.97

.17 .21 .23 .51

4.91 7.22 7.21 4.57

.00 .00 .00 .00

5.08 7.43 7.44 5.08

60

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX A. (Continued ) (33)

(34) Latium

(35)

(36)

(37)

(38) (39) Abruzzi

(40)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

9.34 9.23 5.65 7.68

2.14 2.08 2.22 2.14

.77 .76 .75 .74

12.24 12.07 8.63 10.56

7.00 9.59 9.97 4.86

2.42 2.43 2.10 2.22

.00 .00 .00 .00

9.42 12.02 12.07 7.08

1865 1866 1867 1868 1869

7.23 3.58 2.62 .00 .00

1.97 2.08 2.06 3.29 3.89

.73 .71 .72 .72 .71

9.93 6.37 5.41 4.00 4.60

.00 .00 .00 .00 1.12

2.67 2.47 2.85 2.65 2.68

.00 .00 .00 .00 .00

2.67 2.47 2.85 2.65 3.80

1870 1871 1872 1873 1874

.08 .18 .27 .58 .87

3.98 2.27 5.51 7.91 6.00

.70 .00 .00 .00 .00

4.76 2.45 5.78 8.49 6.87

2.25 4.37 6.54 6.75 4.20

3.54 3.40 4.10 4.63 4.01

.00 .00 .00 .00 .00

5.79 7.77 10.64 11.38 8.21

1875 1876 1877 1878 1879

1.05 1.42 1.45 1.41 1.14

4.72 4.49 5.54 6.52 5.02

.00 .00 .00 .00 .00

5.77 5.91 6.99 7.93 6.16

3.27 .00 .00 1.35 4.02

3.96 3.44 3.75 3.80 4.01

.00 .00 .00 .00 .00

7.23 3.44 3.75 5.15 8.03

1880 1881 1882 1883 1884

1.64 3.58 6.14 9.46 9.76

4.26 2.78 7.74 13.30 14.47

.00 1.05 2.03 2.56 2.08

5.90 7.40 15.92 25.32 26.30

6.59 9.74 10.97 6.40 4.75

4.16 4.30 4.59 4.45 4.95

.00 .00 .00 .00 .00

10.75 14.04 15.56 10.85 9.70

1885 1886 1887 1888 1889

8.02 8.91 7.90 9.56 10.69

13.48 17.63 21.97 22.34 21.60

1.38 .99 .69 .76 .75

22.88 27.53 30.56 32.65 33.04

5.92 7.73 9.24 7.42 1.04

5.42 5.80 6.28 6.01 5.38

.00 .00 .00 .00 .00

11.34 13.53 15.52 13.43 6.42

1890 1891 1892 1893 1894

12.18 14.18 12.63 7.35 8.01

15.68 9.49 8.10 6.61 5.24

.58 .38 .25 .14 .14

28.44 24.05 20.98 14.09 13.39

1.95 3.01 3.33 4.77 9.18

5.18 5.14 4.81 4.15 3.87

.00 .00 .00 .00 .00

7.13 8.15 8.14 8.92 13.05

61

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX A. (Continued ) (33)

(34) (35) Latium

(36)

(37)

(38) (39) Abruzzi

(40)

Railways Other civil Military Total Railways Other civil Military Total 1895 1896 1897 1898 1899

.34 .40 .27 .62 .91

4.23 4.26 4.26 3.94 4.31

.14 .08 .20 .24 .13

4.71 4.74 4.73 4.79 5.35

7.05 8.71 7.36 1.88 2.69

3.73 3.32 3.15 3.55 4.17

.00 .00 .00 .00 .00

10.78 12.03 10.51 5.43 6.86

1900 1901 1902 1903 1904

.96 .93 1.03 .82 1.00

4.22 5.43 6.97 6.40 4.31

.10 .10 .11 .09 .04

5.27 6.46 8.11 7.31 5.35

3.69 4.36 3.42 .02 .06

4.15 3.99 4.12 4.11 4.30

.00 .00 .00 .00 .00

7.84 8.35 7.54 4.13 4.36

1905 1906 1907 1908 1909

1.87 3.80 2.52 3.19 2.74

5.36 5.74 5.74 7.20 10.84

.03 .02 .01 .01 .00

7.25 9.56 8.27 10.40 13.58

.04 .00 .03 .10 .00

6.07 7.49 6.35 5.55 6.01

.00 .00 .00 .00 .00

6.11 7.49 6.38 5.65 6.01

1910 1911 1912 1913

2.41 2.98 3.42 2.16

13.63 14.85 16.34 17.85

.01 .01 .00 .00

16.05 17.84 19.76 20.01

.12 .51 .67 .72

6.32 6.53 6.47 6.51

.00 .00 .00 .00

6.44 7.04 7.14 7.23

(41)

(42) (43) Campania

(46) Apulia

(47)

(48)

(44)

(45)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

6.97 5.14 4.75 3.55

3.36 3.63 3.67 3.00

.00 .00 .00 .00

10.33 8.77 8.42 6.55

9.26 16.22 23.33 28.04

.95 .98 .96 1.00

.00 .00 .00 .00

10.21 17.20 24.29 29.04

1865 1866 1867 1868 1869

6.61 9.46 10.21 6.84 1.35

3.65 4.94 4.37 3.86 4.31

.00 .00 .00 .00 .00

10.26 14.40 14.58 10.70 5.66

21.49 13.33 12.92 9.72 2.73

1.45 1.70 2.13 3.04 3.31

.01 .01 .00 .00 .00

22.95 15.03 15.05 12.76 6.04

62

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX A. (Continued ) (41)

(42) (43) Campania

(44)

(45)

(46) Apulia

(47)

(48)

Railways Other civil Military Total Railways Other civil Military Total 1870 1871 1872 1873 1874

.58 1.02 1.70 2.63 2.86

6.07 6.21 6.51 5.99 5.98

.00 .00 .00 .00 .00

6.65 7.23 8.21 8.62 8.84

.97 1.19 .93 .05 .10

4.07 2.87 3.91 4.48 3.77

.00 .00 .00 .00 .00

5.04 4.06 4.84 4.53 3.87

1875 1876 1877 1878 1879

2.25 1.72 2.12 3.52 4.34

6.40 8.87 12.25 10.78 8.76

.00 .00 .00 .00 .02

8.65 10.59 14.37 14.30 13.11

.16 .21 .18 .16 .12

2.95 3.10 3.70 3.96 3.98

.00 .00 .00 .00 .00

3.11 3.31 3.88 4.12 4.10

1880 1881 1882 1883 1884

4.67 7.60 9.32 9.16 9.67

8.72 7.17 7.12 9.08 12.46

.08 .07 .04 .07 .07

13.47 14.85 16.47 18.31 22.20

.71 3.06 7.27 9.42 11.43

3.64 2.68 2.80 2.99 3.37

.00 .01 .04 .24 .77

4.35 5.75 10.10 12.66 15.57

1885 1886 1887 1888 1889

11.32 9.24 6.69 7.70 8.10

11.04 9.01 9.13 9.36 9.27

.26 .36 .34 .41 .47

22.62 18.61 16.15 17.47 17.84

10.30 6.21 1.99 3.15 4.92

3.20 3.05 3.23 3.78 3.96

1.11 1.03 1.38 2.02 2.17

14.60 10.29 6.60 8.95 11.05

1890 1891 1892 1893 1894

9.33 12.18 12.47 9.85 12.60

8.75 7.94 7.38 7.12 6.50

.44 .32 .20 .07 .07

18.52 20.45 20.04 17.05 19.17

6.19 8.26 7.50 3.93 7.11

3.71 3.25 2.69 2.27 1.86

1.85 1.23 .83 .93 1.05

11.75 12.74 11.02 7.12 10.02

1895 1896 1897 1898 1899

5.18 .15 .19 1.00 1.98

6.27 5.73 5.35 5.24 5.31

.08 .05 .08 .09 .06

11.53 5.93 5.62 6.32 7.35

4.02 .07 .18 .57 .51

1.73 1.71 1.71 1.87 2.10

.63 .23 .16 .15 .16

6.38 2.00 2.05 2.60 2.76

1900 1901 1902 1903 1904

2.32 3.84 3.54 1.39 1.60

5.99 6.79 8.03 9.36 8.89

.05 .03 .03 .02 .01

8.37 10.66 11.60 10.78 10.50

.55 .43 .68 .61 .59

2.13 2.23 2.65 2.86 2.79

.16 .12 .11 .10 .05

2.84 2.78 3.44 3.57 3.43

63

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX A. (Continued ) (41)

(42) (43) Campania

(44)

(45)

(46) Apulia

(47)

(48)

Railways Other civil Military Total Railways Other civil Military Total 1905 1906 1907 1908 1909

.82 1.37 1.41 2.09 2.13

8.85 11.27 12.85 13.96 15.64

.01 .02 .03 .05 .16

9.68 12.66 14.29 16.10 17.94

.72 1.05 1.00 .92 1.45

2.91 3.57 3.95 4.54 5.15

.06 .10 .20 .34 .71

3.69 4.72 5.15 5.80 7.31

1910 1911 1912 1913

1.92 2.74 3.27 3.24

16.67 17.40 17.34 15.25

.21 .13 .09 .07

18.79 20.28 20.70 18.56

1.69 1.32 3.03 5.40

6.19 6.92 13.12 15.37

1.09 1.11 .85 .49

8.97 9.35 17.00 21.26

(49)

(50) (51) Basilicata

(52)

(53)

(54) (55) Calabria

(56)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

.00 .00 .00 .00

.47 .46 .48 .31

.00 .00 .00 .00

.47 .46 .48 .31

.00 .31 .62 2.00

.38 .38 .43 .48

.00 .00 .00 .00

.38 .69 1.05 2.48

1865 1866 1867 1868 1869

.72 1.47 2.02 2.53 1.84

.56 .55 .68 .68 .82

.00 .00 .00 .00 .00

1.28 2.02 2.70 3.21 2.66

3.99 6.78 9.04 10.36 8.45

1.04 .95 1.18 1.72 1.88

.00 .00 .00 .00 .00

5.03 7.73 10.22 12.08 10.33

1870 1871 1872 1873 1874

.00 .65 1.73 3.21 4.56

1.06 1.07 1.81 2.31 2.10

.00 .00 .00 .00 .00

1.06 1.72 3.54 5.52 6.66

7.74 6.87 7.84 10.87 12.77

3.27 3.58 3.10 2.66 3.00

.00 .00 .00 .00 .00

11.01 10.45 10.94 13.53 15.77

1875 1876 1877 1878 1879

4.72 3.68 3.20 3.44 4.82

2.07 2.13 2.72 2.93 3.09

.00 .00 .00 .00 .00

6.79 5.81 5.92 6.37 7.91

9.40 3.95 2.50 .72 .86

3.62 3.95 4.76 4.98 5.00

.00 .00 .00 .00 .00

13.02 7.90 7.26 5.70 5.86

64

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX A. (Continued ) (49)

(50) (51) Basilicata

(52)

(53)

(54) (55) Calabria

(56)

Railways Other civil Military Total Railways Other civil Military Total 1880 1881 1882 1883 1884

3.55 .00 .00 .00 .00

3.35 2.72 2.80 2.25 2.48

.00 .00 .00 .00 .00

6.90 2.72 2.80 2.25 2.48

.82 1.30 1.83 2.34 2.16

6.32 5.56 4.84 4.36 4.43

.00 .01 .04 .12 .14

7.14 6.88 6.71 6.82 6.73

1885 1886 1887 1888 1889

.00 .00 .35 1.98 3.20

2.36 2.27 2.42 2.44 2.48

.00 .00 .00 .00 .00

2.36 2.27 2.77 4.42 5.68

1.88 1.60 1.91 2.25 1.89

3.88 3.41 3.53 3.77 3.88

.19 .22 .33 .43 .57

5.95 5.23 5.76 6.46 6.34

1890 1891 1892 1893 1894

4.46 5.36 4.03 2.05 3.69

2.45 2.24 2.04 1.76 1.43

.00 .00 .00 .00 .00

6.91 7.60 6.07 3.81 5.12

3.61 7.51 11.18 15.61 21.34

3.84 3.70 3.50 3.27 3.22

.59 .45 .27 .08 .05

8.03 11.67 14.95 18.96 24.61

1895 1896 1897 1898 1899

2.65 3.39 2.66 .00 .00

1.32 1.16 1.04 1.09 1.22

.00 .00 .00 .00 .00

3.97 4.55 3.70 1.09 1.22

8.22 .69 .30 .40 .36

3.51 3.47 3.38 3.97 4.35

.04 .03 .03 .03 .03

11.78 4.19 3.71 4.40 4.74

1900 1901 1902 1903 1904

.00 .00 .00 .00 .00

1.12 .96 .93 .86 .93

.00 .00 .00 .00 .00

1.12 .96 .93 .86 .93

.03 .05 .07 .08 .09

3.99 3.86 3.89 4.01 4.19

.04 .03 .03 .02 .01

4.06 3.94 3.99 4.11 4.30

1905 1906 1907 1908 1909

.00 .00 .00 .00 .00

1.17 2.14 2.68 3.85 4.17

.00 .00 .00 .00 .00

1.17 2.14 2.68 3.85 4.17

.08 .05 .06 .06 .15

4.90 6.99 7.92 6.68 7.47

.01 .02 .03 .04 .12

4.99 7.06 8.00 6.78 7.74

1910 1911 1912 1913

.00 .00 .00 .04

5.72 7.21 5.77 2.63

.00 .00 .00 .00

5.72 7.21 5.77 2.67

.45 1.21 2.07 4.16

8.75 8.43 8.05 8.84

.16 .13 .09 .07

9.36 9.77 10.21 13.07

65

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX A. (Continued ) (57)

(58) Sicily

(59)

(60)

(61)

(62) (63) Sardinia

(64)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

1.40 2.89 4.85 5.59

2.44 2.17 2.59 2.90

.00 .00 .00 .00

3.84 5.06 7.44 8.49

.00 .00 .00 .00

1.27 1.35 2.04 2.18

.00 .00 .00 .00

1.27 1.35 2.04 2.18

1865 1866 1867 1868 1869

7.36 10.15 9.68 10.07 10.89

3.88 4.38 6.05 6.75 6.40

.00 .00 .00 .00 .00

11.24 14.53 15.73 16.82 17.29

.00 .00 .93 3.59 5.85

1.76 1.85 1.93 1.92 1.77

.00 .00 .00 .00 .00

1.76 1.85 2.86 5.51 7.62

1870 1871 1872 1873 1874

8.68 4.71 4.24 7.01 8.04

7.47 6.27 6.37 5.81 6.02

.00 .00 .00 .00 .00

16.15 10.98 10.61 12.82 14.06

8.94 10.75 7.53 3.24 3.08

2.49 2.21 2.79 3.18 2.55

.00 .00 .00 .00 .00

11.43 12.96 10.32 6.42 5.63

1875 1876 1877 1878 1879

7.08 7.89 8.63 13.79 19.11

6.25 7.28 9.11 8.85 8.84

.00 .00 .00 .00 .02

13.33 15.17 17.74 22.64 27.97

1.20 3.93 6.56 8.75 10.74

2.18 2.40 3.02 2.68 2.23

.00 .00 .00 .00 .00

3.38 6.33 9.58 11.43 12.97

1880 1881 1882 1883 1884

17.78 7.73 1.64 3.08 4.26

9.57 8.33 8.48 8.42 9.02

.08 .10 .12 .31 .35

27.43 16.16 10.23 11.81 13.63

9.04 2.49 1.79 1.45 .00

2.16 2.08 1.68 1.56 1.58

.00 .00 .00 .00 .00

11.20 4.57 3.47 3.01 1.58

1885 1886 1887 1888 1889

6.41 7.11 8.66 12.16 15.87

8.59 8.25 8.62 8.39 7.23

.47 .55 .81 1.08 1.43

15.47 15.90 18.10 21.63 24.52

.00 .30 1.37 1.68 .67

1.59 1.95 2.09 1.94 2.08

.00 .00 .58 1.51 2.02

1.59 2.25 4.05 5.13 4.77

1890 1891 1892 1893 1894

17.40 18.46 15.11 11.59 8.04

7.97 8.64 7.20 7.28 7.79

1.47 1.13 .67 .20 .13

26.83 28.23 22.98 19.07 15.96

.15 .39 1.24 1.41 .30

2.05 1.55 1.15 .97 .98

1.62 .71 .31 .22 .16

3.81 2.65 2.70 2.60 1.44

66

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX A. (Continued ) (57)

(58) Sicily

(59)

(60)

(61)

(62) (63) Sardinia

(64)

Railways Other civil Military Total Railways Other civil Military Total 1895 1896 1897 1898 1899

3.51 .13 .14 .47 .96

6.56 6.90 8.48 8.10 7.82

.13 .10 .08 .08 .09

10.19 7.12 8.70 8.64 8.87

.00 .02 .04 .04 .03

.91 1.12 1.39 1.24 1.28

.12 .07 .03 .01 .00

1.03 1.20 1.46 1.29 1.31

1900 1901 1902 1903 1904

.95 .60 .78 .56 .53

8.26 8.30 8.48 8.76 8.61

.12 .09 .08 .07 .04

9.33 8.99 9.34 9.39 9.18

.04 .04 .05 .03 .04

1.43 1.32 1.36 1.60 1.75

.00 .01 .04 .06 .04

1.47 1.38 1.46 1.69 1.83

1905 1906 1907 1908 1909

.93 .60 .74 .79 .85

8.96 9.74 10.67 12.47 18.88

.03 .02 .03 .05 .13

9.92 10.36 11.44 13.31 19.86

.04 .06 .07 .07 .06

1.81 1.97 2.08 2.21 2.83

.01 .00 .00 .00 .00

1.87 2.04 2.15 2.28 2.89

1910 1911 1912 1913

1.67 3.06 2.02 1.89

22.12 19.41 17.22 16.41

.17 .13 .09 .07

23.96 22.60 19.34 18.37

.08 .09 .18 .61

3.05 3.10 3.50 3.54

.00 .00 .00 .00

3.13 3.19 3.68 4.15

APPENDIX B. SOCIAL-OVERHEAD MAINTENANCE: VALUE ADDED AT 1911 PRICES (MILLION LIRE) (1)

(2) (3) Piedmont

(4)

(5)

(6) Liguria

(7)

(8)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

.51 .51 .49 .56

4.05 4.40 4.67 4.58

.28 .31 .32 .34

4.85 5.22 5.47 5.48

.08 .08 .09 .12

1.80 1.95 1.84 1.57

.28 .31 .32 .34

2.16 2.34 2.25 2.03

67

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX B. (Continued ) (1)

(2) (3) Piedmont

(4)

(5)

(6) Liguria

(7)

(8)

Railways Other civil Military Total Railways Other civil Military Total 1865 1866 1867 1868 1869

.62 .64 .63 .68 .73

4.71 4.87 4.84 4.72 4.87

.27 .23 .19 .21 .21

5.61 5.74 5.65 5.61 5.81

.12 .12 .12 .23 .25

1.50 1.37 1.44 1.31 1.33

.27 .23 .19 .21 .21

1.89 1.72 1.75 1.75 1.79

1870 1871 1872 1873 1874

.83 .94 1.00 1.08 1.14

4.80 4.72 4.69 4.75 4.74

.19 .23 .20 .24 .22

5.82 5.90 5.89 6.07 6.11

.27 .29 .55 .60 .76

1.25 1.35 1.53 1.45 1.56

.19 .23 .20 .24 .22

1.72 1.87 2.28 2.29 2.55

1875 1876 1877 1878 1879

1.17 1.25 1.26 1.26 1.31

4.48 4.69 4.97 5.19 5.00

.24 .25 .26 .26 .30

5.90 6.18 6.49 6.71 6.61

.77 .80 .81 .81 .83

1.48 1.44 1.69 1.65 1.94

.24 .25 .26 .26 .30

2.49 2.49 2.76 2.71 3.07

1880 1881 1882 1883 1884

1.43 1.56 1.71 1.87 2.03

5.23 5.32 5.19 5.24 5.26

.32 .31 .34 .35 .39

6.98 7.19 7.24 7.46 7.69

.89 .93 .94 .96 .98

1.71 1.63 1.59 1.65 1.84

.32 .31 .34 .35 .39

2.92 2.87 2.87 2.96 3.21

1885 1886 1887 1888 1889

2.10 2.17 2.39 2.50 2.57

5.38 5.51 5.68 5.82 5.92

.43 .45 .47 .49 .49

7.90 8.13 8.54 8.81 8.98

.95 .94 .95 1.00 1.07

1.82 1.94 2.05 2.07 2.30

.43 .45 .47 .49 .49

3.20 3.33 3.47 3.57 3.85

1890 1891 1892 1893 1894

2.62 2.64 2.64 2.77 2.75

5.90 5.97 6.04 6.17 6.07

.47 .48 .48 .50 .52

8.99 9.08 9.15 9.43 9.34

1.07 1.04 1.01 1.02 1.09

2.49 2.51 2.51 2.44 2.47

.47 .48 .48 .50 .52

4.03 4.03 3.99 3.96 4.07

1895 1896 1897 1898 1899

2.73 2.81 2.86 2.90 3.05

6.11 6.25 6.35 6.43 6.42

.52 .54 .57 .57 .55

9.36 9.60 9.78 9.91 10.02

1.07 1.12 1.18 1.20 1.25

2.60 2.51 2.46 2.44 2.47

.52 .54 .57 .57 .55

4.19 4.17 4.20 4.21 4.27

68

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX B. (Continued ) (1)

(2) (3) Piedmont

(4)

(5)

(6) Liguria

(7)

(8)

Railways Other civil Military Total Railways Other civil Military Total 1900 1901 1902 1903 1904

3.12 3.19 3.28 3.37 3.48

6.19 6.24 6.37 6.39 6.53

.53 .55 .55 .54 .57

9.84 9.98 10.20 10.31 10.57

1.31 1.35 1.40 1.45 1.52

2.44 2.43 2.51 2.61 2.71

.53 .55 .55 .54 .57

4.28 4.33 4.46 4.60 4.80

1905 1906 1907 1908 1909

3.70 3.97 4.03 4.27 4.37

6.60 6.33 6.32 6.32 6.45

.58 .54 .56 .59 .61

10.88 10.85 10.91 11.17 11.43

1.53 1.63 1.65 1.74 1.81

2.79 2.72 2.72 2.65 2.67

.58 .54 .56 .59 .61

4.90 4.89 4.93 4.98 5.09

1910 1911 1912 1913

4.54 4.62 4.86 5.00

6.65 7.11 7.22 7.34

.66 .74 .75 .77

11.84 12.46 12.83 13.11

1.90 1.94 1.99 2.05

2.72 2.85 2.82 2.92

.66 .74 .75 .77

5.28 5.52 5.56 5.74

(10) (11) Lombardy

(12)

(13)

(14) (15) Venetia

(16)

(9)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

.48 .56 .66 .70

6.01 6.52 6.58 6.50

.05 .06 .06 .06

6.54 7.14 7.30 7.26

.27 .27 .26 .27

6.27 6.60 7.08 6.78

2.74 2.72 2.70 2.64

9.28 9.59 10.03 9.68

1865 1866 1867 1868 1869

.76 .90 .92 .96 1.03

6.60 5.98 5.67 6.03 6.30

.05 .04 .04 .04 .04

7.41 6.93 6.62 7.03 7.37

.28 .34 .34 .35 .38

6.75 6.31 10.79 8.26 7.17

2.60 1.27 .19 .21 .21

9.62 7.92 11.32 8.81 7.75

1870 1871 1872 1873 1874

1.16 1.25 1.32 1.50 1.55

6.10 6.33 6.68 6.44 7.02

.04 .04 .04 .04 .04

7.30 7.62 8.04 7.99 8.61

.39 .42 .45 .48 .46

6.69 6.93 7.73 7.35 8.15

.19 .23 .20 .24 .22

7.28 7.59 8.38 8.07 8.83

69

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX B. (Continued ) (9)

(10) (11) Lombardy

(12)

(13)

(14) (15) Venetia

(16)

Railways Other civil Military Total Railways Other civil Military Total 1875 1876 1877 1878 1879

1.57 1.71 1.72 1.78 1.98

6.64 6.64 7.09 7.25 7.31

.04 .04 .05 .04 .05

8.26 8.40 8.85 9.07 9.34

.50 .63 .83 .84 .89

7.46 7.90 8.19 8.51 8.56

.24 .25 .26 .26 .30

8.20 8.78 9.28 9.61 9.75

1880 1881 1882 1883 1884

2.17 2.32 2.64 2.73 2.98

7.81 7.78 7.66 8.28 7.22

.05 .05 .05 .05 .06

10.03 10.15 10.35 11.07 10.26

.97 1.02 1.04 1.08 1.15

8.93 8.38 11.69 10.41 8.90

.32 .31 .34 .35 .39

10.22 9.72 13.07 11.84 10.45

1885 1886 1887 1888 1889

3.18 3.30 3.55 3.89 3.98

7.04 7.70 7.79 7.72 8.27

.06 .06 .06 .07 .06

10.28 11.06 11.40 11.67 12.31

1.29 1.51 1.58 1.72 1.78

8.58 9.37 9.32 9.35 9.79

.43 .45 .47 .49 .49

10.30 11.33 11.38 11.56 12.06

1890 1891 1892 1893 1894

4.04 4.00 3.99 4.15 4.21

8.24 8.06 8.28 8.44 8.05

.06 .06 .06 .06 .06

12.34 12.12 12.32 12.64 12.32

1.81 1.77 1.74 1.78 1.76

9.46 9.29 9.52 9.63 9.16

.47 .48 .48 .50 .52

11.74 11.54 11.74 11.91 11.44

1895 1896 1897 1898 1899

4.17 4.31 4.46 4.58 4.73

8.22 8.78 8.93 9.37 9.39

.06 .06 .06 .06 .05

12.44 13.15 13.44 14.00 14.17

1.73 1.79 1.84 1.86 1.95

9.26 9.63 9.69 9.96 9.67

.52 .54 .57 .57 .55

11.52 11.97 12.09 12.39 12.17

1900 1901 1902 1903 1904

4.80 4.90 5.13 5.28 5.55

8.99 9.21 9.62 9.76 10.01

.05 .05 .05 .04 .04

13.84 14.16 14.80 15.09 15.60

1.99 2.02 2.08 2.15 2.25

8.98 8.96 9.36 9.36 9.40

.53 .55 .55 .54 .57

11.50 11.53 11.99 12.05 12.22

1905 1906 1907 1908 1909

5.65 6.01 6.15 6.50 6.79

10.29 10.04 10.11 10.10 10.29

.04 .04 .04 .04 .04

15.98 16.09 16.30 16.64 17.12

2.27 2.39 2.43 2.67 2.79

9.51 9.06 9.00 8.94 9.16

.58 .54 .56 .59 .61

12.36 11.99 12.00 12.20 12.57

70

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX B. (Continued ) (9)

(10) (11) Lombardy

(12)

(13)

(14) (15) Venetia

(16)

Railways Other civil Military Total Railways Other civil Military Total 1910 1911 1912 1913

7.08 7.44 7.64 7.82

(17)

10.60 11.25 11.36 11.42

(18) Emilia

.04 .04 .04 .04

17.72 18.73 19.04 19.28

3.03 3.16 3.36 3.57

9.52 10.09 10.21 10.17

.66 .74 .75 .77

(19)

(20)

(21)

(22) (23) Tuscany

13.21 13.99 14.31 14.50

(24)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

.35 .44 .52 .54

5.88 6.31 6.19 6.19

.05 .06 .06 .06

6.29 6.81 6.77 6.79

.35 .39 .55 .69

4.36 4.78 5.02 4.67

.18 .19 .20 .21

4.89 5.36 5.77 5.57

1865 1866 1867 1868 1869

.56 .56 .55 .58 .62

6.17 6.62 6.43 6.52 6.64

.05 .04 .04 .04 .04

6.78 7.22 7.01 7.14 7.29

.73 .80 .80 .84 .90

4.54 4.90 4.66 4.93 5.18

.17 .14 .12 .13 .13

5.44 5.84 5.57 5.89 6.20

1870 1871 1872 1873 1874

.65 .70 .79 .85 .82

6.35 6.50 6.56 6.62 7.25

.04 .04 .04 .04 .04

7.03 7.24 7.39 7.52 8.11

.94 1.03 1.15 1.25 1.21

5.30 5.27 5.20 5.01 5.24

.12 .15 .13 .15 .14

6.36 6.45 6.48 6.41 6.59

1875 1876 1877 1878 1879

.83 .86 .86 .86 .88

6.77 6.77 7.10 7.06 6.59

.04 .04 .05 .04 .05

7.64 7.67 8.00 7.97 7.52

1.23 1.30 1.31 1.31 1.34

4.86 4.75 4.85 4.85 4.58

.15 .15 .15 .15 .17

6.24 6.20 6.31 6.31 6.09

1880 1881 1882 1883 1884

.95 .99 1.02 1.36 1.54

6.98 6.46 7.27 6.95 6.87

.05 .05 .05 .05 .06

7.98 7.50 8.34 8.37 8.46

1.45 1.51 1.53 1.55 1.59

4.65 4.49 4.33 4.43 4.24

.17 .17 .18 .18 .20

6.28 6.17 6.03 6.16 6.03

71

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX B. (Continued ) (17)

(18) (19) Emilia

(20)

(21)

(22) (23) Tuscany

(24)

Railways Other civil Military Total Railways Other civil Military Total 1885 1886 1887 1888 1889

1.50 1.57 1.88 2.09 2.33

6.89 7.29 7.45 7.77 8.13

.06 .06 .06 .07 .06

8.45 8.92 9.40 9.92 10.53

1.57 1.63 1.65 1.85 1.86

4.28 4.63 4.74 4.87 5.06

.21 .22 .22 .22 .22

6.06 6.47 6.61 6.94 7.13

1890 1891 1892 1893 1894

2.36 2.34 2.32 2.44 2.51

8.31 8.46 8.12 7.96 7.88

.06 .06 .06 .06 .06

10.73 10.86 10.49 10.46 10.44

1.96 1.93 1.93 2.03 2.02

5.13 5.14 5.16 5.32 5.23

.20 .20 .19 .20 .20

7.29 7.27 7.28 7.55 7.45

1895 1896 1897 1898 1899

2.52 2.60 2.66 2.70 2.79

7.90 8.24 8.82 8.54 8.33

.06 .06 .06 .06 .05

10.47 10.89 11.53 11.29 11.17

1.98 2.07 2.12 2.19 2.29

5.32 5.39 5.40 5.40 5.32

.20 .20 .20 .20 .18

7.49 7.66 7.72 7.78 7.79

1900 1901 1902 1903 1904

2.84 2.92 3.05 3.18 3.30

7.88 8.01 8.40 8.47 8.60

.05 .05 .05 .04 .04

10.77 10.97 11.49 11.69 11.94

2.34 2.39 2.52 2.59 2.70

5.08 5.11 5.26 5.31 5.41

.17 .17 .17 .16 .16

7.59 7.67 7.95 8.06 8.27

1905 1906 1907 1908 1909

3.34 3.59 3.59 3.80 4.03

8.78 8.51 8.56 8.61 8.93

.04 .04 .04 .04 .04

12.17 12.14 12.19 12.44 13.00

2.72 2.87 2.88 3.06 3.16

5.49 5.30 5.30 5.28 5.42

.16 .14 .14 .15 .15

8.37 8.31 8.32 8.48 8.72

1910 1911 1912 1913

4.24 4.48 4.59 4.79

9.41 10.10 10.40 10.43

.04 .04 .04 .04

13.69 14.62 15.03 15.25

3.37 3.51 3.63 3.85

5.66 6.03 6.12 6.17

.15 .16 .16 .16

9.18 9.70 9.91 10.18

72

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX B. (Continued ) (25)

(26) (27) Marches

(28)

(29)

(30) (31) Umbria

(32)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

.06 .06 .14 .15

1.20 1.31 1.34 1.38

.05 .06 .06 .06

1.31 1.43 1.54 1.60

.00 .01 .01 .01

.62 .67 .68 .71

.04 .04 .04 .04

.65 .72 .73 .76

1865 1866 1867 1868 1869

.16 .22 .22 .23 .24

1.36 1.47 1.44 1.40 1.48

.05 .04 .04 .04 .04

1.57 1.73 1.69 1.67 1.76

.04 .23 .23 .24 .25

.76 .81 .79 .80 .86

.03 .03 .02 .03 .03

.83 1.07 1.04 1.07 1.13

1870 1871 1872 1873 1874

.25 .27 .29 .33 .31

1.41 1.37 1.43 1.37 1.46

.04 .04 .04 .04 .04

1.69 1.68 1.76 1.74 1.82

.26 .28 .30 .33 .36

.89 .93 .93 .90 .94

.02 .03 .03 .03 .03

1.18 1.24 1.26 1.25 1.33

1875 1876 1877 1878 1879

.32 .33 .33 .33 .34

1.52 1.50 1.53 1.58 1.50

.05 .05 .05 .05 .05

1.88 1.87 1.90 1.95 1.90

.40 .41 .41 .41 .42

.93 .92 .94 .96 .88

.03 .03 .03 .03 .03

1.36 1.36 1.38 1.40 1.33

1880 1881 1882 1883 1884

.37 .38 .38 .39 .45

1.48 1.61 1.61 1.58 1.60

.06 .06 .06 .06 .07

1.91 2.05 2.05 2.04 2.12

.45 .47 .48 .60 .62

.89 .86 .84 .86 .86

.04 .03 .04 .04 .04

1.38 1.37 1.36 1.50 1.52

1885 1886 1887 1888 1889

.45 .51 .51 .57 .57

1.63 1.66 1.66 1.69 1.77

.08 .08 .08 .09 .08

2.15 2.25 2.25 2.35 2.42

.60 .71 .72 .74 .74

.90 .95 .99 1.03 1.04

.04 .05 .05 .05 .05

1.55 1.71 1.76 1.81 1.83

1890 1891 1892 1893 1894

.57 .56 .54 .55 .54

1.76 1.82 1.90 1.97 1.94

.08 .08 .08 .08 .09

2.41 2.46 2.52 2.61 2.57

.74 .72 .70 .71 .70

1.04 1.07 1.09 1.11 1.10

.04 .04 .04 .04 .04

1.82 1.83 1.83 1.87 1.84

73

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX B. (Continued ) (25)

(26) (27) Marches

(28)

(29)

(30) (31) Umbria

(32)

Railways Other civil Military Total Railways Other civil Military Total 1895 1896 1897 1898 1899

.58 .59 .61 .69 .72

1.90 1.95 2.04 2.02 2.03

.09 .09 .09 .09 .09

2.56 2.63 2.74 2.81 2.84

.69 .71 .73 .74 .77

1.09 1.12 1.13 1.13 1.11

.04 .04 .04 .04 .04

1.82 1.87 1.91 1.91 1.92

1900 1901 1902 1903 1904

.73 .75 .77 .80 .83

1.98 1.95 1.99 2.04 2.09

.09 .09 .09 .09 .09

2.80 2.79 2.84 2.92 3.01

.79 .81 .84 .86 .90

1.07 1.07 1.08 1.09 1.11

.04 .04 .04 .04 .04

1.90 1.92 1.96 1.99 2.04

1905 1906 1907 1908 1909

.84 .92 .92 1.09 1.13

2.13 2.07 2.06 2.03 2.10

.09 .08 .09 .09 .09

3.06 3.07 3.07 3.21 3.33

.91 .96 .96 1.02 1.05

1.12 1.08 1.08 1.08 1.12

.04 .03 .03 .03 .04

2.07 2.07 2.07 2.14 2.20

1910 1911 1912 1913

1.18 1.20 1.23 1.26

2.22 2.36 2.39 2.45

.10 .11 .11 .12

3.50 3.67 3.73 3.83

1.10 1.12 1.14 1.17

1.16 1.23 1.26 1.27

.04 .04 .04 .04

2.29 2.39 2.44 2.48

(33)

(34) Latium

(35)

(36)

(37)

(38) (39) Abruzzi

(40)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

.08 .16 .16 .17

1.63 1.72 1.84 1.81

.74 .73 .73 .71

2.45 2.61 2.73 2.69

.00 .00 .05 .12

1.24 1.35 1.41 1.38

.04 .04 .04 .04

1.28 1.39 1.50 1.55

1865 1866 1867 1868 1869

.24 .24 .27 .28 .31

1.86 1.79 1.80 1.78 1.81

.70 .69 .70 .70 .69

2.80 2.72 2.77 2.76 2.82

.12 .12 .12 .13 .14

1.36 1.36 1.33 1.37 1.38

.03 .03 .02 .03 .03

1.52 1.51 1.47 1.53 1.55

74

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX B. (Continued ) (33)

(34) (35) Latium

(36)

(37)

(38) (39) Abruzzi

(40)

Railways Other civil Military Total Railways Other civil Military Total 1870 1871 1872 1873 1874

.32 .34 .36 .39 .38

1.87 1.95 2.11 2.21 2.47

.68 .09 .08 .09 .08

2.87 2.37 2.55 2.69 2.94

.14 .16 .16 .25 .24

1.40 1.39 1.38 1.42 1.52

.02 .03 .03 .03 .03

1.56 1.58 1.57 1.70 1.79

1875 1876 1877 1878 1879

.39 .41 .41 .42 .44

2.56 2.67 2.71 2.75 2.58

.09 .09 .10 .09 .11

3.04 3.17 3.22 3.26 3.13

.31 .32 .32 .32 .33

1.61 1.56 1.62 1.57 1.51

.03 .03 .03 .03 .03

1.95 1.91 1.97 1.92 1.88

1880 1881 1882 1883 1884

.48 .50 .51 .51 .61

2.83 2.79 2.62 2.73 2.54

.12 .11 .12 .13 .14

3.43 3.40 3.25 3.36 3.29

.36 .37 .48 .57 .62

1.64 1.58 1.62 1.65 1.58

.04 .03 .04 .04 .04

2.04 1.98 2.13 2.26 2.24

1885 1886 1887 1888 1889

.60 .64 .70 .74 .78

2.62 2.91 3.17 3.39 3.65

.15 .16 .16 .17 .17

3.37 3.70 4.03 4.30 4.59

.60 .61 .62 .78 .78

1.59 1.66 1.76 1.77 1.78

.04 .05 .05 .05 .05

2.24 2.31 2.43 2.59 2.61

1890 1891 1892 1893 1894

.79 .80 .93 .94 1.05

3.85 4.01 4.01 4.13 4.05

.16 .16 .16 .17 .17

4.80 4.97 5.10 5.23 5.27

.78 .76 .77 .78 .81

1.77 1.87 1.90 1.83 1.80

.04 .04 .04 .04 .04

2.59 2.68 2.71 2.65 2.65

1895 1896 1897 1898 1899

1.04 1.08 1.10 1.12 1.17

3.96 3.83 3.75 3.76 3.62

.17 .18 .19 .19 .18

5.18 5.09 5.04 5.06 4.97

.80 .83 .99 1.01 1.05

1.84 1.88 1.92 1.91 1.86

.04 .04 .04 .04 .04

2.68 2.75 2.96 2.96 2.95

75

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX B. (Continued ) (33)

(34) (35) Latium

(36)

(37)

(38) (39) Abruzzi

(40)

Railways Other civil Military Total Railways Other civil Military Total 1900 1901 1902 1903 1904

1.20 1.26 1.30 1.33 1.39

3.49 3.53 3.66 3.79 3.91

.17 .18 .17 .17 .18

4.86 4.96 5.13 5.29 5.48

1.09 1.11 1.21 1.24 1.29

1.80 1.80 1.83 1.83 1.86

.04 .04 .04 .04 .04

2.93 2.95 3.07 3.11 3.19

1905 1906 1907 1908 1909

1.40 1.50 1.54 1.64 1.68

4.01 3.93 3.99 4.04 4.21

.18 .17 .17 .18 .19

5.59 5.60 5.71 5.86 6.08

1.32 1.40 1.40 1.48 1.53

1.88 1.83 1.84 1.87 2.04

.04 .03 .03 .03 .04

3.23 3.26 3.27 3.39 3.60

1910 1911 1912 1913

1.76 1.79 1.93 2.00

4.43 4.76 4.88 4.98

.20 .22 .23 .23

6.39 6.77 7.04 7.21

1.59 1.61 1.77 1.89

2.18 2.32 2.34 2.37

.04 .04 .04 .04

3.81 3.97 4.15 4.30

(41)

(42) (43) Campania

(44)

(45)

(46) Apulia

(47)

(48)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

.12 .12 .16 .17

2.36 2.58 2.54 2.62

.16 .17 .18 .19

2.64 2.87 2.88 2.98

.00 .00 .00 .08

1.42 1.54 1.61 1.63

.04 .04 .04 .04

1.46 1.58 1.65 1.76

1865 1866 1867 1868 1869

.17 .18 .20 .28 .32

2.85 3.31 3.39 3.49 3.62

.15 .13 .10 .12 .12

3.18 3.62 3.70 3.89 4.06

.19 .21 .23 .31 .35

1.69 2.07 2.15 2.22 2.35

.03 .03 .02 .03 .03

1.91 2.30 2.40 2.56 2.72

1870 1871 1872 1873 1874

.34 .36 .38 .41 .41

3.77 3.71 3.64 3.59 3.63

.11 .13 .11 .13 .13

4.22 4.20 4.13 4.13 4.17

.37 .40 .43 .47 .45

2.36 2.66 2.33 2.35 2.55

.02 .03 .03 .03 .03

2.75 3.09 2.79 2.84 3.03

76

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX B. (Continued ) (41)

(42) (43) Campania

(44)

(45)

(46) Apulia

(47)

(48)

Railways Other civil Military Total Railways Other civil Military Total 1875 1876 1877 1878 1879

.44 .46 .47 .47 .51

3.58 3.73 3.90 4.06 3.90

.13 .13 .14 .13 .15

4.15 4.32 4.50 4.66 4.56

.45 .47 .47 .47 .48

2.51 2.65 2.57 2.68 2.55

.03 .03 .04 .04 .05

2.99 3.15 3.08 3.19 3.08

1880 1881 1882 1883 1884

.55 .58 .64 .70 .73

4.15 4.07 3.91 3.75 3.87

.16 .16 .17 .17 .19

4.86 4.81 4.72 4.62 4.78

.52 .54 .55 .59 .64

2.61 2.66 2.65 2.63 2.67

.05 .05 .06 .06 .07

3.18 3.26 3.26 3.29 3.38

1885 1886 1887 1888 1889

.78 .84 .88 .93 .96

4.19 4.20 4.25 4.43 4.60

.20 .20 .21 .22 .21

5.17 5.25 5.34 5.58 5.77

.67 .72 .74 .76 .77

2.74 2.88 3.09 3.27 3.32

.08 .09 .10 .11 .11

3.49 3.69 3.92 4.14 4.20

1890 1891 1892 1893 1894

.97 1.00 1.06 1.11 1.14

4.63 4.82 4.83 4.75 4.64

.20 .20 .19 .20 .20

5.80 6.01 6.08 6.06 5.98

.76 .76 .81 .82 .81

3.30 3.44 3.46 3.45 3.42

.11 .11 .11 .12 .13

4.17 4.31 4.39 4.40 4.36

1895 1896 1897 1898 1899

1.21 1.25 1.28 1.30 1.36

4.66 4.80 4.79 4.84 4.88

.20 .20 .21 .20 .19

6.06 6.25 6.27 6.35 6.43

.86 .89 .91 .92 .96

3.57 3.71 3.62 3.59 3.59

.14 .14 .15 .16 .15

4.57 4.74 4.68 4.67 4.70

1900 1901 1902 1903 1904

1.40 1.44 1.52 1.59 1.74

4.70 4.64 4.68 4.73 4.80

.18 .18 .18 .17 .18

6.28 6.26 6.38 6.49 6.72

1.02 1.04 1.08 1.13 1.18

3.46 3.39 3.42 3.47 3.53

.15 .16 .16 .16 .17

4.63 4.59 4.66 4.76 4.88

1905 1906 1907 1908 1909

1.78 1.87 1.88 1.98 2.04

4.81 4.66 4.66 4.68 4.76

.18 .16 .16 .17 .17

6.77 6.69 6.70 6.82 6.97

1.24 1.31 1.41 1.50 1.54

3.55 3.41 3.37 3.30 3.36

.18 .17 .18 .19 .20

4.97 4.89 4.96 4.99 5.10

77

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX B. (Continued ) (41)

(42) (43) Campania

(44)

(45)

(46) Apulia

(47)

(48)

Railways Other civil Military Total Railways Other civil Military Total 1910 1911 1912 1913

2.19 2.25 2.30 2.48

4.94 5.25 5.28 5.40

.18 .20 .19 .19

7.30 7.69 7.78 8.07

1.61 1.75 1.79 1.84

3.50 3.70 3.86 4.19

.22 .25 .26 .27

5.33 5.70 5.91 6.30

(49)

(50) (51) Basilicata

(52)

(53)

(54) (55) Calabria

(56)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

.00 .00 .00 .00

.53 .58 .61 .57

.04 .04 .04 .04

.57 .62 .64 .61

.00 .00 .00 .00

.86 .93 .98 1.00

.04 .04 .04 .04

.89 .97 1.02 1.04

1865 1866 1867 1868 1869

.00 .00 .00 .00 .02

.56 .55 .52 .53 .60

.03 .03 .02 .03 .03

.60 .58 .54 .55 .65

.00 .01 .01 .04 .07

1.01 1.17 1.13 1.15 1.33

.03 .03 .02 .03 .03

1.04 1.20 1.17 1.22 1.43

1870 1871 1872 1873 1874

.02 .03 .03 .03 .03

.62 .63 .65 .64 .67

.02 .03 .03 .03 .03

.66 .69 .70 .70 .72

.11 .15 .17 .19 .22

1.37 1.31 1.29 1.30 1.37

.02 .03 .03 .03 .03

1.50 1.49 1.49 1.52 1.62

1875 1876 1877 1878 1879

.06 .08 .09 .09 .10

.69 .74 .72 .88 .84

.03 .03 .03 .03 .03

.78 .85 .84 1.00 .97

.30 .33 .36 .36 .38

1.36 1.52 1.48 1.52 1.49

.03 .03 .03 .03 .04

1.69 1.88 1.87 1.91 1.90

1880 1881 1882 1883 1884

.17 .17 .18 .18 .18

.79 .89 .86 .88 .98

.04 .03 .04 .04 .04

1.00 1.10 1.08 1.10 1.20

.41 .42 .43 .45 .47

1.50 1.52 1.47 1.56 1.52

.04 .04 .04 .04 .05

1.95 1.98 1.94 2.06 2.04

78

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX B. (Continued ) (49)

(50) (51) Basilicata

(52)

(53)

(54) (55) Calabria

(56)

Railways Other civil Military Total Railways Other civil Military Total 1885 1886 1887 1888 1889

.18 .17 .18 .18 .18

1.06 1.07 1.04 .99 .97

.04 .05 .05 .05 .05

1.29 1.28 1.26 1.22 1.20

.46 .46 .47 .49 .50

1.62 1.67 1.70 1.76 1.77

.05 .05 .05 .06 .06

2.13 2.18 2.23 2.30 2.32

1890 1891 1892 1893 1894

.18 .19 .24 .24 .25

.99 1.03 1.07 1.08 1.06

.04 .04 .04 .04 .04

1.22 1.26 1.35 1.36 1.35

.50 .50 .49 .51 .59

1.73 1.81 1.89 1.93 1.90

.05 .05 .05 .06 .06

2.29 2.37 2.44 2.50 2.55

1895 1896 1897 1898 1899

.25 .25 .30 .31 .32

1.08 1.09 1.10 1.09 1.07

.04 .04 .04 .04 .04

1.37 1.38 1.44 1.44 1.43

.68 .71 .73 .74 .78

1.92 1.96 2.01 2.02 2.03

.06 .06 .06 .06 .06

2.66 2.73 2.80 2.83 2.87

1900 1901 1902 1903 1904

.33 .33 .34 .35 .37

1.03 1.03 1.03 1.01 1.01

.04 .04 .04 .04 .04

1.40 1.39 1.40 1.40 1.42

.79 .81 .84 .86 .90

1.97 1.97 2.00 2.01 2.04

.06 .06 .06 .06 .06

2.81 2.84 2.89 2.92 3.00

1905 1906 1907 1908 1909

.37 .39 .39 .42 .43

1.00 .95 .96 1.00 1.07

.04 .03 .03 .03 .04

1.41 1.37 1.39 1.46 1.54

.91 .96 .96 1.02 1.05

2.05 1.97 1.95 1.91 1.91

.06 .06 .06 .06 .06

3.02 2.98 2.97 2.99 3.03

1910 1911 1912 1913

.45 .45 .46 .48

1.19 1.33 1.39 1.41

.04 .04 .04 .04

1.67 1.82 1.89 1.92

1.09 1.12 1.15 1.18

2.09 2.21 2.22 2.25

.07 .08 .08 .08

3.25 3.41 3.45 3.51

79

Social-Overhead Construction in Italy’s Regions, 1861–1913

APPENDIX B. (Continued ) (57)

(58) Sicily

(59)

(60)

(61)

(62) (63) Sardinia

(64)

Railways Other civil Military Total Railways Other civil Military Total 1861 1862 1863 1864

.00 .00 .01 .01

3.71 4.06 4.33 4.34

.11 .12 .12 .13

3.82 4.18 4.46 4.47

.00 .00 .00 .00

1.19 1.29 1.32 1.39

.04 .04 .04 .04

1.23 1.33 1.36 1.43

1865 1866 1867 1868 1869

.01 .04 .06 .06 .09

4.55 5.15 5.22 5.09 5.17

.10 .09 .07 .08 .08

4.66 5.28 5.35 5.22 5.34

.00 .00 .00 .00 .00

1.43 1.47 1.57 1.51 1.53

.03 .03 .02 .03 .03

1.47 1.50 1.60 1.53 1.55

1870 1871 1872 1873 1874

.13 .18 .19 .20 .22

4.92 4.83 4.75 4.56 4.53

.07 .09 .08 .09 .08

5.12 5.10 5.02 4.84 4.84

.00 .01 .03 .03 .04

1.47 1.65 1.69 1.60 1.81

.02 .03 .03 .03 .03

1.49 1.69 1.75 1.65 1.87

1875 1876 1877 1878 1879

.23 .30 .30 .30 .31

4.50 4.55 4.66 5.12 4.84

.09 .09 .09 .09 .11

4.82 4.94 5.05 5.52 5.25

.04 .04 .04 .05 .05

1.84 1.89 2.27 2.21 1.98

.03 .03 .03 .03 .04

1.91 1.96 2.34 2.29 2.07

1880 1881 1882 1883 1884

.45 .56 .57 .58 .59

4.88 5.29 4.93 4.75 4.75

.11 .11 .12 .12 .13

5.44 5.95 5.62 5.45 5.47

.09 .10 .10 .11 .11

2.05 2.36 2.16 2.10 2.05

.04 .04 .05 .05 .05

2.19 2.50 2.31 2.26 2.21

1885 1886 1887 1888 1889

.58 .64 .67 .69 .72

4.81 5.03 5.25 5.37 5.40

.14 .15 .16 .16 .16

5.54 5.82 6.08 6.22 6.28

.11 .11 .11 .16 .20

2.10 2.16 2.27 2.30 2.22

.06 .06 .06 .07 .07

2.26 2.33 2.44 2.52 2.49

1890 1891 1892 1893 1894

.73 .79 .81 .90 .90

5.47 5.60 5.58 5.64 5.52

.15 .15 .15 .15 .16

6.35 6.54 6.53 6.69 6.58

.20 .20 .19 .24 .26

2.14 2.18 2.17 2.26 2.27

.07 .07 .07 .07 .07

2.40 2.45 2.42 2.57 2.60

80

CARLO CICCARELLI AND STEFANO FENOALTEA

APPENDIX B. (Continued ) (57)

(58) Sicily

(59)

(60)

(61)

(62) (63) Sardinia

(64)

Railways Other civil Military Total Railways Other civil Military Total 1895 1896 1897 1898 1899

1.00 1.03 1.06 1.07 1.11

5.70 5.85 5.86 5.78 5.72

.16 .16 .17 .17 .16

6.86 7.04 7.08 7.02 6.99

.26 .26 .27 .28 .29

2.35 2.51 2.42 2.39 2.44

.07 .08 .08 .08 .08

2.68 2.84 2.77 2.75 2.81

1900 1901 1902 1903 1904

1.16 1.19 1.23 1.30 1.35

5.55 5.50 5.57 5.64 5.75

.15 .16 .15 .15 .16

6.86 6.84 6.95 7.09 7.25

.29 .30 .31 .32 .33

2.39 2.34 2.37 2.44 2.51

.08 .08 .08 .08 .08

2.76 2.71 2.76 2.84 2.92

1905 1906 1907 1908 1909

1.37 1.45 1.45 1.54 1.58

5.78 5.53 5.45 5.46 5.56

.16 .15 .15 .16 .16

7.31 7.13 7.05 7.15 7.30

.33 .35 .35 .37 .39

2.53 2.41 2.36 2.28 2.29

.08 .08 .08 .09 .09

2.95 2.84 2.79 2.74 2.77

1910 1911 1912 1913

1.68 1.75 1.85 1.90

5.76 6.13 6.23 6.36

.17 .19 .19 .19

7.61 8.07 8.26 8.46

.40 .41 .42 .43

2.36 2.46 2.45 2.53

.10 .11 .11 .12

2.86 2.98 2.98 3.07

BILATERAL TRADE FLOWS IN EUROPE, 1857–1875: A NEW DATASET Markus Lampe ABSTRACT This study constructs a comprehensive, internationally comparative set of foreign trade data for the period 1857–1875. The dataset is constructed using information at the commodity group-level and contains import and export values for the UK, France, the Zollverein, the Netherlands, Belgium, Austria-Hungary, and the United States, itemised by trade partner. The study tackles three basic problems related to the heterogeneity in national statistics of the period: different definitions of aggregates, inadequate ‘official’ pricing, and the ‘proximity bias’, i.e. the misleading practice of crediting imports to bordering countries from where they physically entered, but where they did not originate. After passing successfully a consistency test, the resulting dataset contains harmonised and country of origin-corrected bilateral trade values for 7 central importers, 10 points in time, and 21 commodity groups, along with ad valorem tariff rates for all commodity groups and countries. They offer new detailed insights into the composition and evolution of trade and tariffs in the third quarter of the 19th century. Furthermore, a basic implementation of the gravity equation shows that as a consequence of the proximity bias estimates using uncorrected data are to be taken with care, especially when assessing border effects and the impact of policy variables. Research in Economic History, Volume 26, 81–155 Copyright r 2008 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(08)26002-7

81

82

MARKUS LAMPE

1. INTRODUCTION Much has been written about inaccuracy and inconsistency of international trade statistics predating the foundation of international statistical bodies like the UN, the IMF, and the OECD. The most telling criticism was provided by Oskar Morgenstern who qualified foreign trade statistics for the first half of the 20th century as ‘‘exceedingly poor’’ (Morgenstern, 1963, p. 138). He stated that the degree of inaccuracy was such that the chance to obtain reliable results from any kind of empirical tests of international trade theory was doubtful. Hence, any attempt of economic historians to make sense of 19th century international trade statistics and to use them for statistical analysis seems to be even less promising, as that period predated statistical unification and modern data procession even more.1 Still, especially for the decades after 1870 economists and economic historians are making intensive use of international trade statistics, as is proven by high citation frequencies for the accessible data compendiums of Angus Maddison (2006) and Brian R. Mitchell (2003). This practise finds backing by Paul Bairoch, who published central and authoritative works on the structure of European trade in the 19th century (Bairoch, 1973, 1974, 1976) and proceeded from the assumption that foreign trade statistics for the later part of the 19th century were not or not substantially less reliable than those for the 1970s (with exception of Dutch statistics). While he provided no direct evidence for this claim, Giovanni Federico and Antonio Tena (1991) re-examined early 20th century trade data comprehensively and gave a positive verdict on data quality, at least for aggregate figures, but suggested further case studies with disaggregate data. Anna Carreras Marı´ n (2005) followed their recommendation and in a study on geographical distribution of international trade in textiles in 1913 found that even on the aspect Federico and Tena had ruled out by aggregation, origin, and destination of trade flows, data was fairly reliable if individual country pair comparisons are weighted by importance in total trade. Whether results for 1913 are representative for the decades before has never been investigated. Reliable data especially for the decades before 1870 is highly needed, as according to the cited works by Bairoch as well as Walt W. Rostow (1978) and Arthur Lewis (1981), they were marked by accelerated growth in European and world trade and what looks like the highest growth rates ever.2 Studies on this issue and its causes have been limited mostly to the investigation of integration in commodity prices and decreases in transport costs. These approaches do not allow tackling the contribution of increasing international cooperation in trade and monetary politics, nor do they explain

Bilateral Trade in Europe, 1857–1875

83

trade growth directly.3 For this, reliable bilateral data is needed. Two central aims of this paper therefore are to construct a comprehensive, internationally comparable set of foreign trade data for the period from 1857 to 1875, to solve reported problems with historical data and to test the consistency of results. By this, the present paper evaluates the arguments that led a large group of historians to heavy scepticism and even dismissal of trade data as a historical source, and it provides a dataset along with qualified information on its limits to prevent the shortcomings of unconsidered use of potentially biased data. The approach follows D.C.M. Platt’s (1971) advice and combines the statistical investigation of the numbers gathered from original sources, historical foreign trade tables, with contemporaries’ and historians’ insights on how these statistics were elaborated. The result is an attempt to harmonize idiosyncrasies resulting from the fact that international unification in data gathering and publication was far from being implemented in the third quarter of the 19th century. National publications reflected a variety of differing ways of collecting, sampling, and presenting data according to domestic preferences and needs.4 Three central and basic problems have to be dealt with: Which volumes were gathered? How were prices and values obtained and calculated? And, in which way should and could origins and destinations be recorded?5 The following section presents the construction principles of the new dataset that comprises data for 7 central players in the North Atlantic economy of the mid-19th century and a bundle of 21 commodity groups of central importance in intra-European trade. Section 2 then conducts a survey of the general challenges that 19th century foreign trade statistics pose to the investigator, describe the most important idiosyncrasies of the individual statistics covered and apply strategies to correct shortcomings and overcome inconsistencies that can be treated at the national level. Next, the consistency of the dataset is analysed by systematic comparisons of pair-wise records from partner’s statistics on their respective import and export volumes. By this, the challenge posed by Morgenstern (1963, p. 180) that ‘‘writers of all phases of foreign trade will have to assume the burden of proof that the figures on commodity movements are good enough to warrant manipulation and the reasoning to which they are customarily subject’’ is accepted. The tests show that the dataset has a fair, although far from perfect consistency and that it can be used for the investigation of international trade on bilateral, commodity structure and aggregate level. Subsequently, the main problem of geographical misassignment of trade flows arriving after transit through third countries is tackled by a comprehensive and systematic correction. The final section evaluates the magnitude of the proximity bias in

84

MARKUS LAMPE

empirical work applying a basic version of the gravity equation. We see that the border bias in uncorrected data is of considerably magnitude and as a consequence models using uncorrected data are of lower quality than models estimated with corrected data. Apart from the border bias, this leads to problems in the estimation of the impact of policy variables, in our case tariff levels. It is therefore recommended to be cautious about the results of estimations made with uncorrected data. Furthermore, the last section presents descriptive statistics on three central aspects of the dataset: bilateral trade shares, main providers of specific commodities to Western Europe, and commodity specific tariff rates.

2. THE SAMPLE: COUNTRIES AND COMMODITIES Because of their importance in world trade and the quality of their statistics a core group of seven countries and customs areas was selected for investigation: the United Kingdom, France, the German Zollverein – since 1871 the German Customs Area –, Austria-Hungary, Belgium, the Netherlands, and the United States. All national statistics refer to national territories only, excluding colonies and other foreign dependencies. The seven countries and customs areas included in the dataset accounted for around 62% of world exports in 1860 (calculated from Hanson, 1980, Table A.3). Taking into account the commodity group selection, the dataset covers about 36% of world exports, but a far higher share of trade in manufactures. On a European scale, according to Bairoch (1973, Table 3), the countries in the dataset undertook around 79% of Europe’s total exports, of which at least 58% formed part of the commodity sample described below. This means that the dataset covers at least 46% of total European exports.6 Judging the quality of data gathering and processing, four general criteria appear: (a) completeness of the reports, (b) reliability and consistency in recording and reporting in systematic schemes, (c) recording and/or calculation of values on the basis of realistic prices, and (d) systematic and reliable determination and publication of origin and destination of the merchandise. The last two criteria are treated in detail in Section 3, while this section aims to provide a general overview of virtues and shortcomings of the original statistics used in the research presented here. As a start, Table 1 informs about the differing degrees of detail in the national statistics, as well across countries as over time. French statistics were consistently the most detailed, while the Belgian ranked last in terms of number of differentiated items. In all countries, import records were more

85

Bilateral Trade in Europe, 1857–1875

Number of Individual Items Reported in the for Import, Export, and Transit and Re-export Statistics (Commodity Tables) in Selected Years. Country

United Kingdom

France

Belgium

The Netherlands

Zollverein/Germany

Austria-Hungary

USA

Import Records

Export Records

1857: 952 (515 under other articles)

1857: 303 (94 under other articles)

1865: 798 (334 under other articles)

1865: 407 (71 under other articles)

1875: 351 (incl. ‘from all countries’ as origin)

1875: 207 (incl. ‘to all countries’ as destination)

1858: 824 1865: 1100 1875: 1009 1857: 554 1865: 141 1875: 221 1857: 459 1865: 484 1875: 506 (special), 72 (general) 1857: 263 1865: 271 (I), 340 (II) 1875: 498 1857: 279 1865: 417 1875: 575 1856/1857: 356

1858: 796 1865: 877 1875: 1008 1857: 542 1865: 141 1875: 218 1857: 458 1865: 478 1875: 478 (special), 72 (general) 1857: 249 1865: 258 (I), 336 (II) 1875: 498 1857: 258 1865: 377 1875: 530 1856/1857: 126

1865/1866: 729

1865/1866: 261

1875/1876: 203

1875/1876: 232

Transit/Re-export Records 1857: Re-export 601 (384 under other articles), transit 28 1865: Re-export 608 (474 under other articles), transit 33 1875: Re-export 335 (incl. ‘to all countries’ as destination), transit 7 1858: 443 1865: 402 1875: 485 1857: 542 1865: 141 1875: 218 1857: 368 1865: 317 1875: 68 1857: 247 1865: 249 (I), 324 (II) 1875: (498) 1857: 86 1865: 26 1875: 80 1856/1857: 273 (re-export) 1865/1866: 220 (re-export) 1875/1876: 168 (re-export), 162 (in transitu)

Notes: In Belgian statistics, exports and transit are reported in joint tables. Zollverein 1865: separate statistics were published for each semester, 1875: imports and exports in transit were included in imports and exports tables. Although counts for Austria-Hungary are dependent on the interpretation of subheadings and special provisions for certain individual commodities from specific trade partners as individual items or not, the figures presented here are consistent over time. Dutch transit counts are for ‘Doorvoer met overlading’ (i.e. transit with reloading); after 1871, imports for home consumption and exports of domestic production are reported in separate tables (‘special’) from total (‘general’) imports and exports. Source: Own count.

86

MARKUS LAMPE

detailed than export and transit records. They also were assumed to be far more reliable than export and transit records. The common reason for both phenomena is that the former were subject to duties and attentive customs control, while only a small fraction of exports was dutiable, and transit was free in general, and hence controls were much less strict.7 This is why volumes and values of bilateral foreign trade in the dataset were taken and reconstructed exclusively from importers’ records. Export, re-export and transit records were used only to correct geographical misassignments, and therefore were not used as absolute amounts, but for the information on shares of exports of domestic production and transit from third countries only. The pair-wise comparison of importers’ and exporters’ records in Section 4 shows that this is feasible. When overall data quality is referred to, the foreign trade statistics of the United Kingdom generally ranked as best practice both among contemporary and modern scholars. To German statisticians who in the 1860s were looking for models for the reform of the Zollverein’s statistics, the British way of gathering quantities and values was appreciated as being ‘‘of a high degree of perfection’’ (Hirth, 1870, p. 426).8 Also eminent British economists valued British statistics critically as largely complete and not systematically distorted (Bourne, 1872; Giffen, 1882; Bateman, 1892–1893; Imlah, 1958).9 In France, the majority of declarations were made in quantities, although for some high value articles also value declarations occurred. The literature on the quality of French trade statistics is small (Hirth, 1870; Richter, 1901a), but there are no hints that French statistics contained systematic errors, except for doubts on the accuracy of export prices (see below). Publications were extremely detailed and the classification did not change substantially over time. Belgian foreign trade statistical publications were quite similar to the French ones, although far less detailed. Contrary to almost all other historical foreign trade accounts, the consistency and reliability of Belgian publications has been systematically assessed recently. Daniel Degre`ve (1982) published data series for the 19th century with some remarks on its quality, but was criticised by Edwin Horlings (2002, p. 111) for uncritically collecting and processing historical data while failing ‘‘to construct consistent and reliable figures on foreign trade’’, as original Belgian statistics were systematically distorted due to the presence of ‘disguised transit’, i.e. re-exports of low-duty commodities from entrepoˆts after ‘nationalisation’ that leaded to double-counts of imports and exports. (The problem is described in Section 3 and comprehensively treated in Section 5.)

Bilateral Trade in Europe, 1857–1875

87

Dutch foreign trade statistics have been called ‘‘a stubborn source’’ (een weerbarstige bron) by J. Thomas Lindblad and Jan van Zanden (1989, p. 231) and have been heavily criticised by everybody who worked with them.10 Apart from general doubts about the accuracy of bookkeeping, especially for exports and transit, the main problems with Dutch trade statistics were unreliable ‘official values’ and the presence disguised transit, which had to be corrected. When talking about stubborn sources, the Zollverein’s CommerzialNachweisungen, i.e. ‘Commercial Records’, championed all other statistics, or as Platt (1971, p. 125) expressed it, ‘‘[ . . . ] nobody can attempt to make sense of German statistics before 1880’’. Until 1871, the statistics of the Zollverein were mere quantity statistics (mostly in Zentners) that made impossible any statement on values of trade. Additionally, the practice of including national harbours and the independent Hanseatic Towns (Hamburg, Bremen, and Lu¨beck until 1868) in the list of trading partners made information on geographical distribution almost useless. Some alleviation can be obtained from the accounts by experts like Georg Hirth (1870, pp. 418–429; 1869, pp. 68–69) and Adolf Soetbeer (1875, p. 748) who regarded the collected quantity figures as not less reliable than those of other countries.11 In the subsequent sections it is shown that German statistics can be worked with after introducing prices and comprehensive treatment of records on origin and destination, although certain caveats remain. The main provisions for the elaboration of Austro-Hungarian trade statistics were given in 1855 and remained generally unchanged until 1875 (Richter, 1900a, p. 244). Of the countries in the dataset, the Habsburg Monarchy is the only territory for which contemporaries attest a considerable amount of smuggling (Hirth, 1869, pp. 69–70).12 Nevertheless, the main challenges are the detail of records and the valuing practices discussed in Section 3. Some European countries had to be excluded from the dataset despite their importance in European trade, because their statistical publications were to sparse for our needs and/or their trade portfolio and degree of detail in statistics proved to be too limited. Unavailability of regular and systematic trade statistics led to the exclusion of Russia, Italy, Switzerland, Turkey and Romania and Wallachia, while Denmark, Sweden, and Norway were excluded because comprehensive information on transit was not available and the statistics would have required intensive treatment on prices. Furthermore, the exports of these countries were concentrated in few sectors and their absolute market size was small.13 The main reasons for the exclusion of Spain were the large amount of (mostly unsorted) individual

88

MARKUS LAMPE

items in reviewed statistical publications, reported criticism on the official values and smuggling problems, as well as the extremely limited availability of the statistics in Germany. Of these countries, missing data for Russia, the country with the highest population in Europe, and for Switzerland, for its centrality between France, Germany, and Austria, are most deplored. For research designs interested in international comparisons the United States, Europe’s most important trading partner, especially as a market for manufactured commodities, have been included into the dataset. Their statistics differed from those of all other countries in that they were published for fiscal years that dated from July, 1st to June, 30th. The values contained in the dataset thus have been calculated as averages of the two fiscal years involved, i.e. data was gathered for all fiscal years from 1856/1857 to 1875/ 1876. As to the quality of US statistics, Alexander Del Mar (1868), since 1866 Director of the newly formed Bureau of Statistics, wrote in the preface to the 1867 statistical tables that in the period when the Secretary of the Treasury had elaborated the tables, they had been of very low prestige among businessmen and scholars.14 Still, the reasons he enumerated – the very incomplete recording of exports via railway to Canada15 and merchandise recorded as imports that were actually of national produce, as well as certain exports of national production that were recorded partially as re-exports16 – were found to have had no substantial impact on the quality of the present dataset, whose focus lies on intra-European trade. As is shown below, the resulting inaccuracies for certain commodity groups were of a magnitude that did not systematically bias the trade figures we deal with. Finally, it should that due to wars the territorial coverage of several national statistics changed in the period covered. The most affected territory was that of Austria-Hungary: During the whole period it included the main parts of the Habsburg Monarchy, namely Cisleithania and Hungary, while Dalmatia remained outside the main customs area until 1880. From February 1853 to October 1857, the politically independent Duchies of Modena and Parma were included in Austrian statistics as they formed an ephemeral customs union with the Habsburg Monarchy. In 1859, Lombardy was lost to France (and subsequently annexed to Sardinia); in 1866 the remaining part of the Veneto-Lombardian Kingdom fell to Italy. From 1867 to 1879 the customs area remained unchanged. As a consequence of the Austrian loss of Lombardy, in 1860 the formerly Italian territories of Nice and Savoy became part of France. In 1871, France lost Alsace and Lorraine to the nascent German Kaiserreich. The German Zollverein’s area had already been extended before, when the former Danish Duchies of Schleswig and Holstein became a Prussian province in 1867 and when the Mecklenburgs and Lu¨beck

Bilateral Trade in Europe, 1857–1875

89

entered the Zollverein in 1868.17 In 1871, Alsace and Lorraine were separated from France and annexed to the German Kaiserreich, and hence included into the German customs area. Different treatment of Alsace and Lorraine in French and German statistics in the single year of 1871 might have caused some distortions in the international comparability of statistics for this year.18 Finally, US statistics lack records on trade for ports of the Confederate States during the Civil War (1861–1865), as they were officially closed by the Union Blockade. As we deal with a period predating the Brussels Tariff (1913), the Standard International Trade Classification (1950), and the Harmonized System (1996) no unified commodity classification was used across national statistics, and national classification schemes were derived almost directly from national tariff schemes which were elaborated according to domestic industry, import and protection structures and prevailing political convictions.19 To constitute a commodity sample of reasonable coverage, complete import and export figures were gathered for all items enumerated in the trade statistics of all included countries for the year 1865. Then, from the 50 items with the highest individual import and export values for every country, harmonised commodity groups were constituted for imports and exports, 40 and 26 respectively. In the next step, some commodity groups were discarded despite being important in international trade because they were incompatible to the central aim to reconstruct bilateral trade among Europe’s major trading nations for the following specific reasons:  Products of tropical agriculture and extraction, namely cotton, tobacco, tea, coffee, cocoa, guano, indigo, cabinet woods, and dyewoods, have not been covered to avoid problems of transit and re-export through Western and Central European countries where they were not produced.  Mining products in general, such as zinc, tin, copper, coal, and crude oil (petroleum) have been excluded as they could not be produced and exported by countries without natural deposits. This caused empty fields in the international trade matrix, because specific mining minerals were important for one or two countries only.20  Butter and cheese made up an important share of Dutch exports (about 7% of uncorrected special exports), but did not rank high in other countries.  Sugar, unrefined and refined, was excluded for two reasons: it proved difficult to distinguish non-European cane sugar from European beet sugar, and additionally sugar was one of the first sectors were export subsidies were implemented on a big scale, especially in France and Austria-Hungary.21

90

MARKUS LAMPE

 Finally, haberdashery and the like were excluded because it proved impossible to establish a definition that would have been even remotely homogeneous across differing national classifications.22 Tables 2 and 3 show the 21 remaining commodity groups along with their shares in every country’s total imports and exports. A more detailed list of items can be found in Appendix 1. Especially due to the exclusion of tropical goods and mining products, the share of imports covered in the sample decreased to about 37% of total trade on average, while for most countries’ exports it covers around 60–70%. The shares of exports of the US, the Netherlands, and Belgium are below 50% due to their extensive produce of and trade in tropical and subtropical goods, and due to methodological differences in the calculation of commodity group values and total trade data taken from other sources (see table note). For the investigation of intraEuropean trade, this is not problematic, but one should be aware of these limitations in other contexts. Most of the other more important excluded categories (except haberdashery and dyewoods) consisted of only one or two individual items, for which additional data could be gathered relatively easily.23 In the process of matching the classification on national schemes, some specific problems occurred that mainly had to do with changes in national classifications between 1857 and 1875. The most important of these changes occurred in Belgium in 1865 when the number of items in the foreign trade statistics was cut down to 141 for both imports and exports. This had noticeable consequences for records on cloth, where former detailed records were merged into summary accounts for the items ‘cottons’, ‘woollens’, ‘linens’, ‘silk wares’ and ‘other cloth’ in 1865. This presents a problem because in the classification elaborated for this paper, the ‘silk wares’ group also contains high-value items of other materials like jaconet, tulles, embroidery and laces, which were more similar to their equivalents of silk then to simple and heavy items of the same material. Furthermore, fe´cules (mostly potato starch) and son (bran) were grouped with meals, flours, and milling products from 1865 onward, but did not belong to this commodity group before 1865 and in the other countries covered. Data was adjusted for all years after 1864 based on the average share of non-fe´cules and non-son products in the sum of the new broad category from 1859 to 1863. For textiles and other items, since 1866 records were again more detailed (cf. Table 1). For 1865, textiles figures were adjusted to the general scheme using the totals of individual items given in the more detailed tariff revenue records. As these did not contain information on geographical distribution,

Importer Commodity Group

UK

France

Zollverein/ Germany

Belgium uncorr.

Belgium corr.

The Netherlands uncorr.

The Netherlands corr.

AustriaHungary

USA

Arithmetic Mean

Coefficient. of Variation

Wheat Rye Milling production Wines Spirits and liqueurs Hides, skins, and leather Articles of leather/rubber Wood Wool Woollen yarn Woollens and worsteds Cotton yarn Cottons Linen yarn Linens Silk Silk wares Glass and glassware Pig iron Bar iron and steel Ironware

6.62 0.04 1.78 2.27 0.73 1.88 0.84 7.90 5.81 0.67 1.10 0.04 0.33 0.66 0.07 4.23 5.74 0.40 0.05 0.39 0.26

1.73 0.00 0.03 0.18 0.19 3.77 0.13 4.53 9.20 0.50 1.25 0.43 0.18 0.38 0.47 13.44 0.86 0.04 0.56 0.64 0.14

3.23 2.55 0.74 0.87 0.15 3.68 0.52 6.15 9.37 6.83 2.32 2.73 0.25 2.72 0.94 2.28 2.51 0.31 1.11 0.29 0.37

3.13 0.52 0.63 2.14 0.08 4.73 0.10 3.09 11.12 0.71 2.51 1.19 1.67 0.66 0.05 0.90 3.21 0.06 0.27 0.55 0.18

3.41 0.56 0.68 2.33 0.09 1.24 0.11 3.36 12.08 0.78 2.73 1.29 1.81 0.72 0.06 0.98 3.49 0.07 0.29 0.60 0.20

1.12 1.95 0.73 1.15 0.07 1.03 0.13 2.31 2.57 2.63 1.86 5.56 1.93 0.98 0.68 1.24 0.90 0.25 0.99 2.70 0.80

1.12 1.95 0.73 1.15 0.07 0.33 0.13 2.31 1.29 0.54 1.86 3.10 1.93 0.98 0.68 0.06 0.90 0.25 0.00 1.62 0.80

0.77 0.34 1.01 0.33 0.12 4.62 0.56 0.72 5.41 2.48 1.54 5.87 0.25 1.30 0.05 1.03 2.43 0.44 0.14 0.38 0.69

0.81 0.06 1.10 0.83 0.17 1.40 0.73 1.64 2.78 0.14 7.89 0.30 3.55 0.38 4.80 0.82 5.97 0.50 0.42 3.74 0.88

2.52 0.79 0.87 1.13 0.22 2.42 0.43 3.80 6.56 1.71 2.67 1.97 1.19 1.02 1.01 3.27 3.13 0.29 0.37 1.09 0.48

0.84 1.31 0.61 0.76 1.06 0.66 0.71 0.67 0.59 1.40 0.89 1.07 1.09 0.80 1.69 1.44 0.67 0.63 1.04 1.15 0.64

41.81

38.63

49.92

37.52

36.87

31.58

21.78

30.49

38.92

36.92

0.24

Total trade

91

Notes: Totals for the Zollverein are from Bondi (1958, p. 145; see discussion in Section 3). Belgium corrected includes correction for disguised transit (see Section 5) and uses revised totals by Horlings (2002, p. 138), on a comparable basis. Netherlands uncorrected includes updated current prices (see Section 3), Netherlands corrected additionally includes correction for disguised transit (see section 5). For both Dutch figures, ‘total trade’ was taken from Smits et al. (2000, Table H.1). As they correct for much less disguised transit than here (they do not correct trade volumes of yarns and raw silk), Tables 2 and 3 understate the percentage of imports and exports covered in the sample to a substantial extent. Austria-Hungary totals are uncorrected, while data on some categories was calculated with corrected price data. As prices especially for cotton products were set revised upward, percentage shares of other commodities might be biased upward. Arithmetic mean and coefficient of variation include corrected series only for Belgium and the Netherlands.

Bilateral Trade in Europe, 1857–1875

Commodity Groups and their Share in Countries’ Special Import Totals (Excluding Precious Metals; Per cent).

UK

France

Zollverein/ Germany

Belgium uncorr.

Belgium corr.

The Netherlands uncorr.

The Netherlands corr.

AustriaHungary

USA

Arithmetic Mean

Coefficient of Variation

Wheat Rye Milling prod. Wines Spirits and liqueurs Hides, skins, and leather Articles of leather/ rubber Wood Wool Woollen yarn Woollens and worsteds Cotton yarn Cottons Linen yarn Linens Silk Silk wares Glass and glassware Pig iron Bar iron and steel Ironware

0.02 0.00 0.01 0.00 0.15 0.60

1.62 0.43 2.89 8.43 1.92 2.37

5.84 1.13 1.20 1.01 1.29 2.04

0.28 0.49 0.03 0.01 0.14 5.25

0.28 0.50 0.03 0.01 0.14 0.47

0.39 0.02 0.03 0.00 0.54 0.91

0.39 0.02 0.03 0.00 0.54 0.24

5.05 1.43 3.36 1.35 0.51 1.59

3.78 0.07 6.75 0.08 0.36 0.49

2.42 0.51 2.04 1.55 0.70 1.11

1.00 1.11 1.23 1.98 0.95 0.78

1.52

2.95

2.04

0.30

0.30

0.04

0.04

2.60

0.55

1.43

0.81

0.00 0.66 3.27 11.35

1.04 1.07 0.69 8.40

7.39 2.87 1.13 13.58

0.63 0.71 4.17 8.02

0.63 0.71 4.20 8.07

0.14 2.82 2.48 0.20

0.14 1.60 0.48 0.20

7.08 14.15 0.76 4.42

2.44 0.13 0.04 0.09

2.68 3.03 1.51 6.59

1.20 1.65 1.04 0.80

6.69 22.62 1.99 5.16 0.47 7.66 0.45 0.97 5.84 2.47

0.11 2.58 0.39 0.83 4.64 17.44 0.72 0.00 0.57 2.46

1.28 4.59 0.52 2.59 0.95 13.06 0.86 0.07 0.46 1.20

0.83 2.60 3.12 3.41 0.21 0.31 2.91 0.15 3.58 1.64

0.84 2.62 3.14 3.43 0.21 0.31 2.93 0.15 3.60 1.65

2.99 2.33 0.14 0.27 1.35 0.03 0.10 1.05 1.13 0.25

0.64 2.33 0.14 0.27 0.22 0.03 0.10 0.10 0.10 0.25

0.17 1.25 1.50 2.17 2.56 3.15 3.77 0.02 0.79 1.31

0.00 0.57 0.00 0.15 0.00 0.07 0.27 0.01 0.04 0.82

1.39 5.22 1.10 2.09 1.29 5.57 1.30 0.19 1.63 1.45

1.71 1.49 1.06 0.88 1.32 1.13 1.11 1.83 1.36 0.57

71.91

61.53

65.08

38.79

34.22

17.23

7.88

59.00

16.71

44.80

0.56

Total trade

Notes: see Table 2.

MARKUS LAMPE

Exporter Commodity Group

92

Commodity Groups and their Share in Countries’ Special Export Totals (Excluding Precious Metals; Per cent).

Bilateral Trade in Europe, 1857–1875

93

countries of origin were allotted in proportion to their share in the broader items of the import statistics. A similar problem occurred with the notoriously broad textile items in the Zollverein’s statistics prior to the second semester of 1865: All cotton textiles (light, heavy, dyed, undyed, mixed, or pure) were merged into one heading and reported in quantities only. To improve international comparability, shares for high-value cottons have been extrapolated from reported cotton records for the period from 1857 to the first semester of 1865, and subsequently added to the silk wares group.24 In the US, in the later 1860s due to the formation of broader items in the statistical classification, yarns disappeared from the statistical tables on trade by commodities and partner countries, and were summed up with ‘other manufactures’ of a certain textile fibre. Fortunately, the statistics for imports for domestic consumption (that did contain no information on origins of imports and reported totals only) continued to provide information on the amount of yarns imported. Hence it was possible to use the share of special imports of yarn in the broader category to reconstruct overall yarn imports and assign countries of origin proportionately. For exports, the problem could not be solved this way, but is of minor importance in the light of the small share of yarns in total exports of 1865 as given in Table 1b. Two additional problems mentioned by Del Mar (1868) were of minor importance in for the present dataset: Wood originating from Maine was credited as imported from Canada after being transported via New Brunswick, and tin-plates were sometimes classified as re-exports instead of exports of national production, despite having received substantial modification in the US.25 Minor changes had to be dealt with for the Netherlands, where prior to 1863 rubber shoes, other articles of gutta-percha and certain mirrors were reported in the sum of ‘other articles’. Because of their negligible importance, mirrors (0.7% of all glass imports in 1863) were not reconstructed. For articles of rubber, 7.8% was added to the 1857 to 1861 values of the corresponding category based on their 1863 shares and geographical composition. For transit tables, comparable problems occurred in the UK, the Netherlands, and Austria-Hungary, where the degree of detail of the statistics was unsuitable and/or reduced over time. In the UK, prior to 1869 a large part of transit of manufactures was summed up under the broad heading of ‘Silk, Woollen, Cotton, and Leather Manufactures, and other Manufactured Goods not separately specified’. This amount was assigned to the commodity groups woollens and worsteds, cottons, linens, articles of leather and rubber and silk wares in proportion to the import share of each

94

MARKUS LAMPE

of these groups in their sum for ever country.26 Since 1869, the statistics merged this broad heading with coffee, unrefined and refined sugar, wine, and other articles into ‘other articles not separately specified’ Hypothetical amounts of the manufactures category and wine were reconstructed for 1869, 1871, 1873, and 1875 based on these headings’ 1867 share in the sum of the items in the new residual category. In the Netherlands, after 1871 general and special imports as well as general and special exports were reported in four separate tables instead of two. For imports and exports for home consumption and of domestic production, statistics remained fairly detailed, but for overall imports and exports and the transit share of it, the degree of detail was substantially lowered, and was reported in quantities (kg) only. To reconstruct trade for every commodity group, general trade and transit accounts for ‘yarns’, ‘textiles’ (manufacturen), ‘unwrought metal’, and ‘wrought metal/metal wares’ had to be broken down into the more specific items reported for special trade. To achieve this, the share of each item in the total of the group in 1871 was used to reconstruct hypothetical general trade values for the corresponding commodity groups in 1873 and 1875. For transit, an analogous method was used, based on the transit export shares, to which transit import shares were assigned proportionately.27 In Austria-Hungary, transit tables from 1863 to 1869 were divided into very broad categories, the 22 tariff-classes, and mainly reported quantities.28 This means, we lack detailed accounts for transit in individual commodity groups. The shares for quantities in the more detailed transit records of 1859 and 1871 were used to estimate them.29 For some commodity groups, transit records even in the more detailed period were broader than the commodity groups used here even in years with more detailed tables. Those records have been broken down according to each commodity group’s quantity shares in bilateral trade with the Zollverein for the corresponding year. The reconstructed commodity-group-specific transit figures were converted to values using mean prices of imports over every commodity group’s total.

3. METHODOLOGICAL CHALLENGES OF HISTORICAL FOREIGN TRADE STATISTICS Apart from differing degrees of detail and changing commodity composition, three methodological issues have to be treated when dealing with trade statistics for the 1850s to 1870s: (a) general principles on what is foreign

Bilateral Trade in Europe, 1857–1875

95

trade and in which subaggregates it should be divided, (b) practice of pricing and valuation, and (c) assignment of countries of origin and destination. This section discusses in which way idiosyncratic characteristics of national statistical publications present challenges to the comparability of the original trade data, and how these were dealt with during the harmonisation of the present dataset. Regarding definitions of foreign trade and its main aggregates, ‘foreign trade’ in merchandise is defined in principle as the total of all operations of commodity exchange that transcend domestic borders. For the statistical recording of these transactions, concepts named ‘general trade’, ‘special imports and exports’, ‘transit trade’, and ‘re-exports’ are employed. To provide a general definition, ‘special trade’ is what goes into or comes from the domestic market, while ‘general trade’ is a wider concept embracing all (recorded) trade passing a country, including transit, transhipment and reexport, i.e. merchandise passing national territory without being actually imported. Differences in conception between national statistics in the period under study stemmed from the presence of two different systems of sorting and displaying imports, exports and transit: the Anglo-Saxon and the Continental model.30 The general definition of special and general trade employed in the present paper is derived from the Continental model that was prevalent in Central and Western Europe, i.e. in all countries of the sample except the UK and the US. In customs houses on the Continent, incoming merchandise was directly declared as being imported for the home market (special trade) or as transit, while exports were reported as being of domestic production or recognised as transit.31 This led theoretically to consistent statistics that had general imports as equivalent to the sum of special imports and imports in transit, while general exports were equivalent to the sum of special exports and exports in transit, with imports in transit being equivalent to exports in transit. In annual accounts, imports into and from customscontrolled warehouses could temporarily impair these identities. In difference to the Continental Model, the Anglo-American model in principle recorded no special imports as such. Records were made for general imports, and imports for home consumption were devised only for commodities subject to duty. Furthermore, the ‘general imports’ recorded were not equal to those in the Continental Model, because ‘pure’ transit was recorded and published in separate tables. For exports, goods of ‘national growth and produce’, i.e. special exports, were distinguished from goods ‘of foreign and colonial growth and produce’, i.e. re-exports, either from bond or from the market (duty-free goods). The distinction between transit and re-export was somewhat arbitrary and mostly dependent on administrative procedures and

96

MARKUS LAMPE

the tariff scheme: merchandise in transit was exempt from import duties, an advantage that was not useful for imports of duty-free goods that could be reexported duty-free without being subject to bonded warehouse-restrictions.32 To sum up, the statistical tables under the Anglo-Saxon system reported special exports, re-exports, and transit, but did not inform about the share of ‘general imports’ that were re-exported, i.e. the countries of origin of special imports and re-exports were merged in one account. To achieve comparability with the Continental system, special imports then were reconstructed by deducting the amount of re-export from the ‘general imports’ amount.33 Countries of origin for these reconstructed special imports were assigned proportionately to those reported for ‘general imports’.34 The same shares were also used to reconstruct the countries of origin of re-exports. Finally, reexports and reported transit were then added up in one figure to achieve comparability with the transit figures published under the Continental model. Additionally to general system differences, both under the Continental and the Anglo-American system, the re-export of processed imports was an issue. In general, the ‘substantial modification’ rule was applied. Problems of comparability resulted most prominently from special regulations for dutyfree import contingents earmarked for export processing. This ‘improvement trade’, in principle was created on the Continent to foster economic activity in border regions, but was extended widely over time (von Kalchberg, 1871). It was most diffused in textiles, milling, sugar refining, and iron and steel processing. The main challenge during dataset construction resulted from the fact that these ‘temporary admissions’ were given different statistical treatment in different countries. In France, the commodities involved were booked as transit and therefore showed up in general, but not in special trade records. As they underwent substantial modification, the import and export classes of this ‘transit trade’ did not coincide, and thus at the commodity group level statistics were distorted. I therefore modified the official figures and treated imports and exports under this regime as special imports and special exports, as was the case in Belgium. In Germany and AustriaHungary, imports and exports for ‘improvement trade’ were not recorded in separate tables. Because these tables lacked detail and systematic information on places of origin, improvement trade for Germany and Austria was not included in the dataset, although in some cases (most notably iron and steel processing and textiles) improvement trade was not negligible.35 Additionally, the dataset does not include separately reported ‘transit without reloading’ in the Netherlands and ‘transit on short roads’ just touching German territories as well as trade with fairs, postal transport and

Bilateral Trade in Europe, 1857–1875

97

different classes of entrepoˆts that were given as ‘places of origin’ without further information in the Zollverein’s statistics until 1871.36 Apart from issues of aggregation, a practical problem for the separation of special and transit trade resulted under the Continental model. For duty-free and low-duty commodities in transit the benefits of the transit regime were outbalanced by additional costs and restrictions associated with bureaucratic transit and bond procedures. This led to imports in transit being declared as special imports and re-exported as formally ‘nationalized’ products from the domestic market. As this form of re-export was not recorded as such (hence the denomination ‘disguised transit’), it distorted official figures. In consequence, the reported share of special trade in general trade was too high and the share of transit too low, thus leading to inflated figures for trade with neighbouring countries at the expense of countries of ‘real’ origin and destination. This phenomenon is supposed to have been most prominent in the statistics of the transit countries Belgium (Horlings, 2002) and the Netherlands (Lindblad & van Zanden, 1989), but suspicions also existed for French (Hirth, 1870, p. 427), German (Soetbeer, 1875, p. 747), Austro-Hungarian (Richter, 1900, pp. 249–250; Don, 1968, p. 83), and even British (Fuchs, 1893, p. 90) statistics. The problem was of a magnitude that could not be ignored. Therefore systematic tests for and corrections of disguised transit are performed and presented in Section 5. The second major problem consisted in methodological differences in recording prices and values. Of course, the most obvious way to generate values is to gather them directly from the declarations of importers and exporters. Nevertheless, this was practised a comprehensive way only by the UK and the US. The British method of pricing was regarded as the best practice available. From 1853 to 1870, imported quantities were converted to values using ‘computed real values’ that were estimated on an annual basis from monthly reports of a group of around 50 important trading houses in London and additional information from special officers in Liverpool and Hull.37 From 1871 onwards, declared values were used for imports and exports, as already had been practice for exports since 1798.38 It was not known even to contemporary experts whether the switch from computed real values to declared values for imports led to some kind of structural break; Stafford, Maton, and Venning (1953, p. 291) stated that contemporaries ‘‘regarded as at least equal’’ computed real to declared values. Otto Richter (1901b, p. 20), supposed that the change led to a rise import values, but was unable to state the amount or prove his proposition.39

98

MARKUS LAMPE

In the US, during the whole period all prices and values were obtained from declarations. Import values were given as ‘‘sworn specie values at foreign places of export’’, as were re-exports ‘‘[mainly]’’, while exports were reported with their declared free on board (fob) value as ‘‘sworn currencyvalue at domestic places of export’’.40 In contrast to the values of exports the declared values of imports comprised the whole cost upon arrival in the US, together with a commission of at least 2.5%, therefore coming close to the concept of cost, insurance, freight (cif).41 In the other European countries, due to the level of training of 19th century customs officials’ and merchants’ unwillingness to declare values, duties were mostly specific, and original records on imports and exports were kept in hundredweights, meters, dozens, loads, etc. To obtain values, official valuing bodies within or outside statistical offices had to provide ‘official prices’ that were multiplied ex post with recorded quantities. Apart from data collection restraints, the main reasons given in favour of official pricing stemmed from suspicions about the accuracy of value declarations, especially referring to strategic underreporting to evade ad valorem duties and laziness to declare correctly.42 Nevertheless, while value declarations had to be made for each individual consignment, official prices necessarily were estimated mean prices for aggregates whose constituent parts differed in form and quality. Reliability of official prices therefore depended not only on the competence of the price-estimating body, but even more on the conciseness and homogeneity of underlying categories and the frequency of revisions.43 Best practise in the field of official prices – apart from the mentioned British ‘computed real values’ – were the French ‘actual values’ estimated since 1849. They were based on current prices and annually revised by a ‘‘Permanent Valuing Commission’’.44 Soetbeer (1864, pp. 26–27) concluded from his comparison of prices and values from French and Hamburg’s statistics that official import prices were ‘‘remarkably correct’’, but exports were valued quite certainly at too high prices. Based on Soetbeer, Hirth (1869) and other sources, Bodo von Borries (1970, p. 9) estimated that exports were overvalued by about 20% in the 1850s and early 1860s. The reason was mainly that export prices were fixed as equal or higher than import prices, despite the principle that trade costs (cif) should have been included in import prices. Also in Belgium official prices were annually revised by a price commission, but as statistical categories were much broader, estimated prices were much less detailed and can hardly have been as accurate as the French ones.45 The bigger part of prices was re-estimated annually, and imports of many manufactures were declared directly in values, although

Bilateral Trade in Europe, 1857–1875

99

official prices for certain items remained unchanged for a long time.46 For the dataset, import values have been taken to be fairly reliable although they might have been too high in some cases (see Section 4), while doubts on the accuracy of export prices remained.47 As the dataset is based mainly on import values only corrected by export shares, Belgian price data have not been corrected, except for one commodity in one year.48 The main work in price corrections has been done for the statistics of the Netherlands, Austria-Hungary, and especially the German Zollverein. In the Netherlands and Austria-Hungary, the revision of official prices presented a considerable problem, while in German statistics prior to 1871 prices and values were completely lacking. In the Netherlands, official ‘benchmark prices’ (matstaaf) remained almost unrevised from 1846 until World War I, and even their accuracy for the initial year was doubted.49 I followed the example of Lindblad and van Zanden (1989), Smits (1995), and Smits, Horlings, and van Zanden (2000) and reconstructed trade values for all years at the commodity level. British prices were used where available for comparable unities; otherwise (in many manufacturing positions), Hamburg prices were used.50 In Austria-Hungary, the first official prices of 1852 were only occasionally updated until the publication of the 1862 statistics. In 1864, the then president of the Austrian Commission for Statistical Administration, Karl von Czoernig-Czernhausen, undertook a comprehensive survey and gathered individual prices for 1.546 tariff items. Unfortunately, after Czoernig’s departure from the Commission in 1865, the 1863 prices again were only occasionally updated until 1875. The official prices were revised again for the statistics of 1875 when additionally ‘real commercial values’ (wirkliche Handelswerthe) were introduced.51 Based on these facts, Yehuda Don (1968, p. 86) claimed that Austro-Hungarian value statistics before 1875 were ‘‘[ . . . ] of no service for any analytical purpose’’. Sir Robert Giffen (1882, pp. 190–191) stated that the introduction of commercial instead of official values had changed trade aggregates by 10–20%. As prices generally were revised upward, foreign trade would have been undervalued by this amount before 1875. Official prices of 1857–1873 have been revised by comparing the values per kg from Austro-Hungarian statistics with the new estimated prices for the Zollverein (see below), Austria-Hungary’s main trading partner for each commodity group.52 Where differences were very high and trends antidromic, Austrian prices have been corrected conservatively.53 The ‘Commercial Records’ of the German Zollverein prior to 1871 were quantity statistics that did not contain any values at all. Since 1872, the Imperial Statistical Office estimated official prices based on prices quoted at

100

MARKUS LAMPE

principal German markets, mainly Hamburg and Bremen, complemented by the expertise of competent Chambers of Commerce, especially when tariff items containing heterogeneous manufactured goods were concerned. These prices were taken to be more or less reliable, at least for imports.54 The Hanseatic Towns of Hamburg and Bremen had a long history of recording their trade in values based on declarations. Contemporaries regarded their average prices as comparable in quality to those of the UK (Soetbeer, 1864, p. 10). Because the Hanseatic Towns were Germany’s main exchange centres, the officers of their statistical bureaus could easily crosscheck declarations with current market prices, thereby improving reliability even more (Hirth, 1870, pp. 416–418).55 For years prior to the publication of official prices, different scholars and leading officials presented unofficial estimates for the value and balance of the Zollverein’s foreign trade in one or more years, mainly based on Hamburg prices supplemented by other sources.56 In 1958, Gerhard Bondi joined these and other estimates and offered calculations of foreign trade totals and other aggregates for the entire life span of the Zollverein (1834–1871). Unfortunately, his figures suffer from severe shortcomings, as von Borries (1970, pp. 5–7) convincingly argued: Bondi mixed up contemporary estimations despite them not being really comparable, as some worked with fixed, some with current and some with outdated prices from earlier estimates.57 For the present dataset, prices and values were re-estimated for the period prior to 1871 based on prices quoted in Hamburg and Bremen where possible, complementing them in some cases with British prices. Unfortunately, the broad headings of the Zollverein’s tariff rendered the estimation of appropriate prices for imports and exports difficult for commodity classes of heterogeneous composition whose internal composition remained unknown. In these cases, price series were extrapolated backwards from the official prices of the Kaiserreich’s statistics for 1872 and 1873. In most of the items in question, the detailed official prices for comparable categories in French statistics could be employed, thus combining the movement of French prices with the absolute values of the German Imperial Statistical Office at the end of the period. For some minor items that were reformed by the tariff reform of 1865, values and/or price relations for earlier years were taken from Hu¨bner’s (1861, 1859) prices for 1857 and 1859 and Bienengra¨ber’s (1868) prices for the early 1860s. As seen in Section 4, the resulting estimated values are far from perfect, but fairly useable for statistical analysis. Still, according to contemporary and modern accounts, the most severe shortcoming of historical statistics was the incomparability of records on bilateral trade flows due to differing practices and limited capabilities in

Bilateral Trade in Europe, 1857–1875

101

determining countries of origin and destination. In principle, the country of origin of a merchandise can be defined as (a) the country from which the merchandise was directly received, i.e. the country of last land border crossed or last port entered before arriving at the final destination; (b) the country of consignment, i.e. from where the goods were sold; or (c) the country of original growth or production; i.e. the production/extraction country for raw materials and the country of last substantial transformation for manufactures. Analogous considerations apply for exports. Despite being the least satisfying alternative, the ‘border approach’ (a) was prevailing for imports and exports until at least the 1890s in most countries. For exports, in general, the problem of misassignment was larger for more complex trading routes, and for exports possible reselling by exporters’ commercial counterparts made proper declaration and recording of final destinations even more difficult. Transhipment and re-export were most prominent in trade with products of tropical agriculture that were distributed in Europe through central commodity exchanges, but were also present for many other articles.58 As Section 6 will show, the share of merchandise imported after transit through other countries was between 26 and 29% of total trade and thus far from negligible. Thus, the ‘proximity bias’ has to be overcome to reconstruct a meaningful matrix of bilateral trade flows between ‘real’ countries of origin and destination, because otherwise bilateral trade between non-bordering countries would be systematically understated. To deal with this, a comprehensive ‘transit correction’ is performed in Sections 5 and 6. Its basic idea is to use information from partner countries’ export and transit record to reconstruct the ‘real origin’ of recorded imports. The following literature survey aims to provide the necessary background concerning the quality of geographic assignments in each countries’ foreign trade statistics. In the UK, countries devised as countries of origin or destination generally were the last/next ports visited by a ship (Bourne, 1872, p. 202). From 1874 on, the attempt was made to record the initial harbour, while overland transit on the way to this harbour was not accounted for. This led to wide missasignments of countries of origin and destination, e.g. for trade with Germany transiting the Netherlands via the Rhine-Maas system, trade with Austria-Hungary via Hamburg or with Switzerland via Germany, Belgium, and France.59 In the US, also the origins of incoming and the destinations of outgoing ships were recorded as initial ports, not taking into account overland transport before or after.60 In Continental European countries the pure border approach or a methodological mix was applied. French authorities started to record ‘real’

102

MARKUS LAMPE

countries of origin and destination for landward trade in 1857 and for sea trade in 1870.61 In Belgium, since 1841 it was intended to use the ‘real origin/ destination’ approach. Nevertheless, contemporary experts in the Commission for Further Development of the Statistics of the Zollverein reported with indirect proofs that this approach had been comprehensively implemented neither in France nor in Belgium.62 This means that records on origins and to a greater extent those on destinations in French and Belgian statistics resulted from a methodological mix. The reliability tests in Section 4 confirm this, possible distortions could only be dealt with partially, as is discussed in the final part of that section. Dutch, Austro-Hungarian and German statistics were elaborated in principle strictly according to the border approach, although in the Netherlands small amounts of landward imports from non-bordering France were reported occasionally.63 Austria-Hungary and the Zollverein/Germany presented a second particularity, because their statistics treated the free ports outside the customs border as points of origin and destination. Hence, most seaward trade with the Zollverein/Germany was accounted to as coming from or going to the free-ports of Hamburg and Bremen and smaller freeports in the states of Hannover (since 1866 Prussia) and Oldenburg.64 For the smaller share of sea trade through national ports that were not free ports, i.e. Prussian ports in the North and Baltic Sea, geographical distribution on origins and destinations was published since 1858 in separate tables, whose sums except for supposed transmission and printing errors matched the sum of trade reported in the main tables. I matched these statistics into the main tables for imports, exports, and transit. The actual countries of origin and destination of trade flows through Hamburg and Bremen for Germany, and Trieste, Fiume and Venice for Austria-Hungary have been reconstructed from the information on the geographic distribution for trade in every commodity group and year contained in the separate foreign trade statistics published by these free-ports. The underlying approach treats the Hanseatic Towns and the Austro-Hungarian ports as entrepoˆts without proper imports or exports for domestic consumption or of domestic production of these cities. Due to the small size of the ports in comparison with production and consumption in respectively the Zollverein and Austria-Hungary, possible errors resulting from this approach should be acceptable facing the lack of better alternatives (Buchheim, 1982, p. 25). For the Austrian ports, Trieste’s statistics had to be taken as representative for all ports, as for the other harbours comprehensive records were lacking. For Hamburg, there were no export statistics for the

Bilateral Trade in Europe, 1857–1875

103

years 1856–1871. Hence, it was impossible to reconstruct complete records on the destinations of Zollverein’s exports.65 Finally, as a technical note: For countries whose statistics were lacking crosstabulations showing the share of commodities in transit or re-exported which was transferred from each single country of origin to each single country of destination (UK, US, France, Belgium), I constructed hypothetical transit tables by assigning countries of origin to countries of destination proportionately.66

4. STATISTICAL ASSESSMENT OF THE QUALITY OF HISTORICAL BILATERAL TRADE DATA Before we can perform the comprehensive transit correction outlined above we have to assure that import, export and transit statistics serve our purpose. This is because notwithstanding the magnitude of the problem, one might doubt the feasibility of the proposed correction of importers’ statistics with information from partners’ corresponding export and transit records. Corrections might yield inadequate results if corresponding import and export records were unreliable, inaccurate, and internationally incomparable, as judged by authors cited above. If bilateral import and export records did not coincide reasonably, it would be arbitrary to use the latter to correct the former, however well thought the correction might be. Although Sections 2 and 3 to some extent qualified overall criticisms and tried to manage resulting problems, incoherencies remained. Therefore, this section provides a systematic inquiry into the comparability of general export and import records. It compares general trade records because they accounted for all goods leaving exporting countries and entering importing countries, whether in transit or not, and the proposed transit correction is based on the identity of overall bilateral trade flows, not on the coincidence of special imports and special exports (that is explicitly doubted by stating the need for transit correction).67 Although the reliability of foreign trade statistics cannot be measured in absolute terms,68 I follow the literature and employ three ways of assessing the consistency of bilateral records. The first considers the magnitude of the difference between individual bilateral import and export records on the same flow from country A to country B. If we knew for certain that always importers’ records are accurate we could use them as the denominator to

104

MARKUS LAMPE

assess the absolute quality of exporters’ reports, or vice versa. The literature review above suggests that in our case neither importers nor exporters reported perfectly, which is why I use the Implicit Minimal Measurement Error (IMME) ratio, which computes the absolute value of the ratio of the difference of importers’ and exporters’ records relative to their sum. IMME assumes that potentially both reporters contribute to non-identity and thus informs about the (minimal) relative dissimilarity of figures.69 Nevertheless, it does not inform us whether differences are systematic and significant (Folchi & Rubio, 2005). Therefore, in a second step I calculate Pearson’s r for corresponding series of imports and exports (in cross-section and time series) to see if importers’ and exporters’ records systematically diverge in the order and magnitude of bilateral flows. Third, the statistical significance of differences in magnitude between importers’ and exporters’ records – whether importers or exporters tended to report higher figures – are assessed using p-values from Wilcoxon’s non-parametric signed rank pair tests proposed by Marcelo Folchi and Mar Rubio (2005) for this aim. Additionally, information on whether importers’ or exporters’ records tended to be higher is provided, assessed from the comparison of the rank sums of positively and negatively signed differences used to compute the p-values. Table 4 reports the results of the three tests for all commodity groups and importing countries, respectively, in every single year. The tests have been done on two levels of aggregation: the left part of the table reports results for the sum of all commodity groups, i.e. the ‘aggregate’ bilateral trade for every country-pair, while the right part contains results on the pool of all individual trade flows in individual commodity groups. From 1857 to 1869, Zollverein’s exports were omitted as exports via Hamburg could not be directly reconstructed. The overall picture is encouraging: the average IMME for sums is 0.108, which implies that importers’ and exporters’ records differ by 24.2%, on average. This is below the 25% Morgenstern (1963, p. 177) established as the border of ‘good’ results for the first half of the 20th century, i.e. after 30–80 years of progress in statistical unification and recording methods. It is also almost in line with the ‘freight factors’ Federico and Tena (1991, p. 263), assumed as reasonable margins of error based on data for German exports in 1951, 75 years after the end of the period under study. Even more important are the high correlations well above 0.9 for the sums and 0.8 for the pool of commodity groups. Nevertheless, the weighted averages of IMMEs at the commodity group level hint at a (weighted) mean error of about 50%, and the results of the signed rank pair test show that in no way import records are always higher than export records. The finding that in most cases records did

Year

Sum Over all Commodity Groups (Aggregate)

All Commodity Groups Individually (Pool)

IMME

r (im, ex)

p-value Signed Rank Pair Test

Higher Figures Recorded (Rank Sum)

N

IMME

r (im, ex)

p-value Signed Rank Pair Test

Higher Figures Recorded (Rank Sum)

N

1857 1859 1861 1863 1865 1867 1869 1871 1873 1875

0.117 0.153 0.116 0.112 0.099 0.085 0.098 0.126 0.098 0.076

0.935 0.922 0.962 0.957 0.974 0.966 0.965 0.924 0.959 0.978

0.654 0.094 0.383 0.610 0.070 0.514 0.418 0.793 0.453 0.582

Exporters Exporters Importers Importers Exporters Exporters Exporters Exporters Importers Importers

36 36 36 36 35 36 36 42 42 42

0.201 0.238 0.214 0.196 0.180 0.248 0.210 0.220 0.203 0.185

0.841 0.774 0.915 0.931 0.944 0.831 0.899 0.851 0.904 0.923

0.012 0.013 0.898 0.178 0.041 0.135 0.774 0.988 0.446 0.629

Exporters Exporters Importers Exporters Exporters Importers Exporters Exporters Importers Importers

608 609 613 609 594 621 624 750 747 740

Average

0.108

0.954

–

–

–

0.209

0.881

–

–

–

Bilateral Trade in Europe, 1857–1875

Overall Indicators of Data Quality (General Imports, General Exports).

Source: Own calculations from dataset. Prior to 1871, no exporter records are available for Germany due to non-existence of export records for Hamburg.

105

106

MARKUS LAMPE

not differ significantly can be taken as an argument in favour of bilateral trade data.70 Indeed, the only significant results ( po0.05) are obtained for higher export records in 1857, 1859, and 1865. It can be assumed that the high individual differences in bilateral records at the commodity group level stem from two sources that tend to cancel each other out when considering totals. Both have been discussed above: (a) underreporting of exports (quantities) and (b) overvaluing of exports in some countries. To investigate these two problems in greater detail, the analysis was repeated for every country pair in every commodity group over the years. Table 5 presents evidence on the quality of the individual statistics in comparison with all export partners and in comparison with all import partners. Table 6 does the same for statistics for individual commodity groups. Overall, British, US, French and – surprisingly – Dutch statistics score best, while the statistics of Belgium, Austria-Hungary and the Zollverein/ Germany proved to be more problematic. Belgian records were significantly higher than partner records for imports as well as exports, while Austro-Hungarian records were significantly lower. Results for Belgium are driven mainly by seemingly overvalued trade in silk wares, woollens and worsteds, silk, and other articles, mainly textiles. The biggest share of these seemingly overvalued commodities passed Belgium in transit only (as well as wool from the US that also appears as overrecorded import in several years) and therefore might not have been valued with too much care. Furthermore, as special imports were not too affected by trade in these commodities, the problem is less severe for the construction of the dataset than initially appeared. The ‘underreporting’ of Austria-Hungary should be referred to two phenomena: potential underreporting of trade amounts and too low prices, as mentioned above. Furthermore, one should take into account that Austro-Hungarian exports records to all countries except the bordering states of Southern Germany and Saxony are reconstructed from Trieste’s import and export records, which might contain a higher margin of error than other foreign trade statistics. Import statistics for 1867 – the year after the Austro-Prussian War and the loss of Venice – appear as problematic especially for textiles and distort the overall picture to a considerable extent. Especially bulk goods such as wool and wheat were valued by Austrian authorities at lower prices than those estimated for German trade during the dataset construction. Also, landward trade with France and Belgium appears at least partially in these countries’ statistics, causing thereby a bias against Austria-Hungary’s records on these bilateral flows. Finally, there seems to

Country

UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Records as Importer (Pool)

Records as Exporter (Pool)

IMME

r (im, ex) (Average of Individual Years)

p-value Signed Rank Pair Test

Higher Figures Reported (Rank Sum)

N

IMME

r (im, ex) (Average of Individual Years)

p-value Signed Rank Pair Test

Higher Figures Reported (Rank Sum)

N

0.157 0.214 0.252 0.292 0.177 0.197 0.148

0.946 0.881 0.796 0.841 0.936 0.840 0.928

0.000 0.893 0.357 0.000 0.815 0.000 0.871

Exporters Exporters Exporters Importers Exporters Exporters Exporters

1002 960 1234 885 891 632 913

0.162 0.209 0.277 0.229 0.255 0.286 0.149

0.909 0.911 0.701 0.828 0.866 0.874 0.963

0.088 0.524 0.299 0.000 0.526 0.000 0.000

Importers Exporters Exporters Exporters Exporters Importers Exporters

1184 1171 378 1122 1094 713 855

Bilateral Trade in Europe, 1857–1875

Indicators of Data Quality According to Reporting Country.

Source: Own calculations from dataset. Values for Germany are for 1871, 1873, and 1875 only, as for earlier years no exporter records are available for Germany due to non-existence of export records for Hamburg.

107

108

MARKUS LAMPE

Indicators of Data Quality According to Commodity Group. Commodity Group (Pool)

Wheat Rye Milling products Wines Spirits and liqueurs Hides, skins, and leather Articles of leather and rubber Wood Wool Woollen yarn Woollens and worsteds Cotton yarn Cottons Linen yarn Linens Silk Silk wares Glass and glassware Pig iron Bar iron and steel Ironware

IMME

r (im, ex) (Average of Individual Years)

p-value Signed Rank Pair Test

Higher Figures Reported (Rank Sum)

N

0.151 0.189 0.160 0.225 0.203

0.960 0.947 0.942 0.923 0.988

0.223 0.655 0.594 0.000 0.789

Importers Exporters Exporters Exporters Importers

288 214 311 351 353

0.195

0.827

0.014

Importers

345

0.197

0.959

0.210

Importers

318

0.259 0.220 0.120 0.189

0.893 0.910 0.973 0.894

0.427 0.540 0.105 0.001

Importers Exporters Exporters Exporters

309 341 251 322

0.172 0.234 0.209 0.141 0.215 0.298 0.246

0.961 0.895 0.909 0.970 0.923 0.835 0.892

0.154 0.796 0.001 0.001 0.912 0.000 0.908

Exporters Exporters Importers Exporters Importers Exporters Importers

267 340 283 322 257 318 363

0.144 0.137 0.297

0.957 0.979 0.790

0.003 0.244 0.608

Importers Importers Importers

290 325 349

Source: Own calculations from dataset. Prior to 1871, no exporter records are available for Germany due to nonexistence of export records for Hamburg.

have been a problem with the proper recording of improvement trade, i.e. imports to be processed and immediately re-exported to the partner country, between Germany and Austro-Hungary which took place mainly in textiles of cotton and woollen and worsted yarn. This is suspected to have caused consistently lower export records of cotton yarns by Austria-Hungary and generally high margins of error in bilateral trade with cloth.71 A similar phenomenon as for Austria-Hungary appears for the UK, where values are also lower than in partner records. For imports, this is explainable by permanently ‘overvalued’ exports of silk wares and woollens and worsteds

Bilateral Trade in Europe, 1857–1875

109

from France and Belgium and by similar phenomena for silk exports of Belgium and leather exports from France. According to Table 5, the US recorded significantly higher values for its exports than its partners. This result is mainly driven by the fact that the US statistics were geographically more differentiated, and thus exports of small amounts were reported to partners whose statistics merged imports from the US into ‘other countries’. Despite the overall results the main exported commodity of the US, wheat, was reported by importers (most prominently the UK) at higher values than recorded in the US and with comfortingly low IMMEs. German statistics are of acceptable quality, but appear as the most problematic in the sample. This is not surprising given the complex reconstruction process and the valuation based on broad categories. Without going to much into detail about the commodity groups, results can be seen as quite intuitive: data coincidence is highest mainly for homogeneous goods such as wheat, flours, pig iron, bar iron and steel, yarns, silk and spirits, while more complex categories such as iron ware, glass, articles of silk, and other textiles present comparatively higher margins of error. Significant differences between importers’ and exporters’ records were found for 7 categories at the 5% level: while for hides, skins, and leather, pig iron, and yarns of flax, etc. import values are significantly higher, the contrary is found for woollens and worsteds, linens, articles of silk, etc., and wine. The commodities for higher import records have in common that they are mainly homogeneous articles with low tare, while the latter are mostly heterogeneous textiles whose package makes up to more than a quarter of total weight for exports. The ‘undervaluation’ of wine by importers is driven by repeatedly lower values for French wines, which seem not to have been differentiated according to quality in the pricing process of many import countries. Furthermore, in the process of data evaluation three highly distorting data points were identified and corrected by partner values and interpolation from former and latter data: French imports and exports of meals and flours from and to Belgium in 1867 and the Zollverein’s imports of construction wood from Austria-Hungary in 1869.72 In defiance of inconsistencies and problems in detail, the overall picture is highly encouraging for the feasibility of the proposed transit correction. Especially the consistently positive and high correlation coefficients enable us to use export records for the geographical correction of import records to overcome the bias that resulted from the prevailing use of the border approach in recording partner countries. The transit correction is performed in two steps: Section 5 describes the elimination of disguised transit at the

110

MARKUS LAMPE

national level, while Section 6 presents the comprehensive treatment of the proximity bias.

5. DETECTION AND ELIMINATION OF DISGUISED TRANSIT The first problem concerning the omission of transit in import records can be dealt with at the national level. Remember that the double-counting of imports that resulted from the nationalisation of duty-free imports for re-export causes systematic errors, i.e. artificial increases in special trade volumes at the cost of declared transit. Naturally, the presence and – more important – the amount of disguised transit, and therefore its actual impact, can only be estimated or inferred. For Belgium, Horlings (2002) detected disguised transit for wool (since 1850), raw hides and skins (since 1856), wheat (since 1870), and a variety of tropical and semi-tropical goods and minerals that are not part of the sample. The systematic analysis for disguised transit in Dutch statistics by Lindblad and van Zanden (1989, pp. 239–241) found a considerable amount of disguised transit in some items included in the present dataset for the period from 1872 to 1913, namely wool, pig iron, bar iron and steel, and rye. Jan-Pieter Smits (1995, p. 310) named wheat, rye, rye flour, wool, pig iron and bar iron, and steel as commodities with notable disguised transit. To assure that the results in the present study are comparable across countries, a comprehensive analysis of all commodity groups in all countries is done based on the detection method developed by Lindblad and van Zanden (1989, pp. 236–241) in their practical work with Dutch trade statistics. It allows to diagnose disguised transit, at least when it is of overwhelming importance in recorded trade flows: If in a category both the modulus of the difference of special imports and special exports amounts to less than 50% of the maximum of special imports and special exports (((|IMEX|)/max(IM, EX))o 0.5) and the Pearson’s r of special imports and special exports over time is higher than 0.95, the presence of a substantial and distorting share of disguised transit is ‘certain’. In contrast, if the modulus of the difference amounts to more than 70% and/or the correlation coefficient is below 0.7, the existence of considerable disguised transit can be dismissed. Between these limits, disguised transit is possible, but not all-dominant. Lindblad and van Zanden assumed that in commodity groups with highly probable substantial presence of disguised transit, only the

Bilateral Trade in Europe, 1857–1875

111

modulus of the difference between imports and exports was ‘real’ special trade while all other ‘special trade’ was in fact transit. For commodity groups they classify as containing ‘possibly assured disguised transit’ (r W 0.95 and ((|IMEX|)/max(IM, EX))W 0.5) they take 50% of the minimum of special imports and special exports as disguised transit. In his application of the method to Belgian data, Horlings (2002) developed two additional measures in order to classify with greater precision commodities with ‘possible disguised transit’. The first is the ‘propensity to declare transit’, measured as PDT ¼ (T/(Max(IM, EX))þT), where T is recorded transit and IM and EX are recorded special imports and special exports, respectively. The second measure is the correlation of the first differences of special imports and special exports. Horlings assumes that when a high amount of transit is declared, the presence of disguised transit is less likely, and when the correlation coefficient of first differences is small, disguised transit does not move the series. He abstains from Lindblad and van Zanden’s 50%-method for not completely ‘certain’ transit and only corrects – but radically – disguised transit in commodities where he can prove its presence. The results of the analysis for the disguised transit in the present dataset are reported in Table 7. Results for commodity groups and countries were an important presence of disguised transit could be dismissed, are omitted to save space. For the interpretation of Table 7 it should be mentioned that some of the commodity groups are of quite heterogeneous composition. Based on the results, I decided to perform complete transit corrections for raw hides and skins in Belgium and for raw hides and skins, raw silk, pig iron, bar iron and steel (plates and bars) in the Netherlands for the whole period. Furthermore, as results for wool in the Netherlands were not completely conclusive, Lindblad and van Zanden’s 50%-approach was applied throughout the period. Additionally, disguised transit of cotton yarns and woollen and worsted yarns was present in the Netherlands after 1861 when the tariff on these articles was substantially lowered. The evolution of disguised transit in these products can be seen in Tables 8a and b. A similar result has been obtained for the wheat trade of the Zollverein before and after 1865, where the application of the former preferential tariff for Austria-Hungary to all imports led to an increase in disguised transit (Table 8c). In the three cases, I partially corrected for disguised transit in a way that reconstructed constant transit shares for all years following the last ‘non-disguised transit year’, i.e. 88% transit in woollen and worsted yarns and 54% transit in cotton yarns in the Netherlands, and 36% transit in wheat in the Zollverein.73

112

MARKUS LAMPE

Indicators for Disguised Transit (Selected Commodity Groups). Country and Commodity Group

UK/articles of leather and rubber UK/silk wares UK/glass and glassware USA/wood USA/ironware France/hides, skins, and leather France/linen yarn France/bar iron and steel Belgium/hides, skins, and leather Belgium/woollens and worsteds Belgium/cotton yarn Belgium/cottons The Netherlands/ wines The Netherlands/ hides, skins, leather The Netherlands/ wool The Netherlands/ woollen yarn The Netherlands/ cotton yarn The Netherlands/silk The Netherlands/ glass and glassware The Netherlands/pig iron The Netherlands/bar iron and steel ‘Germany’/wheat ‘Germany’/milling products ‘Germany’/wines ‘Germany’/hides, skins, leather ‘Germany’/wood ‘Germany’/linens

((|IMEX|)/Max(IM, EX)) (Average 1857– 1875)

Correlation (IM, EX)

(T/(Max(IM, EX)) þ T) (Average 1857–1875)

Correlation (IM(1), EX(1))

0.47

0.67

0.20

0.32

0.30 0.29

0.52 0.85

0.24 0.14

0.10 0.08

0.35 0.26 0.31

0.36 0.49 0.87

0.02 0.09 0.08

0.16 0.20 0.32

0.41 0.42

0.37 0.09

0.32 0.22

0.23 0.30

0.23

0.97

0.20

0.87

0.48

0.66

0.69

0.64

0.32 0.41 0.98

0.17 0.25 0.96

0.40 0.49 0.58

0.04 0.19 0.83

0.16

0.98

0.38

0.89

0.18

0.55

0.07

0.14

0.45

0.88

0.37

0.81

0.54

0.98

0.23

0.86

0.22 0.47

0.79 0.87

0.23 0.37

0.45 0.55

0.14

0.95

0.02

0.96

0.49

0.98

0.03

0.91

0.38 0.36

0.89 0.96

0.06 0.02

0.95 0.20

0.32 0.55

0.60 0.94

0.25 0.16

0.10 0.74

0.26 0.48 0.48

0.26 0.32 0.96

0.00 0.21 0.40

0.31 0.50 0.66

113

Bilateral Trade in Europe, 1857–1875

(Continued ) Country and Commodity Group

((|IMEX|)/Max(IM, EX)) (Average 1857– 1875)

‘Germany’/glass and glassware ‘Germany’/bar iron and steel ‘Germany’/ironware Austria/woollens and worsteds Austria/linen yarn Austria/silk wares Austria/ironware

Correlation (IM, EX)

(T/(Max(IM, EX)) þ T) (Average 1857–1875)

Correlation (IM(1), EX(1))

0.47

0.59

0.50

0.13

0.49 0.41

0.71 0.73

0.42 0.51

0.51 0.05

0.42 0.44 0.45

0.34 0.14 0.42

0.09 0.42 0.08

0.06 0.81 0.47

Source: Dataset. IM ¼ special imports, EX ¼ Special Exports, and T ¼ declared transit and re-exports

Indicators for Disguised Transit: Woollen Yarn, the Netherlands. Year

Special Imports (d)

Special Exports (d)

Declared Transit (d)

((|IMEX|)/ Max(IM, EX))

(T/(Max(IM, EX))þT)

1857 1859 1861 1863 1865 1867 1869 1871 1873 1875

118,880 74,421 84,071 949,094 1,074,884 1,281,268 1,369,051 1,409,061 1,633,602 1,621,572

2,493 997 1,085 1,016,535 1,059,103 1,161,967 1,229,716 824,227 758,513 1,080,654

628,164 486,689 600,096 145,300 146,111 208,035 233,122 805,347 168,452 235,663

0.98 0.99 0.99 0.07 0.01 0.09 0.10 0.42 0.54 0.33

0.84 0.87 0.88 0.13 0.12 0.14 0.15 0.36 0.09 0.13

Additionally, Horlings found disguised transit of wheat through Belgium since 1870. Owing to the fact that the present dataset ends in 1875, this could not be proven in the present study, but it was found that the propensity to declare transit dropped from 0.19 in 1871 to 0.01 in both 1873 and 1875, and was accompanied by implausible increases of special exports especially to the Netherlands (from d5,711 to d338,777, and d346,350 respectively). A similar trend was found for rye, where total special exports rose from around

114

MARKUS LAMPE

Indicators for Disguised Transit: Cotton Yarn, the Netherlands. Year

Special Imports (d)

Special Exports (d)

Declared Transit (d)

((|IMEX|)/ Max(IM, EX))

(T/(Max(IM, EX))þT)

1857 1859 1861 1863 1865 1867 1869 1871 1873 1875

543,511 708,700 1,007,565 1,298,080 2,269,732 2,921,332 2,673,674 3,695,335 3,605,204 3,011,843

41,977 48,652 93,354 705,357 1,277,830 1,671,587 1,837,890 2,786,104 2,371,583 1,677,741

1,330,143 1,079,670 1,176,631 105,903 72,339 123,403 107,193 310,585 346,038 484,103

0.92 0.93 0.91 0.46 0.44 0.43 0.31 0.25 0.34 0.44

0.71 0.60 0.54 0.08 0.03 0.04 0.04 0.08 0.09 0.14

Indicators for Disguised Transit: Wheat, Zollverein/Germany. Year

Special Imports (d)

Special Exports (d)

Declared Transit (d)

((|IMEX|)/ Max(IM, EX))

(T/(Max(IM, EX))þT)

1857 1859 1861 1863 1865/I 1865/II 1867 1869 1871 1873 1875

1,512,087 1,170,542 4,823,489 1,426,633 838,203 1,254,298 10,593,884 4,695,335 6,655,726 5,087,870 5,677,964

5,447,106 3,750,396 8,750,045 4,275,058 1,258,596 2,324,327 11,831,202 7,452,780 8,198,835 4,793,179 6,547,987

26,136 3,781 1,466,541 1,489,023 894,865 10,624 70,433 26,642 50,185 21,408 27,630

0.72 0.69 0.45 0.67 0.33 0.46 0.10 0.37 0.19 0.06 0.13

0.00 0.00 0.14 0.26 0.42 0.00 0.01 0.00 0.01 0.00 0.00

Source for Tables 8a, 8b, and 8c: Own calculations (dataset).

d75,000 per year on average of the years from 1857 to 1871 to d590,451 in 1873 and d414,743 in 1875). The data have thus been treated in order to raise the PTA for 1873 and 1875 to the average level for 1857 to 1871, which was 0.16 for wheat and 0.45 for rye. In the following sections, the disguised transit-corrected version of the dataset is used, but uncorrected data are available on request.

115

Bilateral Trade in Europe, 1857–1875

6. TACKLING THE PROXIMITY BIAS: CORRECTIONS FOR UNACCOUNTED PARTNER COUNTRY TRANSIT The second part of the transit correction consists in the treatment of the failure of official records to determine correctly countries of origin that were separated from them by more than one common border or direct shipping routes requires comprehensive treatment on an international level. The reconstruction of ‘real’ countries of origin therefore requires going beyond the level of national statistics. To demonstrate the importance of trade that was sent from one country to another while transiting a third one and thus making clear the magnitude of resulting biases, Tables 9a and b depicts the average shares of imports exports in general imports and special exports in general exports for all countries and commodities across years. Two main results can be discerned: The first is that ratios varied widely across countries and commodities: In Belgium and the Netherlands, less than half the merchandise that entered the countries were for home consumption, and only about one third of all goods crossing the border were actually of domestic production. In contrast, in the US transit and re-exports were almost negligible. Bulk commodities like wood, wheat, pig and bar iron, and steel were less likely to cross several

Shares of Special Imports and Exports in General Imports and Exports across Countries (Average for 1857–1875; per cent). Country

Special Imports in General Imports (%)

Special Exports in General Exports (%)

Austria-Hungary Belgium France The Netherlands United Kingdom United States ‘Germany’

60.3 37.0 (41.3) 73.1 50.5 (69.4) 78.1 99.1 63.3 (64.3)

70.9 31.2 (36.1) 79.9 29.1 (57.1) 86.0 89.8 64.5 (65.6)

All (unweighted)

66.0 (69.4)

64.5 (69.3)

All (weighted by special imports)

69.4 (71.2)

73.8 (75.4)

Note: Shares prior to correction for disguised transit in parentheses (if applies). Source: Own calculations from dataset.

116

MARKUS LAMPE

Shares of Special Imports and Exports in General Imports and Exports across Commodity Groups (Average for 1857–1875; Per cent). Commodity Group

Share of Special Imports in General Imports (95.0) (83.5)

Share of Special Exports in General Exports

Wheat Rye Milling products Wines Spirits and liqueurs Hides, skins, and leather Articles of leather and rubber Wood Wool Woollen yarn Woollens and worsteds Cotton yarn Cottons Linen yarn Linens Silk Silk wares Glass and glassware Pig iron Bar iron and steel Ironware

92.9 82.8 95.3 73.0 49.5 73.6 62.8 97.3 74.7 73.5 56.9 62.3 40.1 82.4 80.9 72.0 40.9 60.2 77.6 80.3 63.9

All (unweighted)

71.1 (73.3)

75.9 (78.4)

All (weighted by special imports)

74.5 (76.6)

76.6 (78.0)

(80.9)

(75.9) (78.9) (71.5)

(73.2)

(91.2) (86.2)

83.6 72.0 94.1 78.1 68.6 54.2 83.0 94.6 46.4 82.0 77.0 79.0 83.8 81.2 89.5 43.4 57.2 80.1 76.6 88.8 85.7

(87.6) (73.6)

(66.9)

(48.5) (77.3) (83.8)

(46.0)

(92.7) (91.7)

Notes: see Table 9a.

borders than high-value articles (silk and silk wares, articles of leather and rubber, cottons, woollens) and spirits and liqueurs. Total special export shares are higher than total special import shares because the countries of the sample had a positive balance of trade with the rest of the world in the commodities covered, especially in textiles. The average of the shares of special imports in general imports and special exports in general exports changed slightly between years. It was 70.8% in 1857 and 71.2% in 1875, with a maximum of 74.1% in 1861 and a minimum of 70.8% in 1857 and 1871. In other words, about 26–29% of total imports and total exports were trade in transit and thus subject to the proximity bias, which was unequally distributed among country pairs and commodity groups.

Bilateral Trade in Europe, 1857–1875

117

The attempt to tackle this bias and reconstruct the ‘real countries of origin’ of transited merchandise parts from one basic idea: Every record of importer Ai (e.g. Germany) on its special imports from partner country Bj (e.g. the Netherlands) in a specific commodity (e.g. pig iron) consists of special exports originating in the immediate partner country and transit from third countries (e.g. the UK, see below). The total value of special imports is taken from importer’s records, while the share of each of the two components is calculated from the partner’s export and transit statistics. The amount of trade that originates in the immediate partner country (cf. the last column of Table 9) is ‘direct bilateral special import’ and needs no correction. What remains is trade in transit not originating there, whose ‘real’ countries of origin can be assessed from the immediate partner’s transit statistics. The ‘real special imports’ for every country pair in every commodity group are then found by adding up the direct special imports from the partner country and the imports from the same country that arrive after transit though other countries to whom the original import statistics assigned this trade erroneously. Formally, first for every bilateral special import record of every importer Ai from all direct partner countries BjðIMspc Ai ;Bj Þ, the share of special exports of Bj in its general exports is used to reconstruct Ai’s ‘direct bilateral special imports’ from Bj, i.e. share of bilateral special imports actually originating in gen spc spc Bj, as DIMspc Ai ;Bj ¼ IMAi ;Bj n ðEXBj ;Ai =EXBj ;Ai Þ. Then, the share of transit exports of all other countries Ck through country Bj to in the reported special imports of Ai is from Bj is calculated as T Bj gen spc T Bj IMspc Ai ;Ck ¼ IMAi ;Bj n ðEXCk ;Ai =EXBj ;Ai Þ and named ‘transited bilateral special imports’.74 Of course, the sum of ‘direct bilateral special imports’ and all ‘transited bilateral special imports’ equals, i.e. DIMspc Ai ;Bj þ Pn spc spc T IM ¼ IM . B j k¼1 Ai ;Ck Ai ;Bj Finally, for every dyad of importers Ai and exporters Bj the ‘transited bilateral special imports’ through all Ck’s can be summed up as total indirect Pn T IMspc special imports B j j¼1 Ai ;Ck , which are added to the corresponding ‘direct bilateral special imports’ P to givespc‘corrected bilateral special imports’ spc ¼ DIM þ CIMspc Aj ;Bj ¼C k Ai ;Bj jak T Bj IMAi ;Ck . For countries Ck that do not appear as Bj in the original import records of i (e.g. Switzerland in Pan specific A spc is equivalent to T IM British records), CIMspc B j j¼1 Aj ;Ck Ai ;Ck . For practical application of these considerations, it has to be noted that the dataset only consists of import, transit and export data for seven Ai and Bj, but many more Ck existed and appeared in the statistics. Thus, misassigned trade can be corrected only for transit through these seven countries.

118

MARKUS LAMPE

Although results remain biased for original bilateral special import records from and through other countries, this is of minor importance because the expected amounts of trade conducted between these seven countries in transit through countries can be taken as fairly low. They had the most developed infrastructures in the world, and five of them formed a geographical bloc in Western and Central Europe while for the only geographic outlier, the US, trade received via third countries from countries should have been of limited importance in the sample (territories in the north east might have been supplied in part via Canada in ice-free seasons). Nevertheless, for other countries, especially those outside Europe, statistics might remain biased and cannot be comprehensively treated as systematically assessable detailed statistics are missing. Within Europe uncorrected transit through Switzerland might have caused biases, which are of unknown extent because assessable Swiss statistics are lacking for the period under study. Furthermore, no complete bilateral export records for the Zollverein prior to 1871 were available due to missing export statistics for Hamburg. As Hamburg was Germany’s by far most important export harbour, the problem could not be ignored. I therefore calculated ‘reconstructed Zollverein exports’ for all partners using the separately devised records for imports from Hamburg in the statistics of all countries in the sample. Then, Hamburg (HH) and the Zollverein (ZV) were treated as separate Bj for every importer Ai. DIMspc Ai ;ðZVþHHÞ have been then calculated as ! IMspc EXspc IMspc EXspc Ai ;ZV ZV;Ai Ai ;HH ZV;HH n þ n spc gen spc gen IMspc IMspc Ai ;ZV þ IMAi ;HH EXZV;Ai Ai ;HH þ IMAi ;HH EXZV;HH   spc spc n IMAi ;ZV þ IMAi ;HH where IMspc Ai ;ZV is the recorded bilateral special import of Ai from the Zollverein and Bremen.75 For transit via the Zollverein and Hamburg the formula was n P

EXTCkZV;HH

EXTCkZV;HH k¼1 spc ¼ IM n n T HH IMspc Ai ;Ck Ai ;HH n P EXgen ZV;HH EXTCkZV;HH k¼1

which can be reduced to the product of special imports of Ai from Hamburg and the ratio of transit from Ck to Hamburg via Zollverein to the Zollverein’s total general exports of a commodity to Hamburg. T HH IMspc Ai ;C k

Bilateral Trade in Europe, 1857–1875

119

is added to T ZV IMspc Ai ;Ck , the transit via the Zollverein’s ports in the North and Baltic Sea and Bremen, to perform complete transit corrections for the Zollverein. Except for few cases it was not corrected for transit that passes two countries, as remaining possible biases can be rated as of minor importance in comparison the possible detriment of creating artificial trade flows with very small values. The exceptions were trade of certain goods in the triad Netherlands-Belgium-Germany (mainly on the Rhine-Meuse-Scheldt river system), especially pig iron and steel transited through the Netherlands and then Germany (mostly from the UK), wine, leather articles, silk and silk wares transited through Belgium and than Germany (mostly from France) and leather and glass transited through the Netherlands and then Belgium, as well as some other products in certain years. The criteria applied for necessity of correction was that transits through ‘second transited countries’ from ‘first transited countries’ had to be higher than total special exports of the first transit country. This was done to avoid that transit correction led to an artificial increase in ‘reconstructed special exports’ of non-bordering countries. Transit records were corrected by allotting corresponding gen spc T Bj IMspc Ai ;Ck in proportion to EXBj ;Ai =EXBj ;Ai and the respective transit shares 76 from other countries. An additional caveat had to be taken into account: In fact, Belgium and France tried at least partially to account for ‘real countries of origin and destination’. If properly implemented, this approach would have rendered unnecessary (and even harmful) any type of transit correction. This mix of methods was potentially disturbing for the landward trade of Belgium with Austria-Hungary and Switzerland as well as that of France with the Netherlands and Austria-Hungary. To control their values, the recorded bilateral values for these bilateral trade flows have been compared with reconstructed trade flows from the transit correction as described above. It resulted that recorded bilateral values were markedly lower than bilateral transit recorded in partner countries’ statistics.77 This suggests that only transit on direct river and railway routes and transit with meaningful accompanying documents was recorded in practise. To avoid that trade recorded respectively in Belgium and France and reconstructed by the transit correction would appear as double-counted, the difference of respective Belgian and French special import records from the said countries and corresponding special export records was deducted from the respective 78 This difference was then assigned as bilateral special import to CIMspc Aj ;Ck . all trade partners with transit to Belgium and France, respectively, according

120

MARKUS LAMPE

Pn spc to their share in k¼1 T Bj IMAi ;C k for the special import from those Ck 79 that had to be corrected. While assembling the dataset direct bilateral special imports have been interpolated with bilateral special exports from partner statistics where they were not reported for being merged under ‘other countries’ in importers’ records. This interpolation is limited to small amounts whose omission would not have changed the overall picture substantially, but would have led to less complete trading matrixes at the commodity level. To illustrate the importance of transit correction for certain trade flows, Fig. 1 shows one of the many corrections performed, namely imports of pig iron by the Zollverein/Germany from two sources, the Netherlands and the UK. The series before correction for trade in transit suggest that the Netherlands were at least as important as the UK in providing pig iron to Germany, while the corrected series clearly illustrates that most of the apparently Dutch pig iron originated in fact in the Britain. Because ‘true trade flows’ the quality of the corrected in comparison to the uncorrected data cannot be assessed directly. Alternatively, the performance of both corrected and uncorrected data in empirical analysis can be assessed. To do this, the next section provides basic estimates of the gravity equation

3,000,000 Netherlands uncorrected Netherlands corrected UK uncorrected UK corrected

2,500,000

2,000,000

1,500,000

1,000,000

500,000

0 1857

Fig. 1.

1859

1861

1863

1865

1867

1869

1871

1873

1875

Pig Iron and Scrap Imports into Germany from the UK and the Netherlands. Source: Dataset.

Bilateral Trade in Europe, 1857–1875

121

which show that the corrected data leads to estimates that are better fitted and whose coefficients are more plausible.

7. FINALLY: THE DATASET Owing to detailed and systematic collection, harmonisation and correction of trade values for the most important countries in international trade the resulting dataset is of acceptable quality for empirical analysis. The most common method in research designs that involve bilateral trade flows is the gravity equation. It has been applied for several questions in international economic history of the 19th century, such as the effects of institutions on trade, especially the Latin Monetary Union and the Gold Standard (Flandreau, 2000; Flandreau & Maurel, 2005; Lo´pez-Co´rdova & Meissner, 2003), the bilateral trade treaties following the Cobden-Chevalier Treaty of 1860 (Accominotti & Flandreau, 2008; Lampe, 2008) and empires (Mitchener & Weidenmier, 2008) on trade. Estevardeordal, Frantz, and Taylor (2003) have employed it to study the determinants of international trade growth after 1870, and recently Nikolaus Wolf (2008) used it to shed new light on the degree of economic integration of ‘Germany’ in the late 19th and early 20th century. Except for the studies by Marc Flandreau and co-authors and Lampe (2008), the underlying datasets have 1870 or later as starting points. With the exception of Wolf (2008), who employs transport data on physical volumes of different commodities, and Lampe (2008), who uses the present dataset, all studies rely on aggregate figures only. The present dataset is the first to provide disaggregate figures on a wide range of mainly manufactured goods and, more important, it is the first that is dealing directly with the proximity bias and harmonises foreign trade statistics internationally. In the following, the results of a basic implementation of the gravity equation are presented. To assure comparability with the articles cited above, I use the aggregate figures.80 The application serves two purposes: to assess the possibilities and to clarify the limits of the use of the present dataset for such applications and to trace empirically the performance difference between corrected and uncorrected data, especially the impact of the border bias when using uncorrected figures. In this context, the term ‘uncorrected’ refers to the data prior to corrections for disguised and unrecognised partner country transit, while ‘corrected’ refers to the figures after the treatment described in Sections 5 and 6.81 In the construction of the present dataset the focus has been laid on quality, not on quantity of observations. The core dataset consists of

122

MARKUS LAMPE

42 bilateral cross-Sections (76 countries) and 10 periods in time. An extended version is constructed including imports of the 7 core countries from Denmark, Italy (sum of Sardinia, Tuscany, the Two Sicilies, and the Roman States throughout), Portugal, Russia, Spain, Sweden and Norway, and Switzerland. This augments the number of cross-sections to 88.82 The gravity equation is implemented as by Andrew K. Rose (2004), Olivier Accominotti and Marc Flandreau (2008) and many others. I estimate lnðIMij Þt ¼ b0 þ lnðY i Y j Þt þ b2 lnðDij ÞðtÞ þ Sbm S lnðZ ijm ÞðtÞ þ ijt where unidirectional trade flows (imports of country i from j at time t) are explained by national incomes of importer (Yi) and exporter (Yj) and by the distance that separates them (Dij). S bm Sln(Zijm) is a vector of control variables consisting of dummies for a common border and four types of variables intended to deal with ‘multilateral resistance’ (Anderson & van Wincoop, 2003), e.g. a changing nominal exchange rate, diverging development of producer prices in domestic currency, changing impacts of physical and institutional barriers on trade. As inflation was no important phenomenon in the period under study and exchange rates were more or less fixed by the underlying precious metal standards, I include time-invariant importer fixed effects to model these.83 The only important deviation in exchange rates was presented by the Greenback devaluation during the American Civil War, for which I include a dummy variable which is ‘‘1’’ if the US are an importer or exporter in 1861 to 1865. Furthermore, Table 12 below shows that tariff rates did vary considerablherefore, the export weighted average tariff levels from Table 13 have been included to model tariff levels explicitly.84 All data on values are expressed in 1,000 d Sterling in current prices, distance is measured as great-circle distance between capitals in kilometres. The national income data are the same as in Lampe (2008). Because value data are in current prices, I also include time-period fixed effects. The gravity equations are estimated as Pooled OLS, Panel GLS and as pooled poisson pseudo-maximum likelihood (PPML) regressions, all with robust standard errors. The latter recently have been proposed by Joa˜o Santos Silva and Silvana Tenreyro (2006) as the ‘new workhorse’, because it is less vulnerable than OLS to heteroscedasticity patterns likely present in bilateral trade data and it allows the inclusion of ‘‘0’’ values in the dependent variable.85

Bilateral Trade in Europe, 1857–1875

123

The results of the exercise are shown in Table 10. We see is that despite the inclusion of a number of dummy variables the R2 is always higher for the corrected data, and in the econometrically favoured estimate, PPML core, the difference is about 12% points. Aikaike’s and Schwartz’ Information Criteria always favour the models with corrected data over those with uncorrected data. The coefficient of the border dummy is always higher for the uncorrected than for the corrected data, indicating that the border bias is of visible magnitude. For the OLS core estimate this difference means that bordering countries are estimated to have had 650% higher trade volumes than non-bordering countries, in comparison to 155% according to the corrected figures. For PPML core the difference is lower, equivalent to 94% in comparison to 69% for the corrected figures.86 Furthermore, the impact of distance on trade is always estimated about 15–20% higher for the uncorrected figures, indicating an additional bias. The impact of tariffs on imports cannot be assessed at a confidence interval of 5% for the uncorrected data neither with OLS nor PPML for the core sample. For the extended sample OLS and GLS indicate again a bias of considerable magnitude. The estimates of the ‘old workhorse’, OLS and the proposed new one, PPML, are qualitatively equivalent concerning sign and significance of coefficients. As a general conclusion, the uncorrected data allows estimating meaningful gravity models, but its results are potentially biased in magnitude, especially if it comes to the interpretation of border effects which are considerably magnified by unaccounted transit. Therefore the corrected data is to be preferred in empirical application as well as from the a priori views discussed in previous sections. While the core estimates appear as more precise than the extended ones, data on imports from countries not directly included in the dataset should also be useable.87 For empirical use one main advantage of the dataset lies in the disaggregate figures. Subsamples can be use e.g. to trace the development and geographical distribution of wheat imports into Europe, or to assess the changing fate of textile exporters according to location and fibres used. Although the commodity groups do not represent total trade and understate the participation of exporters of raw materials and other products of tropical and sub-tropical regions in ‘real’ total trade, a comprehensive and new account on the relative importance of countries and commodities in intraEuropean trade can be obtained. To illustrate this and to give a more detailed insight into the dataset, three kinds of information are presented in the following: (a) bilateral trade flows between countries in the form of shares of imports in the totals of all commodity groups covered, (b) the most important providers for each

124

Basic Gravity Model Estimations: Coefficients and p-values (in Parentheses). Variable

Constant ln(YiYj)t ln(Dij) Border

Civil war 1857 1861

OLS Extended

GLS Core

GLS Extended

PPML Core

PPML extended

Uncorr.

Corr.

Uncorr.

Corr.

Uncorr.

Corr.

Uncorr.

Corr.

Uncorr.

Corr.

uncorr.

corr.

11.04 (0.000) 0.99 (0.000) 1.29 (0.000) 2.02 (0.000) 2.97 (0.285) 0.39 (0.428) 0.24 (0.509) 0.15 (0.690)

12.68 (0.000) 1.02 (0.000) 1.12 (0.000) 0.94 (0.000) 3.49 (0.028) 0.02 (0.954) 0.03 (0.882) 0.14 (0.535)

10.92 (0.000) 0.94 (0.000) 1.07 (0.000) 2.59 (0.000) 4.19 (0.030) 0.29 (0.416) 0.33 (0.229) 0.14 (0.635)

9.19 (0.000) 0.86 (0.000) 0.95 (0.000) 1.52 (0.000) 3.27 (0.024) 0.05 (0.858) 0.07 (0.733) 0.68 (0.765)

9.41 (0.172) 0.91 (0.000) 1.24 (0.005) 1.98 (0.001) 3.08 (0.001) 0.13 (0.583) 0.25 (0.136) 0.30 (0.074)

4.20 (0.324) 0.66 (0.000) 0.97 (0.002) 0.82 (0.013) 3.03 (0.000) 0.06 (0.756) 0.08 (0.482) 0.20 (0.127)

7.77 (0.100) 0.81 (0.000) 1.04 (0.005) 2.77 (0.000) 3.85 (0.000) 0.15 (0.443) 0.29 (0.024) 0.15 (0.223)

2.69 (0.491) 0.56 (0.000) 0.78 (0.025) 1.71 (0.000) 2.86 (0.000) 0.14 (0.335) 0.10 (0.241) 0.12 (0.174)

8.21 (0.000) 0.82 (0.000) 0.88 (0.000) 0.66 (0.000) 2.82 (0.114) 0.16 (0.669) 0.05 (0.861) 0.01 (0.973)

21.39 (0.000) 1.25 (0.000) 0.68 (0.000) 0.53 (0.000) 3.44 (0.020) 0.11 (0.714) 0.07 (0.776) 0.01 (0.979)

9.43 (0.000) 0.83 (0.000) 0.72 (0.000) 0.87 (0.000) 2.88 (0.120) 0.07 (0.843) 0.00 (0.991) 0.04 (0.836)

12.06 (0.000) 0.86 (0.000) 0.46 (0.000) 0.70 (0.000) 3.07 (0.051) 0.08 (0.781) 0.01 (0.977) 0.11 (0.594)

MARKUS LAMPE

ln(1þtariff)

OLS Core

1865 1867 1869 1871 1873 1875 N (Pseudo-)R2 Akaike-IC Schwartz’ BIC

0.07 (0.859) 0.37 (0.335) 0.27 (0.491) 0.00 (0.993) 0.18 (0.654) 0.04 (0.929) 0.18 (0.711)

0.08 (0.707) 0.41 (0.064) 0.24 (0.299) 0.20 (0.383) 0.17 (0.450) 0.06 (0.779) 0.01 (0.978)

0.08 (0.783) 0.33 (0.314) 0.25 (0.450) 0.02 (0.946) 0.10 (0.758) 0.16 (0.606) 0.28 (0.437)

0.11 (0.603) 0.28 (0.227) 0.15 (0.515) 0.07 (0.754) 0.28 (0.207) 0.28 (0.186) 0.05 (0.830)

0.10 (0.605) 0.24 (0.170) 0.23 (0.275) 0.00 (0.982) 0.23 (0.343) 0.06 (0.852) 0.13 (0.717)

0.03 (0.819) 0.23 (0.050) 0.04 (0.737) 0.04 (0.770) 0.44 (0.002) 0.40 (0.017) 0.34 (0.042)

0.08 (0.548) 0.29 (0.072) 0.18 (0.272) 0.02 (0.885) 0.22 (0.226) 0.29 (0.186) 0.15 (0.545)

0.02 (0.800) 0.15 (0.112) 0.00 (0.991) 0.11 (0.298) 0.48 (0.000) 0.56 (0.000) 0.33 (0.010)

0.13 (0.564) 0.11 (0.615) 0.06 (0.811) 0.04 (0.879) 0.21 (0.424) 0.14 (0.593) 0.08 (0.756)

0.27 (0.201) 0.33 (0.119) 0.22 (0.311) 0.25 (0.240) 0.12 (0.582) 0.32 (0.128) 0.40 (0.053)

0.12 (0.583) 0.06 (0.787) 0.12 (0.594) 0.05 (0.842) 0.23 (0.354) 0.16 (0.504) 0.07 (0.743)

0.11 (0.572) 0.06 (0.754) 0.08 (0.713) 0.01 (0.943) 0.19 (0.379) 0.10 (0.621) 0.01 (0.947)

416 0.57 1676 1761

420 0.66 1258 1343

826 0.55 3488 3588

880 0.60 3164 3263

416 0.57 n/a n/a

420 0.63 n/a n/a

826 0.54 n/a n/a

880 0.58 n/a n/a

420 0.59 933360 933445

420 0.71 518606 518691

880 0.61 1571875 1571975

880 0.63 1184820 1184920

Bilateral Trade in Europe, 1857–1875

1863

Note: Importer-specific effects not reported. Source: Own calculations with STATA, version 10.

125

126

MARKUS LAMPE

commodity group to the sum of all European countries covered in the sample, and (c) corresponding ad valorem equivalent average tariff rates for all commodity groups. Of course, the dataset contains much more detailed data on the commodity groups, which would exceed the limits of the present study. Tables 11a–g present the trade relative importance of selected European and non-European countries (mainly the US) as providers of exports for other European countries and the US in the total of the commodity sample. The figures also give detailed insights into the magnitude and direction of the proximity bias estimated above, especially when dealing with trade volumes with remote countries and centres of commerce (UK, the Netherlands, and Belgium). Table 12 displays the main providers of imports into the six European countries of the sample for every commodity group, i.e. the share of exporters in the sum of imports of the UK, France, ‘Germany’, Belgium, the Netherlands, and Austria-Hungary.88 The table makes clear that only few countries covered the full range of exports. All of these countries are in the core sample, but even for them the share in individual commodity groups is far from uniform. The countries not individually covered in the dataset – maybe with the exception of Switzerland – appear as providers of a small number of goods only: Scandinavia most notably exported wood and semimanufactured iron. Italy was mainly an exporter of silk, leather and wine (to some degree), while Spain and Portugal almost exclusively exported wines. From today’s point of view it might surprise that the export portfolio of the US was much more similar to that of Russia and Turkey than to the industrialised core.89 The main suggestion for the economic history of trade in the 19th century is thus that foreign trade should be treated more at the commodity level, as total exports for many countries were driven by few commodities, whose production and commerce depended on specific conditions and developments. It might be misleading to look for uniform developments of trade across countries. This picture is reinforced by Table 13 which contains the average tariff rate for each commodity group prevailing in every country in each year covered by the sample, i.e. the ratio of import duties recollected to (corrected) import values. For the use of this table, three remarks have to be made: Firstly, as the overwhelming majority of duties were specific, ad valorem rates not only changed with tariff reforms, but also with price fluctuations. Secondly, the commodity groups consisted of various items and internal compositional changes influenced on the average tariff rate. Thirdly, the tariff rates quoted are averages that neither capture prohibitions nor account for the exact

127

Bilateral Trade in Europe, 1857–1875

Shares of Imports into the UK, 1857–1875. Totals of Commodity Groups in Sample; Per cent. First Line: Uncorrected Figures, Second Line Corrected Figures. Partner

France

1857 1859 1861 1863 1865 1867 1869 1871 1873 1875 Unweighted Mean

11.0 8.3 ‘Germany’ 11.5 13.8 Belgium 2.3 1.2 The Netherlands 2.0 0.5 Austria-Hungary 0.3 0.6 Russia 6.8 6.8 Scandinavia 5.5 5.5 Switzerland 0.0 1.0 Spain 2.2 2.2 Portugal 2.1 2.1 Italy 0.4 0.7 Turkey 0.6 0.7 USA 7.5 7.1 British North 8.7 America 8.8 British India 3.2 3.2 Australia 6.7 6.7

20.7 18.0 10.2 11.1 3.2 1.7 2.3 0.8 0.3 1.6 8.0 8.0 6.5 6.5 0.0 1.6 2.1 2.1 1.5 1.5 0.3 0.8 0.7 0.8 1.2 1.2 8.7 8.7 3.3 3.3 8.3 8.3

16.2 12.7 10.5 11.0 2.1 0.8 1.1 0.2 1.1 1.3 8.1 9.2 5.2 5.3 0.0 2.9 3.9 3.9 1.1 1.1 0.3 0.5 1.1 1.3 17.7 17.6 11.0 11.0 2.5 2.5 6.0 6.0

20.4 16.6 9.6 10.0 3.7 1.9 1.6 0.3 0.3 0.5 7.7 8.5 5.9 5.9 0.0 3.1 2.9 2.9 1.4 1.4 0.2 0.5 0.8 0.8 10.9 10.8 10.6 10.6 2.9 2.8 6.8 6.8

26.5 21.4 9.8 11.1 4.6 2.2 2.0 0.4 0.7 1.0 11.7 12.5 6.7 6.7 0.0 3.3 2.4 2.4 1.6 1.6 0.1 0.5 0.7 0.7 1.8 1.7 7.4 7.4 2.3 2.3 8.9 8.9

20.9 16.0 9.6 11.5 4.0 1.7 2.0 0.3 0.2 1.2 15.5 15.9 5.2 5.2 0.0 2.7 2.8 2.9 1.1 1.1 0.4 0.6 2.1 2.6 5.5 5.3 5.2 5.3 1.7 1.7 9.2 9.2

23.0 18.7 8.1 12.5 6.2 2.6 2.2 0.5 0.9 1.3 9.8 10.0 5.7 5.7 0.0 3.7 2.6 2.6 1.1 1.1 0.3 0.4 1.5 1.7 11.3 11.0 7.4 7.7 1.9 1.9 7.5 7.5

16.3 15.5 7.0 11.5 6.6 2.2 3.1 0.6 0.8 1.3 13.6 13.9 5.4 5.5 0.0 1.3 2.7 2.7 1.6 1.6 0.4 0.5 1.5 1.6 12.0 11.6 6.9 7.3 3.0 3.0 7.8 7.8

20.3 17.1 5.7 8.2 5.0 2.0 2.1 0.6 0.4 0.7 9.7 9.9 7.2 7.2 0.0 3.0 3.4 3.4 1.5 1.5 0.4 0.5 0.6 0.6 15.6 15.2 7.7 8.0 3.2 3.2 7.9 7.9

22.5 19.0 7.4 10.4 5.7 2.4 2.0 0.6 0.4 0.9 9.1 9.5 5.6 5.7 0.0 3.2 2.2 2.2 1.4 1.4 0.3 0.5 0.8 0.9 16.3 15.5 6.3 7.1 3.3 3.3 8.5 8.5

19.8 16.3 8.9 11.1 4.3 1.9 2.0 0.5 0.5 1.0 10.0 10.4 5.9 5.9 0.0 2.6 2.7 2.7 1.4 1.4 0.3 0.5 1.0 1.2 10.0 9.7 8.0 8.2 2.7 2.7 7.8 7.8

Note: Scandinavia: Denmark, Sweden, Norway. Figures for Turkey include Moldavia and Wallachia. Italy includes the Roman States. Spain includes the Canary Islands. Portugal includes Azores and Madeira. British India in some years includes Ceylon, Singapore and Eastern Straits Settlements.

128

MARKUS LAMPE

Shares of Imports into France, 1857–1875. Totals of Commodity Groups in Sample; Per cent. First Line: Uncorrected Figures, Second Line Corrected Figures. Partner

UK

1857 1859 1861 1863 1865 1867 1869 1871 1873 1875 Unweighted Mean

26.9 12.0 ‘Germany’ 8.7 10.5 Belgium 5.8 3.0 The Netherlands 0.5 1.0 Austria-Hungary 1.3 1.2 Russia 6.4 6.7 Scandinavia 4.9 5.0 Switzerland 3.5 3.6 Spain 7.1 7.1 Portugal 0.0 0.1 Italy 9.7 9.8 Turkey 11.1 11.2 USA 2.1 2.3 Algeria 2.4 2.4

30.1 9.3 9.1 7.8 4.8 3.1 1.6 1.5 0.4 3.0 5.0 5.4 7.4 7.5 5.2 5.2 2.7 2.7 0.0 0.1 10.3 10.4 9.7 10.0 0.1 0.1 2.7 2.7

26.4 12.3 9.6 8.8 6.4 3.7 0.4 0.7 0.5 1.4 10.7 12.4 5.4 5.7 3.2 3.2 3.3 3.4 0.1 0.1 6.4 6.4 8.2 8.4 8.1 10.4 3.1 3.1

29.6 12.9 6.5 6.1 6.6 4.1 0.8 0.6 0.8 2.1 5.6 6.0 5.6 5.8 4.1 4.1 1.3 1.3 0.1 0.1 10.7 10.7 11.4 11.7 0.6 0.9 2.8 2.8

27.3 11.1 6.1 4.9 6.4 4.2 0.8 0.7 0.2 1.9 4.7 5.5 6.9 7.1 5.1 5.1 1.3 1.4 0.2 0.3 10.8 10.9 9.2 9.3 0.1 0.2 3.1 3.1

20.0 9.4 8.5 8.2 9.6 2.9 0.6 0.6 0.7 1.5 8.4 9.4 5.4 5.7 4.5 4.7 2.1 2.3 0.1 0.1 11.4 11.9 11.4 12.1 0.2 0.3 1.6 1.7

23.8 11.9 7.1 7.5 5.4 3.0 0.3 0.4 0.7 0.4 4.6 4.9 7.5 7.7 6.2 6.2 2.4 2.4 0.1 0.1 14.3 14.5 8.5 8.6 0.2 0.3 0.9 0.9

26.4 15.8 3.2 5.8 10.3 2.7 0.4 0.7 0.2 0.3 9.4 10.1 2.2 2.4 4.2 4.3 2.2 2.2 0.2 0.3 14.9 15.0 8.6 9.0 2.2 2.9 2.0 2.0

19.0 10.3 7.9 6.7 8.5 2.9 0.8 0.4 0.5 0.4 5.7 6.1 4.3 4.5 4.0 6.1 4.3 4.3 0.4 0.4 11.5 11.6 9.1 9.3 2.1 2.5 3.8 3.8

22.6 14.2 10.2 9.7 9.4 3.0 0.5 0.7 0.6 0.4 7.7 8.1 3.6 3.8 4.1 6.2 1.8 1.8 0.1 0.1 10.4 10.5 5.5 5.6 1.0 1.2 2.8 2.8

25.2 11.9 7.7 7.6 7.3 3.3 0.7 0.7 0.6 1.3 6.8 7.5 5.3 5.5 4.4 4.9 2.8 2.9 0.1 0.2 11.1 11.2 9.3 9.5 1.7 2.1 2.5 2.5

Note: Scandinavia: Denmark, Sweden and Norway. Figures for Turkey include Moldavia and Wallachia. Italy includes the Roman States.

impact of discriminating tariff schemes resulting from commercial treaties. Nevertheless, as the former were practically abolished in the period and the latter extensively generalised due to MFN clauses, the bias should be rather small. Furthermore, duties on spirits based on duties recollected are not strictly comparable, because in some countries tariffs included equivalents to

129

Bilateral Trade in Europe, 1857–1875

Shares of Imports into the Zollverein/German Customs Area, 1857– 1875. Totals of Commodity Groups in Sample; Per cent. First Line: Uncorrected Figures, Second Line Corrected Figures. Partner

1857 1859 1861 1863 1865 1867 1869 1871 1873 1875 Unweighted Mean

UK

28.2 37.8 5.4 7.4 11.6 3.4 12.6 1.1 13.7 12.8 15.9 16.8 1.1 1.1 7.0 7.0 0.2 0.3 0.0 0.1 0.0 0.1 0.0 0.1 0.5 0.6

France Belgium The Netherlands Austria-Hungary Russia Scandinavia Switzerland Spain Portugal Italy Turkey USA

20.8 30.6 4.8 7.6 10.7 3.8 14.1 1.6 13.5 12.3 18.7 19.7 1.5 1.4 6.2 6.3 0.0 0.1 0.0 0.0 0.0 0.1 0.0 0.1 0.3 0.3

19.8 28.6 3.7 5.3 9.1 2.5 13.0 2.1 26.0 25.0 13.8 14.9 1.0 0.8 4.8 5.0 0.1 0.1 0.0 0.0 0.0 0.1 0.0 0.1 0.5 0.8

21.9 29.4 4.6 7.4 10.7 3.1 13.6 3.4 20.6 19.3 13.3 14.5 0.9 0.7 6.5 6.7 0.0 0.1 0.0 0.0 0.0 0.1 0.0 0.2 0.3 0.6

20.6 27.3 5.3 7.9 11.1 2.7 13.0 4.4 22.8 21.6 14.0 14.2 0.9 0.7 4.9 5.1 0.1 0.1 0.0 0.1 0.0 0.1 0.0 0.2 0.1 0.3

14.6 19.3 5.7 7.8 9.4 1.7 11.2 2.5 31.9 27.7 14.2 17.3 1.0 0.4 6.9 7.4 0.1 0.1 0.0 0.0 0.0 0.2 0.0 0.6 0.2 0.3

15.2 21.0 5.3 9.3 14.2 3.0 12.3 3.6 26.0 21.0 12.1 12.3 1.3 1.3 7.9 7.7 0.1 0.2 0.0 0.1 0.0 0.2 0.1 0.4 0.2 0.3

17.3 23.9 2.2 7.6 15.0 4.3 16.1 3.4 20.8 17.8 11.9 14.3 1.2 1.3 7.6 7.9 0.0 0.1 0.0 0.1 0.0 0.5 0.0 0.7 0.2 0.4

13.8 21.2 9.0 13.4 14.9 4.1 16.7 3.0 12.4 9.9 17.6 20.1 1.2 1.5 7.0 7.2 0.1 0.3 0.1 0.1 0.0 1.9 0.0 0.5 0.8 1.5

14.7 19.6 9.6 13.6 14.8 4.1 15.0 3.4 15.7 11.5 13.9 15.8 0.7 0.8 7.8 8.2 0.1 0.2 0.1 0.1 0.0 1.6 0.0 1.2 0.7 1.5

18.7 25.9 5.5 8.7 12.2 3.3 13.8 2.8 20.3 17.9 14.5 16.0 1.1 1.0 6.7 6.9 0.1 0.2 0.0 0.1 0.0 0.5 0.0 0.4 0.4 0.7

Note: Scandinavia: Denmark, Sweden, Norway, and Schleswig and Holstein. Figures for Turkey include Moldavia and Wallachia. Italy includes the Roman States. USA might contain a small share of imports from British North America, as the Zollverein’s port statistics only contain information on imports from North America (but most imports were conducted via Hamburg and Bremen, who distinguished US from Canadian ports). Some European trade was not accounted for as it appeared as entered from fairies, free entrepoˆts and via postal service.

national excises (most prominently the UK, but also in the Zollverein), while in others (the Netherlands and France) these were charge additionally, and in some (Belgium and Austria-Hungary) practise varied across years. The result of a tentative reconstruction of comparable figures excluding excises and equivalents for domestic production taxes is reported as ‘Spirits and liqueurs alternative’ in Table 13.90 Reported tariff rates in France for wheat,

130

MARKUS LAMPE

Shares of Imports into Belgium, 1857–1875. Totals of Commodity Groups in Sample; Per cent. First Line: Uncorrected Figures, Second Line Corrected Figures. Partner

1857 1859 1861 1863 1865 1867 1869 1871 1873 1875 Unweighted Mean

UK

21.3 12.8 26.7 25.8 15.5 15.6 6.5 4.6 0.1 2.2 3.2 4.0 2.1 2.4 0.6 0.9 0.0 0.2 0.0 0.2 0.3 0.5 0.1 0.2 1.7 1.5

France ‘Germany’ The Netherlands Austria-Hungary Russia Scandinavia Switzerland Spain Portugal Italy Turkey USA

18.4 12.1 38.4 40.1 11.2 11.7 8.4 4.4 0.0 1.8 1.9 2.7 2.9 3.3 0.3 0.9 0.0 0.5 0.0 0.1 0.1 0.4 0.0 0.7 0.2 0.3

16.6 8.4 22.2 25.3 18.8 19.6 7.4 4.8 0.1 1.3 5.6 8.1 5.5 6.3 0.4 0.5 0.0 0.2 0.0 0.1 0.0 0.1 0.7 0.9 3.9 4.6

14.9 11.9 34.7 36.7 18.3 17.4 6.3 4.0 0.5 3.7 2.3 3.6 5.4 6.2 0.9 1.8 0.0 0.1 0.0 0.0 0.1 0.4 0.8 1.1 2.9 3.5

15.9 13.7 31.2 33.4 9.2 9.1 3.9 2.6 0.1 0.7 2.5 3.1 4.0 4.4 2.3 2.7 0.0 0.1 0.0 0.1 0.0 0.1 0.1 0.1 0.4 0.2

17.2 11.9 29.3 30.0 15.3 13.8 5.0 2.7 0.3 1.5 5.5 7.3 2.9 3.2 0.7 1.6 0.0 0.2 0.0 0.0 0.1 0.1 1.4 2.1 0.6 0.3

17.7 11.2 31.1 33.9 14.4 13.9 4.4 2.8 0.2 0.8 0.9 1.6 3.4 3.9 0.9 1.9 0.0 0.1 0.0 0.1 0.0 0.1 0.7 0.8 0.4 0.4

19.3 12.4 23.3 24.8 21.6 19.1 4.5 2.3 0.3 1.7 3.5 5.3 5.7 6.4 0.1 2.1 0.1 0.1 0.0 0.1 0.0 0.1 1.5 2.0 3.0 3.2

21.7 13.2 22.7 24.9 10.5 10.1 3.8 2.7 0.0 0.5 6.1 6.6 4.4 5.0 0.2 0.8 1.0 1.2 0.2 0.3 0.1 0.2 0.1 0.3 5.7 5.8

23.5 12.0 30.5 33.1 12.8 12.5 3.8 3.1 0.0 0.5 6.3 6.4 4.8 5.6 0.2 0.8 0.1 0.2 0.1 0.1 0.0 0.1 0.8 0.9 3.6 3.2

18.7 12.0 29.0 30.8 14.8 14.3 5.4 3.4 0.2 1.5 3.8 4.9 4.1 4.7 0.7 1.4 0.1 0.3 0.0 0.1 0.1 0.2 0.6 0.9 2.2 2.3

Note: Scandinavia: Denmark, Sweden, and Norway. Figures for Turkey include Moldavia and Wallachia. Italy includes the Roman States.

pig iron and steel include duty free imports under the regime of temporary admission for re-export manufacture; actual tariff levels for imports administratively declared for home consumption were higher.91 Reported rates for imports into the US do not include duty-free imports of raw materials from British North America under the Reciprocity Treaty of 1854 (wheat, rye, meals and flours, hides and skins, wood, and wool), as European exporters and the duties they faced form the focus here; average tariffs including these imports naturally were lower.92 At the end of the table we find two aggregate measures. The first is the weighted average of the preceded commodity-group rates, where the

131

Bilateral Trade in Europe, 1857–1875

Shares of Imports into the Netherlands, 1857–1875. Totals of Commodity Groups in Sample; Per cent. First Line: Uncorrected Figures, Second Line Corrected Figures. Partner

UK

1857 1859 1861 1863 1865 1867 1869 1871 1873 1875 Unweighted Mean

30.7 26.8 France 7.2 9.7 ‘Germany’ 33.4 30.3 Belgium 10.8 7.2 Austria-Hungary 0.0 0.7 Russia 10.4 12.9 Scandinavia 4.2 5.2 Switzerland 0.0 2.3 Spain 0.2 0.3 Portugal 0.1 0.1 Italy 0.3 0.2 Turkey 0.2 0.3 USA 0.2 0.3 Java 1.0 0.3

29.9 26.9 10.6 12.9 32.8 30.6 13.1 6.7 0.0 1.1 6.3 8.1 4.6 5.4 0.0 3.8 0.2 0.4 0.1 0.1 0.0 0.1 0.3 0.7 0.1 0.1 0.7 0.1

31.0 27.3 6.6 8.5 33.0 31.5 10.4 5.8 0.0 0.5 8.4 11.9 4.6 5.2 0.0 2.6 0.1 0.2 0.0 0.1 0.0 0.0 0.7 0.9 1.6 1.6 0.8 0.3

41.8 31.4 6.9 10.3 28.3 26.8 11.9 6.7 0.1 0.9 2.4 9.2 4.1 5.8 0.0 1.8 0.1 0.2 0.0 0.0 0.0 0.1 0.8 1.2 1.1 1.7 0.6 0.1

45.7 34.0 5.7 10.6 23.8 26.3 13.5 8.1 0.0 0.9 1.6 4.2 4.9 7.0 0.0 1.9 0.1 0.1 0.0 0.0 0.0 0.1 1.2 1.8 0.0 0.1 0.6 0.3

45.0 33.0 5.7 9.0 23.9 26.4 13.0 6.8 0.0 1.8 7.7 12.5 3.2 4.5 0.0 1.7 0.1 0.2 0.0 0.0 0.0 0.0 0.3 0.7 0.1 0.1 0.5 0.1

46.2 31.9 4.7 12.4 26.0 30.5 14.3 6.4 0.1 0.9 2.4 4.2 3.5 5.2 0.0 1.3 0.1 0.2 0.0 0.0 0.0 0.0 1.3 3.0 0.1 0.2 0.7 0.1

46.1 33.1 5.2 9.9 22.4 23.6 11.3 5.5 0.2 1.0 7.9 12.2 2.7 3.7 0.0 2.7 0.3 0.4 0.0 0.1 0.1 0.1 1.6 2.5 0.8 1.6 1.3 0.6

50.1 32.1 4.8 10.8 17.3 21.5 13.3 6.6 0.0 0.8 5.7 9.5 1.7 2.6 0.0 1.2 0.5 0.9 0.1 0.2 0.2 0.2 0.6 1.0 2.7 4.9 1.7 1.0

41.3 28.6 4.7 10.1 26.1 29.1 14.4 5.7 0.0 0.5 7.4 11.8 1.5 2.2 0.0 1.6 0.4 0.6 0.0 0.1 0.0 0.1 0.9 1.4 1.4 2.6 1.7 1.0

40.8 30.5 6.2 10.4 26.7 27.7 12.6 6.5 0.0 0.9 6.0 9.7 3.5 4.7 0.0 2.1 0.2 0.4 0.0 0.1 0.1 0.1 0.8 1.4 0.8 1.3 1.0 0.4

Note: Scandinavia: Denmark, Sweden, and Norway. Figures for Turkey include Moldavia and Wallachia. Italy includes the Roman States.

unweighted averages of every country’s corrected 1865 export shares from Table 3 serve as weights. The alternative estimates for tariffs and liqueurs have been used in the calculation of these figures. The second is the classical and much-criticised ‘average tariff’, which was calculated from total imports and total duties recollected as quoted in official statistics or the most accepted revisions. The data illustrates well that tariff reductions during the ‘European free-trade era’ that is conventionally dated from 1860 to the late

132

MARKUS LAMPE

Shares of Imports into Austria-Hungary, 1857–1875. Totals of Commodity Groups in Sample Per cent. First Line: Uncorrected Figures, Second Line Corrected Figures. Partner

UK

1857 1859 1861 1863 1865 1867 1869 1871 1873 1875 Unweighted Mean

2.3 21.7 France 0.8 3.7 ‘Germany’ 63.6 28.4 Belgium 1.1 1.8 The Netherlands 0.1 0.2 Russia 6.9 7.3 Scandinavia 0.0 0.1 Switzerland 2.3 4.2 Spain 0.0 0.0 Portugal 0.0 0.0 Italy 8.8 8.8 Turkey 12.8 12.8 USA 0.4 0.6 Egypt 0.3 0.3

3.9 16.9 0.7 3.1 53.6 27.0 2.0 2.7 0.1 0.2 12.4 12.8 0.0 0.1 0.7 2.7 0.0 0.0 0.0 0.0 6.4 6.4 18.2 18.2 0.0 0.2 0.3 0.3

1.8 22.6 0.3 3.0 60.0 25.1 0.7 1.4 0.1 0.3 7.1 7.6 0.0 0.1 0.4 2.4 0.0 0.0 0.0 0.0 4.8 4.8 23.2 23.2 0.1 0.2 1.0 1.0

1.4 18.9 0.1 5.0 66.1 30.6 0.6 1.3 0.1 0.2 6.0 6.1 0.0 0.2 0.4 4.0 0.0 0.0 0.0 0.0 6.4 6.4 17.9 17.9 0.0 0.1 0.4 0.4

1.1 20.2 0.2 3.4 68.2 33.0 0.8 1.9 0.1 0.4 4.2 4.6 0.0 0.1 0.5 3.7 0.0 0.0 0.0 0.0 5.0 5.0 19.0 19.0 0.0 0.0 0.4 0.4

0.5 3.8 3.5 2.0 0.5 27.0 26.5 23.8 14.0 14.7 0.1 0.7 0.2 0.2 0.1 2.3 4.4 1.9 3.1 2.6 69.3 75.8 76.0 75.4 80.8 28.2 32.4 33.5 38.2 41.6 0.0 0.2 0.4 0.2 0.1 1.1 4.0 6.7 5.4 5.3 0.0 0.1 0.0 0.0 0.0 0.5 0.5 0.9 0.7 1.4 2.4 3.4 2.4 6.4 2.7 2.7 4.5 2.5 6.6 3.0 0.0 0.0 0.0 0.0 0.0 0.1 0.3 0.2 0.2 0.1 0.5 0.3 0.3 0.2 0.5 2.5 2.0 4.6 8.0 7.2 0.0 0.1 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 9.3 6.8 7.9 6.7 6.2 9.3 6.8 7.9 6.7 6.2 17.3 8.4 8.1 7.6 8.0 17.3 8.4 8.1 7.6 8.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 0.4 0.4 0.2 0.1 0.1 0.3 0.4 0.2 0.1 0.1 0.3 0.4

2.1 20.6 0.3 3.3 68.9 31.8 0.6 3.2 0.1 0.5 5.4 5.8 0.0 0.2 0.6 4.1 0.0 0.0 0.0 0.0 6.8 6.8 14.0 14.0 0.0 0.2 0.4 0.4

Note: Scandinavia: Denmark, Sweden, and Norway. Figures for Turkey include Moldavia and Wallachia. Italy includes the Roman States.

1870s were most prominent in manufacturing, as tariffs on industrial raw materials and foodstuffs generally were low already at the beginning of that period, mostly due to previous unilateral tariff reforms. Tariff levels of the US during and after the Civil War illustrate that it was far from participating in the European liberalisations, neither unilaterally nor by its integration into the institutional framework of bilateral MFN treaties that spread all over Europe during that period.

133

Bilateral Trade in Europe, 1857–1875

Shares of Imports into the United States, 1857–1875. Totals of Commodity Groups in Sample; Per cent. First Line: Uncorrected Figures, Second Line Corrected Figures. Partner

UK

1857 1859 1861 1863 1865 1867 1869 1871 1873 1875 Unweighted Mean

55.3 48.8 France 19.2 18.7 ‘Germany’ 5.4 6.2 Belgium 1.6 2.9 The Netherlands 0.7 0.8 Austria-Hungary 0.0 0.5 Russia 0.7 0.8 Scandinavia 0.5 0.5 Switzerland 0.0 3.3 Spain 0.4 0.4 Portugal 0.1 0.2 Italy 0.1 0.2 Turkey 0.3 0.4 British North (6.1) America (6.1)

56.4 47.0 16.8 19.4 6.0 6.5 0.8 2.0 0.9 1.0 0.1 0.7 0.4 0.4 0.3 0.4 0.0 2.5 0.3 0.4 0.1 0.1 0.1 0.1 0.3 0.3 (5.7) (5.7)

56.6 46.6 12.4 15.2 5.9 6.1 0.9 2.0 0.9 1.2 0.0 0.8 0.4 0.7 0.5 0.6 0.0 1.8 0.3 0.4 0.1 0.1 0.1 0.1 0.4 0.4 11.9 11.9

61.3 47.7 4.5 12.5 4.5 4.5 0.8 1.5 0.4 0.4 0.0 0.1 0.6 0.8 0.3 0.5 0.0 0.1 0.3 0.4 0.1 0.1 0.1 0.1 0.5 0.6 15.6 15.6

67.6 55.5 5.8 10.7 5.9 6.5 0.9 2.1 0.4 0.7 0.0 0.1 0.5 0.6 0.4 0.5 0.0 0.7 0.3 0.3 0.0 0.1 0.0 0.0 0.1 0.1 11.0 11.0

63.2 53.9 10.4 13.0 6.5 6.4 1.3 2.2 0.3 0.7 0.0 0.1 0.9 1.0 0.7 0.8 0.0 1.1 0.2 0.5 0.0 0.1 0.1 0.1 0.1 0.2 7.0 7.0

61.4 66.1 57.1 48.2 53.8 56.0 50.1 39.7 13.0 9.0 11.3 22.4 13.4 12.1 12.2 20.5 6.4 6.7 10.7 9.9 6.3 5.2 9.2 10.1 1.1 1.3 1.3 2.1 1.7 2.1 1.5 1.8 0.3 0.4 0.4 0.5 1.1 1.2 0.7 0.9 0.0 0.0 0.0 0.1 0.4 0.5 0.5 0.6 0.6 0.4 0.6 0.4 0.6 0.5 0.7 0.5 0.6 0.8 0.9 0.2 0.8 1.0 1.3 0.6 0.0 0.0 0.0 0.0 2.7 1.9 3.2 4.3 0.3 0.3 0.3 0.3 0.5 0.4 0.4 0.4 0.0 0.0 0.0 0.1 0.1 0.1 0.1 0.1 0.0 0.1 0.1 0.0 0.1 0.1 0.1 0.4 0.2 0.1 0.1 0.0 0.3 0.5 0.2 0.4 6.5 5.1 5.6 3.7 6.5 5.1 5.6 3.7

59.3 49.9 12.5 14.8 6.8 6.7 1.2 2.0 0.5 0.9 0.0 0.4 0.5 0.7 0.5 0.7 0.0 2.2 0.3 0.4 0.1 0.1 0.1 0.1 0.2 0.3 7.8 7.8

Note: Scandinavia: Denmark, Sweden, and Norway. Figures for Turkey include Moldavia and Wallachia. Italy includes the Roman States. British North America includes all British Possessions, Provinces and Dominions in North America – imports under Reciprocity Treaty for 1857 and 1859 have been conservatively reconstructed and are probably too small. Russia includes Alaska. Spain includes the Canaries. Portugal includes Azores and Madeira, and in some years Cap Verde.

Exporters’ Mean Shares in the Sum of Imports of the UK, France, ‘Germany’, Belgium, the Netherlands, and AustriaHungary for every Commodity Group, 1857–1875 (Unweighted Average of Per cent Shares in each Year, Corrected Figures). Commodity Group Exporter

UK France ‘Germany’ Belgium The Netherlands Austria-Hungary Russia Scandinavia Switzerland Spain Portugal Italy Turkey USA

Wheat Rye Milling Wines Product

1.2 4.5 15.6 0.6 1.0 6.4 27.9 2.6 0.4 0.7 0.1 1.2 7.2 18.2

0.1 9.9 7.5 2.0 2.0 13.4 52.3 2.0 0.2 0.1 0.0 0.4 5.2 1.1

0.9 28.7 16.9 0.7 0.9 11.8 1.6 1.2 0.4 2.5 0.0 0.7 0.5 22.8

0.1 42.3 4.4 0.1 0.1 1.5 0.0 0.0 0.4 31.0 13.3 2.6 0.4 0.0

Spirits and Liqueurs

2.2 49.4 8.3 0.5 2.2 0.1 0.6 0.2 0.3 0.1 0.0 0.1 0.0 1.1

Hides, Articles Wood Wool Woollen Woollens Cotton Cottons Linen Linens Silk Silk Glass Pig Bar Iron- Sum Yarn yarn wares and Iron Iron ware Yarn and of Skins, and GlassWorsteds Leather and Steel ware and Leather Rubber 7.7 6.5 10.1 2.1 1.5 1.8 2.3 0.8 2.3 1.0 0.1 3.5 5.3 3.3

15.8 50.8 12.7 3.3 0.3 4.0 0.2 0.4 1.4 0.0 0.0 2.0 0.0 1.7

0.1 1.7 11.1 0.8 0.3 4.2 24.9 28.7 2.4 0.0 0.0 0.7 0.1 2.2

4.7 4.0 6.4 1.6 2.6 7.6 6.0 0.4 0.4 1.3 0.4 0.6 5.6 0.3

62.1 8.4 9.3 12.8 1.7 2.4 0.0 0.0 1.1 0.0 0.0 0.1 0.0 0.0

39.1 23.7 19.7 3.9 1.1 1.1 0.1 0.0 0.4 0.0 0.0 0.2 0.5 0.1

74.4 0.9 6.5 2.0 2.2 0.2 0.2 0.1 6.5 0.0 0.0 0.2 0.0 0.0

40.5 14.2 14.1 5.6 2.3 0.5 0.0 0.0 11.8 0.0 0.0 0.2 0.0 0.8

37.8 7.1 3.5 14.6 2.1 12.9 16.0 0.0 0.1 0.0 0.0 0.7 0.0 0.0

25.3 3.3 9.3 30.3 1.4 22.0 2.9 0.1 0.3 0.0 0.0 0.5 0.0 0.4

4.7 3.9 1.4 0.3 0.4 0.6 0.6 0.0 13.0 0.6 0.1 20.4 8.0 0.0

6.1 39.5 17.6 0.6 0.2 1.1 0.0 0.0 21.3 0.0 0.1 1.2 0.1 0.0

1.8 12.6 15.9 25.4 2.4 31.4 0.1 0.0 0.4 0.0 0.0 3.2 0.0 0.3

68.8 0.5 7.0 7.4 1.6 0.3 0.3 5.0 0.5 0.1 0.0 0.4 0.0 0.1

34.9 2.5 13.8 16.4 1.3 1.5 0.9 23.3 0.2 0.1 0.0 0.2 0.0 0.0

33.9 10.4 17.3 16.9 2.2 2.4 0.5 0.5 2.4 0.0 0.0 0.4 0.1 3.6

11.8 10.7 10.2 2.7 1.2 4.7 10.3 4.3 4.1 1.8 0.6 3.3 3.6 4.5

Note: Scandinavia: Denmark (including imports of Zollverein members from Schleswig-Holstein), Sweden and Norway. Figures for Turkey include imports from Moldavia and Wallachia. Italy includes the Roman States. Spain in most cases includes the Canary Islands. Portugal in most cases includes Azores and Madeira.

135

Bilateral Trade in Europe, 1857–1875

Ad valorem Equivalent Average Tariff Rates for Every Commodity Group and Country, 1857–1875 (Per cent of Import Value). 1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

1.8 0.7 1.6 0.0 0.0 2.7 15.2

2.3 1.1 2.2 2.5 0.0 1.0 15.0

1.8 1.1 0.2 1.7 0.0 1.2 20.2

2.3 1.1 0.5 2.2 0.6 3.1 23.7

2.7 0.1 0.2 2.4 0.6 1.2 12.0

1.7 1.0 0.0 1.6 0.4 0.5 11.8

0.7 0.7 0.0 2.1 0.5 0.4 16.5

0.0 1.5 0.0 1.5 0.5 0.8 14.1

0.0 0.9 0.0 0.0 0.4 0.5 14.8

0.0 1.5 0.0 0.0 0.5 0.2 17.7

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

2.8 1.1 0.7 0.0 0.0 5.7 15.0

3.2 1.7 0.7 3.6 0.1 2.5 15.0

2.8 1.0 0.7 2.6 0.0 2.3 15.0

3.4 0.4 0.6 3.2 0.8 2.2 23.8

4.0 3.0 0.2 3.7 1.0 4.4 14.5

2.9 0.2 0.0 2.3 0.7 0.5 18.3

1.6 0.0 0.0 2.9 0.7 0.6 18.8

0.0 0.0 0.0 1.9 0.8 3.6 21.5

0.0 0.3 0.0 0.0 0.7 1.8 13.1

0.0 0.0 0.0 0.0 0.7 0.7 9.6

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

2.1 1.4 2.5 0.0 0.7 1.7 15.3

2.6 2.2 2.4 3.6 1.0 1.2 13.3

2.3 1.5 0.8 2.7 0.7 0.6 14.9

2.7 3.3 1.2 3.3 2.8 0.3 10.0

2.8 1.2 0.2 3.6 2.9 0.1 19.6

1.8 2.3 0.0 2.4 2.2 0.1 19.3

0.8 2.4 0.0 2.7 2.4 0.0 18.3

0.0 1.9 0.0 2.5 2.0 0.1 17.7

0.0 1.4 0.0 0.0 1.8 0.1 18.0

0.0 1.8 0.0 0.0 2.2 0.1 15.5

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

68.6 0.4 39.3 0.5 0.1 32.8 35.7

89.4 0.5 62.9 0.6 0.1 27.6 30.0

36.2 0.7 47.6 0.8 0.1 31.4 34.9

36.6 0.6 53.1 0.6 0.0 31.6 40.0

40.9 0.7 41.3 0.5 0.0 30.5 81.7

32.8 0.8 37.6 0.5 0.0 23.6 77.1

32.8 0.8 24.8 0.5 0.0 16.4 77.0

26.1 6.0 27.3 0.5 0.0 18.3 81.5

25.8 11.0 22.1 0.5 0.0 18.1 69.0

26.8 10.7 19.8 0.5 0.0 18.5 68.5

Wheat UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Rye UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Milling products UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Wines UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

(Continued ) 1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

178.5 16.7 53.7 6.2 0.0 51.4 65.6

294.7 12.3 71.1 9.2 0.0 58.2 30.0

251.5 16.3 66.6 27.0 0.1 59.5 47.3

289.9 10.1 74.0 65.1 16.0 44.7 107.5

320.8 8.6 60.5 64.4 10.4 44.8 273.8

300.3 6.2 52.5 58.3 8.7 36.4 283.3

265.0 11.2 43.0 43.7 7.0 31.5 272.6

242.7 8.9 49.4 48.9 7.6 26.8 192.0

237.0 4.1 41.8 99.5 6.4 31.1 168.2

237.7 2.8 42.7 112.2 6.9 24.5 158.0

1861

1863

1865

1867

1869

1871

1873

Spirits and liqueurs UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

1857

1859

1875

Spirits and liqueurs alternative (excluding equivalents for domestic excises and production taxes) UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

120.4 16.7 32.4 6.2 0.0 32.7 65.6

12.4 16.3 44.9 11.9 0.1 37.9 47.3

12.7 10.1 52.4 16.9 16.0 28.6 78.0

14.0 8.6 42.7 14.7 10.4 28.6 128.9

11.9 6.2 30.9 12.5 8.7 25.2 73.4

1859

1861

1863

1865

0.0 0.6 0.6 0.2 0.7 1.3 7.0

0.0 0.8 0.4 0.2 1.0 1.2 6.7

0.0 0.1 0.3 0.3 1.0 1.0 7.7

0.0 0.2 0.2 0.4 0.2 1.3 12.3

1857

1859

1861

1857

176.0 12.3 49.5 9.2 0.0 37.1 30.0

10.4 11.2 21.4 7.8 7.0 14.5 218.4

4.5 8.9 27.8 12.3 7.6 11.9 148.3

9.0 4.1 20.2 11.7 6.4 14.1 116.1

10.4 2.8 21.1 12.7 6.9 10.7 94.6

1867

1869

1871

1873

1875

0.0 0.2 0.3 0.3 0.2 0.8 20.8

0.0 0.2 0.2 0.3 0.2 0.6 16.2

0.0 0.2 0.2 0.3 0.2 1.0 16.0

0.0 0.3 0.3 0.2 0.2 1.3 14.9

0.0 0.2 0.6 0.2 0.1 1.5 6.3

0.0 0.3 0.7 0.3 0.2 1.4 5.4

1863

1865

1867

1869

1871

1873

1875

Hides, skins, and leather UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Articles of leather and rubber UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

14.8 8.2 4.3 18.8 5.2 2.7 27.0

12.6 11.6 2.1 24.8 5.1 2.5 24.0

0.0 8.8 3.7 17.4 5.2 3.3 24.6

0.0 10.8 3.8 10.2 5.0 3.8 33.7

0.0 8.5 2.5 10.7 5.0 3.0 44.6

0.0 7.9 1.4 9.9 5.0 3.1 46.2

0.0 4.3 1.5 9.3 5.0 2.4 46.3

0.0 5.8 1.5 5.8 5.0 1.9 46.3

0.0 5.0 2.7 9.9 5.0 2.3 44.5

0.0 4.9 2.3 9.6 5.0 2.5 47.8

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

7.8 0.9

8.1 0.8

2.5 0.6

2.5 0.5

2.6 0.1

0.0 0.1

0.0 0.0

0.0 0.0

0.0 0.1

0.0 0.1

Wood UK France

137

Bilateral Trade in Europe, 1857–1875

(Continued )

‘Germany’ Belgium The Netherlands Austria-Hungary USA

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

0.7 4.8 1.2 0.6 29.1

0.7 5.8 1.3 0.3 24.0

0.6 5.8 1.3 0.2 17.8

0.6 6.5 1.3 0.2 20.0

0.1 6.3 1.3 0.3 20.0

0.0 7.2 1.3 0.2 20.7

0.0 7.2 1.3 0.2 19.9

0.0 5.6 1.4 0.2 20.0

0.0 4.4 1.4 0.1 16.4

0.0 5.0 1.6 0.2 16.2

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

0.0 6.8 0.0 0.0 0.0 0.0 11.1

0.0 5.9 0.0 0.0 0.0 0.0 1.2

0.0 0.2 0.0 0.0 0.0 0.0 3.5

0.0 0.1 0.0 0.0 0.0 0.0 8.2

0.0 0.1 0.0 0.0 0.0 0.0 32.6

0.0 0.1 0.0 0.0 0.0 0.0 38.7

0.0 0.1 0.0 0.0 0.0 0.0 37.3

0.0 0.2 0.0 0.0 0.0 0.0 49.4

0.0 0.2 0.0 0.0 0.0 0.0 39.5

0.0 0.2 0.0 0.0 0.0 0.0 35.5

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

1.5 70.0 1.6 7.4 1.6 3.8 25.0

1.0 80.0 1.7 8.1 2.4 3.6 19.0

0.0 4.0 1.4 7.2 2.4 3.5 24.6

0.0 3.3 0.9 3.5 0.2 2.4 35.7

0.0 3.6 0.7 2.7 0.2 2.4 48.7

0.0 4.3 0.8 3.2 0.2 1.2 57.6

0.0 4.5 1.0 3.6 0.2 0.9 79.0

0.0 3.7 0.9 3.4 0.2 0.9 80.7

0.0 4.3 0.9 2.8 0.2 1.1 66.2

0.0 4.7 1.0 2.5 0.2 1.2 70.9

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

0.1 45.5 10.2 12.7 3.6 17.7 24.9

0.1 48.1 10.4 14.9 3.3 17.5 21.7

0.0 14.9 11.7 14.5 3.5 15.9 24.3

0.0 15.0 11.6 9.1 5.1 14.9 38.1

0.0 10.2 8.8 10.0 5.0 13.2 60.6

0.0 10.0 6.7 10.1 5.0 11.6 58.3

0.0 10.0 8.3 10.0 5.0 10.4 67.1

0.0 10.0 8.5 10.0 5.0 8.5 67.4

0.0 9.9 7.6 10.0 5.0 8.3 59.1

0.0 10.0 8.1 10.0 5.0 7.7 63.1

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

0.0 36.1 10.7 2.6

0.0 36.1 10.1 4.6

0.0 10.7 8.9 4.9

0.0 6.5 4.4 1.6

0.0 5.5 4.4 1.7

0.0 10.0 4.4 2.5

0.0 9.3 5.3 2.0

0.0 6.5 5.8 2.1

0.0 7.3 5.5 2.6

0.0 9.3 5.9 2.3

Wool UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Woollen yarn UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Woollens and worsteds UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Cotton yarn UK France ‘Germany’ Belgium

138

MARKUS LAMPE

(Continued )

The Netherlands Austria-Hungary USA

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

1.2 6.6 22.4

1.1 6.5 19.0

1.0 7.2 24.1

0.1 3.0 40.0

0.1 4.4 31.1

0.1 6.2 63.5

0.1 5.6 61.6

0.1 5.6 51.7

0.1 5.6 45.0

0.1 4.5 51.0

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

0.1 11.4 40.9 14.5 4.1 27.7 21.7

0.7 21.3 56.5 19.1 4.1 29.1 24.0

0.0 15.5 62.4 19.4 4.1 29.3 25.6

0.0 12.0 42.9 7.8 5.0 27.3 32.7

0.0 12.0 22.1 3.7 5.0 18.7 47.3

0.0 13.1 17.3 8.6 5.0 15.8 42.5

0.0 14.0 17.0 9.0 5.0 12.2 44.6

0.0 13.6 18.9 8.8 5.0 9.0 40.6

0.0 13.0 18.9 8.2 5.0 5.0 36.7

0.0 13.6 17.2 7.8 5.0 9.9 39.0

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

0.0 18.9 4.0 0.8 1.0 3.6 18.5

0.0 21.5 4.8 0.5 1.1 3.4 15.0

0.0 9.1 3.5 0.4 1.1 2.5 23.4

0.0 6.8 2.3 0.2 0.3 2.4 34.0

0.0 6.7 2.7 0.3 0.3 2.5 36.6

0.0 8.9 4.1 0.2 0.3 2.7 36.9

0.0 9.4 2.5 0.2 0.3 1.4 34.0

0.0 8.5 2.1 0.5 0.3 1.4 30.3

0.0 7.7 2.0 0.2 0.4 1.4 34.0

0.0 7.8 2.1 0.1 0.5 1.1 35.4

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

0.9 22.8 2.8 23.5 1.2 19.8 17.9

0.4 19.1 3.7 31.5 1.2 15.2 15.0

0.0 14.9 4.1 16.2 1.2 19.8 18.1

0.0 12.2 3.0 8.6 3.8 20.4 29.8

0.0 10.8 2.4 9.4 4.0 12.8 38.4

0.0 10.0 2.2 9.3 3.9 4.2 37.7

0.0 11.2 1.4 7.7 3.7 4.0 37.9

0.0 10.5 2.9 7.7 3.7 4.0 34.4

0.0 11.1 4.2 7.0 3.6 4.3 33.7

0.0 11.7 4.2 7.7 3.7 3.5 35.5

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

0.0 0.3 0.0 0.2 0.1

0.0 0.4 0.2 0.3 0.0

0.0 0.3 0.2 0.2 0.2

0.0 0.2 0.2 0.0 0.1

0.0 0.1 0.2 0.0 0.2

0.0 0.1 0.0 0.0 0.2

0.0 0.1 0.1 0.0 0.1

0.0 0.1 0.1 0.0 0.4

0.0 0.2 0.0 0.0 0.3

0.0 0.1 0.0 0.0 0.3

Cottons UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Linen yarn UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Linens UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Silk UK France ‘Germany’ Belgium The Netherlands

139

Bilateral Trade in Europe, 1857–1875

(Continued )

Austria-Hungary USA

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

0.5 4.5

0.5 3.9

0.5 11.2

0.6 1.2

0.6 6.6

0.4 5.5

0.1 1.6

0.1 1.1

0.1 0.5

0.1 0.3

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

14.0 21.4 10.1 4.6 4.7 14.1 22.8

10.7 9.0 9.4 5.1 3.8 14.0 19.0

0.0 10.7 9.2 5.4 3.4 12.3 21.2

0.0 8.7 9.0 3.9 5.0 11.1 30.3

0.0 5.9 6.1 3.8 5.0 9.9 59.1

0.0 4.3 4.1 2.3 5.0 3.2 58.6

0.0 4.6 4.4 2.0 5.0 3.4 54.9

0.0 2.3 4.4 2.1 5.0 2.9 53.8

0.0 3.8 4.1 2.5 5.0 2.5 53.3

0.0 4.4 4.0 3.7 5.0 4.7 53.7

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

1.9 44.4 6.6 11.7 5.0 3.0 24.3

2.0 42.5 5.9 12.1 5.8 2.5 21.0

0.0 26.1 4.1 7.4 5.0 2.8 26.4

0.0 18.6 6.4 6.0 4.7 5.3 52.8

0.0 8.0 6.8 8.7 4.8 3.5 48.8

0.0 7.5 6.9 7.9 4.9 6.4 48.1

0.0 6.0 4.5 6.9 4.9 4.1 51.2

0.0 5.7 6.1 9.7 4.8 4.2 53.8

0.0 4.0 7.9 9.4 4.8 4.3 44.5

0.0 7.7 7.7 9.6 4.2 4.1 43.0

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

0.0 19.1 23.1 18.2 0.0 15.0 27.4

0.0 19.2 30.4 10.0 0.0 15.5 24.0

0.0 17.6 32.3 5.6 0.0 15.4 29.8

0.0 18.7 26.9 9.8 0.0 15.6 47.0

0.0 8.9 21.9 6.7 0.0 16.2 52.6

0.0 13.1 24.5 5.4 0.0 16.1 50.5

0.0 0.7 14.9 4.6 0.0 10.0 50.1

0.0 9.5 6.9 4.7 0.0 10.0 34.6

0.0 5.2 4.0 3.2 0.0 10.0 21.2

0.0 7.5 0.0 3.5 0.0 10.0 28.9

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

0.0 10.9 29.4 2.9 0.0 15.7 23.1

0.0 10.1 31.2 3.3 0.0 21.1 17.8

0.0 5.8 25.7 3.2 0.0 20.1 21.8

0.0 7.0 27.8 4.1 0.0 16.9 37.5

0.0 2.6 16.5 2.4 0.0 11.1 36.4

0.0 2.4 17.8 1.5 0.0 8.6 40.2

0.0 2.1 15.5 1.9 0.0 22.1 43.6

0.0 5.8 14.2 2.0 0.0 22.1 40.2

0.0 4.7 11.8 2.0 0.0 20.6 27.1

0.0 6.2 8.8 2.4 0.0 19.6 23.2

Silk wares UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Glass and glassware UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Pig iron UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Bar iron and steel UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

(Continued ) 1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

12.4 23.7 37.8 18.0 3.6 11.9 24.1

10.8 33.2 31.0 22.6 3.6 16.5 19.8

0.0 12.5 29.4 19.6 3.7 15.4 26.3

0.0 13.1 34.9 13.3 3.9 10.0 37.2

0.0 11.5 31.3 5.4 3.1 10.4 42.9

0.0 11.5 15.9 2.8 3.4 12.8 42.7

0.0 13.2 12.9 3.1 3.1 17.5 42.1

0.0 12.3 15.0 4.0 3.2 13.6 41.7

0.0 12.0 7.1 3.3 2.2 13.4 35.0

0.0 11.7 7.0 4.3 2.4 13.5 35.4

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

Ironware UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

Export weighted average tariff UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

7.4 17.8 12.1 7.1 2.1 10.9 21.4

8.9 18.7 14.7 9.3 2.0 10.9 17.6

1.7 8.2 14.4 8.1 2.0 10.8 20.6

1.8 7.2 12.6 5.2 2.9 10.0 28.7

2.0 5.4 8.4 4.6 2.8 8.0 41.2

1.4 5.6 6.6 4.8 2.7 6.2 40.8

1.2 5.7 5.9 4.6 2.7 5.7 45.7

0.8 5.5 6.5 4.6 2.7 5.0 44.1

0.9 5.5 5.8 4.2 2.6 4.4 38.5

0.9 5.9 5.4 4.3 2.6 4.9 38.8

1857

1859

1861

1863

1865

1867

1869

1871

1873

1875

16.3 11.8 7.1 3.2 1.2 7.3 15.8

12.9 5.3 6.2 3.0 0.9 6.1 21.0

11.9 6.9 6.8 2.7 0.8 5.6 32.2

10.0 4.7 5.9 2.3 0.8 4.9 42.8

9.8 4.0 3.7 2.4 0.8 4.1 46.9

8.9 3.8 3.1 2.1 0.8 4.8 45.7

7.6 4.4 3.3 1.9 0.7 4.5 39.8

6.7 6.1 2.5 1.7 0.7 4.5 27.9

6.3 6.5 2.5 1.6 0.7 3.6 30.8

Total trade average tariff UK France ‘Germany’ Belgium The Netherlands Austria-Hungary USA

14.0 9.9 7.5 2.8 1.0 7.8 18.2

Notes: On ‘Spirits and liqueurs alternative’ and ‘Export weighted average tariff’ see text. ‘Total trade average tariff’: Ratio of overall duties recollected (excluding transit and navigation fees) to total imports for home consumption. Sources: Database for all commodity groups. Total trade average tariffs calculated using duty figures from national statistics except for the US, where Mitchell’s (2003, The Americas, Table G.6) figures were used. For years where import duties were accessible from the original statistics, values quoted by Mitchell were $3–4 Mio higher, which led to no substantial change in figures. As always, data for the US is the mean of two fiscal years, ending June 30 each. Imports for home consumption (excluding precious metals in coins and bullions) were taken from original national statistics for France, the UK, the US and Austria-Hungary. For the UK in all years and for the US prior to the fiscal year 1867/1868 imports for home consumption were calculated as imports – re-exports; from the fiscal year 1868/1869 newly reported imports for home consumption were used. Austria-Hungary totals are unrevised original totals and subject to a higher margin of error. Revised import totals were used for Belgium from Horlings (2002, p. 138) and for the Netherlands from Smits et al. (2000, Table H.1). For the Zollverein from 1857 to 1871 estimates by Bondi (1958, p. 145) were used, complemented with official values for 1873 and 1875.

Bilateral Trade in Europe, 1857–1875

141

NOTES 1. Other frequently cited works in sceptical or dismissive accounts on 19th century trade data are Don (1968) and Platt (1971). This is not the place to discuss the even wider use of today’s trade data by international economists, although authors such as Yeats (1978), Parniczky (1980), Kostecki and Tymowski (1984), Rozanski and Yeats (1994), and Makhoul and Otterstrom (1998) have pointed out that there still remains a considerable amount of error. 2. See Accominotti and Flandreau (2008, pp. 155–157), for a comparison of different estimates of world trade growth rates. 3. The central motivation for the construction of the presented dataset was the investigation of the causes and effects of the Cobden-Chevalier Network of bilateral trade agreements on international trade flows. 4. Bateman (1892–1893), Sevenig (1915), and Flux (1923) give early accounts on the struggle for unification since 1853, whose first result was the establishment of an International Convention for the Publication of Customs Tariffs in 1890, and the adoption of the Brussels Tariff Classification in 1913 as a general scheme for aggregation of commodity and tariff classes. See Estevardeordal (1997, App. 1) for a discussion of problems in the collection of comparable data on the eve of World War I. 5. See Federico and Tena (1991), Platt (1971), von Borries (1970, pp. 7–14), Don (1968), Morgenstern (1963, p. 141), Flux (1923), Bateman (1892–1893), and Bourne (1872), to name the more prominent works on historical international trade statistics only. 6. Sample shares for Belgium and the Netherlands have been calculated conservatively, see below. 7. Cf. Hirth (1870, pp. 426–429), Soetbeer (1864), Buchheim (1982, pp. 19–20), and Platt (1971). On details see Bourne (1872, pp. 202–206) and Fuchs (1893, pp. 88–94) for the UK, Hirth (1870, p. 415) for France, Hirth (1869, p. 106; 1870, pp. 416–417) for the Netherlands, Kaiserliches Statistisches Amt (1873a, pp. 125–130, 137–138; 1873b, pp. XI–XII), and Soetbeer (1875, pp. 747–749) for Germany. 8. See also Soetbeer (1864), who showed in a detailed analysis of bilateral trade records on exports of the UK imported by the Hanseatic Towns in 1862 that figures reported in British and Hamburg and Bremen trade statistics matched to a remarkable extent. 9. See also Soetbeer (1864, p. 20), Hirth (1870, p. 409), and Soetbeer (1882, p. 484). On details about how declarations were obtained and controlled see Bourne (1872) and Hirth (1870, pp. 410–414). In the process of data gathering, all items listed under ‘other articles’ (see Table 1) had to be excluded due to lack of information on geographical distribution. In practice, possible distortions should be negligible as the ‘other articles’ contained no items of major importance, and their share in imports was only 1.7% of total imports in 1865. 10. To cite just a few: Smits (1995, ch. IV and app. VI); Bairoch (1973, pp. 30–31 and various table notes), and de Jonge (1968, pp. 13–16). 11. See von Borries (1970, pp. 30–33), for the period prior to 1858. 12. Richter (1900, pp. 249–250) and Don (1968, p. 83). Furthermore, an unknown amount of disguised transit in duty-free items has been claimed for the late 19th

142

MARKUS LAMPE

century. This constituted no major problem as only very few items in the sample were duty-free and the tests for disguised transit (section 5) found no significant amounts. 13. See section 7 for detailed figures on the importance of excluded European countries. 14. Literally: ‘‘[ . . . ] it was painfully evident that, [ . . . ] instead of being relied upon as authority in such matters, our official reports, though distributed gratuitously, and in large numbers, were but rarely quoted, except to be confuted by the less pretentious, but obviously more correct, statistics of boards of trade, chambers of commerce, and other local organizations’’ (p. VIII), and that furthermore, a revision of the quality of import statistics had shown that, ‘‘[i]n two words, then, their condition was even worse – much worse – than was publicly supposed [ . . . ]’’ (Del Mar, 1868, p. XLIII). On Del Mar’s report see Ely (1953, pp. 270–272). 15. The legislation on obligatory declarations dated from 1820 and provided (until 1893) that exports had to be declared only if made in vessels. This led to a clear understatement of exports to the Canadian territories; see Del Mar (1868, pp. V, IX, XXII–XXIII), Young (1875, pp. XV–XVII), and Simon (1960, p. 631). 16. Del Mar (1868, pp. XXIV–XXIX and XLV–XLIX). 17. The main bias that arises from these territorial changes in the Zollverein’s customs area concerns trade with Scandinavia and probably Russia via the Baltic Sea, as Lu¨beck was a major port in this area and trade of the Zollverein with Lu¨beck and Schleswig-Holstein included an unknown amount of transit trade to Denmark and other places at the Baltic Sea. 18. In French statistics, the two provinces might have been retained part of the customs area in at least some months, while in the Zollverein’s statistics for 1871, trade with Alsace and Lorraine was reported in annotations to the main tables (as were exports to troops in France); see Richter (1901a, pp. 286–289) and Soetbeer (1875, p. 769). On distortions that might have been caused to international trade in the early 1870s by the French war indemnity see Giffen (1882, p. 195), and Soetbeer (1875, pp. 740–742). 19. On inconsistency in classifications see Bateman (1892–93) for the 1880s and Estevadeordal (1997, app.) as well as Flux (1923, pp. 298–300) for even later periods. 20. The most prominent mining product was coal, which also was the central fuel for transportation. As national practices on recording varied, additional distortions evolved from exports of coal to steamships and trains that burnt it on their way; e.g. export records of Bremen reported for 1865 that 69% of coal exports as ‘‘ships’ demand’’. In contrast, the UK did not consider coal exports to ships as exports; see Richter (1901b, p. 17). 21. See e.g. La´ng (1906, pp. 121–128) or Richter (1901a, p. 295). 22. Haberdashery might have included a wide variety of fancy articles, perfumery, and small items of metal, porcelain, wood and other materials, such as buttons, soaps, costume jewellery and many other handicraft products. 23. In contrast, the 21 included commodity groups on average consist of 6–7 items each, with wide variation between countries and from one commodity to another. The most detailed accounts are found for linen and silk textiles in France, where more than 20 different items are recorded, while Belgium sums them up one broad category in 1865 and to 3–5 in other years. In contrast, wheat and rye consist of only

Bilateral Trade in Europe, 1857–1875

143

one item, just being differentiated in Austria-Hungary according to special tariff preferences. 24. Their values were reconstructed according to their average shares of highvalue cottons in the values for each trading partner in the second semester of 1865 and in 1867. The old and the new scheme have been consolidated using the concordance provided by Bra¨mer (1868). 25. Tin-plates made up a small fraction of total exports of the ‘Bar iron and steel’ group only and were almost exclusively exported to non-European countries (Cuba, Haiti, Canada, etc.), and about 99.9% of all wood imported into the US came from Canadian Territories, and the wood originating from Maine caused no major distortion in this pattern. Furthermore, trade with Canada lies outside the focus of this paper. 26. The remaining small amount was interpreted as representing other articles not covered in the sample. 27. For silk, no general trade and transit data were published after 1872, so the general trade/special trade ratio for 1871 was used to reconstruct general trade and transit volumes. 28. In addition, for the totals of the most important subclasses (c. 80 so-called Waarengruppen) official values were reported. 29. The 1871 records were used to reconstruct detailed transit records for 1869 and 1867, and transit for 1863 was reconstructed from 1859 shares. For 1865, shares of 1871 were combined with shares for the port of Venice for 1859. 30. Cf. Federico and Tena (1991, pp. 261–262), Don (1968, p. 83). 31. In the customs house where incoming merchandise was declared as transit, the outer packing of transit loads was sealed to prevent fraudulent imports for the home market. 32. See Bourne (1872), Stafford et al. (1953), Richter (1901b, pp. 21–22), and Hirth (1870, p. 426) for the UK. For the US we are lacking transit tables prior to the fiscal year 1867/1868. Therefore, prior to the calendar year 1869, only re-exports could be included for the US, not transit. See Del Mar (1868, pp. X, XLV, XLIX–LI, LXLIX) for a detailed discussion of the issue. 33. Del Mar (1868, p. XVII), practised the method of deducting re-exports from general imports in an abstract of US statistics prior to 1867. On British records, Giffen (1882, p. 182), remarked that some newspaper and journal articles had failed to deduct re-exports from general exports, and therefore reported an alarming British trade deficit that obviously did not exist as such. 34. Where data on the amount and/or geographical distribution of goods ‘entered for home consumption’ existed, they were used instead. This was the case for imports subject to duty in the UK and from tables of overall amounts of imports for home consumption and amount of duties received in the US from 1869 to 1875. To the latter countries of origin were assigned proportionately to their share in ‘general imports’ for which geographical distribution was published. 35. Only duty-free imports for shipbuilding into the Zollverein have been included because they were reported in the main tables for imports and exports. 36. After 1871, the geographical distribution of trade via entrepoˆts was included in the tables while postal transport continued to be recorded as such and an increasing amount of trade with ‘not determined’ partners appeared which should have accounted for the rest of the imports reported separately before; see

144

MARKUS LAMPE

Kaiserliches Statistisches Amt (1873a, pp. 133–134, 137–138, 141–142, 145), Committee of the British Association Appointed to inquire into the Accuracy and Comparability of British and Foreign Statistics of International Trade (1904), and Richter (1901b, pp. 25, 29). In addition, in Germany some duty-free goods imported overland were not recorded, mostly due to ‘small border trade’ provisions. The only items that matter to us were firewood and construction wood, and amounts were supposed to be small; see Kaiserliches Statistisches Amt (1873a, pp. 130, 137; 1873b, p. X) and von Borries (1970, p. 29). Other articles were hay, straw, forage, fresh fish, peat, beehives, etc. 37. See Soetbeer (1864, pp. 17–20), Hirth (1869, pp. 84, 97–99; 1870, pp. 429–439), and Stafford et al. (1953, pp. 289–293). For an important share of imports, mainly manufactured goods, declared values were directly recorded, and no quantities were published in the statistics. 38. See Bourne (1872, pp. 206–209, 213), Hirth (1869, pp. 100–101), Richter (1901b, pp. 16, 19); Platt (1971, pp. 123–124), and Don (1968, pp. 84–85) for details and evaluation. 39. See Giffen (1882, p. 188). Bateman (1892–93, pp. 533–534) stated that the British method of obtaining declared values was the best available even in the 1890s. Nevertheless, Fuchs (1893, pp. 88–89) blamed the cited eminent British statisticians for not having treated the subject at all. He expressed scepticism about British statistics in general. 40. Quotes are taken from Table 1 (p. 4) of the Annual Report of the Chief of the Bureau of Statistics on the Commerce and Navigation of the United States for the Fiscal Year ended June 30, 1875; see also Hirth (1869, p. 106). ‘Specie value’ referred to the gold value of exports from countries with a depreciated currency; see Bateman (1892–93, pp. 533–534). 41. This was changed in 1883, when import values had to be declared as invoice values only, excluding insurance and transport costs and commissions. See Simon (1960, pp. 639, 642–43), Committee of the British Association Appointed to Inquire into the Accuracy and Comparability of British and Foreign Statistics of International Trade (1904, 442), Bateman (1892–93, pp. 533–534), and Flux (1923, p. 302). 42. Cf. Soetbeer (1882, pp. 480–481; dismissing these advantages), Flux (1923, p. 302), and Don (1968, pp. 84–85). It should be noted that Simon (1960, pp. 642–644) concluded for the imports of the US that distortions caused by underdeclaration and smuggling together made up about 3% of import values, and therefore presented no major problem. His judgment was based on estimates by the then Director of the Statistical Office in 1874. 43. This applies especially when prices of certain market places were used to value the entire import and export volumes, cf. Paasche (1874) and Bourne (1872, p. 206). Hirth (1869) gives comprehensive accounts of official pricing in the UK, France, Belgium, Italy, Austria, Russia, the Netherlands, and the Hanseatic Towns. See also von Borries (1970, pp. 7–9) and Flux (1923, p. 301). 44. See Soetbeer (1864) and Hirth (1869, pp. 101–102; 1870, p. 431). 45. See Hirth (1869, pp. 103–104) and Platt (1971, p. 124). The annual reports of the price commission were included in the introductory part of each year’s foreign trade statistics.

Bilateral Trade in Europe, 1857–1875

145

46. For example, the price for cullet in the official statistics of 1865 was still the same as in 1833; in 1875, anchors and marine chains were valued at their 1857 prices. 47. Soetbeer (1864, pp. 29–30) notes that Belgian prices might have overvalued exports by about 50%; see also Horlings (2002, pp. 115–117), Buchheim (1982, p. 130), and Sydow (1977, p. 216). 48. The extremely low price of 1.90 bfs per kg for wool was changed to 2.35 bfs per kg in 1869. 49. In 1868 prices were newly fixed, but remained unchanged for the vast majority of items. See Lindblad and van Zanden (1989, p. 232), Platt (1971, p. 124), Hirth (1869, p. 106), and Soetbeer (1864, pp. 28–29). 50. Smits (1995, app. VI) used prices from Spiethoff (1955), Posthumus (1943) and apparently unpublished work by Alain Callewaert I was unable to locate for most manufactures. Smits et al. (2000) additionally used the unpublished PhD thesis by Arthur van Riel (Postponed conformity, University of Utrecht), which I also have not assessed. German and UK prices have also been used to correct Dutch statistics by Lindblad and van Zanden (1989), Buchheim (1982), and de Jonge (1968). Where possible, I crosschecked prices between sources and with prices from Posthumus (1943). 51. See v. Neumann-Spallart (1876, pp. 95–99), Richter (1900, pp. 246–248), and Hirth (1869, pp. 104–105). 52. I compared average prices per kg of bilateral exports to Germany from Austro-Hungarian statistics with those of bilateral imports from Austria-Hungary as reported in German statistics, and vice versa, to avoid composition biases due to different internal structure of trade in commodity groups. 53. As a general rule, I moved up official revisions to previous years and avoided to constructed artificially high or low prices that were never quoted in the publications. Revisions were clustered in certain items like wine, spirits, hides and skins, cotton and linen yarns, cottons, linens, and silk. 54. Soetbeer (1875, pp. 751–3, 756–769) compared the official prices of the German tables for 1872 and 1873 to those of Hamburg and France and came to the conclusion that they were not the reason for the big German trade deficit reported for 1872 und 1873; he affirmed that half of reported deficit was real, while the other half was caused by statistical understatement of German exports, especially in textiles. 55. Values in Bremen were ‘invoice values’ for imports and exports, including for imports freight and assurance ‘as prevailing here’. Prices were for net quantities when goods were sold by weight and for gross quantities when goods were sold by pieces, i.e. most manufactures (Hirth 1869, pp. 76–82; Soetbeer 1882, p. 484). Hamburg values were values as quoted at the commodity exchange, i.e. including all trade costs and importers’ gross revenue. 56. Von Borries (1970, pp. 33–46) presented a review of those early attempts; see also Soetbeer (1864, p. 31), and the discussion of Bienengra¨ber’s (1868) and other attempts in Hirth (1869, pp. 110–113). 57. Von Borries (1970) and Rosenberg (1960) accused Bondi also of systematically distorting values for some trade flows in order to prove Marxist arguments about the evolution of capitalist economies. See also Dumke (1994, part II, pp. 2–16) for a general review of the discussion on the quality of existing estimates.

146

MARKUS LAMPE

58. On the general problem see e.g. Federico and Tena (1991, pp. 262, 269), Platt (1971), Don (1968, pp. 88–91), Flux (1923, pp. 303–304), and Giffen (1882, p. 189). 59. Switzerland did not show up at all in British statistics. On the Problem see Platt (1971, pp. 122–124), Stafford et al. (1953, p. 293), Committee of the British Association Appointed to inquire into the Accuracy and Comparability of British and Foreign Statistics of International Trade (1904, p. 444), Richter (1901b, pp. 22–30), Fuchs (1893, p. 90), Giffen (1882, p. 189), and Bourne (1872, pp. 202–204, 208–209). 60. Additional statistics for ‘indirect trade’ were published which recorded ‘real countries’ of origin for the share of imports that was not directly received from its ‘original origin’ (e.g. Switzerland). 61. See Platt (1971, p. 122), Richter (1901a, p. 292), Giffen (1882, p. 191), and Hirth (1869, p. 103). 62. Kaiserliches Statistisches Amt (1873a, pp. 131–132), Hirth (1870, p. 427), von Neumann-Spallart (1876, pp. 102–103), and Richter (1900, pp. 292–294). 63. For the Netherlands see Lindblad & van Zanden (1989, p. 232) and Platt (1971, p. 122); for Germany Kaiserliches Statistisches Amt (1873a, pp. 131–132; 1873b, p. XII); for Austria-Hungary v. Neumann-Spallart (1876, pp. 103–104), Richter (1900, pp. 250–252), and Don (1968, pp. 88–91). 64. Cf. among others Hirth (1869, p. 107), Soetbeer (1882, pp. 485–487), and Zimmermann (1908, pp. 312–317). 65. From 1872 on, tentative export statistics for Hamburg (quantities) were provided in a special table of the official German statistics. Additional errors for the Zollverein might result from the omission of imports via Altona in calculating the shares in Hamburg’s foreign trade. Altona is now a part of Hamburg, but was a complete free-port without comprehensive statistics in period under study. As a large share of Altona’s trade was supplied to or from its dominating neighbour (von Borries 1970, pp. 107–109, Krawehl 1977, pp. 134–145), it was assumed as a best guess that its composition was comparable to that of Hamburg, and hence omission of this trade did not matter for shares. The small free ports of Brake and Geestemu¨nde (now a part of Bremerhaven) have not been corrected due to negligible shares of imports coming through them, except of yarns of wool traded through Geestemu¨nde that originated 100% from the UK (Hull), see Speerschneider (1866) and Statistik des Deutschen Reichs, vol. 5, p. II. 10 (1872); vol. 11, p. II. 114 (1873); vol. 23, pp. II. 244–245 (1875). Trieste’s statistics reported a considerable amount of trade with other Austrian ports. I interpret these amounts as internal trade and foreign trade distributed via Trieste, and accordingly omitted them in the calculation of shares of countries of origin and destination. Furthermore, shares calculated from Trieste’s statistics were based on unrevised fixed official prices that have not been revised for the present study. 66. In France, information on origins and destinations was less comprehensive in transit tables than in import and export tables. To enhance comparability, I disentangled the ‘other parts’ residual proportionately to the countries’ shares in the difference of respective general and special imports and exports of categories in whose transit tables ‘other parts’ accounted for more than 10% of transit or 80.000 Francs in value.

Bilateral Trade in Europe, 1857–1875

147

67. The tests were exploratively conducted also for records on special import and exports. They performed far worse and confirmed the perceived need to correct those records systematically. 68. Morgenstern (1963, p. 132) expressed this tellingly: ‘‘If the series disagree, something is at fault; if they agree, they may, of course, still both be wrong.’’ 69. Following van Bergeijk (1995, p. 5), IMME is defined as IMME ¼ ((IMij EXji)/(IMijþEXji)). As proposed by Carreras Marı´ n (2005) import weighted averages are presented. 70. Similar encouraging findings on the non-significance of differences between import and export records have been obtained by Folchi and Rubio (2005) on petroleum and coal trade with Latin American countries in 1926. 71. For the complex see von Kalchberg (1871, pp. 105–118). 72. French records on flours and meals also appear as too high in 1868 and led French Customs Authorities to calculate long run records for foreign trade development without ‘‘grains, flours, and meals’’ in 1869. They might have to do with improvement trade (see von Kalchberg, 1871). Zollverein’s import records for dutyfree construction wood in 1869 most certainly are distorted by a transmission error for imports to Saxony. 73. I suppose that the low values in ‘propensity to transit’ for wheat in Germany in 1857 and 1859 were caused by limited transport facilities and not by disguised transit. Furthermore, milling products in Germany have not been corrected, as there is strong suspect that imports and exports were conducted locally with neighbouring countries instead comprehensive disguised transit. The share of meals and flours in German trade was very low, and therefore possible distortions due to disguised transit should be negligible. 74. As for the US and the UK separate statistics for re-exports and transits existed, separate calculations for transits and re-exports have been summed up in one figure in practise. Where disguised transit was corrected, the corrected figures excluding disguised transit were used in this section. 75. In France and Belgium (and in some years in the US and the UK), Bremen is merged with Hamburg under ‘Hanseatic Towns’. This presents no bigger problem, as trade with Hamburg accounts for more than 90% of trade with Hanseatic Towns and thus the possible bias resulting from the inclusion of Bremen (and Lu¨beck) in ‘Hamburg figures’ is small. 76. In the original transit corrections this trade would have been erroneously allotted to the first transit country as country of origin. In general, this concerns small amounts of trade only, but most notably in trade in iron and steel through the Netherlands and Germany to Austria-Hungary substantial distortions would have resulted, i.e. blown-up special export records for ‘Dutch’ iron and steel. According to Dutch transit tables this trade originated in the UK, but in the correction of German-Austrian through German transit tables it would be reassigned as being Dutch, because it entered Germany from the Netherlands, not from the UK. 77. Similar evidence was reported by the Commission for Further Development of the Statistics of the Zollverein in Kaiserliches Statistisches Amt (1873a, pp. 131–132). 78. The fact that there is no export data for Switzerland did not harm this approach as Switzerland was also unable to conduct direct seaward exports to Belgium due to the lack of ports.

148

MARKUS LAMPE

79. Similar corrections for transit from those Ck though Belgium and France have not been conducted. 80. For an application of the disaggregate figures see Lampe (2008). 81. ‘Uncorrected’ includes price corrections and other adjustments at the national level described in Section 3. 82. The theoretical number of cross-sections would be 91, but due to difficulties in the treatment of imports of the Zollverein from Schleswig, Holstein (Altona), Lauenburg and via the Hamburg-Lu¨beck railway trade flows imports of ‘Germany’ and Austria-Hungary (where trade arrived after transit through Germany) from Denmark have been excluded. Because of the separation of Lombardo-Venetia from Austria-Hungary trade figures of the latter with Switzerland (largely in raw silk prior to 1866) suffered a structural break and have been excluded. 83. An ‘econometrically correct’ shortcut implementation of the Anderson–van Wincoop specification including ‘multilateral resistance’ would involve the inclusion of time-varying importer and exporter effects (Baldwin & Taglioni, 2006). In the present context, the limited total number of observations (420 and 880, respectively) limits the degrees of freedom in the model. The inclusion of the time-varying effects would result in a number of 2NT dummy variables, where N is the number of countries and T the number of periods. This would mean the inclusion of 140 dummy variables to be estimated from 420 observations. Thus, while the Baldwin–Taglioni specification involves no problems when ‘‘T and N are large’’ (p. 23), it is unfeasible in the present case. 84. Following Geraci and Prewo (1977) the variable has been coded as ln(1þtariff rate). 85. In the corrected dataset there are no 0 values, but in the uncorrected figures there are some. 86. Percent equivalents of the coefficient can be calculated as exp(b)–1. 87. For some country pairs the margin of error might be higher due to the possible importance of transit through countries not covered in the dataset, e.g. Switzerland in the case of trade with Italy or Turkish and Russian ports in the Black Sea and Sweden and Denmark in the Baltic Sea. 88. As countries do not import from themselves, exports of the European countries in the sample are calculated as the sum of exports to the other five countries as recorded in their import statistics. For all other countries imports into all six European countries in the sample were summed up. 89. Of course, as mentioned above, the US was also a major exporter of raw cotton, tobacco, and the like. 90. Excises and production taxes were assessed directly from the trade statistics (Austria-Hungary) and from the seminal works by Julius Wolf (1884, 1887). For France and the Netherlands, no adjustments had to be made. 91. These higher rates were: 1857 1859 1861 1863 1865 1867 1869 1871 1873 1875 Wheat 1.2 Pig iron 23.6 Bar iron and steel 9.1

1.6 34.4 42.1

1.2 21.9 22.8

1.2 22.3 23.2

0.5 22.2 15.4

1.7 21.9 11.1

2.4 11.7 7.4

1.6 19.4 9.8

1.4 15.2 8.2

2.1 18.9 9.6

149

Bilateral Trade in Europe, 1857–1875

92. Including duty-free imports under the Reciprocity Treaty leads to the following alternative average tariffs when it was in force:

Wheat Rye Milling products Hides, skins and leather Wood Wool

1857

1859

1861

1863

1865

0.1 0.1 0.0 7.0 0.2 10.2

0.1 0.0 0.2 6.6 0.0 1.1

0.0 0.0 0.1 7.6 0.1 3.3

0.1 0.0 0.0 12.2 0.1 7.6

0.7 0.1 0.7 20.0 2.2 28.7

ACKNOWLEDGMENTS The elaboration of the dataset is part of the research project on ‘‘Multilateral Free Trade in the 19th Century: Causes and Effects of the Cobden-Chevalier Network, c. 1857–1875’’, and would not have been possible without a financial grant from the Fritz Thyssen Stiftung. The data collection benefited from excellent assistance by Hendrik Buhrs, Christian Flick, Elena HeXelmann, Sonja Lohmann, Annabell MaaX, and Hendrik VoX. I thank Carsten Burhop, Ulrich Pfister, Thorsten Lu¨bbers, seminar participants at the University of Oxford, and the anonymous referee for very helpful advice and suggestions. Of course, I assume the sole responsibility for all errors.

REFERENCES Accominotti, O., & Flandreau, M. (2008). Bilateral trade treaties and the most-favored-nation clause: The myth of trade liberalization in the mid-nineteenth century. World Politics, 60, 147–188. Anderson, J. E., & van Wincoop, E. (2003). Gravity with Gravitas: A solution to the border puzzle. American Economic Review, 93, 170–192. Bairoch, P. (1973). European foreign trade in the XIX Century: The development of the value and volume of exports (preliminary results). Journal of European Economic History, 2, 5–36. Bairoch, P. (1974). Geographical structure and trade balance of European foreign trade from 1800 to 1970. Journal of European Economic History, 3, 557–608. Bairoch, P. (1976). Commerce exte´rieur et de´veloppement e´conomique de l’Europe au XIXe sie`cle. Paris–La Haye: Mouton (Civilisations et Socie´te´s 53). Baldwin, R., & Taglioni, D. (2006). Gravity for dummies and dummies for gravity equations. NBER Working Papers 12516. NBER, Cambridge, MA. Bateman, A. E. (1892–1893). comparability of trade statistics of various countries. Publications of the American Statistical Association, 3, 533–550.

150

MARKUS LAMPE

Bienengra¨ber, A. (1868). Statistik des Verkehrs und Verbrauchs im Zollverein fu¨r die Jahre 1842–1864, Nach den vero¨ffentlichten amtlichen Kommerzial-U¨bersichten etc. Berlin: Duncker. Bondi, G. (1958). Deutschlands AuXenhandel, 1815–1870 (Schriften des Instituts fu¨r Geschichte der Deutschen Akademie der Wissenschaften zu Berlin, Reihe I: Allgemeine und Deutsche Geschichte 5). Berlin (East): Akademie. Bourne, S. (1872). The official trade and navigation statistics. Journal of the Statistical Society, New Series, 35, 196–217. Bra¨mer, K. (1868). Beitra¨ge zur Statistik der Production und Consumtion im deutschen Zollverein in den Jahren 1825 bis 1866. Zeitschrift des ko¨niglich-preussischen Statistischen Bureaus, 8, 198–237. Buchheim, C. (1982). Deutsche Gewerbeexporte nach England in der zweiten Ha¨lfte des 19. Jahrhunderts/zur Wettbewerbsfa¨higkeit Deutschlands in seiner Industrialisierungs-phase. Gleichzeitig eine Studie u¨ber die deutsche Seidenweberei und Spielzeugindustrie, sowie u¨ber Buntdruck und Klavierbau (Studien zur Wirtschafts- und Sozialgeschichte 5). Ostfildern: Scripta Mercaturae. Carreras Marı´ n, A. (2005). Geographical effects on the accuracy of textile trade data: An international approach for 1913. UPF Economics and Business Working Paper no. 821. Universitat Pompeu Fabra, Barcelona. Committee of the British Association Appointed to Inquire into the Accuracy and Comparability of British and Foreign Statistics of International Trade. (1904). Extracts from the report of the committee of the British association appointed to inquire into the accuracy and comparability of British and foreign statistics of international trade. Journal of the Royal Statistical Society, 67, 438–447. de Jonge, J. A. (1968). De industrialisatie in Nederland tussen 1850 en 1914 (reprint: Nijmegen: SUN, 1976). Amsterdam: Scheltema & Holkema. Degre`ve, D. (1982). Le commerce exte´rieur de la Belgique 1830–1913–1939: pre´sentation critique des donne´es statistiques (2 vols). Histoire quantitative et de´veloppement de la Belgique 5: Les relations e´conomiques internationales belges, 1830–1913, vol. 1. Brussels: Palais des Acade´mies. Del Mar, A. (1868). Annual Report of the Director of the Bureau of Statistics, in Annual Report of the Director of the Bureau of Statistics on the Commerce and Navigation of the United States, for the Fiscal Year ended June 30, 1867, Washington: Government Printing Office, V-LXXXI. Don, Y. (1968). Comparability of international trade statistics: Great Britain and Austria– Hungary before world war I. Economic History Review, 2nd Series, 21, 78–92. Dumke, R. H. (1994). German economic unification in the 19th century: The political economy of the Zollverein. Munich: Faculty of Social Sciences, Universita¨t der Bundeswehr Mu¨nchen (Diskussionsbeitra¨ge Institut fu¨r volkswirtschaftslehre). Ely, J. E. (1953). United States. In: J. E. Ely & R. D. G. Allen (Eds), International trade statistics (pp. 269–287). New York: Wiley (Wiley Publications in Statistics). Estevadeordal, A. (1997). Measuring protection in the early twentieth century. European Review of Economic History, 1, 89–125. Estevardeordal, A., Frantz, B., & Taylor, A. M. (2003). The rise and fall of world trade, 1870–1939. The Quarterly Journal of Economics, 118, 359–407. Federico, G., & Tena, A. (1991). On the accuracy of foreign trade statistics (1909–1935): Morgenstern revisited. Explorations in Economic History, 28, 259–273.

Bilateral Trade in Europe, 1857–1875

151

Flandreau, M. (2000). The economics and politics of monetary unions: A reassessment of the Latin monetary union, 1865–1871. Financial History Review, 7, 25–44. Flandreau, M., & Maurel, M. (2005). Monetary union, trade integration and business cycles in 19th century Europe: Just do it. Open Economies Review, 16, 135–152. Flux, A. W. (1923). International statistical comparisons. Journal of the Royal Statistical Society, New Series, 86, 297–315. Folchi, M., & Rubio, M. (2005). On the accuracy of Latin American trade statistics: A nonparametric test for 1925. UPF Economics and Business Working Paper no. 879. Universitat Pompeu Fabra, Barcelona. Fuchs, C. J. (1893). Die Handelspolitik Englands und seiner Kolonien in den letzten Jahrzehnten (Schriften des Vereins fu¨r Socialpolitik 57; Die Handelspolitik der wichtigeren Kulturstaaten in den letzten Jahrzehnten 4). Leipzig: Duncker & Humblot. Geraci, V. J., & Prewo, W. (1977). Bilateral trade flows and transport costs. The Review of Economics and Statistics, 59, 67–74. Giffen, R. (1882). The use of import and export statistics, Journal of the Statistical Society, New Series, 45, 181–284. Hanson, J. R., II. (1980). Trade in transition. Exports from the third world, 1840–1900. New York: Academic Press (Studies in Social Discontinuity). Hirth, G. (1869). Beitra¨ge zu einer Deutschen Handelsstatistik I. Waaren-Durchschnittspreise und auswa¨rtiger Handel. Staatshandbuch fu¨r Gesetzgebung, Verwaltung und Statistik des Norddeutschen Bundes und des Deutschen Zollvereins, 2, 67–144. Hirth, G. (1870). Die Methoden der Zoll- und Handelsstatistik in England, Frankreich, Holland, Hamburg, Bremen und im Zollverein. Annalen des Deutschen Reiches fu¨r Gesetzgebung, Verwaltung und Volkswirtschaft, 3, 407–432. Horlings, E. (2002). The International trade of a small and open economy. Revised estimates of the imports and exports of Belgium, 1835–1990. NEHA-Jaarboek, 65, 110–142. Hu¨bner, O. (1861, 1859). Jahrbuch fu¨r Volkswirthschaft und Statistik (Vols. 6 and 7). Leipzig: Hu¨bner. Imlah, A. (1958). Economic elements in the pax britannica: Studies in British foreign trade in the nineteenth century. Cambridge, MA: Harvard UP. Kaiserliches Statistisches Amt (1873). Einleitung zu der Darstellung des auswa¨rtigen und u¨berseeischen Waarenverkehrs des Deutschen Zollgebiets und der Zollausschlu¨sse sowie des Seeschiffahrtsverkehrs im Jahre 1872. (Zu Band III, IV und V der Statistik des Deutschen Reichs.), in Statistik des Deutschen Reichs, 3, VII-XVI. Kaiserliches Statistisches Amt. (1873a). Die Anordnungen des Bundesrathes fu¨r die gemeinsame Statistik der deutschen Staaten mit den vorbereitenden Verhandlungen insbesondere den Protocollen und Berichten der Kommission fu¨r die weitere Ausbildung der Statistik des Zollvereins (Statistik des Deutschen Reichs 1). Berlin: Verlag der Ko¨niglichen Statistischen Bureaus. Kostecki, M., & Tymowski, M. (1984). Inaccuracy of foreign trade statistics of LDCs: The case of the ASEAN area. Intereconomics, 19, 197–201. Krawehl, O.-E. (1977). Hamburgs Schiffs- und Warenverkehr mit England und den englischen Kolonien 1814–1860 (Forschungen zur internationalen Wirtschafts- u. Sozialgeschichte 11). Cologne: Bo¨hlau. Lampe, M. (2008). Effects of bilateralism and the MFN clause on international trade. Evidence for the Cobden-Chevalier Network (1860–1875). Mu¨nster: Institute for Economic and Social History (unpublished Research Working Paper).

152

MARKUS LAMPE

La´ng, L. (1906). Hundert Jahre Zollpolitik. Vienna–Leipzig: Carl Fromme. Lewis, A. (1981). The rate of growth of world trade. In: S. Grassman & E. Lundberg (Eds), The world economic order. Past and prospects. London: Macmillan. Lindblad, J. T., & van Zanden, J. L. (1989). De buitenlandse handel van Nederland, 1872–1913. Economisch- en Sociaal-Historisch Jaarboek, 52, 231–269. Lo´pez-Co´rdova, J. E., & Meissner, C. M. (2003). Exchange-rate regimes and international trade: Evidence from the classical gold standard era. American Economic Review, 93, 344–353. Maddison, A. (2006). The world economy. Paris: OECD. Makhoul, B., & Otterstrom, S. M. (1998). Exploring the accuracy of international trade statistics. Applied Economics, 30, 1603–1616. Mitchell, B. R. (52003). International historical statistics (various vols). Houndmills: Palgrave Macmillan. Mitchener, K. J., & Weidenmier, M. (2008). Trade and empire. NBER Working Papers no. 13765. NBER, Cambridge, MA. Morgenstern, O. (21963). On the accuracy of economic observations. Princeton: Princeton University Press. Paasche, H. (1874). U¨ber die Preisentwicklung der letzten Jahre nach den Hamburger Bo¨rsennotierungen. Jahrbu¨cher fu¨r Nationalo¨konomie und Statistik, 23, 168–178. Parniczky, G. (1980). On the inconsistency of world trade statistics. International Statistical Review, 48, 43–48. Platt, D. C. M. (1971). Problems in the interpretation of foreign trade statistics before 1914. Journal of Latin American Studies, 3, 119–130. Posthumus, N. W. (1943). Nederlandsche prijsgeschiedenis I. Goederenprijzen op de beurs van Amsterdam, 1585–1914. Leiden: Brill. Richter, O. (1900). Grundlagen der Handelsstatistik einiger fremder Staaten I. O¨sterreichUngarn. Vierteljahrshefte zur Statistik des Deutschen Reichs, 9, I.242–I.255. Richter, O. (1901a). Grundlagen der Handelsstatistik einiger fremder Staaten III. Frankreich. Vierteljahrshefte zur Statistik des Deutschen Reichs, 10, I.285–I.304, III.30. Richter, O. (1901b). Grundlagen der Handelsstatistik einiger fremder Staaten III. GroXbritannien und Irland. Vierteljahrshefte zur Statistik des Deutschen Reichs, 10, III.15–III.30. Rose, A. K. (2004). Do we really know that the WTO increases trade? American Economic Review, 94, 98–114. Rosenberg, H. (1960). Review of Deutschlands AuXenhandel 1815–1870. By Gerhard Bondi. Journal of Economic History, 20, 102–104. Rostow, W. W. (1978). The world economy. History and prospect. London: Macmillan. Rozanski, J., & Yeats, A. (1994). On the (in)accuracy of economic observations: An assessment of trends in the reliability of international trade statistics. Journal of Development Economics, 44, 103–130. Santos Silva, J. M. C., & Tenreyro, S. (2006). The log of gravity. The Review of Economics and Statistics, 88, 641–658. Sevenig, J. P. (1915). Die international einheitliche Handelsstatistik. Weltwirtschaftliches Archiv, 5, 234–243. Simon, M. (1960). The United States balance of payments 1861–1900. In: W. N. Parker (Ed.), Trends in the American economy in the nineteenth century (pp. 629–711). Princeton: Princeton University Press (NBER Studies in Income and Wealth 24). Smits, J. P. (1995). Economische groei en structurele veranderingen in de Nederlandse dienstensector 1850–1913. Unpublished PhD thesis, Vrije Universiteit, Amsterdam.

Bilateral Trade in Europe, 1857–1875

153

Smits, J. P., Horlings, E., & van Zanden, J. L. (2000). Dutch GNP and its components, 1800–1913 (GGDC Research Memorandum 5). Groningen: Groningen Growth and Development Centre. Soetbeer, A. (1864). U¨ber die Ermittlung zutreffender Durchschnittspreise, Vierteljahresschrift fu¨r Volkswirthschaft und Kulturgeschichte, year 2, vol. 3, 8–32. Soetbeer, A. (1875). Bemerkungen u¨ber die Handelsbilanz Deutschlands. Annalen des Deutschen Reichs fu¨r Gesetzgebung, Verwaltung und Statistik, 8, 731–784. Soetbeer, A. (1882). Die auswa¨rtigen Verkehrsbeziehungen und die Handelsstatistik des deutschen Zollgebietes. Deutsches Handelsblatt, 40 (5 October), 479–487. Speerschneider, G. (1866). Statistische Uebersicht von Geestemu¨nde’s Handels- & SchifffahrtsVerkehr im Jahre 1865. Hannover: Kindworth’s. Spiethoff, A. (1955). Die wirtschaftlichen Wechsellagen. Aufschwung, Krise, Stockung (2 vols). Tu¨bingen: Mohr (Hand- und Lehrbu¨cher aus dem Gebiet der Sozialwissenschaften). Stafford, J., Maton, J. M., & Venning, M. (1953). United Kingdom. In: J. E. Ely & R. G. D. Allen (Eds), International trade statistics (pp. 288–307). New York: Wiley (Wiley Publications in Statistics). Sydow, H. (1977). Die Handelsbeziehungen zwischen Belgien und dem Zollverein 1830–188., 2 vols. Cologne–Vienna: Bo¨hlau (Dissertationen zur neueren Geschichte 4). van Bergerijk, P. A. G. (1995). The accuracy of international economic observations. Bulletin of Economic Research, 47, 1–20. von Borries, B. (1970). Deutschlands AuXenhandel 1836 bis 1856. Eine statistische Untersuchung zur Fru¨hindustrialisierung (Forschungen zur Sozial- und Wirtschaftsgeschichte 13). Stuttgart: Gustav Fischer. von Kalchberg, V. (1871). Der zollamtliche Veredelungsverkehr in O¨sterreich, dem Zollvereine und in Frankreich. Eine handelspolitische Studie. Jahrbu¨cher fu¨r Nationalo¨konomie und Statistik, 16, 104–142. von Neumann-Spallart, F. X. (1876). U¨ber Handels-Statistik und Handelswerthe mit besonderer Ru¨cksicht auf die neuesten Reformen in O¨sterreich-Ungarn. Jahrbu¨cher fu¨r Nationalo¨konomie und Statistik, 26, 1–14, 88–105. Wolf, J. (1884). Die Branntweinsteuer. Ihre Stellung im Steuersystem und in der Volkswirthschaft, ihre geschichtliche Entwicklung und gegenwa¨rtige Gestalt in den einzelnen La¨ndern, und ihre Erhebungsformen mit einem die Branntweinsteuer in O¨sterreich-Ungarn betreffenden Abschnitt. Tu¨bingen: H. Laupp. Wolf, J. (1887). Die Branntweinsteuer in den europa¨ischen La¨ndern und in den Vereinigten Staaten von Nordamerika von 1884–1886, mit besonderer Ru¨cksicht auf die Bestrebungen zu ihrer Neugestaltung in Deutschland. Finanzarchiv, 4, 320–435. Wolf, N. (2008). Was Germany ever united? Evidence from Intra- and international trade 1885–1933. CEP Working Paper no.870. Centre for Economic Performance, London. Yeats, A. J. (1978). On the accuracy of partner country trade statistics. Oxford Bulletin of Economics and Statistics, 40, 341–361. Young, E. (1875). Report of the Chief of the Bureau of Statistics, in Annual Report of the Chief of the Bureau of Statistics on the Commerce and Navigation of the United States for the Fiscal Year ended June 30, 1874. Washington: Government Printing Office, VII-XLIX. Zimmermann, F. W. R. (1908). Die deutsche Handelsstatistik in ihrer geschichtlichen Entwickelung und ihrem derzeitigen Stand, Jahrbu¨cher fu¨r Nationalo¨konomie und Statistik, III. Folge, 35, 289–324, 433–473.

154

MARKUS LAMPE

APPENDIX 1. COMMODITY GROUPS Wheat; Rye; Milling products (meals and flours); Wood, lumber and timber, planed, sawed (incl. firewood, excl. staves, dyewood & tropical); Hides, skins, and leather (excl. furs); Wool (sheep’s, goats’, etc.); Silk (raw, floss, tram, organzine, sewing); Pig iron (and scrap iron); Bar iron and steel (incl. plates, wires, rails, tinplate, iron girders, etc.); Simple products of iron and steel (chains, tubes, anvils, nails, screws, structures, etc.); Articles of leather and rubber (caoutchouc and gutta-percha); Woollen and worsted yarn; Woollens and worsteds (cloth, stuffs, etc.); Cotton yarn; Cottons (cloth, fabrics, stuffs); Linen yarn and thread (and yarn of hemp, jute, phormium-tenax, etc., incl. twine); Linens (and cloth of flax and jute); Articles of silk (incl. mixed, and laces, trimmings, ribbons, tulles, muslins, passements and embroidery of other materials); Glass and glassware; Wine; and Spirits and liqueurs.

APPENDIX 2. HISTORICAL STATISTICS USED FOR THE DATASET United Kingdom: Statistical Department, Board of Trade (–1869)/Statistical Office, Customs and Excise Department (1871–). Annual Statement of the Trade and Navigation of the United Kingdom with Foreign Countries and British Possessions in the Year [1857–1875]. London: Her Majesty’s Stationary Office, various years. France: Direction Ge´ne´rale des Douanes [et des Contributions Indirectes (–1867)]. Tableau ge´ne´ral du commerce de la France avec ses colonies et les puissances e´trange`res pendant l’anne´e [1857–1875]. Paris: Imprimerie Nationale, various years. Germany (Zollverein): Central-Bureau des Zoll-Vereins. Statistische Uebersichten u¨ber Waaren-Verkehr und Zoll-Ertrag im Deutschen ZollVereine fu¨r das Jahr [1857–1871]. Berlin: Reimer, various years. Germany (Kaiserreich): Kaiserliches Statistisches Amt. Auswa¨rtiger und u¨berseeischer Waarenverkehr des deutschen Zollgebiets und der Zollausschlu¨sse im Jahre [1873, 1875]. Berlin: Verlag der Ko¨niglichen Statistischen Bureaus, various years (Statistik des Deutschen Reichs 9–11, 22–23). Germany (Hamburg): Handelsstatistisches Bureau. Tabellarische U¨bersichten des Hamburgischen Handels im Jahre [1857–1875]. Hamburg: Ku¨mpel, various years.

Bilateral Trade in Europe, 1857–1875

155

Germany (Bremen): Beho¨rde fu¨r die Handelsstatistik. Tabellarische U¨bersicht des bremischen Handels im Jahre [1857–1865]. Bremen: Strack, various years./Bureau fu¨r Bremische Statistik Jahrbuch fu¨r die Amtliche Statistik des Bremischen Staats vol. 1: Zur Statistik des Schiffs- und Waarenverkehrs im Jahre [1867–1875]. Bremen: G.A. v. Halem, various years. Belgium: Ministre des Finances. Tableau ge´ne´ral du commerce avec les pays e´trangers, pendant l’anne´e [1857–1875], Brussels: M. Hayez, various years. Netherlands: Departement van Financı¨ en. Statistiek van den handel en de scheepvaart van het Koningrijk der Nederlanden over het jaar [1857–1875]. ‘s Gravenhage: Gebroeders Giunta D’Albani, various years. Austria-Hungary: Direction der administrativen Statistik im Kaiserl. Ko¨nigl. Ministerium fu¨r Handel, Gewerbe und O¨ffentliche Bauten. Ausweise u¨ber den auswa¨rtigen Handel O¨sterreich’s und der mit O¨sterreich zollvereinten Staaten im Sonnen-Jahre 1857. Vienna: Kaiserlich-Ko¨nigliche Hof- und Staats-Druckerei, 1859/Rechnungs-Departement des Kaiserl.Ko¨nigl. Finanz-Ministeriums (1859)/Direction der Administrativen Statistik (1861)/K.K. Statistische Central-Commission (1863–1865). Ausweise u¨ber den auswa¨rtigen Handel O¨sterreich’s im Sonnen-Jahre [1859–1865]. Vienna: Kaiserlich-Ko¨nigliche Hof- und Staats-Druckerei, 1861–1866./K.K. Statistische Central-Commission. Ausweise u¨ber den auswa¨rtigen Handel der O¨sterreichisch-Ungarischen Monarchie im Sonnen-Jahre [1867–1875]. Vienna: Kaiserlich-Ko¨nigliche Hof- und Staats-Druckerei, 1869–1876. Austria-Hungary (Trieste): Uffizio di Borsa. Movimento della Navigazione e del Commercio in Trieste nell’ Anno Solare [1857–1869]/Commercio di Trieste nel [1871–1873], Movimento Commerciale di Trieste nel [1875]. Trieste: Tipografia Weis (1858–1870), Tipografia Peternelli & Morterra, (1872–1876). United States: Report of the Secretary of the Treasury, transmitting a report from the Register of the Treasury, of the commerce and navigation of the United States, for the [fiscal] year ending June 30, [1857–1865]/Report of the Secretary of the Treasury, transmitting a report from the Director of Statistics, on the commerce and navigation of the United States, for the year ending June 30, 1866/Annual Report of the Director of the Bureau of Statistics, on the Commerce and Navigation of the United States, for the year ended June 30, [1867, 1870–1876]/Annual Report of the Deputy Special Commissioner of the Revenue in Charge of the Bureau of Statistics, on the Commerce and Navigation of the United States, for the fiscal year ended June 30, [1868, 1869]. Washington: Gov. Print. Off, various years.

GENDER, HEALTH, AND WELFARE IN ENGLAND AND WALES SINCE INDUSTRIALISATION Bernard Harris ABSTRACT In recent years, a number of historians have examined the reasons for differences in the height and health of men and women in nineteenthcentury Britain, often drawing on economic studies which link excess female mortality in the developing world to restrictions in women’s employment opportunities. This paper re-examines this literature and summarises the existing literature on sex-specific differences in height, weight and mortality in England and Wales before 1850. It then uses two electronic datasets to examine changes in cause-specific mortality rates between 1851 and 1995. Although there is little evidence to support the view that the systematic neglect of female children was responsible for high rates of female mortality in childhood, there is rather more evidence to show that gender inequalities contributed to excess female mortality in adulthood.

Historians have recently devoted increasing attention to the different ways in which the economic and social changes of the late-eighteenth and nineteenth centuries affected the health of men and women, and adults and

Research in Economic History, Volume 26, 157–204 Copyright r 2008 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(08)26003-9

157

158

BERNARD HARRIS

children. Johansson (1977, 1991, 1996) argued that in poor rural areas, where there were very few opportunities for paid female employment, girls and women experienced disproportionately high rates of mortality in comparison with their male counterparts. Nicholas and Oxley (1993) argued that the decline in female employment opportunities in agricultural areas led to a deterioration in the average heights of women born between 1790 and 1815 (see also Johnson & Nicholas, 1995, 1997). Humphries (1991) concluded that even though economic factors appeared to have little effect on the relative mortality of female children, they exercised a profound effect on the life chances of female adults (see also McNay, Humphries, & Klasen, 1998). In constructing these accounts, Johansson, Nicholas, Humphries and their co-authors referred explicitly to the work of development economists such as Amartya Sen on the reasons for the existence of high rates of female mortality in many parts of the world today. It is now widely recognised that the sex ratio (i.e. the ratio of females to males) in many of the world’s poorer countries is lower than the ratio which one would expect to find on the basis of the figures for the ‘developed’ world, and it has been estimated that between 60 and 100 million women are ‘missing’ from the world’s population (Sen, 1989, 1990, 1992; Coale, 1991; Klasen, 1994; Klasen & Wink, 2002, 2003). Johansson (1977, pp. 165–166) suggested that disproportionately high rates of female mortality were characteristic of ‘traditional’ or ‘pre-modern’ societies and Klasen (1998, p. 433) argued that ‘these current trends have increased the interest of demographers and economic historians in the issue of female/male mortality ratios in Europe to determine whether there was similar gender bias in mortality during early stages of European development’. However, as McNay et al. (1998, pp. 1–2) have pointed out, ‘economic contributions do not determine female status and welfare mechanically’, and therefore one also needs to pay attention to the role of cultural variables in shaping patterns of female disadvantage. This also suggests that one needs to tread very carefully before assuming that there is a direct relationship between the causes of excess female mortality in Britain during the eighteenth and nineteenth centuries and the causes of excess female mortality in other parts of the world today. In addition to looking at the underlying causes of excess female mortality and other health inequalities, it is also important to look at their more immediate causes. In their study of the living standards of English women between 1795 and 1820, Nicholas and Oxley (1993, p. 737) argued that ‘in Africa and Asia today, women are disadvantaged in their access to food and health care’ and that ‘this work on the intrahousehold distribution of

Gender, Health, and Welfare

159

nutrients [emphasis added] reflects labour-market opportunities for men and women’, but many authors have questioned the extent to which inequalities in the distribution of food can be held responsible for differences in health outcomes today. Although Messer (1997, p. 1676) argued that ‘observational methods, sometimes combined with quantitative nutritional data, have offered supporting evidence that females, especially in northern India and southeast Asia, are nutritionally deprived’, Harriss’s (1990) comprehensive review of the intrafamily distribution of hunger in south Asia failed to show any consistent pattern of female disadvantage and DeRose, Das, and Millman (2000, p. 517) did not find any systematic bias against females ‘either in childhood or in adulthood’, although they did agree that ‘pregnant and lactating women are . . . at a disadvantage’. These findings have led many authors to conclude that the main causes of excess female mortality in childhood are less likely to be found in the differential provision of foodstuffs than in access to health care. Hill and Upchurch (1995, p. 144) found that ‘female mortality disadvantage seems unrelated to differences between girls and boys in sickness rates or anthropometric status, but shows a positive association with relative lack of immunisation coverage among girls, whilst showing a paradoxical negative association with a female disadvantage in diarrhoea treatment rates’. These findings have important implications for the debate about the causes of excess female mortality and female health disadvantage in the British past. It is generally accepted that most of the main forms of medical intervention which are likely to make a significant difference to the survival chances of boys and girls in the modern world were unavailable in Britain before the first half of the twentieth century, so inequalities in the provision of health care are unlikely to have exerted the same impact on mortality rates that might be found today (see, e.g., McKeown, 1976, pp. 91–109). However, it is possible that sex-specific differences in the provision of health care may have exerted a larger impact on adult mortality. In Sick, not dead, James Riley (1997, p. 197) argued that the number of people who obtained access to medical care through friendly societies increased substantially during the second half of the nineteenth century and that this may have contributed to the decline in mortality because people learned how to manage their illnesses more effectively. However, the vast majority of friendly society members were male, and there were significant differences in access to medical care between men and women before the creation of the National Health Service in 1948 (Oren, 1974, p. 233). There has also been increasing debate over the extent to which inequalities in the distribution of resources can be directly linked to women’s

160

BERNARD HARRIS

participation in the labour market, and some authors have questioned the assumption that the best way to improve the health of both girls and women is to increase women’s employment opportunities. Hare (1999, p. 1026) argued that even if women are given more opportunities to obtain paid work, ‘the wide gap between men’s and women’s wages may simply serve to reinforce existing differences within households’ and Jackson and Palmer-Jones (1999) highlighted the danger that increased access to certain types of work may tend to exacerbate health inequalities rather than redressing them. Doyal (2002, pp. 240–241) has pointed out that much of the work which women are now called upon to perform is ‘strenuous, monotonous and ergonomically unsound’. She recognises that ‘waged work can offer women important advantages in terms of both material and social support’, but also warns that many of these advantages ‘have to be paid for by threats to health’. One of the most contentious areas of debate concerns the relationship between economics and culture. As we have already seen, many economists have argued that the root causes of female disadvantage are directly related to women’s position in the labour market, but others have argued that it is at least as important, if not more so, to examine the role of cultural factors (see, e.g., Kishor, 1993, pp. 260–262, 1995, p. 30; Sen, 1989, pp. 26–28, 1990, p. 6). Dyson and Moore (1983) suggested that marriage customs and kinship models played a major role in determining sex-differences in mortality in north and south India. In northern India, it was much more common for daughters to marry husbands from outside their own areas, males tended to rely on other males from their own blood families for support, and females were unable either to inherit property, or to be the means through which property rights could be transferred to offspring. These differences meant that parents were much more likely to favour sons, since boys were more likely to be able to support their parents when they were older, dowries tended to be more expensive, and the lower status of women within their husbands’ households meant that they were less able to channel resources to female children. Kishor (1993, p. 261) questioned the extent to which this model could be applied to north–south differences as a whole, but she nevertheless concluded that ‘wherever patrilocal exogamy is practised, regardless of the level of women’s labour force participation, the probability of male survival in enhanced; by contrast, the probability of female survival is actually lowered at mean levels of female labour force participation, and at best is unaffected when rates of female labour force participation are exceptionally high’ (see also Murthi, Guio, & Dre`ze, 1995, p. 753). The impact of kinship systems on female survival chances has also been

Gender, Health, and Welfare

161

highlighted by Dasgupta, Palmer-Jones, and Parikh (1999) in the case of India, and by Elson (1995a, p. 16, 1995b) and Davin (1995, pp. 45–49) in the case of China (see also Agnihotri, Palmer-Jones, & Parikh, 1999). The impact of cultural factors on female mortality rates has also been examined by other authors. Le Grand and Mbacke´ (1995) claimed that the existence of excess female mortality among children in sub-Saharan Africa was directly related to the impact of traditional cultural practices such as female circumcision and sex-specific food taboos. Other writers have highlighted the relationship between excess female mortality and birth order. Das Gupta (1987, p. 96) argued that excess female mortality was much more common among high birth order girls than among lower birth order girls, leading her to conclude that ‘son preference is primarily culturally determined, and scarcity of resources may at most accentuate the effects of sex bias within a given culture’. These findings have since been echoed by Johansson and Nygren (1991) (in the case of China) and by Muhuri and Preston (1991) (in Bangladesh). As Sen (1989, p. 27) himself has argued: ‘the influence of female gainful employment on women’s survival prospects . . . may . . . look like . . . a simple economic explanation of the demographic difference[, b]ut it would be a mistake to see it thus. There is the deeper question as to why such outside employment is more prevalent in, say, sub-Saharan Africa than in North Africa, why more common in southeast and eastern Asia than in western and southern Asia. Here the cultural – including religious – backgrounds of the respective regions will surely have some considerable relevance’. The importance of cultural factors may help to explain some of the differences between eighteenth- and nineteenth-century Britain and the developing world today. Whilst it is tempting to be able to use the experience of many of today’s developing countries to help explain the British past, it is also important to note that there is much less direct evidence of discrimination against female children in the British past than there is of discrimination against female adults, or against female children in many parts of the developing world today (see Harris, 1998, pp. 414–417). There are also important differences in the demographic record. In many parts of the developing world, there is clear evidence of excess female mortality at very young ages (Chen, Huq, & D’Souza, 1981; D’Souza & Chen, 1980; Hammoud, 1977; Harriss, 1990, p. 367; Hill & Upchurch, 1995; Murthi et al., 1995; Tabutin & Willems, 1995; Waldron, 1985). In Britain in the nineteenth century, excess female mortality was much more strongly represented among older children and adults (Harris, 1998, p. 432; Johansson, 1977, p. 166, 1996, p. 55; McNay et al., 1998, pp. 3–4).

162

BERNARD HARRIS

This paper seeks to illustrate some of these points by taking a new look at the relationship between gender, work and welfare in historical Britain. The first section uses anthropometric evidence to investigate differences in the net nutritional status of males and females in the nineteenth and twentieth centuries, and section two examines differences in sex-specific mortality rates. Section three discusses the impact of agricultural and industrial change on the employment of women in both agriculture and industry and sections four and five explore some of the other factors which might help to explain the high ratio of female mortality to male mortality in the past when compared to today. Section six attempts to summarise some of the implications of this work for our understanding of the relationship between work, health and gender in eighteenth- and nineteenth-century Britain as well as in countries which are currently undergoing industrialisation.

1. ANTHROPOMETRIC INDICATORS OF FEMALE DISADVANTAGE As Eveleth and Tanner (1976, p. 1, 1990, p. 1) have pointed out, ‘the average value of children’s heights and weights reflects accurately the state of a nation’s public health and the average nutritional status of its citizens, when appropriate allowance is made for differences, if any, in genetic potential’. During the past two decades, this insight has played an increasingly important role in historians’ efforts to monitor changes in the health and well-being of past generations (Harris, 1994; Steckel, 1995; Engerman, 1997; Komlos & Coll, 1998). Whilst much of this research has tended to focus on the heights of adult males (using military recruitment records), there is now a growing interest in the heights of both women and children. In considering the use of data on height, it is important to remember that height is a cumulative measure of conditions affecting the life of the individual throughout the period of growth. There is evidence to suggest that heights are particularly sensitive to the effects of circumstances operating during the period of fastest growth – i.e. between the ages of 0 and 2, and during adolescence – but heights are also sensitive to the impact of adverse conditions at other ages (Harris, 1994, p. 302). It is also important to remember that height is a measure of net nutritional status, and not merely a function of dietary intakes (although these are clearly a major factor). It can be affected by the impact of external factors, such as

Gender, Health, and Welfare

163

the demands made on the body by the need to keep warm, by the number of calories expended in exercise and work effort, and – perhaps most importantly – by the effects of disease (Eveleth & Tanner, 1976, pp. 241–245, 1990, pp. 194–198; Van Wieringen, 1979, pp. 445–473, 1986, pp. 307–331). There is also some evidence to suggest that heights can be adversely influenced by psychological stress (Tanner, 1962, pp. 135–137, 1989, pp. 147–148). The basic patterns of growth in childhood are shown in Fig. 1. Both boys and girls grow rapidly in early childhood, and then more slowly between early childhood and adolescence. The rate of growth accelerates during adolescence. The onset of adolescence occurs approximately two years earlier in girls than in boys, and growth finishes sooner. The fact that boys experience an extra two years of growth before adolescence is largely responsible for differences between the heights of males and females in adulthood (Steckel, 1992, p. 276). Nicholas and Oxley (1993) sought to apply some of the insights derived from the first ‘wave’ of anthropometric historians to the study of gender differences in the heights of convicts who were born between 1790 and 1820, and transported to Australia between 1826 and 1840. They found that the average heights of rural-born women declined more sharply than the average heights of either urban-born women or rural-born men. 200 180

boys girls

height (cm)

160 140 120 100 80 60 0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5 10.5 11.5 12.5 13.5 14.5 15.5 16.5 17.5 18.5 age (years)

Fig. 1.

Individual-Type (Longitudinal) Growth Standards for Boys and Girls. Source: Tanner et al. (1966, Tables IIIA, IIIB, VA, VB).

164

BERNARD HARRIS

They concluded that this reflected the additional hardships imposed on women by the decline of female employment opportunities in agricultural areas (see also Nicholas & Steckel, 1991; Johnson & Nicholas, 1997). However, as we will see, the actual timing of the decline in female agricultural employment remains unclear. Although this research has received considerable attention, it has not met with universal approval. Jackson (1996) criticised Nicholas and Oxley for ‘attribut[ing] long-run significance to what is essentially a short-run movement in a fluctuating series’, and claimed that the decline which Nicholas and Oxley did observe was largely a function of changes in the average ages of the cohorts under investigation. In their response, Nicholas and Oxley (1996) denied that their results were affected by variations in average age, but they failed to refute the larger point – namely that they were seeking to attach long-run significance to what was essentially a shortterm change (see also Johnson & Nicholas, 1997, pp. 209–210). Nicholas and Oxley have also been criticised for failing to provide an adequate explanation for the trends they observed (Harris, 1998, p. 424). In their original article, they sought to explain the apparent deterioration in the heights of rural-born women by pointing to evidence from contemporary developing countries, which showed that female children often received a smaller allocation of household resources than their more fortunate brothers. However, they were unable to provide any direct evidence of similar discrimination in the English countryside during the period under review (Nicholas & Oxley, 1993, pp. 736–741; Johnson & Nicholas, 1997, pp. 213–218). In view of the doubts which have been raised about these findings, it is interesting to compare Nicholas and Oxley’s original work with the results published by Nicholas and Johnson on the basis of their subsequent examination of the heights of a sample of ‘habitual criminals’ who had been born between 1812 and 1857, and who were incarcerated in Newgate Gaol in 1877. When Nicholas and Johnson compared their results with the earlier figures produced by Nicholas and Oxley, they found that the average height of male criminals born in 1812/1813 (65.61 inches) compared very closely with the average height of male convicts (65.43 inches) (Johnson & Nicholas, 1995, p. 478). However, when they compared the average heights of female criminals with female convicts, they found that the female convicts were substantially shorter (Nicholas & Oxley, 1993, p. 734, 1996, p. 597; Johnson & Nicholas, 1997, pp. 211, 218, 221). The differences between the average heights of male convicts and male criminals and the average heights of female convicts and female criminals imply that at least one of these series

Gender, Health, and Welfare

165

is less representative than the others, and this in turn implies that conclusions regarding the differential impact of economic development on the welfare of males and females before 1815 should be treated with considerable caution. While it is interesting to note that Nicholas and Oxley’s original findings suggested that the average heights of rural-born females declined more rapidly than those of rural-born males, Nicholas and Johnson’s subsequent research into the heights of Newgate criminals showed that male and female heights moved ‘roughly in unison’ after 1815 (Johnson & Nicholas, 1995, 1997, p. 222). These findings are broadly consistent with the results obtained by Floud in his analysis of the trends indicated by a range of published datasets during the course of the nineteenth century. Whilst Floud was rightly cautious about drawing too many conclusions from some extremely limited samples, he was able to observe a rough similarity between the trends for males and females during those periods for which comparisons were possible (Floud, 1998, pp. 10–15). Although neither Johnson and Nicholas nor Floud were able to detect any significant differences in the main trends in the heights of men and women born after 1815, Riggs (1994) has argued that differences did exist in the average heights of men and women who were born between 1800 and 1849 and incarcerated in Glasgow between 1840 and 1880. He claimed that the average heights of both men and women declined between the birth cohorts of the early-1800s and the 1830s, but the average heights of women continued to decline with the birth cohorts of the 1840s, whereas those of men improved. He speculated that this result might reflect the impact of changes in the labour market for men and women in Glasgow during the 1840s, but he also conceded that the results for this decade ‘could . . . be an artefact of the data, perhaps stemming from subtle changes in policing and, therefore, in the initial sampling of the population from which this dataset is derived’ (p. 73, note 47). It may therefore be premature to argue that these data provide unequivocal evidence of sex-specific differences in the welfare of the Scottish population during this period (cf. Komlos & Coll, 1998, pp. 231–233; Komlos & Baten, 2004, pp. 202). In addition to examining the broad trends in height, Floud was also able to investigate the absolute value of differences in height, weight and body mass index when these figures were compared with modern standards.1 In some respects, these results were rather more surprising. It is now generally accepted that women and children often deprived themselves in order to ensure that the father of the household was adequately nourished (see, e.g., Oren, 1974, pp. 227–278; Harris, 1998, p. 418), but Floud (1998,

166

BERNARD HARRIS

pp. 10–16) found that men were more likely to be underweight than women, and that girls were only slightly less heavy, for their age and height, than boys. However, in a more recent study, Horrell, Meredith, and Oxley (2006, pp. 18–19) found that the body mass indices, or BMIs, of female convicts were significantly lower than those of their male counterparts, and the difference increased as the women grew older. Although cautious, their findings support the view that gender did play an increasingly important part in the allocation of household resources among the adult population.

2. GENDER AND MORTALITY The previous section has shown that the available anthropometric evidence does not provide convincing grounds for believing that gender differences in childhood were reflected in height and weight; however, this does not rule out the possibility that such differences may have been reflected in statistics of mortality. This section will consider some of the arguments relating to sex differences in mortality rates in both the pre-industrial and early-industrial periods, before going on to look at changes in age-specific sex mortality ratios from the mid-nineteenth century onwards. Wrigley, Davies, Oeppen, and Schofield (1997, pp. 296–300) examined sex differences in mortality rates in infancy, childhood, young adulthood and older adulthood. They found that there was little change in the ratios of female mortality to male mortality in either infancy or adulthood, but the ratio of female mortality to male mortality among older children (and especially among those aged 10–14) increased dramatically between 1800/1824 and 1825/1837 (see Table 1). They believed that the main reason for this change was a sharp rise in the incidence of respiratory tuberculosis. In 1861, this disease accounted for 16.2 per cent of all male deaths between the ages of 10 and 14, and for 29.8 per cent of all female deaths. The corresponding figures for those aged 15–19 were 38.7 per cent and 51.1 per cent, respectively. In addition to examining changes in the ratio of female mortality to male mortality in childhood, Wrigley et al. also noted the consistently high ratio of female mortality to male mortality among younger adults, and especially among those aged 25–34. It is likely that these mortality rates would also have reflected the differential impact of tuberculosis mortality, but the major cause was almost certainly mortality associated with childbirth. The combined effect of these two causes meant that, for most of the seventeenth and eighteenth centuries, married women of childbearing age experienced

1q0

1580–1599 1600–1624 1625–1649 1650–1674 1675–1699 1700–1724 1725–1749 1750–1774 1775–1799 1800–1824 1825–1837

4q1

5q5

5q10

M

F

F100/M

M

F

F100/M

M

F

F100/M

M

F

F100/M

174.9 162.6 148.8 160.9 195.0 195.0 207.4 172.9 165.9 146.4 151.3

163.0 141.7 132.0 141.9 160.6 182.7 172.3 151.8 146.0 124.5 136.3

93.20 87.15 88.66 88.19 82.36 93.69 83.08 87.80 88.00 85.04 90.09

87.0 76.9 103.3 114.4 107.8 110.3 121.3 100.5 106.0 97.0 98.4

81.2 86.1 95.8 106.9 107.0 105.2 121.0 114.5 108.2 100.3 98.9

93.33 111.96 92.74 93.44 99.26 95.38 99.75 113.93 102.08 103.40 100.51

49.3 39.2 47.5 46.0 46.0 47.7 53.9 41.9 33.7 23.6 37.7

43.2 32.9 48.4 55.7 45.7 45.1 46.4 40.3 37.1 27.8 30.8

87.63 83.93 101.89 121.09 99.35 94.55 86.09 96.18 110.09 117.8 81.70

18.1 20.3 28.6 26.7 26.9 24.2 29.4 26.4 20.2 19.6 23.2

19.7 25.3 25.4 25.6 28.0 30.0 27.0 25.0 26.5 20.6 46.5

108.84 124.63 88.81 95.88 104.09 123.97 91.84 94.70 131.19 105.10 200.43

Gender, Health, and Welfare

Age- and Sex-Specific Infant and Child Mortality, 1580–1837.

Source: Wrigley et al. (1997, 296pp.).

167

168

BERNARD HARRIS

mortality rates which were often twice as high as those of their male counterparts (Wrigley et al., 1997, pp. 301–307). Whilst Wrigley and his co-authors have undoubtedly enhanced our knowledge of the differences between male and female mortality in the preindustrial past, their analysis was largely concerned with the immediate causes of excess female mortality, without necessarily shedding much light on the circumstances that might have lain behind those causes. This is particularly true of the circumstances surrounding changes in male and female child mortality. Did the rise in the ratio of female death to male deaths in childhood reflect the fact that females were physiologically more susceptible to tubercular infections, or was their susceptibility increased by differences in the circumstances of male and female lives? One of the most forceful advocates of the latter view is Sheila Ryan Johansson (1977, pp. 174–178). She argued that the excess mortality of female children was largely related to the decline in employment opportunities for women and older girls in agricultural areas. In order to illustrate this, she compared male and female mortality rates in the rural district of Stratton with the corresponding rates in the industrial district of Redruth. In the rural district, female mortality rates exceeded male mortality rates at all ages between 5 and 19, and between the ages of 55 and 64, whereas, in the more industrial district, male mortality rates exceeded female mortality rates at all ages except between the ages of 5 and 9. This was not an isolated example of a ‘male-favouring’ mortality pattern in rural areas. In the majority of cases, rural mortality rates were normally lower than urban mortality rates, but Kearns (1993, pp. 93–95) found that rural mortality rates exceeded urban mortality rates among females between the ages of 10 and 24 in 1851/1860, and between the ages of 15 and 24 in 1891/1900. Kearns also showed that the ratio of female mortality to male mortality in a selection of ‘rural districts’ was consistently greater than the ratio of female mortality to male mortality in a sample of 25 ‘large towns’ (Table 2). Even at the end of the nineteenth century, the ratio of female mortality to male mortality was significantly higher in rural areas than in urban areas (Table 3). However, even though these figures provide strong evidence of a malefavouring mortality pattern in rural areas, it is clear that this factor alone cannot account for either the pattern or the extent of excess female mortality in Victorian England. Table 4 shows that the ratio of female mortality to male mortality at older ages began to fall from the mid-1860s, but the ratio of female mortality to male mortality at younger ages remained high throughout the whole of the second half of the nineteenth century.

Age Group

Large Towns Male

Rural Districts Female

Male

Female Mortality as % of Male Mortality Female

Large towns

Rural districts

Gender, Health, and Welfare

Male and Female Mortality Rates in Selected Large Towns and Rural Districts in England and Wales, 1851–1900.

1851–1860 1891–1900 1851–1860 1891–1900 1851–1860 1891–1900 1851–1860 1891–1900 1851–1860 1891–1900 1851–1860 1891–1900 0–1 1–4 5–9 10–14 15–19 20–24 25–34 35–44 45–54 55–64 65–74 75 & over

209.50 56.10 11.00 5.80 8.10 9.80 11.10 15.70 24.20 42.20 83.40 188.20

194.50 31.60 4.90 2.70 4.30 5.50 7.70 13.90 23.40 43.30 83.00 169.70

174.00 55.30 10.80 5.40 7.60 8.60 11.00 14.40 19.40 34.00 70.80 170.70

162.90 30.50 4.90 2.70 3.80 4.70 6.80 11.60 18.30 34.70 70.00 153.50

144.40 21.20 6.90 4.30 5.40 8.20 7.90 8.70 11.90 22.00 54.60 157.80

122.10 12.80 3.20 2.00 3.20 4.70 5.50 7.60 11.60 23.20 55.20 153.80

116.20 20.50 7.10 5.50 8.40 10.10 10.10 10.80 12.30 22.10 51.30 149.10

94.00 12.00 3.20 2.40 4.20 5.10 5.70 7.60 10.40 21.20 51.20 142.10

83.05 98.57 98.18 93.10 93.83 87.76 99.10 91.72 80.17 80.57 84.89 90.70

83.75 96.52 100.00 100.00 88.37 85.45 88.31 83.45 78.21 80.14 84.34 90.45

80.47 96.70 102.90 127.91 155.56 123.17 127.85 124.14 103.36 100.45 93.96 94.49

76.99 93.75 100.00 120.00 131.25 108.51 103.64 100.00 89.66 91.38 92.75 92.39

Source: Kearns (1993, pp. 93–95).

169

170

Male and Female Mortality Rates in Urban and Rural Counties in England and Wales, 1891–1910. Age Group

Urban Counties Male

Rural Counties

Female

Male

Female Mortality as % of Male Mortality Female

Urban counties

Rural counties

1891–1900 1901–1910 1891–1900 1901–1910 1891–1900 1901–1910 1891–1900 1901–1910 1891–1900 1901–1910 1891–1900 1901–1910 72.05 4.78 2.61 3.97 5.06 7.10 12.98 21.86 40.35 79.49 166.67

56.78 3.84 2.21 3.21 4.15 5.71 9.98 18.48 36.09 72.53 159.73

61.02 4.83 2.63 3.64 4.46 6.36 10.70 16.83 32.30 67.47 151.92

47.85 3.97 2.29 2.91 3.50 4.91 8.28 14.12 28.00 59.43 141.76

Source: Parliamentary Papers (1919, p. ccxii).

46.41 3.48 2.23 3.55 5.23 6.17 8.87 14.09 27.65 60.80 154.99

37.17 2.89 1.85 2.87 4.34 5.27 7.48 12.32 25.46 56.24 147.81

37.73 3.55 2.53 3.95 4.79 5.95 7.96 11.49 23.06 53.10 141.40

29.91 2.95 2.02 3.07 3.84 4.81 6.43 9.94 20.37 47.68 131.37

84.70 101.13 100.54 91.71 88.27 89.64 82.46 76.97 80.06 84.87 91.15

84.27 103.31 103.58 90.51 84.49 85.87 82.92 76.38 77.57 81.94 88.75

81.29 102.04 113.59 111.30 91.63 96.39 89.77 81.52 83.41 87.34 91.23

80.46 102.11 109.01 106.86 88.42 91.31 85.95 80.66 80.01 84.78 88.87

BERNARD HARRIS

0–4 5–9 10–14 15–19 20–24 25–34 35–44 45–54 55–64 65–74 75 & over

Female Mortality as a Percentage of Male Mortality, 1838/1842 to 1978/1980. Period

1838–1842 1843–1847 1848–1852 1853–1857 1858–1862 1863–1867 1868–1872 1873–1877 1878–1882 1883–1887 1888–1892 1893–1897 1898–1902 1903–1907 1908–1912 1913–1917 1918–1922 1923–1927 1928–1932 1933–1937 1938–1942 1943–1947 1948–1952 1953–1957 1958–1962 1963–1967 1968–1972 1973–1977 1978–1980

Mid-Year of Series 1840 1845 1850 1855 1860 1865 1870 1875 1880 1885 1890 1895 1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1979

0–4

5–9

10–14

15–19

20–24

25–34

35–44

45–54

55–64

65–74

75–84

X85

All Ages

(f100/m) (f100/m) (f100/m) (f100/m) (f100/m) (f100/m) (f100/m) (f100/m) (f100/m) (f100/m) (f100/m) (f100/m) (f100/m) 85.81 85.57 86.26 86.96 86.56 86.80 86.76 84.87 85.11 84.18 83.90 84.39 84.05 83.85 83.58 82.46 81.92 80.69 79.62 79.32 78.83 78.59 77.67 77.78 77.99 78.20 74.89 75.96 79.05

97.27 97.01 97.26 98.05 100.00 95.06 93.05 93.58 94.27 98.52 99.58 101.84 103.00 102.87 101.23 98.25 101.13 92.68 89.08 91.82 83.51 75.76 68.42 65.38 62.50 65.22 75.00 70.59 66.67

Source: Mitchell (1988, pp. 60–65).

108.88 107.29 104.89 100.81 105.75 100.43 99.53 100.54 101.20 105.96 106.77 105.69 105.31 106.86 104.08 101.87 106.25 98.73 93.75 91.30 83.33 80.00 75.86 72.73 70.00 75.00 55.00 66.67 66.67

112.98 110.54 112.22 110.29 110.03 106.60 105.69 103.03 104.74 103.18 100.00 97.37 92.05 94.81 92.47 89.51 95.60 96.06 90.77 89.38 82.05 78.89 74.07 54.76 45.45 40.00 44.44 40.91 34.62

95.72 93.92 97.67 96.01 97.37 93.24 91.87 93.22 96.17 97.59 92.59 89.60 83.06 84.13 85.79 59.29 71.80 90.85 89.02 87.33 61.95 54.12 80.56 50.00 42.86 43.40 44.68 40.00 42.86

103.23 105.31 106.50 103.07 104.58 98.05 92.37 92.37 94.26 96.43 92.45 92.12 85.67 85.05 84.40 62.93 65.85 88.42 92.13 92.99 72.93 66.46 88.10 75.00 66.67 62.96 60.00 58.33 62.96

99.36 99.06 100.61 97.79 95.62 89.66 85.78 83.81 84.92 85.87 84.28 84.23 81.27 82.79 80.98 73.13 73.32 76.04 76.95 80.54 72.51 71.15 78.00 77.34 74.17 72.13 71.30 70.09 69.49

87.36 87.83 87.49 84.85 83.82 81.48 79.63 77.80 77.71 79.23 76.38 78.06 77.19 77.27 77.07 75.10 75.88 74.65 71.92 70.34 66.79 64.38 63.29 61.26 60.99 60.00 61.06 60.73 61.62

88.04 89.66 89.08 88.01 86.62 83.92 83.32 81.65 82.47 82.74 81.41 81.67 79.61 78.01 77.20 73.42 74.51 74.58 73.38 69.98 63.47 59.29 55.21 51.31 49.40 48.27 49.43 51.52 52.50

88.59 90.17 90.28 89.26 90.01 88.62 87.16 86.98 87.35 86.10 85.82 86.09 85.05 83.77 81.51 77.52 76.74 77.31 75.55 75.34 71.97 68.92 64.14 58.91 56.01 54.00 51.66 51.60 52.04

90.81 91.77 91.67 90.92 91.86 92.24 90.89 89.68 88.97 88.83 89.88 89.03 89.09 87.10 84.93 81.91 83.37 82.03 81.66 78.97 75.88 77.71 77.10 71.24 70.26 67.66 64.91 64.28 62.34

93.16 93.36 93.97 92.29 92.86 90.56 90.61 91.67 89.15 88.63 89.04 91.48 90.99 89.30 88.62 87.96 90.95 87.92 86.00 88.38 89.48 88.54 84.16 86.08 86.74 80.55 79.75 80.67 81.41

92.68 93.07 93.60 92.65 91.96 90.09 89.01 88.17 88.92 89.39 88.79 88.76 87.96 88.03 88.34 79.21 83.11 88.24 88.56 89.06 80.20 75.90 87.34 87.03 89.53 89.87 91.26 93.64 93.48

172

BERNARD HARRIS

It is difficult to reconcile these figures with the view that the existence of excess female mortality at younger ages was solely related to the economic condition of women in rural areas, when the proportion of the population which was living in rural areas continued to decline. These arguments gain further support from the work of Jane Humphries (1991, pp. 465–468). Her analysis of the relationship between education, economic development, factory employment and population density and the pattern of age-specific and sex-specific mortality rates in England and Wales in 1841 showed that there was little relationship between any of the economic variables and the pattern of sex-specific mortality among those aged 5–15, although there was a much stronger relationship between these variables and the pattern of sex-specific mortality between the ages of 15 and 30, and 30 and 60. McNay, Humphries and Klasen have since published two more extensive analyses of the relationship between a range of economic and contextual variables and the pattern of sex-specific mortality at all ages from birth to 54. Their findings suggest that neither the contextual variables nor the economic variables had much impact on the pattern of sex-specific mortality between the ages of 0 and 9, and that ‘the relative importance of the economic variables . . . only clearly emerges in the 15–19 group and beyond’ (McNay et al., 1998, p. 24, 2005, p. 654).2

3. WOMEN’S WORK IN THE EIGHTEENTH AND NINETEENTH CENTURIES Although various authors have highlighted the relationship between work and welfare in the nineteenth century, they have done so in different ways. Nicholas and Oxley (1993, pp. 732–739) argued that the position of women in the agricultural labour market declined in the late-eighteenth and earlynineteenth centuries and that this led to a deterioration in the average heights of women born during this period. However, McNay, Humphries, and Klasen (1998. p. 24, 2005, p. 654) found little evidence to suggest that variations in female employment rates had a profound effect on the mortality rates of females below the age of 15. Their findings suggested that inequalities in access to paid employment only began to exercise a significant effect on the health of the female population as they approached adulthood. In view of these differences, it is obviously worth conducting a more detailed investigation into the history of female employment in the eighteenth and nineteenth centuries. This section will argue that

Gender, Health, and Welfare

173

Nicholas and Oxley exaggerated the extent to which there was a decline in female employment opportunities in agricultural areas between 1795 and 1820 and that there is no clear evidence of a consistent decline in female employment across the country as a whole before circa 1850. However, this did not mean that women had equal access to the labour market or that they received an equal reward for their labours. They consistently received lower wages than men and much of their work was physically arduous. These factors may well have affected the health and well being of adult women across the whole of the period. Before moving on to consider the extent to which women’s employment opportunities changed at the end of the eighteenth century, it is important to recognise that the undervaluing of women’s work has a long history. Oren (1974, p. 221) has argued that ‘the tradition of giving males more food was a very old one in England. It applied where both [men and women] were working in the same field and did not seem ‘‘reasonably correlated’’ with energy needs or productivity’. Marshall (1788, pp. 387–389) found that in Pickering, women received 10d a day for reaping, whilst men received two shillings a day for binding the sheaves, and Eden (1797, p. 47) claimed that in Cumberland, even though women performed a large part of the work, they only received half as much pay. When one considers the combined effects of hard work, lower wages and the unequal distribution of both household resources and household responsibilities, it is easy to understand why the women of traditional rural communities should have struck outsiders as ‘thin, sallow and prematurely old’ (Richards, 1974, p. 341). In recent years, a number of authors have attempted to estimate changes in the value of the wages paid to male and female workers in agricultural areas during this period. Verdon (2002, p. 48) published data taken from the General Views of Agriculture for the period 1793–1813. Her research suggests that, in the majority of areas, women’s wages rose less rapidly than men’s wages with the result that the ratio of female wages to male wages fell from approximately 46 per cent to 40 per cent, but there were exceptions. In Gloucestershire, the average value of women’s wages appears to have increased relative to male wages and the figures for Lincoln and Kent show little change. However, in the majority of cases where data exist for the same county at different points in time, a noticeable decline can be observed. These data are consistent with the view that there was a decline in the relative value of women’s wages in agricultural areas between 1790 and 1815, but this may have owed more to the impact of the Napoleonic Wars than to any long-term changes in either wages or employment opportunities. Snell (1985, appendix) suggested that the average value of male wages in a

174

BERNARD HARRIS

range of counties in the south-east of England rose by 101.4 per cent between 1736/1740 and 1796/1800, whilst the average value of female wages rose by only 36.64 per cent over the same period. However, Snell’s figures also showed that the decline in the relative value of female wages rates before 1800 was not universal and that – in those countries where female wage rates did decline in comparison with male rates – the decline did not continue unabated after that date. In the western counties, male wages rose by 58.8 per cent between 1735/1740 and 1796/1800, whilst female wages rose by 100.4 per cent. In the south-eastern counties, male wages fell by 19.3 per cent between 1796/1800 and 1831/1835, whilst female wages rose by just under 11 per cent. It can also be very difficult to reach any general conclusions about the timing of changes in the pattern of female employment in agriculture between 1750 and 1850. Snell (1981, p. 413, 1985, p. 22) argued that the decline in female agricultural employment began ‘almost certainly between 1750 and 1790’, but Pinchbeck (1930, pp. 62–65, 84–86, 110) claimed that female employment increased during the Napoleonic period and again after 1834, and only began to decline consistently from the 1850s onwards (see also Sharpe, 2007, pp. 52–64). At present, it is difficult to reach any definite conclusions about national trends, but it seems clear that women continued to make an important financial contribution to the economic welfare of labouring families in agricultural areas for much of the nineteenth century (Verdon, 2002, pp. 196–197; Goose, 2007, pp. 12–13). Although there seems little doubt that female participation in agriculture declined at some point during the eighteenth and nineteenth centuries, it did not disappear altogether. Miller (1984, p. 152) has argued that ‘in some arable and mixed farms in the county [of Gloucestershire] women were performing between twelve per cent and 33 per cent of the day work of the farm in the 1870s, eighties and early-nineties, and that about one quarter of these women were working on a basis similar to that of their male counterparts’. Bythell (1993a, pp. 41–42) has also pointed out that ‘women and girls continued to be much in demand for part-time and seasonal work – weeding, thinning, stone-gathering and so on. In some eastern districts this led to the organisation of the notorious mobile ‘‘gangs’’ which so outraged respectable opinion around the middle of the century’. Much of the work which was performed by these workers was extremely arduous, and was often carried out under appalling conditions (see, e.g., Samuel, 1977, p. 24). In 1843, one woman told the Royal Commission on the Employment of Women and Children in Agriculture that ‘hay-making is hard work, very fatiguing, but it never hurt me. Working in the fields is not such hard work

Gender, Health, and Welfare

175

as working in the factory’. However, another worker was less sanguine. She told the Commission that ‘in making hay I have been strained with the work. I have felt it sometimes for weeks; so bad I could not get out of my chair. In leasing [i.e. gleaning], in bringing home the corn, I have hurt my head and been made deaf by it’ (Parliamentary Papers, 1843, pp. 65–68; Simonton, 1998, p. 132; Verdon, 2002, p. 72). As employment opportunities in agriculture declined, the importance of domestic service increased. According to Bythell (1993a, p. 42), more than 40 per cent of the employed women whose occupations were recorded in the 1851 census were employed as domestic servants, and – although it is important not to be too dogmatic on this point, because of definitional changes – the proportion remained at around this level throughout the second half of the nineteenth century (see also Simonton, 1998, p. 98; Schwarz, 1999, p. 238). Some authors have suggested that young women chose to enter domestic service because it was a useful way of acquiring domestic skills whilst earning their own living and preparing for married life, whereas others believe it was an occupation of last resort, which young women in their late-teens and early-twenties were compelled to take up because of the absence of suitable alternatives. Bythell (1993a, pp. 43–44) has pointed out that in Rochdale, where local women really did have a choice between domestic service and manufacturing, the vast majority opted for the latter. However, in Halstead, Essex, many women alternated between factory work (in Courtauld’s) and domestic service, and ‘it was not uncommon for young girls to try both’ (Simonton, 1998, p. 101; see also Goose, 2007, pp. 6–7).3 It is perhaps surprising that historians who have investigated the relationship between work and welfare should have failed to pay more attention to the question of domestic service, which was after all the largest single source of paid employment for women in the middle years of the nineteenth century (Goose, 2007, pp. 8–9). Many contemporaries and historians have claimed that the work was relatively well-remunerated. According to Higgs (1986, pp. 137–138), the average annual income of a domestic servant in the 1830s compared quite favourably with that of a female textile worker, once the cash value of room and board had been taken into account. However, the life of a domestic servant was far from easy. In 1873 it was calculated that the average housemaid’s day lasted from 6 a.m. to 10 p.m., with an hour’s break for meals and an hour and a half for ‘needlework’ in the afternoon. The work was both physically arduous and psychologically testing. Many domestic servants found that they were cut off from the outside world, and unable to

176

BERNARD HARRIS

communicate with their employers (Roberts, 1984, pp. 29–33; Lewis, 1984, pp. 56–57; Verdon, 2002, pp. 88–89). Although historians have continued to argue over the precise nature of trends in female agricultural employment, the position regarding manufacturing industry is equally complex. It seems clear that the early stages of industrialisation were almost certainly characterised by an increase in the number of job opportunities for women outside the agricultural sector. This was partly because the majority of the new jobs were based on outwork rather than factory production, and partly because many of the most dynamic sectors of the economy, such as textile production, were sectors which had traditionally relied on female labour (Richards, 1974, pp. 344–347; Berg, 1993). However, by the middle years of the nineteenth century, many of the circumstances, which had encouraged the growth of female employment, had begun to change. The gradual replacement of the domestic system by factory production, the activities of male trade unionists, the growing pervasiveness of the ideology of ‘separate spheres’, and the introduction of protective legislation which specifically targeted women and children, all contributed to a sharp reduction in female labour force participation rates during the second half of the nineteenth century (Jordan, 1989; Horrell & Humphries, 1995a, p. 112). In recent years, economic historians have begun to devote increasing attention to the wages earned by women in industry, as well as agriculture. Maxine Berg (1993, pp. 31–33) has suggested that ‘it is generally assumed that women by custom received one-third to one-half the wages of men’, and, even though the early expansion of proto-industrial employment enabled some women to earn higher wages than male agricultural workers, they almost invariably earned less than men engaged in similar branches of manufacture. Peter Lindert (1994) argued that women’s wages may have increased more slowly than men’s wages, and certainly did not increase more rapidly than men’s wages, between 1750 and 1850, but the overall impression – as Sara Horrell (2006, p. 9) has recently pointed out – ‘is of a decline in women’s relative earnings although there may have been a brief period of improvement in the early-nineteenth century’ (see also Horrell & Humphries, 1997). However, it is important to distinguish between the absolute value of women’s wages and the importance of the contribution they were able to make to the household budget. Horrell and Humphries (1992, pp. 867–871) have shown that there were considerable variations in the relationship between male earnings and total family incomes between high- and low-wage agricultural counties, and between different branches of manufacturing industry. However, their overall conclusion was that

Gender, Health, and Welfare

177

children’s earnings were more important than the earnings of adult women, and that although family incomes played a vital role in enabling English workers to maintain their living standards before 1820, they became less important thereafter (see also Horrell & Humphries, 1995a, pp. 100–108, 1995b, p. 491; Verdon, 2002, pp. 11–12, 43–45, 61–62).4 Although female workers played a declining role in the industrial workforce, many women continued to be employed in manufacturing industry after 1850, particularly as outworkers (Bythell, 1993b, p. 145). As Jane Lewis (1984, p. 57) has shown, the availability of such work played an important role in enabling many women to contribute to the household economy, especially in circumstances where male wage rates were low, or total household income was reduced by sickness, unemployment, or the absence of a male breadwinner. However, the work itself was poorly remunerated, and the combination of long hours and unhealthy working conditions is likely to have imposed its own burden on their health and well-being (ibid., pp. 59–60). Many of the industries, which did employ women, such as matchmaking, textiles and pottery work, were themselves notoriously unhealthy, as the Physician to the North Staffordshire Infirmary noted in a report on conditions in the pottery industry in 1861: The potters . . . both men and women . . . represent a degenerated population, both physically and morally. They are as a rule stunted in growth, ill-shaped and ill-formed in the chest. They are certainly short-lived; they are phlegmatic and bloodless . . . of all disease, they are especially prone to chest diseases, to pneumonia, phthisis, bronchitis and asthma. One form would appear peculiar to them, that which is known as potter’s asthma or potter’s consumption (quoted in Wohl, 1984, p. 274).

In addition to the costs of outwork and the demands made by specific industries, it is also worth remembering the amount of sheer physical effort associated with many of the jobs, which women continued to perform. As T.D. Acland noted in 1851, when he described the gendered division of labour in the Somerset cheese industry, ‘the women do all the work . . . it is a sad sight to see a man standing by doing nothing while his wife or daughter is turning many times a day a weight about half a hundredweight [56 lbs, or 123.2 kilogrammes]’ (quoted in Samuel, 1977, p. 25). Raphael Samuel (1977, p. 31) made a similar point in his account of working conditions in the nineteenth-century paint industry: Paint was a standard product used in great quantities. It should have been a prime candidate for self-acting machinery, especially in view of the known dangers of lead-colic and ‘drop-hand’. But, in fact, machinery was only used for crushing and grinding – in most factories all the other processes were done by hand. The women who did the stacking (men, it seems, could not be found to do the work because it

178

BERNARD HARRIS

had such a bad name) had to carry heavy pots of lead on their heads, from 30 to 50 lbs in weight, and to climb ladders from 10 to 15 feet high. They did the work, as a Newcastle manufacturer gratefully acknowledged, with ‘truly marvellous skill and rapidity’.

4. ANALYSING LONG-TERM CHANGES IN CAUSE-SPECIFIC MORTALITY The previous section has highlighted some of the problems involved in attempting to relate variations in female employment opportunities to differences in the health of males and females at different ages. As Jane Humphries and others have argued, there is little evidence to suggest that economic factors had a direct effect on the health of girls and young women below the age of 15, although they may well have had an effect at older ages. However, in order to continue to this investigation, it is obviously desirable to look more closely at the epidemiological factors associated with high sexmortality-ratios at different ages. In a previous paper, Harris (1998) used Logan’s (1950) data to examine cause-specific changes in mortality for males and females at different ages, but it is now possible to extend this analysis using Woods’ (1997) collection of nineteenth-century mortality data from the Registrar-General’s Decennial Supplements and the twentieth-century mortality statistics published by the Office of National Statistics (1997). It has been known for many years that the Registrar-General’s statistics are an imperfect guide to mortality change. Rumsey (1875) highlighted a range of problems associated with the certification of death and criticised several aspects of the way in which different causes of death were categorised, and Luckin (1980, p. 55) has noted that ‘the ever-changing categorisation of disease during the first twenty years or so of legislation, and the failure to achieve a convincing differentiation between several of the most important infections, involves the researcher in a constant battle to ensure that like is being compared with like’. Carter (1997, p. 197) has also warned that ‘one must resist the false sense of security that comes from reading texts in languages superficially similar to our own’ and that ‘any comparative conclusions based on reports from more remote periods or . . . cultures must be approached with extreme scepticism’. However, despite this, most researchers believe that the information contained in the Decennial Supplements can be used, provided their limitations are recognised (see, e.g., Hardy, 1994, pp. 491–492). As has often been noted, there is an important difference between the certification of death and the allocation of a ‘principal cause’ of death to a

Gender, Health, and Welfare

179

particular disease category. Rumsey (1875, p. 150, note 1) provided a famous illustration of this point when he quoted the following example of a death certificate supplied by a Gloucester Medical Officer in 1874: ‘Causes of death: (1) alcoholism; (2) fracture of the jaw; (3) bronchitis; (4) Morbus Brightii; (5) inguinal hernia; (6) cerebral haemorrhage. Duration of disease – (a) two months and fourteen days; (b) two days’. As Eyler (1976, p. 352) has pointed out, such certificates must have posed a considerable problem for the compiler of medical statistics in London, and even if such difficulties could be overcome, ‘the causes of death in the mortuary registers were at best only modes of death. The real causes might remain unknown or wilfully suppressed’. These problems are compounded by changes in diagnosis, either as a result of changing social mores (such as an increasing willingness to acknowledge the existence of potentially embarrassing diseases) or improved techniques. Hardy (1994) identified a range of conditions, which were affected by diagnostic changes in the second half of the nineteenth century, including tuberculosis. During the 1850s and 1860s, it seems likely that a significant proportion of deaths which might have been attributed to this disease were either unclassified, or attributed incorrectly to causes such as typhoid or bronchitis (p. 490). This implies that the official statistics are likely to underestimate the extent of the reduction in tuberculosis mortality during the second half of the nineteenth century, whilst overestimating the extent of changes in bronchitis and typhoid mortality. This is not an isolated example. Hardy also shows that improvements in the quality of diagnosis are likely to have played a significant role in the recorded incidence of mortality from a number of diseases, including diabetes, diseases of the genito-urinary system and, most importantly perhaps, cancer. She also argues that ‘the certification of deaths associated with childbirth (that is, with pregnancy as well as birth) is another grey area in the statistical series’. During much of the nineteenth century, it seems likely that many medical practitioners sought to conceal evidence of maternal deaths under more general headings in order to conceal evidence of their own incompetence, and Hardy concludes that ‘the tightening up of certification procedures over the last twenty years of the century contributed to a raised profile for deaths in childbirth, and notably to the maximum in deaths from puerperal sepsis recorded in the decennium 1881–90’ (Hardy, 1994, pp. 484–492). It is also necessary to acknowledge the problems, which are likely to arise whenever one tries to translate original cause-of-death data into alternative disease classifications. Alter and Carmichael (1996, p. 48) have argued that

180

BERNARD HARRIS

‘there has been a natural tendency among demographers to try to fit the descriptions of deaths in earlier periods into the current International Classification of Diseases’ and that ‘most historians of medicine reject this approach and find the results unpersuasive for both present and historical concerns’. However, in a more recent paper, Anderton and Leonard (2004) have attempted to test this argument by comparing the ‘literal’ causes of death in two Massachusetts towns between 1850 and 1912 with the results which would have emerged if these data were reclassified according to the Ninth Revision of the ICD. Although they recognised that some information was bound to be lost or distorted in the process of translation, they nevertheless concluded that both approaches had value: ‘In general, literal causes of death offer a more adequate picture of disease as understood at the time’, but ICD codes ‘provide a view of disease trends that were perhaps not fully understood or of recognised importance’ in the eyes of contemporaries (p. 138). One of the most important questions raised by the use of ICD codes concerns the increasing sophistication of cause-of-death classifications. In 1851/1860, the Registrar-General summarised the main causes of death under 26 separate headings (smallpox; measles; scarlet fever; diphtheria; whooping cough; typhus; enteric fever; simple continued fever; puerperal fever; diarrhoea and dysentery; cholera, diarrhoea etc.; cancer; scrofula; tabes mesenterica; phthisis; hydrocephalus; diseases of the nervous system; diseases of the circulatory system; diseases of the respiratory system; diseases of the digestive system; diseases of the urinary system; diseases of the generative system; childbirth; suicide; other violent deaths; and other causes), which were reduced to 23 categories in Woods’ (1997) electronic dataset. However, by 1901/1910 the number of headings had increased to 192, and in 1979/1995 the total was 5,292 (Office of National Statistics, 1997). One of the consequences of this is that when we attempt to express data from the earlier periods in terms of the classification system used in the latest period, we cannot avoid the risk that some of the deaths attributed to particular causes in the first period might have been allocated differently had they been diagnosed more precisely. When Logan examined the causes of mortality in England and Wales between 1848 and 1947, he classified the different causes of death under the following headings: infectious diseases; cancer; diabetes mellitus; anaemia; diseases of the nervous system and sense organs; diseases of the circulatory system; diseases of the respiratory system; diseases of the digestive system; non-venereal diseases of the genito-urinary system and annexa; diseases of pregnancy, childbirth and the puerperal state; suicide; and other violence

Gender, Health, and Welfare

181

(Logan, 1950, pp. 138–139, 146–147, 150–153, 156–159, 162–165; see also Harris, 1998, pp. 435–436, 439–442, 444). In this paper, we have assigned each of the different causes of death identified by the Registrar-General in 1851/1860 and 1901/1910 to one of the 17 ‘chapters’ of the Ninth Revision of the International Standard Classification of Diseases (World Health Organisation, 1977–1978), and then compared these with the pattern of mortality for males and females between 1979 and 1995. In theory, it would have been possible to reclassify the causes of death in all three periods using the Tenth Revision of the ICD, but this would have involved a great deal of extra effort and is unlikely to have made a significant difference to the overall picture (see also World Health Organisation, 1992, pp. 1–2; Anderton & Leonard, 2004, p. 140, note 3).

5. GENDER AND THE CAUSES OF MORTALITY DECLINE Previous accounts of changes in the ratio of female to male mortality have tended to fall into two camps. Both Henry (1987, 1989) and Wrigley et al. (1997) have argued that these changes are largely attributable to epidemiological changes. In other words, they argue that the cause-structure of mortality in the mid-nineteenth century was dominated by diseases which were associated with higher levels of female vulnerability, and that the decline in the importance of these diseases altered the ratio of female to male mortality by causing female mortality to decline more rapidly. The alternative view suggests that even if females are more susceptible to certain diseases than males, higher levels of mortality from these diseases still provide an index of female disadvantage in past societies (Johansson, 1977; Alter, Manfredini, & Nystedt, 2004). Harris (1998) used Logan’s published data to examine changes in causespecific mortality among males and females in six different age groups between 1848/1872 and 1947. The results suggested that the patterns of change among those aged 0–1 were very similar to those found in the age groups 1–4 and 5–14, and the patterns found among those aged 15–24 were similar to those found among people aged 25–44, but it was impossible to disaggregate the figures more fully using the information in Logan’s paper. However, the new data mean that it is now much easier to examine mortality changes in different age groups and to combine the information for those age groups in which the patterns of change are particularly similar.

182

BERNARD HARRIS

Consequently, the current paper only presents data for people aged 0–14, 15–44, 45–64, and 65 and over. A number of studies have highlighted the need to distinguish between the causes of high sex-mortality-ratios in childhood and the causes of high sexmortality-ratios in adulthood (Harris, 1998; Alter et al., 2004), and this impression is reinforced by the analysis of changes in cause-specific mortality between the ages of 0 and 14. Table 5 shows that the death rate from infectious diseases fell by 49 per cent for males and by 50 per cent for females between 1851/1860 and 1901/1910, and by just over 99.6 per cent for both males and females between 1901/1910 and 1979/1995, whilst death rates overall fell by 39 per cent for males and 40 per cent for females in the first period, and by just over 95 per cent for both sexes in the second period. However, the overall reduction in infectious disease mortality did lead to a reduction in the ratio of female mortality to male mortality. This reinforces Louis Henry’s (1987, p. 108, 1989, p. 196) view that the high ratio of female mortality to male mortality among children in the middle years of the nineteenth century was largely associated with physiological differences in the vulnerability of boys and girls to different infections. It is interesting to compare changes in sex-specific mortality rates in childhood with those among adults between the ages of 15 and 44. In 1851/1860, deaths from infectious diseases accounted for 56 per cent of all male deaths in this age group and 59 per cent of all female deaths, but the female death rate declined far more rapidly than the male death rate between 1851/1860 and 1901/1910, and this made a substantial contribution to the decline in the ratio of female deaths to male deaths over the course of this period.5 However, during the twentieth century, the number of male deaths from infectious diseases declined at much the same rate as the number of female deaths from infectious diseases, and the greater reduction in the overall number of female deaths was largely attributable to the greater reduction in the number of female deaths associated with respiratory diseases, the increase in the number of male deaths attributed to cancer, and the dramatic reduction in the number of female deaths associated with ‘complications of pregnancy, childbirth and the puerperium’ (see Table 6). If we ignore the impact of the reduction in pregnancy-related mortality, there are a number of important similarities between the changes in sexspecific mortality between the ages of 15 and 44 and those at higher ages. Among those aged 45–64, the female death rate from infectious diseases fell much more rapidly than the male death rate between 1851/1860 and 1901/1910, but the death rate from these causes fell at a similar rate for both sexes between 1901/1910 and 1979/1995. However, the death rate from

Age 0–14

Males 1851/1860 1901/1910

1. 2. 5. 6. 7. 8. 9. 10. 11.

12. 13.

17. 18.

Infectious diseases Neoplasms Mental disorders Diseases of the nervous system and sense organs Diseases of the circulatory system Diseases of the respiratory system Diseases of the digestive system Diseases of the genitourinary system Complications of pregnancy, childbirth and the puerperium Diseases of the skin and subcutaneous tissue Diseases of the musculoskeletal system and connective tissue Injury and poisoning Other

All causes

Females

% Change 1979/1995 % Change 1851/1860 1901/1910 to 1851/1860 to 1979/1995 1901/1910

1901/1910

% Change 1979/1995 % Change 1901/1910 to 1851/1860 to 1979/1995 1901/1910

12,178 14 4,849 1,249

6,264 25 16 639

48.56 81.67 99.66 48.84

25 48 2 42

99.61 95.65 88.88 93.49

12,170 14 3,835 947

6,026 20 14 544

50.48 38.56 99.63 42.57

20 40 1 34

99.67 99.51 89.82 93.66

309

244

21.04

17

93.18

285

274

3.74

15

94.48

4,474

3,961

11.46

65

98.36

3,824

3,304

13.62

50

98.48

698

1,091

56.29

10

99.11

558

862

54.58

8

99.12

56

100

78.48

4

96.37

34

85

147.71

3

96.56

0

562

0

100.00

0

428

–

0

100.00

68

76

12.37

0

99.71

64

59

7.54

0

99.54

72

28

61.44

1

97.04

56

18

67.89

1

93.48

982 6,334

625 5,718

36.34 9.72

109 632

82.51 88.95

615 5,339

472 4,433

23.26 16.97

63 498

86.74 88.76

31,283

19,350

38.14

953

95.08

27,740

16,538

40.38

734

95.56

–

183

Notes: ‘Other’ includes diseases identified as: endocrine, nutritional and metabolic diseases and immunity disorders; diseases of the blood and blood-forming organs; congenital anomalies; certain conditions originating in the perinatal period; and symptoms, signs and ill-defined conditions in 1901–1910 and 1979–1995, together with diseases listed under the heading of ‘Other’ in 1851/1860 and 1901/1910, and as ‘not specified’ in 1901/1910. Sources: Woods (1997) and Office of National Statistics (1997).

Gender, Health, and Welfare

Mean Annual Death Rates at Ages 0–14 (Deaths Per Million Living).

184

Mean Annual Death Rates at Ages 15–44 (Deaths Per Million Living). Age 15–44

Males

Females

1851/1860 1901/1910 % Change 1979/1995 % Change 1851/1860 1901/1910 % Change 1979/1995 % Change 1851/1860 to 1901/1910 to 1851/1860 to 1901/1910 to 1901/1910 1979/1995 1901/1910 1979/1995 1. 2. 5. 6. 7. 8. 9. 10. 11.

12.

17. 18.

All causes

Note and Source: See Table 5.

5,339 74 685 13

2,508 165 112 215

53.03 122.29 83.70 1,519.95

21 190 23 45

99.17 15.00 79.68 79.23

5,684 207 572 11

1,866 298 87 182

67.16 44.01 84.76 1,510.94

10 252 7 30

99.45 15.43 91.78 83.56

564

722

27.90

211

70.73

637

739

15.91

84

88.68

825

728

11.68

44

93.96

616

437

29.07

28

93.53

503

289

42.49

37

87.07

575

316

44.92

25

92.20

171

214

24.83

6

97.42

196

446

127.92

6

98.59

0

0

–

0

–

680

241

64.52

5

98.05

16

5

68.63

0

91.17

13

5

59.39

0

90.96

71

19

73.27

2

88.08

50

25

49.87

5

78.63

999 280

675 105

32.38 62.49

480 53

28.86 49.62

145 288

141 97

2.78 66.47

138 32

2.14 66.87

9,541

5,758

39.66

1,112

80.69

9,674

4,880

49.55

622

87.25

BERNARD HARRIS

13.

Infectious diseases Neoplasms Mental disorders Diseases of the nervous system and sense organs Diseases of the circulatory system Diseases of the respiratory system Diseases of the digestive system Diseases of the genitourinary system Complications of pregnancy, childbirth and the puerperium Diseases of the skin and subcutaneous tissue Diseases of the musculoskeletal system and connective tissue Injury and poisoning Other

Gender, Health, and Welfare

185

cancer rose (partly as a result of diagnostic changes) far more rapidly among males than females between 1851/1860 and 1901/1910 (Hardy, 1994, p. 491), and it continued to rise among males after 1910, even though the female death rate fell between 1901/1910 and 1979/1995. There were also similar changes in the pattern of sex-specific mortality from diseases of the circulatory system. Both male and female death rates rose between 1851/ 1860 and 1901/1910 (though male death rates rose by more), and after 1910 the female death rate fell more than the male rate (Table 7). These differences highlight the need to consider the factors which have tended to inflate male mortality as well as the factors which have contributed to reductions in female mortality. Many of these points could also be made in connection with the analysis of sex-specific variations in mortality among those aged 65 and over (see Table 8). The female death rate from infectious diseases declined slightly more rapidly than the male death rate in the first period but the main reasons for differences in the overall rate of mortality decline were related to differences in the number of deaths associated with cancer and with diseases of the circulatory and respiratory systems. Even though some of these changes can be attributed to diagnostic developments, they also reflect the impact of differences in lifestyle and behaviour, such as cigarette smoking, which are directly related to mortality outcomes (McPherson & Coleman, 1988, pp. 431–433; Coleman & Salt, 1992, pp. 280–281). It is possible to extend this investigation using shift-share analysis.6 This technique has often been used by economic geographers to identify the extent to which changes in a particular local economy can be attributed to changes in the national economy, changes in the ‘mix’ of economic activities, and local circumstances (see, e.g., Hanham & Banasick, 2000). In the current paper, this technique has been used to identify the contributions made by different factors to the decline of female mortality in Britain since the mid-nineteenth century. It is assumed that the decline in female mortality rates can be attributed to (a) the decline in overall mortality; (b) changes in the cause-structure of mortality; and (c) factors which are specific to female mortality. If we say that m ¼ the female mortality rate, M ¼ the national mortality rate, and i ¼ a particular cause of death, we can summarise changes in the female mortality rate between time t and time t þ n in the following terms:  tþn   tþn   tþn  M tþn M tþn tþn t t M t Mi t mi i þ mi  1 þ mi   mi  mi ¼ m i Mt M ti Mt mti M ti

186

Mean Annual Death Rates at Ages 45–64 (Deaths Per Million Living). Age 45–64

Males

Females

1851/1860 1901/1910 % Change 1979/1995 % Change 1851/1860 1901/1910 % Change 1979/1995 % Change 1851/1860 to 1901/1910 to 1851/1860 to 1901/1910 to 1901/1910 1979/1995 1901/1910 1979/1995 1. 2. 5. 6. 7. 8. 9. 10. 11.

12.

17. 18.

All causes

Note and Source: See Table 5.

6,254 625 2,828 7

4,087 2,477 393 578

34.65 296.59 86.09 7,743.85

58 3,312 33 136

98.58 33.70 91.70 76.46

4,894 1,511 2,547 5

2,093 3,167 320 469

57.23 109.61 87.42 9,121.58

35 2,907 24 120

98.34 8.21 92.40 74.48

2,786

5,942

113.29

4,913

17.31

3,075

5,477

78.13

1,823

66.72

4,494

3,926

12.63

646

83.54

3,263

2,591

20.61

398

84.64

2,208

1,318

40.32

313

76.24

2,158

1,216

43.64

221

81.84

665

1,483

122.98

56

96.25

553

1,054

90.67

49

95.37

0

0

–

0

–

42

11

74.71

0

99.71

60

34

42.45

4

89.44

49

45

8.20

4

91.44

93

101

8.51

25

74.96

74

178

138.85

44

75.01

1,466 1,601

1,418 562

3.29 64.90

457 185

67.78 67.09

317 1,493

355 491

11.79 67.12

222 141

37.36 71.37

23,087

22,319

3.33

10,138

54.58

19,982

17,467

12.59

5,988

65.72

BERNARD HARRIS

13.

Infectious diseases Neoplasms Mental disorders Diseases of the nervous system and sense organs Diseases of the circulatory system Diseases of the respiratory system Diseases of the digestive system Diseases of the genitourinary system Complications of pregnancy, childbirth and the puerperium Diseases of the skin and subcutaneous tissue Diseases of the musculoskeletal system and connective tissue Injury and poisoning Other

Age 65 þ

Males

Females

1851/1860 1901/1910 % Change 1979/1995 % Change 1851/1860 1901/1910 % Change 1979/1995 % Change 1901/1910 to 1851/1860 to 1901/1910 to 1851/1860 to 1979/1995 1901/1910 1979/1995 1901/1910 1. 2. 5. 6. 7. 8. 9. 10. 11.

12. 13.

17. 18.

Infectious diseases Neoplasms Mental disorders Diseases of the nervous system and sense organs Diseases of the circulatory system Diseases of the respiratory system Diseases of the digestive system Diseases of the genitourinary system Complications of pregnancy, childbirth and the puerperium Diseases of the skin and subcutaneous tissue Diseases of the musculoskeletal system and connective tissue Injury and poisoning Other

All causes

4,673 7,008 796 2,029

49.56 345.29 93.36 23,069.04

222 16,949 853 1,035

95.25 141.85 7.19 48.98

7,826 2,348 10,859 7

3,721 7,047 678 1,542

52.45 200.15 93.76 22,407.96

163 9,969 1,254 823

95.62 41.45 85.14 46.65

9,714

25,973

167.39

32,901

26.67

9,633

23,024

139.01

27,084

17.64

15,762

16,276

3.26

10,277

36.86

13,242

15,023

13.45

7,202

52.06

4,926

3,207

34.90

1,726

46.18

4,722

3,044

35.53

1,817

40.32

2,994

5,692

90.10

981

82.77

768

2,321

202.12

765

67.04

0

0

0

0

264

238

157

–

0

–

–

0

–

10.11

54

77.16

234

294

25.47

110

62.52

460

193.77

281

38.84

114

668

486.76

637

4.62

1,969 35,366

2,042 19,692

3.69 44.32

900 1,315

55.90 93.32

1,163 37,689

1,209 19,698

4.03 47.73

861 1,658

28.85 91.58

93,998

88,086

6.29

67,496

23.38

88,606

78,270

11.66

52,342

33.13

187

Note and Source: See Table 5.

9,265 1,574 11,999 9

Gender, Health, and Welfare

Mean Annual Death Rates at Ages 65 and Over (Deaths Per Million Living).

188

BERNARD HARRIS

The following tables present results for all deaths, and for deaths associated with each of the main disease categories, for females between the ages of 0 and 14, 15 and 44, 45 and 64 and Z65 between 1851/1860 and 1979/1995. Table 9 suggests that more than 98 per cent of the decline in the female death rate between the ages of 0 and 14 was associated with factors which resulted in the decline of both male and female mortality. Further reductions were caused by a decline in the relative importance of deaths from infectious diseases and diseases of the nervous system, although the effects of these changes were counterbalanced by increases in the proportions of deaths associated with other causes. ‘Being female’ was responsible for approximately 3 per cent of the overall decline in mortality during this period, but this was largely the result of changes in the number of deaths associated with complications of pregnancy. There is very little evidence to suggest that it had any significant bearing on the level of mortality from infectious diseases. The following table examines the contributions made by different factors to the decline of mortality among women between the ages of 15 and 44. Although nearly 93 per cent of the decline in female mortality could be attributed to causes associated with the decline in mortality between both sexes, changes in the cause-structure of mortality were responsible for 1.65 per cent of the female decline and ‘being female’ accounted for a further 5.38 per cent of mortality decline. The results for infectious diseases suggest that the overall decline in the incidence of deaths from these diseases was once again a significant factor in reducing the ratio of female-to-male deaths, but being female also made a significant contribution (6.64 per cent of the total decline in female mortality from this cause). This may therefore provide a further indication of the extent to which the particular disadvantages of women’s lives helped to raise their mortality levels in the mid-nineteenth century (Table 10). The final two tables present results for women aged between 45 and 64, and for women aged 65 and over. Although Table 11 provides further evidence of the additional extent of the decline in female mortality from infectious diseases, both tables reveal the importance of twentiethcentury trends in mortality from cancer (neoplasms) and circulatory diseases. In each case, the increasing importance of these diseases had the effect of raising both male and female mortality, but the increase in male mortality was much greater, and this led to a further reduction in the ratio of female-to-male mortality within these age groups (Table 12).

1851/1860 to 1979/1995 Age group 0–14

1. 2. 5. 6. 7. 8. 9. 10. 11.

12. 13.

17. 18.

Infectious diseases Neoplasms Mental disorders Diseases of the nervous system and sense organs Diseases of the circulatory system Diseases of the respiratory system Diseases of the digestive system Diseases of the genito-urinary system Complications of pregnancy, childbirth and the puerperium Diseases of the skin and subcutaneous tissue Diseases of the musculoskeletal system and connective tissue Injury and poisoning Other

Total

%

Whole population

Cause-ofdeath

Female mortality

Total

Whole population

10,514 25 1,517 889

1,517 52 2,317 56

119 3 1 34

12,150 25 3,834 912

86.54 97.05 39.58 97.49

373

79

24

270

4,647

919

46

1,040

495

94

Cause-ofdeath

Female mortality

Total

12.48 207.65 60.44 6.19

0.98 10.60 0.02 3.67

100.00 100.00 100.00 100.00

138.28

29.38

8.90

100.00

3,774

123.13

24.35

1.21

100.00

5

550

189.07

89.98

0.91

100.00

48

15

31

300.84

153.92

46.93

100.00

408

974

566

0

–

–

–

100.00

82

24

7

64

127.85

38.04

10.19

100.00

39

14

2

54

71.30

25.72

2.97

100.00

691 6,317

100 1,682

38 205

552 4,840

125.06 130.51

18.18 34.74

6.87 4.23

100.00 100.00

26,635

470

841

27,006

98.63

1.74

3.11

100.00

189

Source: See Table 5.

Deaths Per Million

Gender, Health, and Welfare

The Contribution of Different Factors to the Decline of Female Mortality at Ages 0–14 Between 1851/1860 and 1979/1995.

1851/1860 to 1979/1995 Age group 0–14

1. 2. 5. 6. 7. 8. 9. 10. 11.

12.

17. 18.

Total

Source: See Table 5.

%

Whole population

Cause-ofdeath

Female mortality

Total

Whole population

Cause-ofdeath

Female mortality

Total

4,108 341 329 157

1,189 456 226 180

377 69 9 5

5,674 45 565 19

72.40 757.98 58.30 844.24

20.96 1,012.29 40.07 970.89

6.64 154.31 1.63 26.66

100.00 100.00 100.00 100.00

903

450

101

554

163.11

81.33

18.23

100.00

641

112

58

587

109.12

19.02

9.90

100.00

522

14

14

550

94.95

2.48

2.57

100.00

461

183

89

190

243.13

96.35

46.78

100.00

507

168

1

676

75.01

24.89

0.10

100.00

10

3

1

13

81.77

23.40

5.17

100.00

43

9

8

44

97.14

20.25

17.38

100.00

183 210

111 32

65 14

7 256

2,594.44 81.95

1,573.31 12.45

921.13 5.60

100.00 100.00

8,416

149

487

9,051

92.98

1.65

5.38

100.00

BERNARD HARRIS

13.

Infectious diseases Neoplasms Mental disorders Diseases of the nervous system and sense organs Diseases of the circulatory system Diseases of the respiratory system Diseases of the digestive system Diseases of the genito-urinary system Complications of pregnancy, childbirth and the puerperium Diseases of the skin and subcutaneous tissue Diseases of the musculoskeletal system and connective tissue Injury and poisoning Other

Deaths Per Million

190

The Contribution of Different Factors to the Decline of Female Mortality at Ages 15–44 between 1851/ 1860 and 1979/1995.

1851/1860 to 1979/1995 Age group 0–14

1. 2. 5. 6. 7. 8. 9. 10. 11.

12. 13.

17. 18.

Infectious diseases Neoplasms Mental disorders Diseases of the nervous system and sense organs Diseases of the circulatory system Diseases of the respiratory system Diseases of the digestive system Diseases of the genito-urinary system Complications of pregnancy, childbirth and the puerperium Diseases of the skin and subcutaneous tissue Diseases of the musculoskeletal system and connective tissue Injury and poisoning Other

Total

Whole population

Cause-ofdeath

2,458 1,490 1,056 138

802 1,330 755 142

2,752

%

Female mortality

Total

Whole population

Cause-ofdeath

Female mortality

Total

239 406 9 14

3,500 566 1,820 18

70.24 263.51 58.03 782.45

22.93 235.23 41.50 803.98

6.83 71.71 0.47 78.47

100.00 100.00 100.00 100.00

1,327

454

1,879

146.50

70.66

24.16

100.00

1,986

209

72

2,267

87.60

9.23

3.18

100.00

1,169

312

26

1,507

77.59

20.71

1.70

100.00

515

169

41

387

132.93

43.53

10.59

100.00

19

11

0

30

63.05

36.73

0.22

100.00

31

6

3

34

91.77

17.14

8.90

100.00

80

15

19

45

175.54

34.04

41.50

100.00

223 707

7 327

28 10

188 1,045

118.70 67.69

3.62 31.31

15.08 1.00

100.00 100.00

12,625

567

1,192

13,250

95.28

4.28

9.00

100.00

191

Source: See Table 5.

Deaths Per Million

Gender, Health, and Welfare

The Contribution of Different Factors to the Decline of Female Mortality at Ages 45–64 between 1851/ 1860 and 1979/1995.

1851/1860 to 1979/1995 Age group 0–14

1. 2. 5. 6. 7. 8. 9. 10. 11.

12.

17. 18.

Total

Source: See Table 5.

%

Whole population

Cause-ofdeath

Female mortality

Total

Whole population

Cause-ofdeath

Female mortality

Total

11,804 8,783 12,414 1,370

4,129 8,700 9,863 1,624

282 2,684 3 94

16,214 2,766 22,280 160

72.80 317.54 55.72 857.26

25.46 314.57 44.27 1,016.05

1.74 97.03 0.02 58.79

100.00 100.00 100.00 100.00

30,828

19,552

2,994

14,270

216.03

137.01

20.98

100.00

27,684

1,868

153

25,970

106.60

7.19

0.59

100.00

7,828

1,453

139

9,419

83.11

15.42

1.47

100.00

2,888

753

701

1,433

201.44

52.56

48.88

100.00

0

0

0

0

–

–

–

–

514

12

39

463

111.14

2.64

8.50

100.00

714

416

196

102

703.11

410.10

193.00

100.00

2,334 58,415

112 18,057

10 1,427

2,232 77,899

104.57 74.99

5.01 23.18

0.43 1.83

100.00 100.00

165,574

464

6,849

172,887

95.77

0.27

3.96

100.00

BERNARD HARRIS

13.

Infectious diseases Neoplasms Mental disorders Diseases of the nervous system and sense organs Diseases of the circulatory system Diseases of the respiratory system Diseases of the digestive system Diseases of the genito-urinary system Complications of pregnancy, childbirth and the puerperium Diseases of the skin and subcutaneous tissue Diseases of the musculoskeletal system and connective tissue Injury and poisoning Other

Deaths Per Million

192

The Contribution of Different Factors to the Decline of Female Mortality at the Ages of 65 and Over Between 1851/1860 and 1979/1995.

Gender, Health, and Welfare

193

6. CONCLUSIONS This paper has provided only limited support for the view that the health and life chances of girls were adversely affected by a decline in agricultural employment opportunities for women in the rural areas of England and Wales during the eighteenth and nineteenth centuries. Although some authors have argued that there was a decline in the relative value of the heights of women who were born in rural areas between 1790 and 1815, these arguments have not gone unchallenged, and the heights of men and women appear to have moved in unison during the remainder of the nineteenth century. There is rather more evidence to show that the ratio of female mortality to male mortality in rural areas was higher than the sex-mortality-ratio in urban areas, but it is difficult to argue that this fact in itself can explain the existence of disproportionately high rates of female mortality in the nineteenth century as a whole. If we are to gain a fuller understanding of the factors that have led to changes in the ratio of female mortality to male mortality, it seems important to distinguish between the causes of excess female mortality in childhood and the causes of excess female mortality in adulthood. This paper has shown that, in contrast to the experience of many of today’s developing countries, there is little evidence to support the view that females suffered systematic neglect in childhood or that such neglect resulted in sex-specific differences in mortality rates. It is true that the ratio of female mortality to male mortality among those aged 0–14 was higher in 1851/1860 than in 1979/1995 but this appears to be largely attributable to changes in the epidemiological environment. Death rates from infectious diseases declined at much the same rate for children of both sexes between 1851/1860 and 1901/1910 and between 1901/1910 and 1979/1995, but the dramatic reduction in the importance of infectious diseases has led to a greater improvement in female survival chances. In contrast to the evidence of changing sex differentials in childhood mortality, there are much stronger grounds for believing that the additional hardships faced by women in adulthood did contribute to their experience of excess female mortality. The evidence presented in Tables 6–7 and Tables 10–11 suggests that the ratio of female deaths to male deaths would still have fallen, even if there had been no differences in the rates at which sex-specific mortality rates declined for each cause of death, because the sexmortality-ratio for infectious diseases was higher than the sex-mortality ratio for other causes. However, the fact that female death rates fell more rapidly than male death rates for individual causes of death suggests that

194

BERNARD HARRIS

women’s ability to resist infection was enhanced by an overall improvement in their standard of living. There are a number of reasons why adult women should have experienced a disproportionately low standard of living in the British past. In the first place, it is clear that, when resources were limited, both men and women attached the highest priority to safeguarding the health of the male head of household and principal breadwinner. In 1863, the Government medical inspector, Dr. Edward Smith, claimed that in many parts of rural England ‘the important practical fact is however well-established, that the labourer eats meat or bacon almost daily, whilst his wife and children may eat it but once a week, and that both himself and his household believe that course to be necessary, to enable him to perform his labour’ (Parliamentary Papers, 1864, p. 249). In poor areas, this tendency persisted well into the twentieth century. In 1933, the eminent nutritionist, Sir Frederick Gowland Hopkins, told a Government Committee that ‘when the income of a family falls below a certain limit . . . the calorie intake of the husband was not likely to fall so much in proportion as that of the wife’ – an opinion which was echoed by other commentators of the period (quoted in Mitchell, 1985, p. 111).7 The welfare of adult women may also have been adversely affected by differences in access to and the nature of paid work. According to Oren (1974, p. 221), the tradition of giving a smaller share of household resources to women was well-established in pre-industrial times, even when women performed an equal share of agricultural labour, so this tendency is unlikely to have been caused by the decline in female employment opportunities, even though it may have been exacerbated by it. When we compare the welfare of women in the past with present-day standards, it is also important to consider the amount of physical (and mental) effort associated not only with paid work, but also with unpaid domestic labour. As Clementina Black (1915, p. 8) observed For a variety of reasons, the industry of housekeeping has not undergone that alteration of methods which has transformed other industries; one person performs all the processes, using for their various purposes inadequate hand-driven tools. Even in wellto-do homes the domestic appliances, although in sum total costly, are primitive; the broom, the scrubbing-brush and the duster are not yet replaced by the vacuum cleaner, the washing-up bowl by a machine, nor the individual oven by the great steam-roaster capable of cooking lots of joints cheaply without attention or wastage.

The third factor, which undoubtedly contributed to the disproportionately high rates of female mortality, was the high rate of pregnancy. In the mid-nineteenth century, the birth rate averaged between 30 and 36 births per

Gender, Health, and Welfare

195

thousand living (compared with a figure of 13.2 in 1980), and it has been estimated that, even in the 1890s, the average working class woman spent nearly 15 years of her life either pregnant, or nursing a child under the age of 1 year (Titmuss, 1987, p. 89). When these figures were combined with the high rate of maternal mortality (i.e. the number of deaths associated with childbirth), it was hardly surprising that maternal deaths should have made such a large contribution to the premature mortality of women of childbearing age. At the same time, it is rather more difficult to estimate the long-term effects of multiple pregnancies on the health of those who survived childbirth. A number of writers have argued that there is little or no relationship between the number of children born to women and their survival prospects at the age of 45 and over, whilst others have even claimed that women who experience multiple births experience a mortality advantage (Bideau, 1986, p. 72; Henry, 1987, p. 110, 1989, p. 197; Dorn & McDowell, 1939; Kiser, Grabill, & Campbell, 1968). On the other hand, there is a great deal of anecdotal evidence to suggest that women who did survive childbirth experienced a wide-range of pregnancy-related health problems, including slit cervixes, prolapsed uteruses, varicose veins, backache, haemorrhaging and ‘lassitude’ (Davies, 1915, p. 9; Brookes, 1986, p. 153; see also Kitagwa & Hauser, 1973). Oren (1974, p. 233) has suggested that the health of women may also have been exacerbated, in comparison with that of men, by differences in access to health care, but it is difficult to know precisely how much weight to attach to this. Riley (1997, pp. 47–51) argued that there was a massive increase in the number of working class men who acquired rights to medical care during the second half of the nineteenth century through membership of friendly societies, but these organisations rarely conferred equivalent rights on their members’ wives or children (Digby, 1994, p. 51). If the provision of medical care did make a difference to the decline of mortality – and the evidence on this point is far from clear – then one might have expected the ratio of female mortality to male mortality to have increased.8 However, it was precisely during this period that the ratio of female mortality to male mortality among those aged 25 and over began to fall. It is also important to consider the factors that might have helped to raise male mortality. In the first place, it is reasonable to suppose that the greater level of employment among men in hazardous industries may have helped to raise male mortality in comparison with female mortality during the second half of the nineteenth century (Johansson, 1977, pp. 178–181). Secondly, it is also arguable that the death rates of men, especially during

196

BERNARD HARRIS

the twentieth century, may have been directly affected by their participation in war – not just in terms of the direct impact of war – related mortality, but also in terms of the long-term effects of military action on the health and longevity of those who survived it (Winter, 1982, p. 105). Thirdly, it is also apparent that the death rates of young men – i.e. those aged 15–34 – have been adversely affected by the disproportionately high number of deaths from violence and accidental causes (Coleman & Salt, 1992, p. 268). Finally, and perhaps most importantly, it is also clear that for most of the twentieth century, the gap between male and female mortality at higher ages has been artificially inflated by the fact that men have shown a much greater propensity to consume excessive amounts of alcohol and to smoke cigarettes (Retherford, 1975; United Nations Secretariat, 1988). Indeed, it is noticeable that it is only recently that the gap between male and female mortality has begun to narrow in these age groups, and this appears to be directly related to the adoption of more ‘masculine’ forms of behaviour by women following the end of the Second World War (McPherson & Coleman, 1988, pp. 431–433; Coleman & Salt, 1992, pp. 280–281; Pampel, 2002). These findings have a number of important implications for the study of sex-specific mortality rates in many parts of the developing world today. Murthi et al. (1995, p. 753) have argued that it is virtually impossible to distinguish between the roles played by economic and cultural factors in causing excess female mortality. Nevertheless, the evidence, which has been presented in this paper, clearly suggests that there was a close link between women’s experience of excess female mortality and their experience of poverty. This contrasts with the experience of women in many of today’s developing countries, where the extent of excess female mortality does not necessarily depend upon the level of hardship, and appears to be more strongly influenced by patterns of landholding and kinship structures than by the simple fact of women’s exclusion from paid labour. While much of the evidence that has been examined in this paper points to the distinctiveness of the British case, it is not without implications for health policy in the developing world today. One of the most striking features of the British mortality record is the increase in the number of male deaths from the so-called ‘diseases of affluence’ (see also Johansson, 1991). Economists and demographers have rightly drawn attention to the factors, which are currently associated with high rates of female mortality in many parts of the developing world. However, the experience of Britain in the twentieth century also demonstrates the need to develop policies, which

197

Gender, Health, and Welfare

will help to protect the health of both men and women in these countries in the future.

NOTES 1. The body mass index is a formula used to combine measurements of height and weight in a single index. The body mass index is derived from the following formula: BMI ¼ (weight [kg])/(height [m])2. 2. McNay et al. examined the relationship between age-specific and sex-specific mortality rates and the following contextual variables: number of spinsters aged 20 þ as a proportion of the female population aged 20 þ ; age differences between marriage partners; literacy; population density; infant mortality; and the number of inhabited dwellings per 1,000 population. The economic variables were: proportions of the population engaged in agricultural and in working-class non-agricultural occupations; proportions of women engaged in agriculture, trade, domestic service, heavy industry, manufacturing and public service; proportion of women aged 20 þ who were property-owners; proportion of women aged 20 þ without independent means of support; percentage of employed males who were engaged in occupations providing a family or ‘breadwinner’ wage; percentage of employed males who were engaged in heavy industry; the ratio of able-bodied female paupers aged 20 þ as a proportion of the total female population aged 20 þ to the number of able-bodied male paupers aged 20 þ as a proportion of the total male population aged 20 þ ; and the ratio of non-able-bodied female paupers aged 20 þ as a proportion of the total female population aged 20 þ to the number of non-able-bodied male paupers aged 20 þ as a proportion of the total male population aged 20 þ (McNay et al., 1998, pp. 12–18). 3. Leonard Schwarz (1999, pp. 254–255) concluded that ‘a presentable young woman with the necessary skills in the field of domestic service and the all-important character reference had considerably more choice over her employment than might have been expected’. 4. It is interesting to compare these findings with Gibson and Smout’s assessment of the importance of family earnings in eighteenth-century Scotland. Gibson and Smout (1995, p. 353) concluded that ‘the most substantial increases in the standard of living in late eighteenth-century Scotland came about less through the expansion of wage rates and employment opportunities for males than for their expansion for women and children. This was true even in non-textile areas’. 5. If the female death rate from infectious diseases had declined at the same rate as the male death rate from infectious diseases, the overall female death rate would have been 5,684 deaths per million living, instead of 4,880 deaths per million living. This would have represented a reduction in the overall death rate of 41.25 per cent between 1851/1860 and 1901/1910, as opposed to the actual reduction of 49.55 per cent. 6. I am particularly grateful to Jane Humphries for this suggestion. 7. Other interwar investigators claimed that women deprived themselves (but not necessarily their husbands) in the interests of their children. In 1938, the authors of the Pilgrim Trust report, Men without work, reported that ‘in most unemployed families, the parents, and particularly the mothers, bore the burden of want, and

198

BERNARD HARRIS

in many instances were literally starving themselves in order to feed and clothe the children reasonably well’ (Pilgrim Trust, 1938, p. 111; quoted in Harris, 1988, pp. 165–166). 8. Medical historians have long accepted Thomas McKeown’s (1976, pp. 108–109) influential argument that therapeutic medicine – as opposed to preventive medicine – made very little contribution to the decline of mortality before the start of the twentieth century. However, Riley (1997, p. 197) has recently argued that curative medical intervention may have made a difference by teaching patients to prolong their lives by managing disease more effectively.

ACKNOWLEDGMENTS Earlier versions of this paper were presented at conferences in Amsterdam (European Social Science History Conference, 2006) and Barcelona (COST Action 34, Gender and wellbeing: work, family and public policies). I should like to thank participants at these meetings for their helpful advice. I am particularly grateful to Andrew Hinde, Jane Humphries, John Komlos and Deborah Oxley for some detailed and extremely valuable comments on the text.

REFERENCES Agnihotri, S., Palmer-Jones, R., & Parikh, A. (1999). Missing women in Indian districts: An entitlements approach. Unpublished paper. Alter, G., & Carmichael, A. (1996). Studying causes of death in the past: Problems and models. Historical Methods, 29, 44–48. Alter, G., Manfredini, M., & Nystedt, P. (2004). Gender differences in mortality. In: T. Bengtsson, C. Campbell & J. Z. Lee, et al. (Eds), Life under pressure: Mortality and living standards in Europe and Asia, 1700–1900 (pp. 327–358). Cambridge, MA: MIT Press. Anderton, D., & Leonard, S. H. (2004). Grammars of death: An analysis of nineteenth-century literal causes of death from the age of miasmas to germ theory. Social Science History, 28, 111–143. Berg, M. (1993). What difference did women’s work make to the Industrial Revolution? History Workshop Journal, 35, 22–44. Bideau, A. (1986). Fe´condite´ et mortalite´ apre`s 45 ans. L’apport des recherches en de´mographie historique. Population, 1, 59–72. Black, C. (Ed.) (1915). Married women’s work, being the report of an enquiry undertaken by the Women’s Industrial Council. London: G. Bell. Brookes, B. (1986). Women and reproduction c. 1860–1919. In: J. Lewis (Ed.), Labour and love: Women’s experience of home and family 1850–1940 (pp. 149–171). Oxford: Basil Blackwell. Bythell, D. (1993a). Women in the workforce. In: P. K. O’Brien & R. Quinault (Eds), The industrial revolution and British society (pp. 31–53). Cambridge: Cambridge University Press.

Gender, Health, and Welfare

199

Bythell, D. (1993b). The sweated trades: Outwork in nineteenth-century Britain. London: Batsford. Carter, K. C. (1997). Causes of diseases and causes of death. Continuity and Change, 12, 189–198. Chen, L., Huq, E., & D’Souza, S. (1981). Sex bias in the family allocation of food and health care in rural Bangladesh. Population and Development Review, 7, 55–70. Coale, A. (1991). Excess female mortality and the balance of the sexes in the population: An estimate of the number of ‘‘missing females’’. Population and Development Review, 17, 517–523. Coleman, D., & Salt, J. (1992). The British population: Patterns, trends and processes. Oxford: Oxford University Press. Das Gupta, M. (1987). Selective discrimination against female children in rural Punjab, India. Population and Development Review, 13, 77–100. Dasgupta, I., Palmer-Jones, R., & Parikh, A. (1999). Between cultures and markets: An eclectic analysis of juvenile gender ratios in India. Unpublished paper. Davies, M. L. (Ed.) (1915). Maternity: Letters from working women. London: G. Bell and Sons. Davin, D. (1995). Women, work and property in the Chinese peasant household of the 1980s. In: D. Elson (Ed.), Male bias in the development process (pp. 29–50). Manchester: Manchester University Press. DeRose, L., Das, M., & Millman, S. (2000). Does female disadvantage mean lower access to food? Population and Development Review, 26, 517–547. Digby, A. (1994). Making a medical living: Doctors and patients in the English market for medicine, 1720–1911. Cambridge: Cambridge University Press. Dorn, H. F., & McDowell, A. J. (1939). The relationship of fertility and longevity. American Sociological Review, 4, 234–246. Doyal, L. (2002). Putting gender into health and globalisation debates: New perspectives and old challenges. Third World Quarterly, 23, 233–250. Dyson, T., & Moore, M. (1983). On kinship structure, female autonomy and demographic behaviour in India. Population and Development Review, 9, 35–60. D’Souza, S., & Chen, L. (1980). Sex differentials in mortality in rural Bangladesh. Population and Development Review, 6, 257–270. Eden, F. M. (1797). The state of the poor. (Vol. 2). London: B & J White. Elson, D. (1995a). Male bias in the development process: An overview. In: D. Elson (Ed.), Male bias in the development process (pp. 1–28). Manchester: Manchester University Press. Elson, D. (1995b). Overcoming male bias. In: D. Elson (Ed.), Male bias in the development process (pp. 191–210). Manchester: Manchester University Press. Engerman, S. (1997). The standard of living debate in international perspective: Measures and indicators. In: R. Steckel & R. Floud (Eds), Health and welfare during industrialization (pp. 17–46). Chicago: University of Chicago Press. Eveleth, P., & Tanner, J. (1976). Worldwide variation in human growth (1st ed.). Cambridge: Cambridge University Press. Eveleth, P., & Tanner, J. (1990). Worldwide variation in human growth (2nd ed.). Cambridge: Cambridge University Press. Eyler, J. M. (1976). Mortality statistics and Victorian health policy: Program and criticism. Bulletin of the History of Medicine, 50, 335–355.

200

BERNARD HARRIS

Floud, R. (1998). Height, weight and body mass of the British population since 1820. National Bureau of Economic Research Working Paper Series on Historical Factors in Long Run Growth, Historical Paper 108. Cambridge, MA: National Bureau of Economic Research. Gibson, A. J. S., & Smout, T. C. (1995). Prices, food and wages in Scotland 1550–1780. Cambridge: Cambridge University Press. Goose, N. (2007). Working women in industrial England. In: N. Goose (Ed.), Women’s work in industrial England: Regional and local perspectives (pp. 1–28). Hatfield: Local Population Studies. Hammoud, E. (1977). Sex differentials in mortality: An enquiry with reference to the Arab countries and others. World Health Statistics Report, 30, 174–206. Hanham, R. Q., & Banasick, S. (2000). Shift-share analysis and changes in Japanese manufacturing employment. Growth and Change, 31, 108–123. Hardy, A. (1994). Death is the cure for all diseases: Using the General Register Office Cause of Death statistics for 1837–1920. Social History of Medicine, 7, 472–492. Hare, D. (1999). Women’s economic status in rural China: Household contributions to male– female disparities in the wage-labor market. World Development, 27, 1011–1029. Harris, B. (1988). Unemployment, insurance and health in interwar Britain. In: B. Eichengreen & T. Hatton (Eds), Interwar unemployment in international perspective (pp. 149–183). Dordrecht: Kluwer International Publishers. Harris, B. (1994). Health, height and history: An overview of recent developments in anthropometric history. Social History of Medicine, 7, 323–346. Harris, B. (1998). Gender, height and mortality in nineteenth- and twentieth-century Britain: Some preliminary reflections. In: J. Komlos & J. Baten (Eds), The biological standard of living in comparative perspective (pp. 413–448). Stuttgart: Franz Steiner Verlag. Harriss, B. (1990). The intrafamily distribution of hunger in south Asia. In: J. Dre`ze & A. Sen (Eds), The political economy of hunger (vol. 1). Entitlement and well-being (pp. 351–424). Oxford: Clarendon Press. Henry, L. (1987). Mortalite´ des hommes et des femmes dans le passe. Annales de De´mographie Historique, 87–118. Henry, L. (1989). Men’s and women’s mortality in the past. Population, 44, 177–201. Higgs, E. (1986). Domestic service and household production. In: A. John (Ed.), Unequal opportunities: Women’s employment in England 1800–1918 (pp. 125–150). Oxford: Blackwell. Hill, K., & Upchurch, D. (1995). Gender differences in child health: Evidence from the demographic and health surveys. Population and Development Review, 21, 127–151. Horrell, S., & Humphries, J. (1992). Old questions, new data, and alternative perspectives: Families’ living standards in the industrial revolution. Journal of Economic History, 52, 849–880. Horrell, S., & Humphries, J. (1995a). Women’s labour force participation and the transition to the male-breadwinner family, 1790–1865. Economic History Review, 48, 89–117. Horrell, S., & Humphries, J. (1995b). The exploitation of little children. Child labor and the family economy in the industrial revolution. Explorations in Economic History, 32, 485–516. Horrell, S., & Humphries, J. (1997). The origins and expansion of the male-breadwinner family: The case of nineteenth-century Britain. International Review of Social History, 42, 25–64.

Gender, Health, and Welfare

201

Horrell, S., Meredith, D., & Oxley, D. (2006). Measuring misery: Body mass among Victorian London’s poor. Paper presented to Session 50 (Anthropometrics, markets and disease in historical standards of living: Eurasian and American countries) of the Fourteenth International Economic History Congress, Helsinki, 23 August 2006. Humphries, J. (1991). Bread and a pennyworth of treacle: Excess female mortality in England in the 1840s. Cambridge Journal of Economics, 15, 451–473. Jackson, R. (1996). The heights of rural-born English female convicts transported to New South Wales. Economic History Review, 49, 584–590. Jackson, C., & Palmer-Jones, R. (1999). Rethinking gendered poverty and work. Development and Change, 30, 557–583. Johansson, S. R. (1977). Sex and death in Victorian England: An examination of age- and sexspecific death rates, 1840–1910. In: M. Vicinus (Ed.), A widening sphere: Changing roles of Victorian women (pp. 163–181). London: Methuen. Johansson, S. R. (1991). Welfare, mortality and gender: Continuity and change in explanations for male/female mortality differences over three centuries. Continuity and Change, 6, 135–178. Johansson, S. R. (1996). Excess female mortality: Constructing survival during development in Meiji Japan and Victorian England. In: A. Digby & J. Stewart (Eds), Gender, health and welfare (pp. 32–67). London: Routledge. Johansson, S., & Nygren, O. (1991). The missing girls of China: A new demographic account. Population and Development Review, 17, 35–51. Johnson, P., & Nicholas, S. (1995). Male and female living standards in England and Wales 1812–1857: Evidence from criminal height records. Economic History Review, 48, 470–481. Johnson, P., & Nicholas, S. (1997). Health and welfare of women in the United Kingdom, 1785–1920. In: R. Steckel & R. Floud (Eds), Health and welfare during industrialization (pp. 201–249). Chicago: University of Chicago Press. Jordan, E. (1989). The exclusion of women from industry in nineteenth-century Britain. Comparative Studies in Society and History, 31, 273–296. Kearns, G. (1993). Le handicap urbain et le de´clin de la mortalite´ en Angleterre et au Pays de Galles 1851–1900. Annales de De´mographie Historique, 75–105. Kiser, C. V., Grabill, W. H., & Campbell, A. A. (1968). Trends and variations in fertility in the United States. Cambridge, MA: Harvard University Press. Kishor, S. (1993). May God give sons to all: Gender and child mortality in India. American Sociological Review, 58, 247–265. Kishor, S. (1995). Gender differentials in child mortality: A review of the evidence. In: M. Das Gupta, L. Chen & T. Krishnan (Eds), Women’s health in India: Risk and vulnerability (pp. 19–54). Bombay: Oxford University Press. Kitagwa, E. M., & Hauser, P. M. (1973). Differential mortality in the United States: A study in socio-economic epidemiology. Cambridge, MA: Harvard University Press. Klasen, S. (1994). ‘‘Missing women’’ reconsidered. World Development, 22, 1061–1071. Klasen, S. (1998). Marriage bargaining and intrahousehold resource allocation: Excess female mortality among adults during early German development, 1740–1860. Journal of Economic History, 58, 432–467. Klasen, S., & Wink, C. (2002). A turning-point in gender bias in mortality: An update on the number of missing women. Population and Development Review, 28, 285–312.

202

BERNARD HARRIS

Klasen, S., & Wink, C. (2003). Missing women: Revisiting the debate. Feminist Economics, 9, 263–299. Komlos, J., & Baten, J. (2004). Looking backward and looking forward: Anthropometric research and the development of social science history. Social Science History, 28, 191–210. Komlos, J., & Coll, S. (1998). The biological standard of living and economic development: Nutrition, health and well-being in historical perspective. In: C. E. Nu´n˜ez (Ed.), Debates and controversies in economic history (pp. 219–282). Madrid: Fundacio´n Ramon Areces/ Fundacio´n fomento de la historia econo´mica. Le Grand, T., & Mbacke´, C. (1995). Sex differences in mortality among young children in the Sahel. Population Bulletin of the United Nations, 39, 79–111. Lewis, J. (1984). Women in England 1870–1950. Brighton: Harvester Wheatsheaf. Lindert, P. (1994). Unequal living standards. In: R. Floud & D. McCloskey (Eds), The economic history of Britain since 1700. Volume 1: 1700–1860 (pp. 357–386). Cambridge: Cambridge University Press. Logan, W. P. D. (1950). Mortality in England and Wales from 1848–1947: A survey of the changing causes of death during the past hundred years. Population Studies, 4, 132–178. Luckin, B. (1980). Death and survival in the city: Approaches to the history of disease. Urban History Yearbook, 53–62. Marshall, W. (1788). The rural economy of Yorkshire. (Vol. 1). London: T Cadell. McKeown, T. (1976). The modern rise of population. London: Edward Arnold. McNay, K., Humphries, J., & Klasen, S. (1998). Death and gender in Victorian England and Wales: Comparisons with contemporary developing countries. Working Paper no. 9801. University of Cambridge Department of Applied Economics. McNay, K., Humphries, J., & Klasen, S. (2005). Excess female mortality in nineteenth-century England and Wales: A regional analysis. Social Science History, 29, 649–681. McPherson, K., & Coleman, D. (1988). Health. In: A. H. Halsey (Ed.), British social trends since 1900: A guide to the changing social structure of Britain (pp. 398–461). Basingstoke: Macmillan. Messer, E. (1997). Intrahousehold allocation of food and health care: Current findings and understandings – Introduction. Social Science and Medicine, 44, 1675–1684. Miller, C. (1984). The hidden workforce: Female fieldworkers in Gloucestershire, 1870–1901. Southern History, 6, 139–161. Mitchell, B. R. (1988). Abstract of British historical statistics. Cambridge: Cambridge University Press. Mitchell, M. (1985). The effects of unemployment on the social conditions of women and children in the 1930s. History Workshop Journal, 19, 105–127. Muhuri, P., & Preston, S. (1991). Effects of family composition on mortality differentials by sex among children in Matlab, Bangladesh. Population and Development Review, 17, 415–434. Murthi, M., Guio, A.-C., & Dre`ze, J. (1995). Mortality, fertility and gender bias in India: A district level analysis. Population and Development Review, 21, 745–782. Nicholas, S., & Oxley, D. (1993). The living standards of women during the industrial revolution, 1795–1820. Economic History Review, 46, 723–749. Nicholas, S., & Oxley, D. (1996). Living standards of women in England and Wales, 1785–1815: New evidence from Newgate prison records. Economic History Review, 49, 591–599. Nicholas, S., & Steckel, R. (1991). Heights and living standards of English workers during the early years of industrialization. Journal of Economic History, 51, 937–957.

Gender, Health, and Welfare

203

Office of National Statistics. (1997). Twentieth century mortality statistics: 95 years of mortality in England and Wales. London: HMSO. Oren, L. (1974). The welfare of women in laboring families. In: M. Hartman & L. Banner (Eds), Clio’s consciousness raised: New perspectives on the history of women (pp. 226–244). New York: Harper Torchbooks. Pampel, F. (2002). Cigarette use and the narrowing sex differential in mortality. Population and Development Review, 28, 77–104. Parliamentary Papers. (1843). Reports of special assistant poor law commissioners on the employment of women and children in agriculture (PP 1843 (510) xii, 1). London: HMSO. Parliamentary Papers. (1864). Sixth report of the Medical Officer of the Privy Council, with Appendix, 1863 (PP 1864 (3416) xxviii, 1). London: HMSO. Parliamentary Papers. (1919). Supplement to the 75th annual report of the Registrar-General of births, deaths and marriages in England and Wales. Part III. Registration summary tables (1901-10) (PP 1914-16 Cd. 8002 viii, 1). London: HMSO. Pilgrim Trust. (1938). Men without work. Cambridge: Cambridge University Press. Pinchbeck, I. (1930). Women workers and the industrial revolution 1750–1850. London: Frank Cass. Retherford, R. (1975). The changing sex differential in mortality (International Population and Urban Research, University of California, Berkeley, Studies in Population and Urban Demography, no. 1). Westport: Greenwood Press. Richards, E. (1974). Women in the British economy since about 1700: An interpretation. History, 59, 337–357. Riggs, P. (1994). The standard of living in Scotland, 1800–1850. In: J. Komlos (Ed.), Stature, living standards and economic development: Essays in anthropometric history (pp. 60–75). Chicago: University of Chicago Press. Riley, J. C. (1997). Sick, not dead: The health of British workingmen during the mortality decline. Baltimore: Johns Hopkins University Press. Roberts, E. (1984). A woman’s place: An oral history of working class women 1890–1940. Oxford: Basil Blackwell. Rumsey, H. W. (1875). Essays and papers on some fallacies of statistics concerning life and death, health and disease. London: Smith, Elder and Co. Samuel, R. (1977). The workshop of the world: Steam power and hand technology in midVictorian Britain. History Workshop Journal, 3, 6–72. Schwarz, L. (1999). English servants and their employers during the eighteenth and nineteenth centuries. Economic History Review, 52, 236–256. Sen, A. (1989). Women’s survival as a development problem. Bulletin of the American Academy of Arts and Sciences, 43, 14–29. Sen, A. (1990). More than 100 million women are missing. New York Review of Books, December 20. Available at http://www.nybooks.com/nyrev/wwwarchdisplay.cgi?1990012200061F Sen, A. (1992). Missing women: SOCIAL inequality outweighs women’s survival advantage in Asia and North Africa. British Medical Journal, 304, 587–588. Sharpe, P. (2007). The female labour market in English agriculture during the industrial revolution: Expansion or contraction? In: N. Goose (Ed.), Women’s work in industrial England: Regional and local perspectives (pp. 51–75). Hatfield: Local Population Studies. Simonton, D. (1998). A history of European women’s work: 1700 to the present. London: Routledge. Snell, K. (1981). Agricultural seasonal unemployment, the standard of living, and women’s work in the south and east. Economic History Review, 34, 407–437.

204

BERNARD HARRIS

Snell, K. (1985). Annals of the labouring poor: Social change and agrarian England, 1660–1900. Cambridge: Cambridge University Press. Steckel, R. (1992). Stature and living standards in the United States. In: R. Gallman & J. Wallis (Eds), American economic growth and standards of living before the Civil War (pp. 265–308). Chicago: University of Chicago Press. Steckel, R. (1995). Stature and the standard of living. Journal of Economic Literature, 33, 1903–1940. Tabutin, D., & Willems, M. (1995). Excess female mortality in the developing world during the 1970s and 1980s. Population Bulletin of the United Nations, 39, 45–78. Tanner, J. (1962). Growth at adolescence, with a general examination of the effects of hereditary and environmental factors upon growth and maturation from birth to maturity (2nd ed.; first published 1955). Oxford: Blackwell Scientific Publications. Tanner, J. (1989). Fetus into man: Physical growth from conception to maturity. Cambridge, MA: Harvard University Press. Tanner, J. M., Whitehouse, R. H., & Takaishi, M. (1966). Height, weight, height velocity, weight velocity: British children, 1965. Archives of Disease in Childhood, 41, 454–471, 613–635. Titmuss, R. (1987). The position of women. In: B. Abel-Smith & K. Titmuss (Eds), The philosophy of welfare: Selected writings of Richard M. Titmuss (pp. 87–101). London: Allen and Unwin. United Nations Secretariat. (1988). Sex differentials in life expectancy and mortality in developed countries: An analysis by age groups and causes of death from recent and historical data. Population Bulletin of the United Nations, 25, 65–106. Van Wieringen, J. (1979). Secular growth changes. In: F. Falkner & J. Tanner (Eds), Human growth (vol. 2, pp. 445–473). London: Baillie`re Tindall. Van Wieringen, J. (1986). Secular growth changes. In: F. Falkner & J. Tanner (Eds), Human growth: A comprehensive treatise (vol. 3, pp. 307–331). London: Plenum. Verdon, N. (2002). Rural women workers in nineteenth-century England: Gender, work and wages. Woodbridge: Boydell Press. Waldron, I. (1985). What do we know about causes of sex differences in mortality? A review of the literature. Population Bulletin of the United Nations, 18, 59–76. Winter, J. (1982). The decline of mortality in Britain, 1870–1950. In: T. Barker & M. Drake (Eds), Population and society in Britain 1850–1980 (pp. 100–120). London: Batsford. Wohl, A. (1984). Endangered lives: Public health in Victorian Britain. London: Methuen. Woods, R. (1997). Causes of death in England and Wales, 1851–1860 to 1891–1900: The decennial supplements (computer file) (SN: 3552). Colchester: UK Data Archive, March 1997. World Health Organisation. (1977–1978). Manual of the International Statistical Classification of Diseases, Injuries and Causes of Death. (2 vols). Geneva: World Health Organisation. World Health Organisation. (1992). International statistical classification of diseases and related health problems, tenth revision. (vol. 1). Geneva: World Health Organisation. Wrigley, E. A., Davies, R., Oeppen, J., & Schofield, R. (1997). English population history from family reconstitution 1580–1837. Cambridge: Cambridge University Press.

TAXATION WITH (?) REPRESENTATION: THE POLITICAL ECONOMY OF PUBLIC FINANCE IN ANTEBELLUM CALIFORNIA Mark T. Kanazawa ABSTRACT Many existing studies point to the political contentiousness of attempts by states in the 19th century to impose property taxes, which after midcentury comprised the main source of state revenues. Yet studies fail to establish a convincing connection between interest group political effectiveness and resulting favorable property tax legislation. This paper takes a closer look at one state that adopted property taxation in the mid19th century and documents intense inter-occupational conflicts between miners and ranchers over creation and administration of the system of property taxes. These conflicts occurred for various institutional reasons, including differential costs of enforcing tax collection and the short-lived political ascendance of miners during, and in the years following, the Gold Rush. The empirical results strongly suggest short-term capture by miners of the state legislature, followed by loss of capture ability as gold declined in economic importance in the 1860s.

Research in Economic History, Volume 26, 205–233 Copyright r 2008 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(08)26004-0

205

206

MARK T. KANAZAWA

A common theme in the economic history of American public finance in the 19th century is the political contentiousness of the property tax systems devised by states and localities to raise revenues for state and local governments. Existing studies have identified numerous and varied manifestations of political struggles over the distributional burden of property taxes: in northern states over taxation of commercial wealth, in southern states over taxation of slaves, in Midwestern states over taxation of public works projects and farming real estate, and in western states over taxation of mineral resources, especially hardrock mining. These struggles are widely believed to have exerted real influence over the ultimate form taken by state and local property tax systems through legislative enactments.1 Existing studies, however, commonly suffer from a failure to convincingly connect legislative enactments regarding property taxes to the effective exercise of political power by groups that stand to gain from those enactments. Positive political theory predicts that the ability of constituencies to obtain favorable legislation depends upon their numbers, how much they stand to gain both as a group and individually, and the magnitude of the transaction costs they incur in organizing to be a politically effective constituency. Logically, time series analyses of property tax laws should tie changes in the law to changes in one or more of these factors. It is rarely the case, however, that significant changes in numbers, stakes, or transaction costs occur in sufficiently close conjunction with a change in legislation to provide a clear connection between the two. On the other hand, cross-sectional analyses of property tax laws based on, for example, state data suffer to some extent from lack of comparability both of interest group dynamics and of the property tax provisions that result from competition among these groups. For these reasons, existing studies provide an incomplete picture of the politics of 19th century property taxation. This paper uses detailed historical evidence to examine property taxation in California in the early years of statehood. The objective is to shed light on the political determinants of 19th century property tax systems more broadly. This context is an apt subject of study for a number of reasons, not the least of which was that California relied heavily on property taxes to fund state and local expenditures and was thus very similar to other states of the Union. In addition, it is well known that gold was discovered in California shortly prior to statehood, and the rapid, massive influx of goldseekers had a major impact on state politics. By the early 1850s, miners were present in sufficient numbers to all but capture both houses of the state legislature, which allowed them to enact property tax legislation that was

Political Economy of Public Finance in Antebellum California

207

highly favorable to miners. The period of gold fever, however, quickly subsided so that on the eve of the Civil War, the California economy was already undergoing significant diversification. The rapid ascendance of mining after the discovery of gold and then its descent shortly thereafter provides a unique opportunity to examine both the acquisition and loss by miners of the ability to ‘‘capture’’ the legislature to secure favorable legislation. But the story told here is more complex than a simple interest group dynamic in the legislative arena over the incidence of the property tax burden. A fundamental premise of this study is that a state’s property tax system originates and evolves over time in a complex interaction of interest group dynamics in which interest groups may simultaneously engage in public- and private rent-seeking. Not only do interest groups lobby for favorable legislation, but they operate privately to maximize the benefit and/or minimize the cost of existing policies and in doing so, they may alter the incentives of legislators to retain, modify or repeal those very policies. Consequently, the manner in which policies evolve over time depends both upon the effective political strength of interest groups and their ability to take effective private actions to shift benefits and costs.2 As will become clear in the discussion, the ability to shift benefits and costs will depend upon how costly it is for the government to enforce the terms of its policies. The more costly is enforcement, the greater is the latitude of private agents to shift benefits and costs. Existing studies of property taxation have examined legislation and implementation largely in isolation from each other, and/or have only peripherally considered enforcement costs. As I will argue below, given the existence of significant enforcement costs, we obtain considerably greater insights into property tax policy by considering both private- and public rent-seeking avenues.

THE CREATION OF THE CALIFORNIA TAX SYSTEM We observe the political dynamics of property taxation in California as early as 1849, when a constitutional convention assembled in San Jose to draft a constitution for the prospective state. Ultimately contained in this constitution was a provision that called for ‘‘equal and uniform’’ taxation, that all property in the state ‘‘shall be taxed in proportion to its value.’’ This provision was unremarkable within the context of other state constitutions during this time period, many of which contained similar provisions (Kettleborough, 1918; Einhorn, 2001). Yet in California, its proposed inclusion in the constitution sparked a heated debate within the convention because of a proposed amendment, strongly supported by delegates from

208

MARK T. KANAZAWA

the southern part of the state, which stipulated that property assessors and tax collectors be popularly elected within the locality in which property taxes would be collected. Southern support for this amendment was fed by the recent rapid influx of gold hunters, who by the time of the convention already vastly outnumbered those living in the south, many of who were living on large private landholdings amassed under Mexican rule. A special census taken in 1852 reveals that the mining counties of Calaveras, El Dorado, Mariposa, Nevada, Placer, Sierra, Tuolumne, and Yuba contained nearly 55% of the entire population of the state and nearly 64% of the population outside of San Francisco (Statistical View of the U.S., 1854, p. 394). The discovery of gold, mostly in these counties, occurred on public lands belonging to the federal government and in particular not privately owned and therefore not subject to a property tax. This combination of circumstances led southerners to fear that they would bear the brunt of any system of property taxation set in place. In speaking out in favor of this amendment, one southern delegate noted that there were probably about 100,000 people living in California at the time, and: Of this (sic) hundred thousand persons there may be fifteen thousand native Californians, and of the whole number, not exceeding five thousand are proprietors of landed property, and the great majority of the residue have neither real estate nor personal property that is taxable . . . (Browne, 1850, p. 369)

This delegate also expressed doubt that a system of property taxes would even be able to raise sufficient revenues and recommended the imposition of a poll tax to supplement property tax revenues. It is clear that he believed both that a system based solely on property taxes would entail considerable costs of collection and that such a system would heavily tax his southern constituents. Both of these beliefs would be borne out by subsequent events. Opponents of this amendment countered that it would permit localities to avoid paying taxes by allowing local landowners to control the processes of assessment and collection. As one northern delegate argued: If the northern portion of the country is to be charged with a design to throw an undue burden of the taxation of the State upon the southern portion, why may not, upon the very same foundation, the charge be thrown back; and why may we not aver that these gentlemen, in assessing their own lands, may avoid their proportionate burden of taxation? (Browne, 1850, p. 371)

He added: I consider this proposition to be equivalent to this. Let every man send in his assessment of his property; let every county with these divers local interests choose their assessors, with instructions how they are to assess the property. It is but one remove from the

Political Economy of Public Finance in Antebellum California

209

odious principle that a man shall have a perfect right to fix his property at what he pleases. (Browne, 1850, p. 372)

However, in the end the convention included the amended section in the new state constitution. The consensus among historians appears to be that the southern delegates feared the political power of mining interests and insisted on inclusion of the amendment to protect them from excessive taxation. Northern delegates acquiesced to its inclusion in the constitution rather than risk withdrawal of the southern delegates from the convention.3 Article 11, section 13 as it finally appeared in the constitution read as follows: Taxation shall be equal and uniform throughout the State. All property in this State shall be taxed in proportion to its value, to be ascertained as directed by law; but assessors and collectors of town, county, and State taxes, shall be elected by the qualified electors of the district, county, or town, in which the property taxed for State, county, or town purposes is situated. (Constitution of the State of California, in Statutes of California, 1850, p. XI)

The constitution was subsequently ratified by an overwhelming margin in a popular referendum in November of 1849 (Fankhauser, 1913, p. 118). Within this constitutional framework it was up to the legislature to establish the particulars of the revenue system of the fledgling state, which it did in March 1850 when it enacted a statute creating a system of property and poll taxes.4 Under this system, with certain exemptions all real and personal property was subject to taxation. Real property included land and buildings, while personal property included goods, furniture, watercraft, and public stock in turnpikes, bridges, and insurance companies. Among the exemptions to this property tax were schoolhouses, courthouses, jails, churches, cemeteries, and the property of the federal government. This statute also called for a poll tax to be assessed on every adult male between the ages of 21 and 50 years. This statute also provided a detailed set of procedures for assessment and collection of property taxes, which were locally administered by counties and then shared with the State. County Assessors were charged with compiling rolls of all taxable inhabitants within the county, their taxable property, and the value of such property and any improvements made to it. Valuation decisions made by the Assessor could be appealed to the county Court of Sessions, which served as a Board of Equalization. Property and poll taxes were to be levied by county Auditors, with tax rates for state revenues determined by statute and rates for county revenues set by the Court of Sessions. The actual tax collection itself was to be done by

210

MARK T. KANAZAWA

the County Treasurer, who was empowered to seize and sell property for non-payment of taxes and to assess additional 10% damages for failure to pay. Upon collection, Treasurers were required to settle up with the county Auditor and the state Comptroller and were subject to additional penalties for any revenue shortfalls. True to the constitution, separate statutes enacted by the legislature called for county Assessors and county Treasurers to be locally elected. However, to guard against abuse and dereliction of duty, these statutes also required them to execute bonds as surety for faithful performance of duty.5 The legislature then enacted one other statute that temporarily set the property tax rate at 50 cents per 100 dollars of taxable property for state taxes, and mandated a poll tax of $5 per adult male.6 These laws formed the fundamental basis for the system of property and poll taxation of the state for many years. One other institutional factor would intimately affect the operation of the tax system; namely, that state taxation of public lands belonging to the federal government was proscribed under federal law. This proscription on state taxation was explicitly stated in the law passed by Congress that admitted California to the Union: the said State of California is admitted into the Union upon the express condition that the people of the said State, through their legislature or otherwise, shall never interfere with the primary disposal of the public lands within its limits, and shall pass no law and do no act whereby the title of the United States to, and right to dispose of, the same shall be impaired and questioned; and that they shall never lay any tax and assessment of any description whatsoever upon the public domain of the United States . . . (U.S. Statutes, 31st Cong., 1st Session, Chapter 50, p. 452.)

This law thus affirmed what many southern delegates to the convention had feared; namely, that the mining regions that mainly lay in the central and northern parts of the state were unlikely to bear their proportionate share of the tax burden.

‘‘EQUAL AND UNIFORM?’’ THE POLITICS OF PROPERTY TAXATION After creating its system of property and poll taxes, the state quickly moved to use it to generate tax revenues, which were badly needed to pay for the burgeoning operating expenses of the state. However, it quickly found itself falling short. In the very first fractional fiscal year ending June 1850, not a

Political Economy of Public Finance in Antebellum California

211

single dollar of property or poll taxes was paid into the state treasury, as the state set its administrative system in place. Simultaneously the state was incurring nearly $350,000 in administrative expenses that included little more than paying salaries and expenses of public officials and printing public documents (Annual Report of the Comptroller of State, 1850, pp. 522–23). In order to pay its initial expenses, the state used the proceeds of a $300,000 bond issue while issuing treasurer’s warrants for most of the remainder. Property and poll tax revenues began to flow into the treasury in the following fiscal year, during which the state managed to collect nearly $300,000 in such revenues. By that time, however, annual total state expenditures had climbed to nearly $742,000, forcing the state to float another bond issue and to issue additional warrants. This pattern of spending well in excess of revenues persisted for several years through fiscal year 1856 (See Fig. 1), at which time total civil debt stood at about three million dollars. The early inability of the state to raise revenues to match expenditures was a result of a number of factors, including a loss of federal revenues from never having enjoyed territorial status and the rapidly increasing expenditure demands for schools, hospitals, etc. associated with the rapid influx of gold seekers.7 An important additional factor was broad-based difficulty in tax collection efforts relating both to tax evasion and to corruption among local collecting officials. By the end of the first full fiscal year, for example, it has been estimated that roughly two-thirds of those owing a poll tax were delinquent on their payments (Ignoffo, 1999, p. 69). By 1857, enforcement of the poll tax had not improved much, according to the state Comptroller who reported that the state was receiving less than 38% of its expected poll tax revenue (Annual Report of the Comptroller, 1857, p. 6). But even when taxes were paid, revenues flowed into the state Treasury slowly and by 1853, the state had brought suit against several county treasurers and other local tax officials in Tuolumne, Calaveras, Sonoma, and El Dorado counties for delinquent tax payments into the state treasury (Annual Report of the Comptroller, 1853, pp. 48–9). By the end of the decade, additional suits had been brought against the treasurers of Sacramento, Santa Clara, Sierra, Trinity and Yolo counties (Annual Report of the Comptroller, 1855, p. 32; 1859, p. 5.). Available evidence strongly suggests that tax collection was especially problematic in the mining regions, particularly early in the decade. In his very first annual report in 1850, the state Comptroller commented on the state’s financial difficulties, making obvious reference to the recent

212

MARK T. KANAZAWA 1800

1600

1400

Thousands of Dollars

1200

1000

800

600

400

Receipts Expenditures

200

62

61

18

60

18

59

18

58

18

57

18

56

18

55

18

54

18

53

18

52

18

51

18

18

18

50

0

Year

Fig. 1.

Annual Revenues and Expenditures, State of California: 1850–1862.

rapid influx of miners and its effect on the state’s ability to raise tax revenues: The population and wealth of the State are of such a character that in many cases they cannot be reached by taxation. Our population is more unsettled and changeable than, perhaps, that of any other State in the Union. As a consequence, property is in like condition. (Annual Report of the Comptroller, 1850, p. 546)

Two years later, the situation had not changed much when the Comptroller again bemoaned the sad state of tax revenues, again mentioning the seemingly insuperable problem of taxing a transient

Political Economy of Public Finance in Antebellum California

213

population but adding that the public lands nature of the mining regions exacerbated the difficulties of raising revenues from this population: The real difficulty that affects the finances of the State, arises from the insecurity of Land Titles, the large Territorial possessions in our midst, claimed by the General Government; their consequent immunity from State taxation, and the transient and movable character of our population, rendering the collection of the tax imposed per capita, impracticable and impossible to a certain extent. (Annual Report of the Comptroller, 1852, p. 510)

In 1854, he was more explicit in singling out mining activity as being difficult to tax: During the three first years of our existence as a State, the mining portions of the country paid comparatively little into the Treasury. Although assessments were made in these localities, the collecting officers were in many instances unable, a few months after, to find the parties or property assessed, so migratory and changing was the population. (Annual Report of the Comptroller, 1854, p. 38)8

In response to these difficulties, the legislature instituted a number of reforms in the manner in which property and poll taxes were collected. In 1851, the legislature attempted to improve collection enforcement by putting tax assessment and collection duties in the hands of county Sheriffs and providing them a larger share of collected tax revenues. It also attempted to direct more tax revenues into the state Treasury by explicitly mandating that the state receive two-thirds of all poll-tax revenues and placing a legislative ceiling on the property tax rate that could be set by counties. In the following year the legislature placed the onus on officers of mining companies to provide county assessors with written statements of the value of property owned by their companies, with specific penalties for non-compliance. The same statute also provided for payments for informants on those engaged in tax fraud. A separate statute authorized the State Comptroller to bring suit against county treasurers and other local collection officials for failure to perform their duties. Further laws were enacted by the legislature in 1853 and 1857 that tightened and streamlined procedures for collecting tax revenues. However, even by 1859 the state comptroller was expressing serious concerns about ‘‘the inefficient manner in which the laws for the collection of the revenue are enforced’’ (Annual Report of the Comptroller, 1859, p. 5). One important consequence of these tax collection difficulties was that different portions of the state ended up bearing very different tax burdens, despite the constitutional mandate for equal and uniform taxation. We have already seen that southern delegates to the 1849 convention were concerned

214

MARK T. KANAZAWA

that the southern counties would bear a disproportionate amount of taxation. In 1852, Governor McDougal confirmed these fears when he reported that the tax system was operating unequally upon different sections of the state, with southern counties bearing considerably more than their proportional share of taxes while enjoying much less than proportional legislative representation, and he provided figures to support this assertion. For example, 6 largely agricultural counties in the south enjoyed less than one-third the legislative representation of 12 mining counties, while paying 37 times as much in per-capita property taxes. Similarly, the same 6 southern counties, despite having a population estimated at about 5% of the 12 mining counties, were actually paying slightly more in total poll taxes.9 Governor McDougal concluded: The interests of the southern counties are in many respects inimical to those of the north. To remedy whatever evils exist, seems impossible under the present Constitution, for that instrument provides that there shall be no special legislation, declaring that ‘all laws of a general nature shall have a uniform operation’. It declares, also, that ‘taxation shall be equal and uniform throughout the state.’ This equality now exists only a legal sense; for while the southern counties, which are mostly covered by grants and in the possession of individuals, pay a heavy tax upon every acre of their land . . . , the mining counties, exceedingly prolific in the returns they make to their occupants, being almost entirely the property of the Federal Government, pay nothing, comparatively, into the State Treasury. (Annual Message of the Governor, Journal of the Assembly, 1852, p. 13)

Five years later, the state Comptroller was declaring that the ‘‘burthens of taxation’’ were not being ‘‘equitably borne’’ under the then-current tax system, with regard to both property and poll taxes (Annual Report of the Comptroller, 1857, p. 6).10 The reaction of the southern counties to this unequal incidence of taxes was predictable: they did not at all like it. And this dissatisfaction gave rise to concerted, repeated attempts by southern legislators throughout the remainder of the decade to have the state split up into two or three smaller entities, either states or territories. As early as 1851, representatives of various southern counties convened in San Diego and Santa Barbara and called for division of the state, citing laws oppressive to the south and disparities in taxation (Ellison, 1950, pp. 174–75). Beginning the following year, the issue of division was repeatedly advanced by southern delegates, mostly through attempts to call for a constitutional convention to effect the constitutional changes that would permit a division. In 1852 and 1853, bills enabling the calling of a convention were passed in the assembly but defeated in the senate (Ellison, 1950, pp. 178–81). In 1855, an assemblyman from San Bernardino County introduced a bill creating a new state

Political Economy of Public Finance in Antebellum California

215

consisting essentially of the southern half of the state, which was reported to a select committee. This committee reported back a substitute that would have divided the state into three portions, a southern, a central, and a northern portion. The report argued for division on the basis that the existing state was essentially too big and cumbersome and that the interests of the different regions were ‘‘widely dissimilar’’ (Report of the Select Committee with reference to Division of the State, 1855, p. 6). It also made reference to the redistributional consequences of the existing situation: The center reaps all the benefits, enjoys all the advantages of government favor, while the extremities are compelled to bear a large proportion of the burden of taxation. All improvements of a public character, colleges, asylums and State buildings of every description, are made at or near the seat of government, and the greatest interest the inhabitants of distant counties can have in them, arises from their sufferings as taxpayers. (Report of the Select Committee with reference to Division of the State, 1855, p. 7)

Finally, it added, within the arena of national politics it would not hurt to have six senators instead of two. The legislative session ended, however, before full action could be taken in both houses on the bill. Further legislative attempts proposing division died without action being taken in 1856 and were narrowly defeated in 1857. Finally, in 1859 an assemblyman from Los Angeles introduced joint resolutions in the assembly proposing division of the state and the creation of the Territory of Colorado out of five southern counties. The resolutions argued that the state was too large and diversified and that legislation that was nominally uniform was in fact unfair to the south.11 The resolutions were referred to a special committee (chaired by the same Los Angeles assemblyman) that came out with two reports, a favorable majority report and an unfavorable minority one. The majority report tersely argued that inhabitants of the southern counties had ‘‘good and valid reasons’’ to want to separate from the rest of the state and reported a bill that would effect division (Journal of the Assembly, 1859, pp. 341–42). The minority report, authored by two northern assemblymen, contained a more expansive discussion that questioned the constitutionality of effecting division through a simple legislative act (Journal of the Assembly, 1859, pp. 350–52). The bill accompanying the majority report was narrowly passed in both the senate and assembly and signed by the governor. The final law called for the question of division to be submitted to the voters of the affected southern counties, with a two-thirds majority required for passage. In such an eventuality, the Governor was directed to send the law and the results of the election to the President and to Congress for action. In the popular vote

216

MARK T. KANAZAWA

among residents of the affected counties conducted in September of 1859, the law was comfortably ratified by a nearly three-to-one margin (Ellison, 1950, p. 187). In conformity with this law, Governor Latham transmitted the statute and the election results to President Buchanan in early 1860, along with a letter detailing his views supporting the constitutionality of the statute (Journal of the Assembly, 1860, p. 125). The state was of course never divided and indeed, there is no evidence that division of California was ever even considered in Washington. The 1860–1861 Congressional Globe contains no record of the issue being debated or even raised on the floor of Congress, despite the fact that Latham was present as one of the senators from California, having been named to the position upon the death of the sitting senator, David Broderick. In explanation, historians have argued that after 1860 Congress became increasingly occupied with the Civil War, which left no time for consideration of division (Ellison, 1950, p. 189). I would add that what the southern counties were requesting – the reversion of part of a state to territorial status with the prospect of readmission as a separate state – may well have caused serious political difficulties given the highly charged issue of slavery in newly admitted states that was prominent in national politics in the 1850s. The admission of California into the Union as a free state as part of the Compromise of 1850 had been extremely controversial, and many in Congress may have viewed division as potentially reopening the slavery issue in that portion of the country.12 Indeed, a minority report in the assembly on the Governor’s message to Buchanan expressed concern that division could lead to intramural struggles over slavery similar to those that had occurred in the wake of the Kansas-Nebraska Act of 1854: The civil discord which now so trammels Congress and threatens our Federal Union had its origin in questions growing out of our territorial organization, and California, occupying a position so eminently conservative, should be the last to offer another opportunity for the enactment of new Kansas difficulties. (Journal of the Assembly, 1860, p. 233)

Congress may have lacked both the time and the political will to consider division.

HOW SHAMELESS CAN YOU GET? PROPERTY TAX EXEMPTIONS FOR MINING The story being told here is not, however, merely that a variety of economic and institutional factors raised the costs of collecting property and poll taxes

Political Economy of Public Finance in Antebellum California

217

in an ‘‘equal and uniform’’ way and in doing so almost tore the state asunder. While all this was occurring, another separate but related political story was unfolding; namely, legislative attempts by mining representatives to secure exemption from property taxes for mining claims and mining machinery. They actually managed to accomplish this in 1857 when the legislature, in enacting a revision to the existing revenue law, added mining claims to the list of property explicitly exempted from property taxes.13 This could not have sat well with the southern counties and probably contributed to their desire to separate from the rest of the state. Indeed, in his 1860 communication to President Buchanan, Governor Latham specifically mentions this tax exemption as contributing to the dissatisfaction of the southern counties, who: complain that the taxes upon their land and cattle are ruinous – entirely disproportioned to the taxes collected in the mining region; that the policy of the State, hitherto, having been to exempt mining claims from taxation, and the mining population being migratory in its character, and hence contributing but little to the State revenue in proportion to their population, they are unjustly burdened; and that there is no remedy, save in a separation from the other portion of the State. (Journal of the Assembly, 1860, p. 125)

How the mining representatives managed to secure exemption from property taxes is the issue to which we now turn. The property tax exemptions secured by miners in 1857 were part of an interesting and in some ways quite telling, larger progression in the tax treatment of mining claims and machinery during the Gold Rush. Placed within this context, the 1857 mining exemptions represented the height of tax liberality with regard to miners throughout the entire period from 1849 to 1865. Laws passed early in the 1850s gave mining no special tax treatment, subjecting mining in essence to the same taxes as a wide variety of other economic activities. The revenue law enacted in 1852 specifically subjected gold dust and stock held in mining companies to taxation. The law enacted the following year explicitly subjected mining machinery to taxation, and, reflecting the rapid development of water supplies to service the mining industry, also made canals and water races taxable. The 1857 legislation that expressly exempted mining claims from taxation was for miners a significant liberalization of existing property tax law. This provision was controversial and the Senate Finance Committee, to whom the bill had been referred, proposed striking the mining claims exemption. However, the exemption was retained, and each house ended up passing slightly different versions of the bill by wide margins, which were quickly

218

MARK T. KANAZAWA

reconciled in conference committee. An amendatory law passed the following year also retained the mining exemption. In 1860, however, the legislature began to enact tax legislation that clamped down on mining exemptions. In that year, the legislature amended the existing revenue law by subjecting mining machinery to taxation, while retaining the exemption on mining claims. The exemption was embodied in the following provision contained in Section 2, subdivision 8: Mining claims (shall be excluded from property taxation); provided, that all machinery used in mining claims, and all property and improvements appurtenant to, or upon, mining claims, which have an independent and separate value that can be estimated and taxed, without taxing such mining claims, shall be subject to taxation. (Statutes of California, 1860, pp. 365–401)

Then in 1864 miners faced an even stiffer, and ultimately insurmountable, challenge to their tax-exempt status when a bill was passed by the legislature that simply repealed the exemption for mining claims.14 The legislature had come full circle, through a period of tax liberalization back to its original stance: treating mining claims like any other form of property. The question is: how do we understand this legislative history, and what were the reasons for the legislative reversals observed both in 1857 and again beginning in 1860?

EVIDENCE ON POLITICAL SUPPORT FOR TAX EXEMPTIONS Certainly an important part of the answer is the fact that during this period, mining representation in the state legislature rose and then fell, owing to rapid demographic shifts. The discovery of gold in 1848 triggered, in the early 1850s, a major influx of immigrants mostly headed for the mining regions. The result was dramatically increased representation of mining regions in the state legislature as legislative districts were added and existing districts were subdivided, to more accurately reflect the shifting population realities in the state.15 From 1850 to 1854 the number of state assemblymen from mining counties increased from 29% to nearly 48% of the total state assembly, while mining senators increased from 25% to 47%.16 After 1860, however, declining gold production and the continuing growth and diversification of the state economy resulted in a decline in mining representation. Whereas in 1860 mining counties accounted for nearly

219

Political Economy of Public Finance in Antebellum California

48% of all senators and 43% of all assemblymen, by 1864 these percentages had fallen to 37% and 35%, respectively. To provide a better sense of the overall pattern of representation of miners versus non-miners during this period, Fig. 2 shows the ratio between the number of senators and assemblymen representing mining counties and the number representing non-mining counties for every year from 1850 through 1864. In 1850, the ratio of mining to non-mining representatives was a little over 41% in the assembly and only 25% in the senate, meaning that mining senators were outnumbered four to one. These ratios rose steadily throughout the early 1850’s, reaching nearly 90% in the senate by 1854 and actually exceeding 100% in the assembly in 1855 and 1856. The level of mining representation in the senate remained steady at around 90% until 1860, when there began a noticeable decline. Mining representation in the assembly fell from its peak in 1857 but remained steady at about 75% until 1861 when it, too, suffered a decline. In terms of being able to 1.2

1

Percentage

0.8

0.6

0.4

Senate Assembly

0.2

64

63

18

62

18

61

18

60

18

59

18

58

18

57

18

56

18

55

18

54

18

53

18

52

18

51

18

18

18

50

0

Year

Fig. 2.

Ratio of Number of Representatives from Mining Counties to Representatives from Non-Mining Counties, State of California, 1850–1864.

220

MARK T. KANAZAWA

command a majority coalition within the Assembly, the period from 1854 to 1861 was clearly the ‘‘golden age’’ for miners in California. Throughout this period, mining representatives could succeed in enacting pro-mining legislation if they voted as a bloc and enticed a handful of other representatives to go along. By 1864, however, a solid voting bloc of mining representatives would not have commanded anywhere near the majority necessary to get a bill successfully through the legislature. The rise and fall of mining representation during this period is broadly consistent with the enactment of mining exemptions from property taxes in 1857 and their subsequent repeal.17 The changing numbers of mining representatives alone cannot, however, explain the observed pattern of property tax treatment of miners. It seems anomalous, for example, that mining exemptions were enacted in 1857, after mining representation had declined to 43% of the total assembly, well down from its peak of over 52% in 1855 and 1856. One might expect mining exemptions to have been enacted earlier, when mining counties enjoyed an absolute majority in the assembly and a near majority in the senate. Recall, however, that the state had been suffering sizable budget deficits since 1850, including a massive shortfall in 1856 when revenues were less than half of total expenditures. These continual deficits added up to a $3.2 million civil debt by the beginning of 1857. By that time, however, the financial picture had brightened, and in 1857 the state incurred the smallest deficit in its short history. The easing of financial conditions may well have reduced political opposition to the exemptions among non-mining legislators. Also important to note were some subtle changes in the political dynamics of support for mining exemptions that occurred after 1860. These changes are illustrated most clearly by comparing the legislative histories of the exemption laws enacted in 1860 and 1864. Both laws were hotly contested in the legislature. In 1860, mining representatives in the assembly barely managed to defeat two hostile amendments that would have either abolished or emasculated the mining exemption. One amendment, to strike the mining exemption entirely, was defeated 32–22. The other amendment also struck the exemption but provided that mining claims would be assessed according to the value placed upon them by the owners. This amendment was even more narrowly defeated, 29–25. A closer look at the roll-call votes on these two amendments reveals a sharp division between the votes of representatives of mining districts and those of all other representatives. Columns (1) and (2) in Table 1 lists the 12 mining counties along with the votes of their representatives in the assembly on these two amendments.18 These data reveal that the two

221

Political Economy of Public Finance in Antebellum California

Summary, Roll-Call Votes on Key 1860, 1864 Votes. COUNTY Value of Gold Prod as % of Total Mfg Value, 1860

Mariposa Nevada Tuolumne Sierra Calaveras Amador Trinity Tulare Placer Siskiyou Plumas El Dorado

0.987 0.880 0.852 0.830 0.775 0.726 0.706 0.659 0.610 0.582 0.395 0.295

1860

1864

(1) VOTE1

(2) VOTE2

(3) VOTE3

(4) VOTE4

YES

NO

YES

NO

YES

NO

YES

NO

0 0 0 0 0 0 0 0 0 0 0 0

2 3 0 2 2 1 0 0 4 1 2 6

0 0 0 0 0 0 0 0 0 0 0 0

2 3 0 2 2 1 0 0 4 1 2 6

1 0 2 1 0 0 0 1 1 0 0 0

0 3 0 1 2 2 0 0 2 2 0 4

1 3 1 1 2 2 0 1 1 2 0 4

0 0 1 1 0 0 1 0 1 0 0 0

TOTAL, Mining Counties All Others

0

23

0

23

6

16

18

4

25

6

23

8

36

8

4

37

TOTAL

25

29

23

31

42

24

22

41

VOTE1: To substitute that owners could specify how much their claims were worth, for purposes of assessment. VOTE2: To strike section subjecting mining equipment to taxation. VOTE3: To repeal mining exemption. VOTE4: To reconsider repeal of mining exemption.

amendments (VOTE1 and VOTE2) faced unanimous opposition among representatives of these 12 gold mining counties, while enjoying overwhelming support among non-mining counties. In 1860, miners were apparently able to retain tax-exempt status for their claims because the mining counties enjoyed a substantial majority of representatives in the assembly and voted largely as a bloc. By 1864, however, the solidarity earlier enjoyed by mining representatives began to crumble. Passage of the 1864 bill was perhaps even more vigorously contested than had been the 1860 bill. The 1864 bill survived a proposal to indefinitely postpone consideration before it was brought up for a vote in the Senate, and then after passage survived a vote to reconsider passage. In the Assembly, passage of the bill survived an amendment proposed by a representative from a mining district that would have

222

MARK T. KANAZAWA

stipulated that the law apply only to mines located on Spanish or Mexican land grants, and then after passage survived another vote to reconsider passage. What is even more interesting than their evident contentiousness, however, is that the roll-call votes on the 1864 mining legislation reveal definite cracks in what had previously been virtually monolithic support among mining representatives for tax-exemption. For example, in 1864 when the senate declined to indefinitely postpone consideration of the bill, 7 of the 21 ‘‘No’’ votes were cast by senators who represented mining counties. Had these seven votes gone the other way, the bill would have been indefinitely postponed. Similarly, 3 of the 23 votes in the Senate vote not to reconsider passage were cast by senators from that same group of mining counties. In the Assembly, 6 of the 42 votes for passage of the bill were by assemblymen representing mining counties, as were 4 of the 41 votes not to reconsider passage (See Table 1, VOTE3 and VOTE4). Other factors were apparently at play.

ECONOMETRIC ANALYSIS OF THE 1864 LEGISLATION To understand the reduction and then finally the elimination of taxexemption of mining claims, one must recognize that miners were not the only group that had an interest in the issue. Revenues raised through property taxation could be used to reduce other forms of taxes, or to fund expenditures on a variety of services that could benefit both miners and non-miners alike. Elimination of tax-exemption for mining claims could mean a sizable redistribution of net benefits to non-miners. The main hypothesis I investigate here is whether support for the 1864 law is correlated with reasonable expectations of non-mining groups that they stood to gain from the attendant increase in property tax revenues. I measure these expectations in two ways. The first is using the total population of the legislator’s district, which simply assumes that more populous counties would expect to receive a proportionately greater amount of expenditures by the state government. To capture this effect, I created the variable POPULATION, defined simply as the total population of the county or counties comprising the legislator’s district. A second way to measure expectations would be some index of actual expenditures enjoyed by each legislative district at the time the 1864 law was being considered. Here the assumption is that current total expenditures are

Political Economy of Public Finance in Antebellum California

223

an accurate predictor of future expenditures on the margin. The annual reports of the State Controller contain a breakdown of expenditures by the state, though most items are not apparently allocated to individual counties. One item that is listed by county, however, is state expenditures for public schools. To capture the actual expenditure effect, I created the variable SCHOOLEXP, which is defined simply as the total expenditure by the state on schools in the legislator’s district. In order to test the robustness of my results, I will consider POPULATION and SCHOOLEXP as alternative measures of expectation of benefit by legislators. In addition to expected benefits from increased expenditures, roll-call votes of legislators will be influenced by the interests of miners in retaining their tax-exempt status. Specifically, one would expect that legislators whose districts more heavily depend upon mining would be more likely to support a tax exemption for mining claims. To measure the extent to which a district depends upon mining, I constructed the variable MINING%, defined as the annual value of mining production as a percentage of total annual manufacturing value for the district. Finally, in the regressions I control for the party affiliation of the legislator with the dummy variable PARTY, which equals one if the legislator was a Union Democrat (the most prevalent political party at the time) and zero otherwise. Table 2 reports the results of logit regressions on several key roll-call votes in both the Senate and Assembly on the 1864 bill to end the mining exemption, including one (Vote2 þ Vote4) that combines votes in both the Assembly and Senate on reconsideration of passage of the bill.19 Overall, two key results should be noted. First, the estimated coefficients on MINING% reflect a consistent pattern of strong opposition to the bill by representatives of mining interests. This result is not surprising and indicates that all else equal, miners viewed this bill as definitely contrary to their interests. Second, the estimated coefficients on SCHOOLEXP and POPULATION indicate support for the bill among interests that would have been more likely to benefit from termination of the tax exemption. Though not as significant as the results on MINING%, the pattern is definitely observable, especially in the votes by the Assembly and in the combined vote on reconsideration. These estimated coefficients may be used to calculate the effect of each regressor on the probability that a legislator would have voted ‘‘Yes’’ in any roll-call vote. Table 3 reports various estimated probabilities for a vote for passage of the 1864 exemption bill by the Assembly, based upon the coefficients in column (5) of Table 2. For example, when SCHOOLEXP and MINING% both take on their sample mean values, the predicted

224

Determinants of Roll-Call Voting on Bill Repealing Property Tax Exemption for Mining Claims, 1864. Vote1 Constant Mining% Schoolexp Population Party % Correct N

Vote2

Vote3

Vote4

(Vote2 þ Vote4)

2.08 (1.10) 5.24 (1.91) 0.09 (0.075) – – 1.26 (1.60)

1.90 (1.16) 5.47 (1.90) – – 0.058 (0.047) 1.17 (1.60)

1.17 (0.86) 4.15 (1.46) 0.11 (0.081) – – 0.22 (2.28)

1.24 (0.85) 4.43 (1.46) – – 0.054 (0.036) 0.05 (2.11)

0.99 (0.49) 2.58 (0.90) 0.057 (0.28) – – 0.82 (0.82)

0.98 (0.52) 2.71 (0.92) – – 0.033 (0.19) 0.82 (0.82)

1.15 (0.67) 4.33 (1.15) 0.12 (0.05) – – 0.07 (1.07)

1.16 (0.67) 4.64 (1.15) – – 0.064 (0.026) 0.11 (1.07)

1.16 (0.52) 4.27 (0.90) 0.12 (0.045) – – 0.26 (0.93)

1.18 (0.52) 4.57 (0.89) – – 0.061 (0.021) 0.08 (0.92)

0.862 29

0.862 29

0.794 34

0.794 34

0.727 66

0.727 66

0.825 63

0.825 63

0.814 97

0.814 97

MARK T. KANAZAWA

Figures in parentheses are standard errors.  Significant at 1%.  Significant at 5%.  Significant at 10%. KEY TO VOTES Vote1: Vote in Senate to indefinitely postpone bill. Vote2: Vote in Senate to reconsider passage. Vote3: Vote in Assembly to pass bill. Vote4: Vote in Assembly to reconsider passage. (Vote2 þ Vote4): Combined vote in Senate and Assembly to reconsider passage.

225

Political Economy of Public Finance in Antebellum California

Estimated Probabilities of a Yes Vote in the Assembly to Pass 1864 Bill. MINING%

SCHOOLEXP

MIN MEAN MAX

MIN

MEAN

MAX

0.73 0.83 0.97

0.54 0.68 0.93

0.17 0.27 0.70

probability of a Yes vote is 0.68.20 One thing that is notable about the figures in Table 3 is that changes in SCHOOLEXP can have a large impact on the probability that a legislator would support terminating the tax exemption. Notice that this is particularly true when MINING% takes on higher values. Most relevant for our purposes, legislators from heavily mining counties may have opposed the tax exemption when those counties stood to gain more from the extra tax revenues. This result may help explain the earlier observation that in 1864 a significant number of legislators representing mining counties apparently ‘‘defected’’ and voted to get rid of the tax exemption for mining claims. My results suggest that the demise of the tax exemption on mining claims in 1864 occurred due to two related factors: a decline in legislative representation of mining interests, and an increase in other demands on state funds. Both of these occurred, of course, because of changes in the state’s demographic patterns as mining declined and the economy diversified. Miners apparently fought hard to retain their favored tax status, but by 1864 lacked the legislative numbers to stem the tide. Just as significantly, however, by this time mining counties were themselves divided, perhaps because some stood to gain from termination of the tax-exempt status of mining claims.

HOW UNUSUAL WAS CALIFORNIA? California was by no means the only state to experience difficulties in administering its property tax system in 19th-century America. The particular issues experienced by California during its early statehood seem, however, different from southern states struggling over taxation of slaves, northern states struggling over taxation of real versus personal property

226

MARK T. KANAZAWA

or intangible industrial personalty, or Midwestern states struggling over taxation of public works projects or farming real estate, all of which are issues stressed in the economic history literature. Indeed, on the surface, California’s early tax woes seem unique, stemming from a large historical shock – the Gold Rush – that dominated the state economy for an entire decade and enabled miners to capture the state legislature, while the transient nature of mining and its location on public lands exacerbated its difficulties in administering its system of property taxes. One thing that certainly was not unique about 19th-century California was the large concentration of public lands in the state. Indeed, during the 19th century the federal government owned a great deal of public lands throughout the vast majority of the country west and south of the original 13 colonies. These lands were acquired both through treaties and purchases from foreign countries and also by the ceding of claims to certain lands by some of the original states. By 1850, the federal government owned some 1.2 billion acres of public lands and as late as 1880 still owned about 900 million acres in the lower 48 states. (Atack & Passell, 1994, p. 251). Throughout the late 19th century, nearly every state outside the original 13 colonies contained significant amounts of federal land. As a general rule, like California these states were prohibited by their acts of admission to the Union from levying taxes on those lands. In addition, prior to 1820, acts of admission for states containing federal lands commonly prohibited states from levying taxes on lands for 5 years after passing into private hands.21 This exemption from property taxes was apparently a source of continual early tension between the federal government and public lands states, which periodically agitated for the right to tax federal lands. Beginning in the 1820s, Congress debated the desirability of retaining the provision that exempted federal lands from state taxation for 5 years after privatization. In 1825, the Senate Committee on Public Lands reported favorably on a bill that would have repealed this exemption, arguing: In most of the new states with the greatest abundance of real estate, there is very little personal property . . . the consequences of this has been, and must continue to be, that the revenue of their governments have been derived almost exclusively from land taxation. (cited in Hibbard, 1924, p. 86)

During this time, congressmen from public lands states repeatedly called for the federal government to relinquish its legal right to federal land within their boundaries. In doing so they argued, among other things, that cession

Political Economy of Public Finance in Antebellum California

227

of the federal lands would benefit farmers, who were bearing a disproportionate tax burden under the current system. The public lands in the hands of the new States would put those States on an equal footing with the original States. They would be sources of revenue for the improvement of those States, and they would relieve their agricultural interests from the heavy burdens they at present bear to sustain their treasury. Few of the old States are reduced to the necessity of a land tax to meet their current expenditures, but in the new States, having few objects of taxation, lay heavy contributions on their agricultural interests for that purpose. (Senator Hendricks of Indiana, cited in Winter, 1979, p. 57)

Extensive hardrock mineral strikes occurred on federal lands in various western states in the 19th century, including Nevada, California, Colorado, South Dakota, Idaho, and Montana (Marks, 1994, pp. 21–51). Evidence suggests that the rise of mining sometimes generated political tensions between the newly ascendant miners and other occupational groups such as farmers, ranchers, and merchants. In at least some cases, miners were able to wield significant political influence in order to obtain their desired political outcomes, including favorable tax policies. This occurred in Nevada after the discovery of the Comstock Lode in 1859 and in Montana during its territorial period, to give just two examples (Johnson, 1982; Mayer & Riley, 1985, pp. 71–72; Spence, 1975). Indeed, the 1884 Montana constitution expressly exempted mining claims for ‘‘valuable mineral deposits’’ from property taxation (Constitution of the State of Montana, 1884, Article XII, p. 25). One observer has described this provision as ‘‘supported by a well organized mining interest and opposed by a bewildered and weak agrarian group’’ (Smurr, 1955, pp. 223–24). And though this provision was amended 5 years later to eliminate the exemption, Spence (1975) has concluded that ‘‘nowhere was the impact of the mineral interests more apparent than in the tax structure of the [Montana] territory’’ (Spence, 1975, p. 205). Johnson (1982) documents the role of lode silver miners in the creation of the new Nevada constitution in 1864. According to Johnson, this constitution was a heavily politicized document whose final form was shaped by political struggles involving large industrial miners, small miners, ranchers, farmers, merchants, businessmen, and laborers (Johnson, 1982). A key issue in the convention was the treatment of mining taxes and specifically, whether mines should be taxed like any other form of property. Miners argued that a distinction should be drawn between producing and not-yet-producing mines so as not to discourage mining development, and they favored taxing the net proceeds from mining operations, as opposed to the market value of mines. Other groups countered that taxing mining proceeds would create a loophole that would permit mining companies to

228

MARK T. KANAZAWA

reduce the amount of taxes they paid. Miners apparently prevailed, as the final constitution taxed only mining proceeds (Kettleborough, 1918, p. 890). This evidence speaks to the political influence of miners under circumstances similar to the California Gold Rush, as the Comstock Lode discovery in proportionate terms was arguably as great an exogenous shock to the Nevada economy, if not greater.22

CONCLUSIONS The case of antebellum California provides a number of object lessons regarding the politics of property taxation in 19th-century America. Within the institutional framework of the time, key features of the state property tax system were visibly shaped by the rapid rise and fall of a key interest group: gold miners. They exercised their influence both through tax avoidance and through their elected representatives, who managed for a time to secure highly favorable tax treatment for their constituents. Tax avoidance was made possible by difficulties both in assessing mining property and in the actual enforcement and collection of taxes. Two salient factors combined to make uniform collection of property taxes extremely difficult. One was the fact that prosecution of mining was inherently much more transitory than agriculture. The second was that public lands belonging to the federal government were not taxable. Since mining occurred largely on public lands, these two factors combined to dramatically lower the relative burden of property taxes on miners. Successful attempts by mining interests to obtain selective exemptions from property taxes only exacerbated political divisions, which was almost responsible for a division of the state by 1860. Future research will broaden the focus and examine the property tax experiences of other states during this period. The evidence provided here hints at the richness of the strategies of states in devising methods for taxing mines. Though antebellum California seems to have considered only ad valorem property taxation or exemption therefrom, other states such as Nevada and later on, Montana, opted for systems based upon taxation of mining proceeds. Indeed, historical inspection of western mining states indicates wide variation with regard to reliance on ad valorem property taxation of mining. By 1944, for example, Warren Roberts reported that a strong minority of mining states relied on ad valorem property taxes, the remainder relying on other types of taxes, most notably taxes on mining

Political Economy of Public Finance in Antebellum California

229

proceeds.23 The interesting political question concerns how and why the differential tax treatments came to pass under different political and economic circumstances. Characterizing the dynamics that resulted in these varied outcomes in the cross-section of western mining states should go a long way toward providing a better understanding of the true causes and consequences of property taxation in the 19th century.

NOTES 1. The literature on the politics of 19th century property taxation is voluminous. The classic treatment of taxation of commercial wealth is by Richard Ely (1888) (see also Benson, 1965; Howe & Reeb, 1997). See Einhorn (2001) and the studies of Wallenstein (1984, 1985) for important treatments of the politics of taxation of slaves. Scheiber (1969) provides an excellent discussion of the political economy of taxation of internal public works projects during the antebellum period, and Buck (1969) and Hicks (1961) discuss the politics of taxation of farming real estate during the latter 1800s. See Spence (1975) and Johnson (1982) for good discussions of mining taxation. Recent important studies by Wallis and others characterize the political determinants of the long-term evolution of state and local tax systems within a public choice framework. See Wallis (2000a, 2000b, 2005), Sylla and Wallis (1998), and Heckelman and Wallis (1997). 2. I explored this idea within the context of 19th century public lands policy in my 1996 paper. See Kanazawa (1996). 3. See, for example, (Bancroft, 1888, pp. 297–98; Hunt, 1895, p. 44; Fankhauser, 1913, pp. 122–23; Ellison, 1950, p. 37). 4. ‘‘An Act prescribing the mode of Assessing and Collecting Public Revenue,’’ Chapter 52, Statutes of California, 1st Session, 1850, pp. 135–144. 5. ‘‘An Act concerning the office of County Treasurer,’’ Chapter 42, Statutes of California, 1st Session, 1850, pp. 115–17; ‘‘An Act concerning the office of County Assessor,’’ Chapter 43, Statutes of California, 1st session, 1850, p. 117. 6. ‘‘An Act defining the amount of revenue to be collected to defray the expenses of the Government of the State of California, for the year eighteen hundred and fifty,’’ Chapter 18, Statutes of California, 1st Session, 1850, p. 65. 7. Report of Committee on Ways and Means, 1849 CA Constitutional Convention, cited in (Fankhauser, 1913, p. 120); Annual Report of the Comptroller of State (1850), pp. 545–46. 8. See also (Annual Message of the Governor, Journal of the Assembly (1856), p. 20). 9. Annual Message of the Governor, Journal of the Assembly (1852), p. 13. See also (Caughey, 1940, p. 336). 10. See also (Cleland, 1975, pp. 121–24). 11. Sacramento Union, February 5, 1859; Los Angeles Star, February 19, 1859; cited in (Ellison, 1950, p. 184).

230

MARK T. KANAZAWA

12. Fogel recounts that the position of the U.S. Supreme Court in the 1850s was that neither Congress nor a territorial legislature could exclude slavery from a territory. See (Fogel, 1989, p. 343). 13. ‘‘An Act to provide Revenue for the Support of the Government of this State,’’ Chapter 261, Statutes of California (1857), p. 326. 14. ‘‘An Act supplementary to an Act entitled an Act to provide Revenue for the Support of the Government of this State, approved May seventeenth, eighteen hundred and sixty-one,’’ Chapter 416, Statutes of California (1864), p. 471. 15. During this period, the state underwent legislative reapportionments in 1851, 1852, 1853, 1857, 1861, 1863, and 1864. See (Allen, 1965, pp. 81–91). 16. Here ‘‘mining counties’’ is defined as: Amador, Calaveras, El Dorado, Mariposa, Nevada, Placer, Plumas, Sierra, Siskiyou, Trinity, Tulare, Tuolumne, and Yuba counties. These were the most-heavily gold-producing counties in the state in 1860, with the exception of Yuba. See (Manufactures of the United States in 1860 (1865), pp. 23–33). Yuba is included because histories indicate it was an important mining county in the early 1850s, though the 1860 census does not reflect this. See, for example, (Paul, 1947, p. 92). 17. It is worth mentioning that in the days prior to ‘‘one-person, one-vote’’, the frequent legislative reapportionments did not guarantee to miners representation in the legislature directly proportional to their numbers. In 1860, for example, of the 21 legislative districts in the Assembly, the 9 districts that represented mining areas enjoyed one representative for every 4,300 residents whereas the remaining 12 districts averaged one representative for every 5,300 residents. The suggestion is that reapportionments lagged behind population changes, as gold production was well past its peak by 1860 and many miners had moved on. This may have enabled miners to stave off loss of their tax exempt status marginally longer. 18. Yuba County is now excluded because by 1860, its mining industry had all but disappeared. 19. A series of logit regressions of (Vote2 þ Vote4) that included a legislative dummy variable ( ¼ 0 if Assembly, ¼ 1 if Senate) in various permutations revealed no significant voting differences between the two legislative chambers. The reported results are for the regression that omitted the dummy. 20. PARTY is assumed to equal zero, which represents Union Democrats, the vast majority of the party observations. 21. These prohibitions on post-disposal sale stemmed from the fact that under federal land disposal policy prior to 1820, settlers could buy federal lands on credit and were given five years to complete payments. 22. In addition, some evidence suggests the existence of political rent-seeking by oil companies over property taxation of oil company property during the 19th century. Ralph Hidy, for example, has argued that during this period, low property taxes on oil companies ‘‘undoubtedly contributed materially to the rapid development of the oil business’’, as oilmen engaged in a ‘‘running fight . . . with legislatures to stave off taxes’’ (see Hidy, 1950, p. 89). 23. Roberts (1944) lists Arizona, New Mexico, Alabama, Michigan, and Minnesota as states relying on ad valorem property taxation, while Nevada, Utah, Idaho, Montana, Louisiana, Arkansas, South Dakota and Colorado used non-ad valorem systems (see Roberts, 1944, pp. xi–xiii).

Political Economy of Public Finance in Antebellum California

231

ACKNOWLEDGMENTS This research was funded in part by grants from the National Science Foundation, the American Philosophical Society, and Carleton College. Earlier versions of this paper were presented at Carleton College and at the Bill Lane Center for the Study of the North American West at Stanford University. I thank participants at those seminars, as well as Karen Clay, Jenny Wahl, and Gavin Wright for particularly helpful comments. I am also grateful to librarians, archivists, and volunteers at the Ahmanson Room of the Huntington Library, the Bancroft Library at UC-Berkeley, the California State Library, and the historical societies of Nevada, Placer, and Tuolumne Counties.

REFERENCES Allen, D. A. (1965). Legislative sourcebook: The California Legislature and reapportionment, 1849–1965. Sacramento: California State Assembly. Atack, J., & Passel, P. (1994). A new economic view of American history from colonial times to 1940 (2nd ed.). New York: Norton. Bancroft, H. H. (1888). History of California (Vol. VI). San Francisco: History Co. Benson, S. (1965). A history of the general property tax. In: G. C. S. Benson, S. Benson, H. McClelland & P. Thomson (Eds), The American property tax: Its history, administration, and economic impact (pp. 11–81). Claremont: Men’s College. Browne, J. R. (Ed.) (1850). Report of the debates in the convention of California on the Formation of the state constitution. Washington: John Towers. Buck, S. J. (1969). The granger movement (reprint). Lincoln: University of Nebraska Press. California Laws. (1850). Statutes of California. San Jose: Winchester. California Laws. (1857). Statutes of California. Sacramento: Allen, State Printer. California Laws. (1864). Statutes of California. Sacramento: Clayes, State Printer. California Legislature. (1852). Journal of the assembly, 3rd session. San Francisco: Fitch & Geiger, State Printers. California Legislature. (1856). Journal of the assembly, 7th session. Sacramento: Allen, State Printer. Caughey, J. W. (1940). California: History of a remarkable state. New York: Prentice-Hall. Cleland, R. G. (1975). The cattle on a thousand hills. San Marino: Huntington. Constitution of the State of Montana, as adopted by the constitutional convention of the Territory of Montana. (1884). Helena: Fisk. Einhorn, R. L. (2001). Species of property: The American property-tax uniformity clauses reconsidered. Journal of Economic History, 61, 974–1008. Ellison, W. H. (1950). A self-governing dominion: California, 1849–1860. Berkeley: University of California Press. Ely, R. T. (1888). Taxation in American states and cities. New York: Crowell.

232

MARK T. KANAZAWA

Fankhauser, W. C. (1913). A financial history of California: Public revenues, debts, and expenditures. Berkeley: University of California Press. Fogel, R. W. (1989). Without consent or contract. New York: Norton. Heckelman, J. C., & Wallis, J. J. (1997). Railroads and property taxes. Explorations in Economic History, 34, 77–99. Hibbard, B. H. (1924). The history of the public lands policies. New York: Macmillan. Hicks, J. D. (1961). The populist revolt (reprint). Lincoln: University of Nebraska Press. Hidy, R. W. (1950). Government and the petroleum industry of the United States to 1911. Journal of Economic History, 10, 82–91. Howe, E. T., & Reeb, D. J. (1997). The historical evolution of state and local tax systems. Social Science Quarterly, 78, 109–121. Hunt, R. D. (1895). The genesis of California’s first constitution. In: A. Herbert (Ed.), Johns Hopkins University studies in historical and political science. Baltimore: Johns Hopkins Press. Ignoffo, M. Jo. (1999). Gold rush politics: California’s first legislature. Sacramento: CA State Senate. Johnson, D. A. (1982). Industry and the individual on the far western frontier: A case study of politics and social change in early Nevada. Pacific Historical Review, 51, 243–264. Kanazawa, M. T. (1996). Possession is nine points of the law: The political economy of early public land disposal. Explorations in Economic History, 33, 227–249. Kettleborough, C. (1918). The state constitutions. Indianapolis: Bowen. Marks, P. M. (1994). Precious dust: The American gold rush era, 1848–1900. New York: Morrow. Mayer, C. J., & Riley, G. A. (1985). Public domain, private dominion. A history of public mineral policy in America. San Francisco: Sierra Club Books. Paul, R. W. (1947). California gold: The beginning of mining in the far west. Lincoln: University of Nebraska. Roberts, W. A. (1944). State taxation of metallic deposits. Cambridge: Harvard University Press. Scheiber, H. N. (1969). Ohio canal era: A case study of government and the economy, 1820–1861. Cambridge: Harvard University Press. Smurr, J. W. (1955). The Montana tax conspiracy of 1889. Montana, the Magazine of Western History, 5, 46–53. Spence, C. C. (1975). Territorial politics and government in Montana, 1864–89. Urbana: University of Illinois Press. Statutes of California, passed at the First Session of the Legislature, Constitution of the State of California. (1850). San Jose: Winchester. Sylla, R., & Wallis, J. J. (1998). The anatomy of sovereign debt crises: Lessons from the American state defaults of the 1840’s. Japan and the World Economy, 10, 267–293. U.S. Department of Interior. Census Office. (1854). Statistical View of the United States: Compendium. Washington: Government Printing Office. U.S. Department of Interior. Census Office. (1865). Manufactures of the United States in 1860. Washington: Government Printing Office. Wallenstein, P. (1984). Rich man’s war, rich man’s fight: Civil war and the transformation of public finance in Georgia. Journal of Southern History, 50, 15–42. Wallenstein, P. (1985). More unequally taxed than any people in the civilized world: The Origins of Georgia’s Ad Valorem tax system. Georgia Historical Quarterly, 69, 459–487.

Political Economy of Public Finance in Antebellum California

233

Wallis, J. J. (2000a). State constitutional reform and the structure of government finance in the nineteenth century. In: J. C. Heckelman, J. C. Moorhouse & R. M. Whaples (Eds), Public choice interpretations of American economic history. Boston: Kluwer. Wallis, J. J. (2000b). American government finance in the long run: 1790 to 1990. Journal of Economic Perspectives, 14, 61–82. Wallis, J. J. (2005). Constitutions, corporations, and corruption: American states and constitutional change, 1842 to 1852. Journal of Economic History, 65, 211–256. Winter, C. E. (1979). Four hundred million acres (reprint). New York: Arno Press.

AN ANATOMY OF A CARTEL: THE NATIONAL INDUSTRIAL RECOVERY ACT OF 1933 AND THE COMPLIANCE CRISIS OF 1934 Jason E. Taylor and Peter G. Klein ABSTRACT This paper explores the nature and causes of the cartel compliance crisis that befell the National Industrial Recovery Act (NIRA) one year after its passage in 1933. We employ a simple game-theoretic model of the NIRA’s cartel enforcement mechanism to show that the compliance crisis can largely be explained by changes in expectations, rather than a change in enforcement policy. Specifically, firms initially overestimated the probability that defection would be met with sanction by the cartel’s enabling body, the National Recovery Administration – including a consumer boycott resulting from loss of the patriotic Blue Eagle emblem – and complied with the industry cartel rules. As these expectations were correctly adjusted downward, cartel compliance was lost. We support this hypothesis empirically with industry-level panel data showing how output and wage rates varied according to consumer confidence in the Blue Eagle. The analysis provides insight about cartel performance more generally.

Research in Economic History, Volume 26, 235–271 Copyright r 2008 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(08)26005-2

235

236

JASON E. TAYLOR AND PETER G. KLEIN

Modern cartel theory is typically seen as one of the great successes of gametheoretic economics. However, despite the appeal of the simple models that fill intermediate theory textbooks, many aspects of real-world cartels remain mysterious. Genovese and Mullin’s (2001) examination of the early twentieth century US sugar cartel, based on the recorded minutes of member meetings, found that cartel members rarely behaved as standard repeated-game models predict – defections were forgiven, punishments were inconsistent, rules were frequently adjusted, and so on. Moreover, contemporary cartel analysis mostly relies on exogenous shocks to explain cartel formation and breakdown. Comparative-static models identify conditions under which cartel behavior is or is not consistent with optimizing behavior by member firms, relying on changes in these conditions to explain changes in cartel characteristics. For example, the breakdown of previously stable cartels is attributed to such factors as reduced demand growth, increased demand uncertainty, the appearance of new entrants, and exogenous technological change (Eckbo, 1976; Rotemberg & Saloner, 1986; Griffin, 1989; Marquez, 1994; Dick, 1996; Levenstein & Suslow, 2004, 2006; Suslow, 2005). Fewer studies examine the internal dynamics of cartel behavior exclusive of changes in demand, technology, and other external factors. Following Genovese and Mullin, among others who have employed historical episodes to gain insight into the contemporary theory of cartel success and failure, we study a hitherto neglected aspect of the cartels created by the US National Industrial Recovery Act (NIRA). Under this legislation, which lasted from June 1933 to May 1935, hundreds of industries in the manufacturing sector drew up specific cartel codes – all of which were public record and whose provisions were legally enforceable through government fines and imprisonment. Many politicians and journalists, and even a few economists, argued that enforced cartelization would boost prices and wages, stimulating recovery from the Great Depression. Empirical evidence suggests that the cartels were initially successful in facilitating collusive outcomes (Hawley, 1966; Weinstein, 1980).1 However, several recent studies of the NIRA have noted that the cartel agreements systematically broke down less than a year after they were put into place. Irons (1982), Brand (1988), Alexander (1994, 1997), Krepps (1997), and Taylor (2002, 2007), in addition to contemporary accounts of the legislation, describe a widespread ‘‘compliance crisis’’ that rendered the legislation ineffective long before it was struck down by the US Supreme Court in 1935. Understanding the nature and causes of this breakdown provides valuable insight into the internal dynamics of cartel enforcement.

237

Anatomy of a Cartel

We develop a model highlighting the changes in cartel effectiveness before and after the compliance crisis of Spring 1934. Our analysis emphasizes the particular importance of changing expectations in cartel enforcement and punishment – including consumer and firm perceptions of the Blue Eagle compliance emblem – as a key contributor to cartel breakdown. Because the government was supposed to punish defectors through a combination of fines and removal of the Blue Eagle (which was to have brought a consumer boycott), there was no need for the firms themselves to punish defection. However, as it became apparent to firms that government punishment was unlikely, the few initial unpunished defections begat further defections, which soon became a title wave of defections so large that even a few scattered incidences of belated government enforcement could not stop. The broader implications of the study are relevant not just to formal, state-enforced cartels, but also to regulatory and other policies that facilitate collusion (e.g., Rothbard, 1984).

THE LOGIC OF THE NIRA The NIRA was passed by Congress June 16, 1933. Supporters claimed that inter-firm coordination would promote industrial recovery by ending the ‘‘ruinous’’ or ‘‘cut-throat’’ competition widely seen as the cause of the Depression. Industry cooperation was viewed as having been largely successful during World War I and was subsequently portrayed as the best hope for recovery. Donald Richberg, General Council of the National Recovery Administration (NRA), the NIRA’s enabling body, explained that ‘‘individual self-protection must give way to an orderly common effort’’ and ‘‘that thousands of businessmen themselves should know better than any small group of lawmakers’’ what specific collective efforts would best stimulate economic recovery (Irons, 1982, p. 97). Following this logic, manufacturing industries were required under the Act to write codes of fair competition, subject to some governmental oversight, specifying detailed rules, and regulations for firm behavior. The NIRA established 765 industry and supplemental codes, each containing several trade-practice provisions. The most common provision – contained in over 400 codes, including those for the pig iron, steel, coal, newsprint, lead, and woodworking machinery industries – was open price filing, which required firms to file their prices with code authorities and give advance notice of any price changes. As is well known in the modern

238

JASON E. TAYLOR AND PETER G. KLEIN

oligopoly literature, open price filing discourages competition by revealing firms’ pricing policies to rivals, allowing them to match the price or otherwise ‘‘retaliate’’ against a price-cutting firm.2 Most industrial codes contained much more specific provisions controlling prices, quantities, capacity, advertising, hiring, and other policies. A 1937 study found approximately 130 categories of trade-practice provisions contained within the NIRA codes.3 To illustrate, the Fishing Tackle code ordered that ‘‘no manufacturers shall sell or offer for sale any product at a price less than its ‘reasonable cost of production and distribution’ ’’ (article II, Section 1). The Boot and Shoe Manufacturing code mandated that price increases accompany any cost-raising actions such as the use of special shoe boxes or labels (article VIII, Section 4). The Hosiery code required firms to submit weekly shipping reports and monthly reports on ‘‘production, production and selling costs, shipments, stocks on hand, change[s] in equipment, [and] wages and hours of labor’’ (article VIII). The Iron and Steel code restricted the production of new capacity: ‘‘none of the members of this code shall initiate the construction of any new blast furnace or open hearth or Bessemer steel capacity’’ (article V, Section 2). The Handkerchief code monitored quality: ‘‘No member of the industry shall use the words ‘Hand Rolled Hem’ to designate that class of handmade hem known as ‘Whipped Edge,’ which latter term means any hem or edge on which the thread used to fasten same is whipped or looped around and encloses the entire rolled edge’’ (article VII, Section 14). The Ice code forbade the ‘‘enticement of competitor’s employees’’ in an attempt to limit labor-market competition (article IX, Section 2, part i). Although the President was directed to reject codes that ‘‘promoted monopolies,’’ the basic provisions of many codes did just this. As Brand (1988, p. 117) puts it, ‘‘attempts to mitigate competition also opened the door for businessmen seeking to dispense with competition altogether.’’4 While the scope of the regulatory aspects of the specific industry codes was largely up to the cartel participants, all codes had to include wage-rate increases. Although this would presumably only exacerbate the unemployment problem, there was a widespread belief among policy makers, businessmen, and economists that higher wages would increase purchasing power, demand, production, and ultimately, employment (O’Brien, 1989; Vedder & Gallaway, 1993; Taylor & Selgin, 1999). The high-wage doctrine claimed that an increase in wage rates would give workers more money to spend, which would in turn increase the demand for goods and services. This expanded demand would allow firms to increase production and,

Anatomy of a Cartel

239

in the end, hire more workers.5 A 1937 study explains the background of the NIRA’s wage provisions this way: [There was a] growing emphasis on the idea that business, in its own interest, needed to pay liberal wages in order to provide the widely distributed purchasing power which was regarded as necessary if the output of mass production was to find an adequate market . . . [The high-wage doctrine] found influential adherents among prominent businessmen, and apparently had appreciable effect on [firms’] actual wage policies (House Document No. 158, 75th Congress, 1st Session, p. 2).

To more quickly put the NIRA’s labor provisions into effect, in July 1933 the President’s Reemployment Act created a temporary ‘‘blanket code’’ which set wage and hour guideposts that firms were asked to follow until they could pass their own industry codes. Individual firms would sign a pledge to President Roosevelt stating that they would pay workers at least $0.35 to $0.40 per hour – a minimum wage that exceeded the average wage paid to common labor at the time. Industry-specific codes were more flexible, but one Congressional study of 578 codes found that 338 codes, covering 55 percent of all code employees, specified $0.40 an hour or more. In contrast, only 14 codes, covering under 5 percent of coded employees, specified wages of $0.30 an hour or less. In all cases, the law required the industries’ average wage rates to be increased. Overall, the average wage rose from $0.35 an hour when the NIRA was passed in June 1933 to $0.43 an hour one year later, an increase of 23 percent.6 The NIRA’s codes of fair competition allowed firms to maintain industry cartels far beyond what would have been feasible on the free market, where cartels generally break down in the absence of an effective enforcement mechanisms. As is well known, firms producing similar products can benefit by coordinating to reduce output, increase prices, or achieve other strategic objectives. However, although decisions to coordinate in this fashion may be collectively beneficial, they may not be individually rational: firms can generally increase their own profit by deviating from any agreement, tacit, or explicit. Following Stigler (1964), economists have sought to identify defection and punishment mechanisms that facilitate collusion. While Stigler focused on the ability to detect deviations from the collusive outcome, more recent contributions, following Green and Porter (1984), focus on optimal retaliation, taking the probability of deviation as given. Still, Stigler’s basic insight – that cartel members structure their agreements, explicitly or tacitly, to discourage individually rational but collectively harmful behavior – remains one of the best-established results of modern industrial organization.

240

JASON E. TAYLOR AND PETER G. KLEIN

The NIRA altered the costs and benefits of coordination by specifying how firms would collude and providing legal sanctions for deviations from the cartel codes. In a voluntary cartel, deviations are typically ‘‘punished’’ through retaliation; if one firm cuts its price, the others cut their prices in response, leading to price competition that harms the defector. However, because such a punishment harms the retaliating firms as well, the threat of strong retaliation may not be credible. Indeed, recent studies of early twentieth-century cartels in the bromine (Levenstein, 1997) and sugar (Genovese & Mullin, 2001) industries show that deviations were rarely met with massive retaliation, but rather with small, matching deviations or renegotiation. In a government-enforced cartel, deviation is against the law, so there may be no need for rivals to retaliate. Moreover, under the NIRA the active regulators of the codes were generally the industrial executives themselves, who had proper incentives to monitor participation and could report a rival firm’s defection to the local compliance board for punishment. The NIRA enforced cartelization in two ways. First, the NIRA codes, once signed by the President, were law. Section 10(b) of the NIRA states that ‘‘any such violation of any such rule or regulation shall be punishable by fine of not to exceed $500 or imprisonment for not to exceed six months, or both.’’ Each day a violation continued was considered a separate offense and hence subject to a new round of penalties. Compliance boards also required that violators pay wage reparations – usually in the form of back wages to underpaid laborers – before they were considered in compliance.7 Assuming firms paid employees the NIRA-suggested $0.40 minimum wage, a $500 fine, even without reparations, would be the equivalent of 1,250 hours of labor – a severe penalty, particularly for a small firm. Likewise, the cost to a business owner of spending up to six months in jail was certainly not trivial. The overall magnitude of this punishment also depends, of course, upon the expected probability of being caught, the business owner’s discount factor, the business owner’s subjective aversion to prison, and the expected probability that imprisonment will be imposed for the violation.8 The second – and, in the view of the Roosevelt Administration, far more important – enforcement mechanism was the Blue Eagle emblem. Firms deemed to be in full compliance with their industry’s code of fair competition were allowed to display the emblem on their storefronts, products, or in their advertisements.9 Intermediate, wholesale, and retail goods also had to be purchased from NIRA-compliant firms to avoid losing the Blue Eagle. The text of the NIRA encouraged consumers and producers (and required

Anatomy of a Cartel

241

the federal government) to ‘‘support and patronize establishments which have also signed this agreement and are listed as members’’ of the NRA. The Blue Eagle emblem then supposably allowed consumers to differentiate between firms that had signed codes and remained in compliance with their cartels and firms that either refused to sign or lost the emblem as punishment for a violation. By encouraging consumers to sign a ‘‘Statement of Cooperation’’ reading ‘‘I will cooperate in the reemployment by supporting and patronizing employers and workers who are members of the NRA,’’ Roosevelt essentially called for a nationwide boycott of firms that did not display the Blue Eagle (U.S. National Recovery Administration, 1933, p. 278). Upon signing this pledge, individuals often were given a ‘‘Consumer’s Badge of Cooperation,’’ usually in the form of a small lapel pin or button displaying Blue Eagle that could be worn as a patriotic symbol of support for the President’s economic recovery plan. Fig. 1 displays these NRA enforcement devices as well as the Blue Eagle emblem. Contemporary anecdotal evidence suggests that many consumers heeded Roosevelt’s call, making the Blue Eagle an important part of the NIRA’s enforcement mechanism. Despite the Administration’s large emphasis on the Blue Eagle as its primary means of enforcement, the emblem has received almost no serious attention in the economics literature on the New Deal.

Fig. 1. NIRA Paraphenalia. A ‘‘Consumer’s Statement of Cooperation’’ card with lapel pin ‘‘Badge of Cooperation,’’ a Blue Eagle emblem sticker, and a Wrigley’s chewing gum wrapper displaying the Blue Eagle. Source: authors’ personal collection.

242

JASON E. TAYLOR AND PETER G. KLEIN

To enforce the cartels, the NRA Compliance Division was established in October 1933. The division had 54 state and branch offices and about 1,400 personnel. Each office had both a director of labor compliance and a director of trade-practice compliance. Enforcement proceeded in stages. After receiving a written complaint, the director of the relevant compliance office would assign a field officer to investigate the alleged violation, and, if appropriate, try to convince the firm to come into compliance. If violation continued, the firm owner would be invited to come into the office to discuss the case with the local compliance director. The state and branch office personnel had little real authority, however; they could simply explain the rules and threaten further action. Only the National Compliance Director in Washington could impose formal sanctions after reviewing the evidence in a closed hearing. (The accused was invited to attend the hearing to rebut evidence of the alleged violation, but the identity of the original accuser was supposed to be never revealed.) If the hearing determined that a code violation had occurred, the firm would be invited to sign a certificate promising future compliance. If the firm continued not to comply, the firm owner would receive a telegram from the NRA Administrator’s office ordering him to ‘‘cease displaying the Blue Eagle and deliver all NRA insignia in his possession to the local postmaster’’ (U.S. Committee of Industrial Analysis, 1937, p. 70). The NRA Litigation Division was then charged with reviewing the case and referring it to the Department of Justice or the Federal Trade Commission for institution of fines and imprisonment. The NRA itself had no authority to impose penalties beyond stripping the Blue Eagle from violators. On October 11, 1933 the first use of the NRA Compliance Division’s enforcement process made national headlines when Theodore G. Rahutis of Gary, Indiana was sent a telegram from Washington ordering him to give up the Blue Eagle for violating the NRA’s wage and hour provisions. The Associated Press quoted a contrite Rahutis saying his restaurant would ‘‘take immediate steps to regain our status with the administration.’’10

CARTEL BEHAVIOR UNDER THE NIRA: A STRATEGIC ANALYSIS The NIRA cartel-code structure and enforcement mechanism can be illustrated with a simple oligopoly model. The NIRA codes attempted to facilitate profit-maximizing collusion on price, quantity, quality, and other

243

Anatomy of a Cartel

forms of competition, but also had to include higher wage rates. Participation was mandatory; firms that violated code provisions risked possible fine, imprisonment, and the loss of the Blue Eagle emblem, leading to a potential consumer boycott. Consider first the simplest case, a one-period, n-firm symmetric Cournot game in which pni ¼ ð1=nÞpm represents firm i’s payoff from colluding (pm is the industry’s monopoly profit), pci firm i’s profit under Cournot competition, pdi firm i’s profit from defecting while the remaining n  1 firms cooperate, and pzi firm i’s profit from cooperating while another firm defects. Under the standard assumptions that pdi 4pni and pci 4pzi , the only Nash equilibrium is the one in which each firm defects. The NIRA changes the payoff structure as follows. Let g be the cost of complying with the NIRA codes – primarily the cost of paying higher wage rates, but also any other compliance costs, including reporting requirements, the costs of using a less-than-optimal technology, and the additional cost of buying from NIRA-compliant supplier cartels.11 Let F be the value of the government fine and prison sentence from defecting from the cartel, and BE be the perceived cost of lost business from not being able to display the Blue Eagle. Defectors risk being caught and punished by having their Blue Eagle taken away and being fined or imprisoned. Assume that defectors (instantaneously) lose BE with expected probability y1 and F with expected probability y2 but avoid having to pay g.12 Cooperation is a Nash equilibrium as long as pni  g  pdi  y1 BE  y2 F

(1)

This expression can usefully be rewritten as y1 BE þ y2 F  ðpdi  pni Þ þ g

(2)

where the left-hand side of the inequality represents the expected loss from defecting and the right-hand side represents the expected gain. This formulation emphasizes the importance not only of enforcement (y1 and y2), but also of the substitutability among the sources of loss from punishment (rivals’ reactions ½pdi  pni , the government’s fine [F ], and consumer pressure [BE]). For instance, if social pressure to conform to the codes of fair competition is high – giving the Blue Eagle compliance mechanism ‘‘bite’’ – the fine from defecting need not be substantial. Ceteris paribus, a compliance crisis could thus be caused by a change in firms’ beliefs about of the cost of losing the Blue Eagle, even without a reduction in the fine for violating the NIRA provisions and the likelihood of being caught.

244

JASON E. TAYLOR AND PETER G. KLEIN

Expressing pni and pdi in terms of cost parameters, demand parameters, and n shows how changes in market conditions, firm heterogeneity, and other factors affect the ability of firms to cartelize under the NIRA, holding other factors, such as fines and a potential boycott, constant. Appendix A presents a simple case with linear demand and constant marginal cost. As seen in the Appendix, the right-hand side of inequality (2) – the gain to firm i from violating its cartel agreement, ceteris paribus – is increasing in n, decreasing in industry demand, and (generally) decreasing in industry cost. In other words, as in the original Stigler (1964) model, the marginal gain from defecting rather than cooperating increases with the number of firms in an industry, suggesting that NIRA-sponsored collusion should have been easier to sustain in highly concentrated industries. This squares with Alexander’s (1994) estimates of ‘‘critical concentration ratios’’ at which industries were able to sustain collusion before and after the NIRA. It also suggests that collusion becomes harder to sustain as industry demand falls. Moreover, as shown for the general case by Schmalensee (1987) and Rothschild (1999), and others, cost asymmetries make collusion harder to sustain; unless output quotas (continuing with the Cournot example) are set relative to costs, high-cost firms have a greater incentive to deviate.13 The cross-sectional implication is that the ability to sustain collusion in a cartel with mandatory wage increases is sensitive to variation in member firms’ capital-labor ratios. The time-series implication is that compliance will tend to break down if capital-labor ratios fall, consistent with Alexander’s (1997) general conclusion that firm heterogeneity played an important role in the failure of the NIRA. The NIRA was originally set to expire after two years, suggesting a finitely repeated game. However, many business owners and policy makers expected the NIRA’s cartel-enabling provisions to be renewed and maintained indefinitely. Given the great deal of uncertainty about the legislation’s duration, it may be useful to view competition under the NIRA as an infinitely repeated game. For simplicity consider an n-firm infinitely repeated game in which firm i plays a simple trigger strategy (Friedman, 1971): In period t, cooperate if all firms j6¼i cooperated in period t  1; otherwise, revert to Cournot output in period t and all subsequent periods. Each firm faces a per-period discount rate of d. Without the NIRA, firm i gets pni  1=ð1  dÞ by cooperating and pdi þ pci  d=ð1  dÞ by defecting. As a condition for sustaining cooperation as a subgame-perfect Nash equilibrium, pni  1=ð1  dÞ must be greater than or equal to pdi þ pci  d=ð1  dÞ, which reduces to the standard condition that there is not ‘‘too much’’

245

Anatomy of a Cartel

discounting, i.e., d

pdi  pni pdi  pci

(3)

With the NIRA, the cooperating firm i gets ðpni  gi Þ  1=ð1  dÞ. By defecting, assuming that BE and F are imposed with probabilities y1 and y2 respectively in the period immediately following defection, firm i gets pdi þ ðpci  y1 BE  y2 FÞ  d=ð1  dÞ.14 The condition for sustaining cooperation is now ðpni  gi Þ 

1 d  pdi þ ðpci  y1 BE  y2 FÞ  ð1  dÞ ð1  gÞ

(4)

which reduces to dNIRA 

ðpdi  pni Þ þ g ðpdi  pci Þ þ y1 BE þ y2 F

(5)

As in the one-shot case, the effect of the NIRA on the ability to sustain cooperation depends on the relative magnitudes of g, y1, y2, BE, and F. The comparative-statics results are the same in both the one-shot and the repeated games: the likelihood of successful cartelization is increasing in y1, y2, BE, and F and decreasing in g.15 Of course, the grim-trigger strategy, in which punishment continues indefinitely, may not be realistic. As mentioned above, industry studies by Levenstein (1997) and Genovese and Mullin (2001) suggest that firms rarely respond to defections with massive retaliation. Small defections, even if noticed, are often ignored, while more substantial defections are typically matched in kind. Under the NIRA, firms face an additional disincentive to retaliate: the punisher itself risks fine, imprisonment, and loss of its own Blue Eagle. An alternative modeling strategy is to assume that rivals do not respond to defections at all, relying on government sanction to bring the defector back in line. In this case, the defecting firm’s payoff is pdi  y1 BE  y2 F: Inequality 5 thus becomes dNIRA 

ðpdi  pni Þ þ g y1 BE þ y2 F

(6)

The comparative-statics results are the same as above: the likelihood of successful cartelization is increasing in y1, y2, BE, and F and decreasing in g.16

246

JASON E. TAYLOR AND PETER G. KLEIN

Of course, interest rates also affect the ability to sustain collusion in a repeated game. As interest rates rise, the short-term gain from defecting increases and the long-term loss from being punished falls, ceteris paribus, making collusion more difficult to sustain. Nominal interest rates remained fairly stable throughout the NIRA period, however, falling slightly from 4.72 percent in July 1933 to 4.36 percent during the April 1934 compliance crisis.17 The crisis thus does not appear to have been driven by rising interest rates. The Good-Patriot Effect As noted above, firms do not always respond to defections with substantial retaliation. Moreover, the NIRA compliance mechanism provided an important disincentive to retaliate – retaliation itself invites legal sanction. Moreover, if consumers take the Blue Eagle seriously, then firms may prefer to respond to a defection by highlighting to consumers their own continued compliance. To see this, imagine that if a defecting firm is caught by the authorities and loses its Blue Eagle, rivals that stick to the cartel code reap windfall profits equal to GP – what we call the ‘‘good-patriot’’ effect that a firm receives for cooperating in the face of a defection by a rival firm who now loses the patriotic emblem and invites a consumer boycott. In a symmetric model in which m firms defect, the remaining n  m firms enjoy GP ¼ (m/(n  m))BE.18 Under what conditions is GP large enough to prevent firms from retaliating against firms that deviate from the NIRA’s provisions? To simplify the analysis assume that only one firm defects and that the remaining firms act in concert. In other words, each remaining firm compares its payoff under the Cournot equilibrium in which all firms have abandoned the NIRA with the ‘‘sucker’s payoff ’’ in which the defection goes unpunished. Matching the defection, as we saw above, yields a payoff of pci  y1 BE  y2 F (again, assuming that all other firms ‘‘punish’’ by defecting as well).19 Allowing the defection to go unpunished yields pzi  g þ y1 GP, where GP ¼ (1/(n  1))BE. Thus, continuing to cooperate while one other firm defects is a best response to that defection if pci  y1 BE  y2 F  pzi  g þ y1 GP or, substituting for GP and rearranging terms,  n  g  y2 F  y1 BE  pzi  pci n1

(7)

(8)

247

Anatomy of a Cartel

Cournot profit can be expressed as pci ¼ ð1=ðn þ 1Þ2 Þðða  cÞ2 =bÞ: To solve for pzi assume that cooperating firms continue to produce their NIRA-sanctioned levels of output ðxni Þ, while the price adjusts to pd. Each cooperating firm thus earns a sucker’s payoff equal to ðpd  c0 Þxni . That is,    1 a  c  2g þ ða þ 3c þ 2gÞn 1 a  c  g cþg : pzi ¼ (9) 4 n n 2b which reduces to pzi ¼

1 ða  c  gÞða  c  2g þ ða  c þ 6gÞnÞ 8 bn2

(10)

The right-hand side of inequality (8), pzi  pci , therefore equals 1 ða  c  gÞða  c  2g þ ða  c þ 6gÞnÞ 1 ð a  c  gÞ 2  b 8 bn2 ðn þ 1Þ2 To simplify this expression, let a ¼ 1, b ¼ 1, c ¼ 0, and g ¼ 1. The expression above reduces to 1 1  4n 8n þ 4n2 þ 4 and inequality (8), the condition under which firms will continue to cooperate in the face of a defection, becomes  n  1 1 (11) g  y2 F  y1 BE   n1 4n 8n þ 4n2 þ 4 Inspection of inequality (11) reveals that the left-hand side is increasing in g, decreasing in y1 and y2, decreasing in BE, and decreasing in F. The left-hand side is increasing in n, while the right-hand side is decreasing in n. In other words, the inequality is more likely to hold (i.e., the cartel is maintained), ceteris paribus, when g is small, y1 and y2, are large, F large, BE large, and n small. These make intuitive sense: The smaller the compliance cost, the more likely a firm will stick with the NIRA agreement. Similarly, a large fine, a high probability of being caught, and a substantial cost from lost business if the Blue Eagle is lost each make continued compliance more likely. Finally, again consistent with Stigler’s proposition, collusion is easier to sustain in more concentrated industries, ceteris paribus. In short, then, incorporating the good-patriot effect into the analysis reinforces our main results. Indeed, the possibility of good-patriot rewards increases the likelihood that

248

JASON E. TAYLOR AND PETER G. KLEIN

firms will allow defections to go unpunished, maintaining partial cooperation with the NIRA codes instead of a complete breakdown.

The High-Wage Doctrine So far we have ignored the high-wage doctrine, the belief that paying higher wages would increase profits by increasing the demand for firms’ products. We have modeled the higher wage rates mandated by the NIRA as a cost, specifically as the main component of the compliance cost g. What happens to the ability to sustain cooperation under the NIRA if firms regard higher wages, consistent with the high-wage doctrine prevalent at the time, as a benefit to themselves rather than a cost? If all firms believe in what we might call an ‘‘individual high-wage doctrine’’ – the belief that a firm benefits from paying higher wages to its own workers, regardless of what other firms in the industry are doing – then firms will continue to pay high wages even if they defect from their cartel and cease abiding by the NIRA’s other provisions (lowering the perceived cost of g, making cartels easier to sustain, without changing the other comparative-statics results). If firms accept the high-wage doctrine but believe the marginal effect of their own wage payments on industry demand is small, then, defectors will free ride on other firms’ wage policies by paying lower wages to their own workers, making g the opportunity cost to firm i of paying higher wages instead of free riding.20 While the actual (nominal) wage rates mandated by the NIRA were rarely amended throughout the period, an analysis of the high-wage doctrine can help explain a compliance crisis in at least two ways. First, even if no one believed in the high-wage doctrine, a fall in the market wage would constitute an increase in the wage component of g, the cost of paying the NIRA-mandated wage scale, decreasing the likelihood of cooperation.21 Second, and perhaps more importantly, firms may have initially believed in the high-wage doctrine, preferring to pay higher wages whether mandated or not, but then lost faith in the merits of high wages. Changes in beliefs about the benefits of paying higher wages thus could help explain the breakdown of the NIRA cartels. Taylor and Selgin (1999) cite contemporary evidence strongly suggesting that some firm owners and policy makers indeed lost faith in the high-wage doctrine as the depression continued throughout the 1930s.

Anatomy of a Cartel

249

THE COMPLIANCE CRISIS OF 1934 The analysis above provides a framework for explaining the NIRA’s enforcement mechanism and its potential for success or failure in the achievement of cartel outcomes. In particular the model can shed light on the causes of the compliance crisis that took place in the late winter and early spring of 1934, less than a year into the cartel experiment. If we assume that each firm’s optimal strategy in absence of any cartel enforcement mechanism is to defect, the model demonstrates that the NIRA’s ability to maintain cartelization hinged primarily on firms’ expectations of the costs of being caught and punished. These potential punishment costs include fine, imprisonment, and the stripping of the Blue Eagle emblem. Is there evidence that these expectations changed? Consider first the loss of the Blue Eagle.

The Effectiveness of the Blue Eagle Emblem The importance of the Blue Eagle depended primarily on consumers’ willingness to adhere to Roosevelt’s call to boycott firms lacking the compliance indicator. If firms felt that consumers would not take the Blue Eagle into account when deciding where to shop, BEi and GPi are zero, making it more difficult to maintain cartels, ceteris paribus. As the Blue Eagle compliance badge began appearing on products and in store windows, the Roosevelt Administration continued a daily push of households to ‘‘buy under the Blue Eagle.’’ In an August 7, 1933 speech, NIRA Chief Administrator Hugh Johnson said, ‘‘Where should you spend? . . . You should spend under the Blue Eagle. If you spend there you are spending for increased employment. If you spend elsewhere you are hurting the chance’’ for economic recovery.22 Contemporary journalist and Roosevelt critic John T. Flynn (1948, pp. 40–41) reported that ‘‘Every instrument of human exhortation opened fire on business to comply [with the NIRA codes] – the press, pulpit, radio, movies. Bands played, men paraded, trucks toured the streets blaring the message through microphones . . . The President went on the air: ‘In war in the gloom of attack,’ he crooned, ‘soldiers wear a bright badge to be sure that comrades do not fire on comrades. Those who cooperate in this program must know each other at a glance. That bright badge is the Blue Eagle.’ ’’ Wolvin’s (1968) speech-communications dissertation examined the Blue Eagle campaign as a program of ‘‘persuasion and coercion.’’ Interestingly,

250

JASON E. TAYLOR AND PETER G. KLEIN

he notes that the NRA Publicity Division focused much of its attention on women in promoting the Blue Eagle emblem and the boycott of firms lacking it. According to Wolvin, ‘‘The bugle call had been sounded to rally the female population of America armed with their pocketbooks’’ (p. 127). For example, an August 26, 1933, NRA press release stated: If the women who control the purse strings of the nation use this mighty instrument of mass buying power to support the Blue Eagle, they can assist to bring about a new order in industry . . . Women should insist on following the sign of the Blue Eagle in all their buying. Every time they spend under the Blue Eagle, they are making jobs for the butcher, the baker, and the candlestick maker and, thereby, helping to achieve security for themselves and to build a better and happier America (quoted in Wolvin, 1968, pp. 132–33).

Moreover, the Administration regularly compared the nation’s battle against economic depression to war, noting that during wartime, citizens are asked to make economic sacrifices. The specific sacrifice being asked of consumers was simply to patronize only those firms that bore the patriotic Blue Eagle emblem, even if they charged higher prices. In this vain, Ruth Bryan Owen, U.S. Minister to Denmark, broadcast the following appeal to American women on CBS radio in late August 1933: In other wars the women’s part has been to wait and weep. In this great effort toward recovery women are playing a vital and important part as the industries hoist their banners with the Blue Eagle . . . If every woman within the sound of my voice should make a resolve, ‘‘God helping me, I shall make every effort to assure the success of this program,’’ just that is enough to insure the victory of the greatest of all wars . . . Mr. President, we will justify the trust you have put in us, gladly and with a high heart we women will do our part (quoted in Wolvin, 1968, pp. 126–27).

Contemporary accounts suggest that consumer respect for the Blue Eagle emblem was substantial and widespread at the onset of the NIRA in the late summer and autumn of 1933. The ubiquitous Blue Eagle, accompanied by the slogan, ‘‘We Do Our Part,’’ graced store windows, newspapers, chewing gum wrappers, and was even ‘‘reverse-sunburned’’ on the backs of movie starlets Frances Drake and Toby Wing (Leuchtenburg and the Editors of LIFE, 1964, p. 14). Johnson went on a national tour, complete with motorcades and brass bands, promoting the Blue Eagle. In September 1933, a ticker tape parade in New York City drew two million Blue Eagle supporters (Johnson, 1935, p. 267). This prompted the September 12, 1933 edition of U.S. Law Week, a leading legal periodical, to call the Blue Eagle and its consumer’s badge of cooperation, ‘‘a most effective extra-legal weapon.’’23 On September 4, 1933 Johnson elaborated the Administration’s approach to cartel enforcement: ‘‘Our first and chief reliance is on public

Anatomy of a Cartel

251

opinion . . . we know that to take away the Blue Eagle is more severe than any puny fine’’ (quoted in Wolvin, 1968, pp. 194–95). Relating this to our model of firm behavior under the NIRA, the contemporary anecdotal evidence clearly suggests that BE and GP helped to secure compliance during the initial months of the cartel experiment. The case of the Ford Motor Company, which refused to sign the NRA automobile code, helps illustrate the potential costs of a consumer boycott. Henry Ford was vilified as a traitor in a series of scathing editorials published in many of the nation’s leading newspapers. The Louisville Courier-Journal, for instance, stated that by not signing onto the NRA Ford was ‘‘displaying an astonishing stupidity . . . which may be largely explained by his inordinate egotism’’ (September 6, 1933). The Milwaukee Journal wrote: ‘‘If Henry Ford is too great or too good to wear a Blue Eagle, his is out of tune with his country and out of touch with his fellow Americans’’ (September 7, 1933). Federal, state, and local government agencies – important potential buyers of Ford products – joined consumers in the boycott of Ford. For example, Tennessee state purchasing agent Emmett Hunt announced on August 31, 1933, that his state would not purchase Ford products until the company signed and abided by the NRA automobile code. Governor Pinochot of Pennsylvanian and Governor Brann of Maine announced the next day that their states would also boycott Ford products and many other state and local governments followed suit (Wolvin, 1968, pp. 193–94). This was precisely the kind of publicity and powerful consumer sentiment the NRA had hoped to create. As Johnson noted, ‘‘The time is coming when somebody is going to take one of these Blue Eagles off someone’s window, and that’s going to be a sentence of economic death.’’24 While the Blue Eagle campaign appears to have initially succeeded in convincing firms to comply with the cartels or face the consequence of a painful boycott, the wave of patriotic support apparently did not last. Brand (1988, p. 106) notes that by early 1934 consumers had experienced a ‘‘decline in moral enthusiasm’’ for the NRA and its provisions. Hawley (1966, p. 68) likewise notes there was a ‘‘reversal in public sentiment’’ toward the NRA around the same time. After this reversal the Blue Eagle appears to have lost much, if not all, of economic significance. As a contemporary government study noted: ‘‘The loss of the right to display the Blue Eagle, to the extent that public interest in patronizing only enterprises which displayed it waned, gradually became a penalty of little consequence’’ (U.S. Committee of Industrial Analysis, 1937, p. 70, emphasis added). Thus firms’ perception of reduced consumer enthusiasm for the emblem, represented in our analysis by a reduction in BE (and hence GP), could have played an important role in causing the compliance crisis of early 1934.

252

JASON E. TAYLOR AND PETER G. KLEIN

Empirical Evidence on the Role of the Blue Eagle Systematic empirical evidence on firms’ beliefs regarding the value of the Blue Eagle is difficult to obtain. One proxy for a firm’s attitude toward the importance of the emblem is its willingness to pay to display it. Firms complying with the NRA cartel codes had the right not only to display the Blue Eagle emblem on their products and in their store windows, but also to include it in their advertisements. Beginning in late July 1933, the Blue Eagle emblem began appearing in print advertisements of NRA-covered firms. To measure its importance, we examined a diverse sample of eight daily newspapers, Atlanta Journal-Constitution, Chicago Tribune, Christian Science Monitor, Lansing State Journal, New Orleans Times-Picayune, New York Times, San Francisco Examiner, and Washington Post. We observe the first 25 advertisements each Thursday between August 3, 1933 and June 6, 1935, just after the NIRA was ruled unconstitutional, and record how many of these displayed the Blue Eagle emblem. Thus each weekly observation records the percentage of 200 advertisements that display the patriotic emblem. Fig. 2 shows a five-week central moving average of the percentage of ads carrying the patriotic emblem during the pertinent months.25 The data suggest that firms’ beliefs in consumer enthusiasm for the Blue Eagle emblem were strongest in September and October of 1933 when as Percent of Advertisements Displaying Blue Eagle Emblem in Eight Newspapers (5-Week Central Moving Average) 50 40 30 20 10

6/3/35

5/3/35

4/3/35

2/3/35 3/3/35

1/3/35

12/3/34

11/3/34

9/3/34

10/3/34

8/3/34

7/3/34

6/3/34

5/3/34

4/3/34

2/3/34 3/3/34

1/3/34

12/3/33

11/3/33

9/3/33

10/3/33

8/3/33

0

Fig. 2. Appearance of the Blue Eagle. Data are based on a sample of the first 25 advertisements in eight newspapers (200 ads per observation) from every Thursday between August 3, 1933 and June 6, 1935. Newspapers: Atlanta Journal-Constitution, Chicago Tribune, Christian Science Monitor, Lansing State Journal, New Orleans Times-Picayune, New York Times, San Francisco Examiner, and Washington Post.

253

Anatomy of a Cartel

many as 44 percent of ads contained the emblem (a significant number considering that the NIRA applied only to manufacturing and retail firms, with a few minor exceptions). A noticeable drop-off began in November and December 1933. Table 1 shows the monthly averages, by newspaper, for the last five months of 1933. In all but one case the monthly average declined between October and November, and in all eight cases it declined between November and December. During this time the NIRA faced a wave of challenges. For example, on November 27, 1933, 85 percent of Chambersburg, Pennsylvania restaurants voluntarily surrendered their Blue Eagles saying that the restaurant code was bringing financial ruin. Then on December 2, 1933, the entire Lincoln, Nebraska compliance board resigned in protest of a lack of enforcement resources from Washington. A few days later, members of the Lowell, Massachusetts compliance board likewise resigned. Such episodes, which made national news, gave the NRA and its Blue Eagle symbol a black eye and likely played an important role in the apparent decline in enthusiasm for the emblem. By January 1934, only around 22 percent of ads, or around half of the number at its earlier peak, displayed the Blue Eagle. Another small wave of the emblem’s disappearance began in late May and June of 1934. Interestingly, this is correlated with the release of a report by the National Recovery Review Board, headed by Clarence Darrow, which painted the NRA in an unfavorable light as having supported monopolies at the expense of small businesses. By January 1935, and for the remaining 5 months in which the Blue Eagle continued to be the official symbol of compliance with the NIRA, only around 5 percent of all ads carried the emblem.

Percent of Advertisements Displaying the Blue Eagle Emblem, August–December 1933. Monthly Averages

Lansing State Journal Chicago Tribune New York Times Washington Post Christian Science Monitor Atlanta Journal-Constitution New Orleans Times-Picayune San Francisco Examiner

August

September

October

November

December

0.19 0.50 0.20 0.24 0.17 0.30 0.29 0.41

0.43 0.65 0.34 0.26 0.19 0.35 0.55 0.65

0.44 0.57 0.20 0.39 0.30 0.33 0.54 0.52

0.36 0.60 0.15 0.30 0.22 0.26 0.42 0.47

0.14 0.56 0.13 0.20 0.21 0.29 0.38 0.36

254

JASON E. TAYLOR AND PETER G. KLEIN

That so many firms went to the initial effort of including the emblem in their advertisements – and in many cases, the major theme of the ad was simply to proclaim the firm’s compliance with the NIRA rather than information about the quality or price of their product – in the fall and winter of 1933–1934 suggests that firms believed that the Blue Eagle carried weight with consumers. Such a belief, coupled with a fear that defection from the NIRA could result in loss of the emblem, would clearly have made compliance with the cartels more likely. Perhaps more telling, the fact that the percentage of advertisements carrying the emblem fell so dramatically in the Winter and Spring of 1934, even though it was still the official emblem of compliance, strongly suggests that the Blue Eagle lost much of its behavioral significance and hence lost its ability to help firms sustain collusion. An examination of the Blue Eagle in advertisements of some specific firms across this time period may also be of value. While collecting our weekly observations, we tracked the ads of several specific firms looking for any patterns in behavior. We searched for two patterns explicitly. First, we wanted to confirm that what Fig. 2 measures is the adoption and subsequent dropping of the Blue Eagle in advertisements of the same firms rather than a change in the composition of advertising firms in the sample. It was clear from our analysis that indeed it was the case that individual firms in our sample displayed and then dropped the emblem. For example, Moore’s Retail consistently displayed the Blue Eagle in its San Francisco Examiner advertisements until February 8, 1934, and then dropped it in all subsequent advertisements. Likewise Fanny Farmer Candies regularly displayed the emblem in its advertisements in the Christian Science Monitor until March 29, 1934 after which time it stopped. There were some exceptions as Holmes Retail regularly included the Blue Eagle in its advertisements in the New Orleans Times Picayune all the way until June of 1935, when we stopped collecting data since the legislation had been ruled unconstitutional. The second pattern we explicitly looked for was whether the dropping of the emblem from advertisements generally corresponded to the firm employing a new advertisement or whether the emblem was removed from an existing advertisement. It seems a far stronger statement of a firm’s lack of enthusiasm for the emblem for it to actually remove the Blue Eagle from an existing ad than it is to simply decide not to include it in a new one. While we looked for this pattern anecdotally in all the papers, we were a bit more systematic with respect to this question for the Christian Science Monitor. We tracked seven companies that both had relatively stable advertisements in the Monitor and displayed and dropped the Blue Eagle during the sample. Of these seven, five dropped the Blue Eagle from an existing advertisement,

Anatomy of a Cartel

255

while for the other two the dropping of the emblem corresponded with a change in advertisement. Generally speaking, from our analysis of advertisements from all eight papers we conclude that the emblem’s disappearance was by no means limited to a change in advertisements. To see if the fading importance of the Blue Eagle contributed to cartel failure, as described in our model, we construct an industry-level panel of monthly data on output and wage rates, the former as a measure of firms’ success in obtaining collusive outcomes and the latter as a measure of compliance with the NIRA’s wage provisions. For the output regressions, we use the 66 industries in the National Bureau of Economic Research Macrohistory Database that were covered by a specific NIRA code from January 1927 to December 1937. Our wage regressions use the 30 of these industries for which average hourly earnings data are also reported. For each industry in our sample we obtain from the corresponding NIRA code of fair competition the date in which the industry code was passed. The opening pages of each code generally listed the specific subcategories of firms or production that fell under the jurisdiction of the code, helping us to match each industry to its specific cartel code. Table 2 lists the 66 industries in the sample, marking with an asterisk the subsample of 30 industries for which we also have average hourly wage data, and the date that each industry’s code was passed. From this information we create dummy variables for each industry’s pre-code months (from July 1933 to the month the industry code was passed [if a code was passed after the 16th of the month we counted that month as a pre-code month]) and the code months (from the month of code passage to May 1935, when the NIRA was ruled unconstitutional). To control for important non-NIRA factors that could have influenced output and hourly earnings, each regression includes the real money supply, real government spending, real government revenue, and time trend variables.26 We also include fixed effects to control for omitted variables that vary across industries but do not change over time. One important factor that both could have affected industry decisions and varied over time is capacity utilization. For example, an industry operating near 100 percent of its capacity would likely have a very different optimal response to changing economic conditions or institutions (such as cartelization) than one operating at only 20 percent capacity. We create a capacity utilization proxy by dividing monthly production by the industry’s maximum level of monthly production between January 1927 and December 1929. In each regression, we include our capacity utilization proxy lagged one period so that its value is exogenous to movements within a particular month.

256

JASON E. TAYLOR AND PETER G. KLEIN

Date of Industry Code Passage, Sampled Industries. Industry Alcohol American cheese Asphalt Auto carts Beef and veal Bituminous coal Book Brick Butter Cement Cheese Chemicals Condensed milk Construction Copper Corn grindings Cotton Cotton goods Crude petro, application Crude petro, RTS Douglas fir lumber Evaporated milk Fertilizers Fine paper Glass Ice cream Large cigars Lead ore Leather Locomotives Lubricants Lumber Machinery Meat Men’s shoe Merchant pig iron Metal Milk, New York Newsprint Paper and pulp Paper production Passenger cars Pig iron Pork

Date Code Passage 8/21/1934 2/2/1935 11/6/1933 11/8/1933 1/4/1934 9/18/1933 2/17/1934 3/26/1934 1/4/1934 11/27/1933 2/2/1935 2/10/1934 1/4/1934 1/31/1934 4/21/1934 1/4/1934 7/9/1933 11/17/1933 8/19/1933 8/19/1933 8/19/1933 1/4/1934 10/31/1933 11/17/1933 10/3/1933 1/4/1934 6/19/1934 5/24/1934 9/7/1933 2/16/1934 8/19/1933 8/19/1933 3/17/1934 1/4/1934 10/3/1933 8/19/1933 11/2/1933 1/4/1934 11/17/1933 11/17/1933 11/17/1933 8/26/1933 8/19/1933 1/4/1934

Both Wage þ Output Data













   

     

257

Anatomy of a Cartel

(Continued ) Industry Raw silk Rayon Rayon yarn Refined lead Rice Rubber SþC tires Slab zinc Small cigarettes S. pine lumber Steel ingot Steel sheet Tire pneumatic case Tire tubes Total shoe Trucks Wheat flour Women’s shoes Woodworking machinery Wool Wrapping paper Zinc ore

Date Code Passage 10/7/1933 8/26/1933 8/26/1933 5/24/1934 1/4/1934 12/15/1933 12/21/1933 3/26/1935 2/9/1935 8/19/1933 8/19/1933 8/19/1933 12/21/1933 12/21/1933 10/3/1933 8/26/1933 1/4/1934 10/3/1933 5/14/1934 7/26/1933 11/17/1933 3/26/1935

Both Wage þ Output Data  

 

         

Notes: We employ output data for all 66 industries above. For the 30 industries with an asterisk, we also have average hourly earnings data.

Table 3 reports the results of GLS panel regressions employing fixed effects and cross-section weights. To control for serial correlation we use log differences (percentage changes) rather than levels. The results of specifications (1) and (3) suggest that output growth fell and real average hourly earnings growth rose during the pre-code months, when firms were free to negotiate over the contents of their industry’s cartel code, and during the months in which the code was in effect (the earnings increases are not statistically significant during the industry-code months, however). This suggests that the NIRA was generally successful in its quid pro quo of facilitating cartels in exchange for higher average hourly earnings. In specifications (2) and (4) we add our monthly measure of the Blue Eagle’s importance as proxied by its appearance in newspaper advertisements. Since our regressions employ monthly data, we use the moving average of the first observation of each month for our analysis. As the table shows, output growth fell more sharply, and real average hourly earnings

258

JASON E. TAYLOR AND PETER G. KLEIN

Impact of the Blue Eagle Emblem on Output and Wage Rates. Dependent Variables Log difference industry output (1) (2) Constant Pre-code NIRA months (July 1933-code passage) NIRA code months (industry code passage through May 1935) DLog money supply DLog government Spend DLog government revenue Time trend Industry capacity Utilization (–1) Percent of advertisements with Blue Eagle emblem Cross-sections Observations Durbin–Watson statistic

Log difference industry real average hourly earnings (3) (4)

0.07142 (11.02) –0.01182 (–2.42) –0.00641 (–3.16)

0.07155 (11.10) 0.01949 (2.61) 0.01373 (3.26)

–0.00623 (–4.87) 0.02713 (6.69) 0.00104 (1.18)

–0.00647 (–5.07) 0.01879 (4.10) –0.00416 (–3.48)

–0.11432 (–2.09) 0.00715 (2.30) 0.00073 (0.45) 0.00006 (2.14) –0.11638 (–11.98)

–0.10717 (–1.92) 0.00870 (2.77) 0.00046 (0.29) 0.00006 (1.96) –0.11603 (–12.03) 0.00130 (–5.50) 66 8372 1.88

0.12310 (5.97) 0.00320 (5.97)  0.00041 (  1.53) 0.00002 (2.92) 0.00888 (5.42)

0.12058 (5.86) 0.00288 (5.89)  0.00040 (  1.47) 0.00002 (3.18) 0.00904 (5.53) 0.00034 (4.86) 30 3823 1.82

66 8372 1.88

30 3823 1.81

Dependent variables as defined in the text. Monthly observations from January 1927 to December 1937. T-statistics in parentheses.  and  denote statistical significance at the 1 and 5 percent levels, respectively. All regressions employ industry fixed effects and cross-section weights. Standard errors calculated under a Period SUR covariance method.

grew significantly faster, when the Blue Eagle was viewed by firms as being economically important. Moreover, when one controls for the prominence of the Blue Eagle emblem, output growth rose (rather than fell) and earnings fell (rather than rose) during the NIRA code months. These results offer empirical support to the notion that perceptions of the Blue Eagle emblem played an important role in cartel compliance and breakdown under the NIRA. As with any empirical study of this period the results should be interpreted with caution as there were several important institutional changes that cannot easily be measured and controlled for. Moreover,

259

Anatomy of a Cartel

the causality could run in the opposite direction – as the NIRA cartels fell apart, the Blue Eagle became less important. (Unfortunately, we lack sufficient time-series observations under the NIRA to reliably conduct Granger causality tests.) However, we also find that the Blue Eagle’s importance is negatively correlated with firms’ defections from their cartels. Fig. 3 reports the number of weekly complaints about trade-practice provisions received in compliance field offices between November 1933 and January 1935. We take this to be a reasonable proxy for number of defections. The correlation coefficient between trade-practice complaints and the percent of advertisements carrying the Blue Eagle is –0.608 and is statistically significant at the 5 percent level. Thus the more important the Blue Eagle emblem, the fewer the defections. This offers further evidence that firms’ perceptions of the value of the Blue Eagle emblem helped drive cartel success and failure under the NRA.

Government Enforcement through Fine and Imprisonment According to a 1937 government study the NRA’s methods of ‘‘education and conciliation, at first relatively successful in bringing about compliance, gradually became less effective [so that] greater emphasis was placed by [the] NRA upon recourse to actions in the courts.’’27 Fig. 3 reveals that complaints of code violations increasingly flooded NRA compliance offices, spiking in April and May of 1934. Johnson had a strong aversion to Average NRA Trade Practice Provision Complaints Per Week, November 1933 to January 1935 1200 1000 800 600 400

1/1/1935

12/1/1934

11/1/1934

10/1/1934

9/1/1934

8/1/1934

7/1/1934

6/1/1934

5/1/1934

4/1/1934

3/1/1934

2/1/1934

1/1/1934

12/1/1933

11/1/1933

200 0

Fig. 3. Trade-Practice Provision Complaints. Source: U.S. National Recovery Administration, Research and Planning Division, Charts on the Operation of the National Industrial Recovery Act, Chart 71.

260

JASON E. TAYLOR AND PETER G. KLEIN

litigation and had hoped all along that the Blue Eagle emblem, in and of itself, would be sufficient to facilitate code compliance. Furthermore, the NRA’s lawyers generally hesitated in pursuing violators too vigorously, fearing the legislation might not hold up in court. As Irons (1982, pp. 34–35) notes, this combination of factors made the Litigation Division slow to pursue cartel violators. In early 1934, it became obvious that some new action was needed to stop widespread defections. Blackwell Smith, the NRA’s chief lawyer, prepared a staff memo outlining what he called a ‘‘Machiavellian’’ strategy of ‘‘threats’’ and ‘‘tricks’’ designed to ‘‘bring to swift justice locally well known chiselers.’’28 This strategy – perhaps best called ‘‘smoke and mirrors’’ – entailed pursuing a handful of high-profile targets to give the illusion that punishment of violators was widespread. To illustrate this strategy in practice, on February 28, 1934, Cornelia Bryce Pinchot, wife of the Pennsylvania governor Gifford Pinchot, charged that ‘‘Blue Eagles are adorning sweat shops in hundreds of towns in Pennsylvania’’ because of the lack of NRA enforcement against violators. At the same event, Louis Waldman, spokesman for the Socialist Party said, ‘‘We support the NRA . . . but it has failed . . . because of the methods Gen. Johnson has used for its enforcement.’’29 The very next day, the NRA responded by publicly calling for the removal of the Blue Eagle emblem from 12 large firms and referred 10 of these cases to the Department of Justice and one to the Federal Trade Commission for pursuit of fines and imprisonment.30 While this flurry of NRA activity made national headlines, in hopes of spreading the illusion of widespread enforcement, the reality was that such broad action was the exception and not the rule. The NRA Compliance and Litigation Divisions also employed what we call a ‘‘pick-your-battles-wisely’’ strategy. When compliance officers pursued a cartel violator in court, they generally attempted to do so in jurisdictions with judges expected to be NIRA supporters. Of the cases that made it to district courts on constitutional grounds, 71 percent of Republican judges ruled against the NRA, while 80 percent of Democratic judges ruled in its favor (Irons, 1982, p. 56). Because court losses could hurt cartel compliance (represented in our model by a drop in y2, NRA compliance officers had to consider carefully not only the severity of a defection, but also the likelihood that punishment could be secured in court when considering whether or not to take legal action. By the time the legislation was ruled unconstitutional in May 1935, NRA state and branch offices had processed a total of 155,102 complaints. Of these, 88,872 were deemed valid but handled by ‘‘education and

Anatomy of a Cartel

261

[threatened] coercion’’ in state offices. Only 7,136 cases were referred to the National Compliance Board in Washington and of these, 2,064 cases were referred to the Litigation Division. Only 564 cases reached the courts and many of these were still pending, and hence dropped, after Schechter.31

The Nature and Causes of the Compliance Crisis: Implications of the Model Incorporating these facts into the game-theoretic model presented above strongly suggests that changing perceptions, rather than changes in enforcement, are most responsible for the compliance crisis. The stated penalties of a $500 fine and up to 6 months in jail for cartel violations remained in place throughout the duration of the NIRA, so our F can be viewed as constant. Furthermore, from the above figures on litigation and code violation complaints, the actual probability of being caught and fined or imprisoned does not seem to have changed significantly over time either. As the case numbers suggest, the NRA Compliance Division rarely acted on complaints beyond politely asking violators to come into compliance. Instead, y1 and y2, the expected probabilities of being caught and punished, appear to have fallen from incorrectly high levels to more realistic ones during late 1933 and early 1934. This raises an important question: which firms defected first? The obvious answer is that the first firms to defect were those with better information about the true values of y1 and y2. Suppose, for example, we allow firms to have different beliefs about the likelihood of punishment. Assume that each firm i has beliefs yi1 and yi2 drawn randomly from normal distributions of y1 and y2 with means y1 and y2 and standard deviations si1 and si2 , respectively. In the initial periods of the game, the firms most likely to defect are those whose beliefs yi1 and yi2 are at the bottom of the respective distributions. In subsequent periods, firms in continued compliance observe that defectors are punished less frequently than expected, and these firms adjust their beliefs about punishment by adopting lower values of yi1 and yi2 , leading to additional defections, and so on. Industry case studies strongly support such a sequence of events. According to David T. Mason, executive officer of the Lumber Code Authority, the ‘‘code worked well for many months, during which time undoubtedly the vast majority of the industry did comply [mainly] because they feared the action which would be taken upon failure to comply.’’ However, Mason noted, a few violations sowed ‘‘the seeds of

262

JASON E. TAYLOR AND PETER G. KLEIN

trouble . . . As time went on with lack of action by government enforcement agencies . . . [violators] became more brazen and outspoken, encouraging their neighbors in similar violations’’ (Mason, 1935, p. 23). Mason noted that by June 1934, a full-fledged crisis left the code virtually ineffective. Although the NRA Litigation Division did belatedly push the Department of Justice to prosecute a few lumber code violators, Mason suggested that this should have been done from the start. ‘‘Such action would have been in accordance with the military theory that the first action toward a raw recruit is to teach him discipline and respect for his organization’’ (p. 24). Similar events occurred in the macaroni industry. Alexander (1997) notes that head of the macaroni code, G. G. Hoskins, pleaded with government compliance administrators throughout 1934 to prosecute large macaroni firms who were allegedly violating the macaroni code’s price provisions. However, according to Alexander (p. 331), the NRA was ‘‘divided against itself, with various departments rendering conflicting opinions about the legality of the code provisions.’’ Alexander cites a January 1935 memo from Hoskins, in which he noted that the cartel ‘‘membership at large has come to believe that all our statements that the Code could be enforced have been false’’ (p. 332). The lack of enforcement action from Washington also served to alienate members of local compliance boards, further exacerbating the compliance crisis. The December 1933 episodes mentioned earlier whereby the NRA Compliance Boards of Lincoln, Nebraska and Lowell, Massachusetts resigned in protest illustrate the point. Members of the Lincoln board noted that while the local board had relayed several cases of non-compliance to the National Compliance Board in Washington for removal of the Blue Eagle and further legal action, ‘‘there has been nothing done.’’ The local board estimated that only 30 percent of Lincoln firms were in compliance with the codes even though almost all firms displayed the Blue Eagle.32 In Lowell, the Board’s letter of resignation noted, ‘‘We repeatedly have requested cooperation and assistance which has not been afforded us [from Washington] which has been annoying and embarrassing.’’33 Because cases such as these made national news, they clearly had an impact upon both the perceived probabilities that government enforcement would be forthcoming in the face of a violation and consumers’ actual response to ‘‘buy under the Blue Eagle’’ as measured in our model by BE and GP. Essentially, it appears that in the summer and autumn of 1933 our key inequalities – equation (1) for the one-shot or finitely repeated game and equation (6) for the infinitely repeated game – held, implying that collusion was a Nash equilibrium. Over time, however, some firms’ subjective

Anatomy of a Cartel

263

evaluations of y1, y2, and BE changed (recall the actual value of F itself did not change) so that the expected payoff from defection came to exceed the expected payoff from compliance. Interestingly, this does not appear to have happened all at once. Some business owners appear to have continued to view the NIRA enforcement mechanism as having weight so that the inequality in equation (8), which provides the condition under which firms will continue to abide by a cartel agreement in the face of a defection, held. Suggesting an important role for y2 in particular, a 1937 study concluded that ‘‘compliance tended to break down in industries, and in areas generally, in which an alleged violator was allowed to continue non-compliance with impunity’’ (U.S. Committee of Industrial Analysis, 1937, p. 72). As defection became more common it became increasing clear that y1, y2, BE, and GP were actually quite small, and cartel maintenance became exceedingly difficult. In short, the NIRA’s enforcement mechanisms appear to have worked initially, but began to fail as firms realized that punishment, both from the government and from patriotic consumers, was unlikely. Of course, the fact that firms were still exempt from antitrust law could have allowed some industries – those for which collusion was a Nash equilibrium even in the absence of the NIRA codes – to maintain effective cartels after the compliance crisis.

The Model in Light of Existing NIRA Literature Our analysis of the NIRA cartel enforcement mechanism and compliance crisis strongly complements recent empirical studies of the NIRA. Alexander (1994, 1997) and Krepps (1997), both using industry-level data on price-cost margins, find that the effects of the NIRA varied widely across industries. Both authors attribute this largely to differences in enforceability of certain provisions of codes and to structural differences within industries affecting the gains to defection. Krepps, for example, finds that industries with codes that contained the relatively easily enforceable provision of openprice filing fared much better in raising price-cost margins than those that did not. This suggests that the NIRA compliance mechanism itself was not able to enforce most other code provisions. Alexander (1997), using the percentage change in establishment size between 1929 and 1933 as a proxy for cost heterogeneity, finds that cost heterogeneity is negatively correlated with profit (proxied by percent changes in dollars available for fixed costs between 1933 and 1935 and percentage change in price cost margin for same time).34 Consistent with the

264

JASON E. TAYLOR AND PETER G. KLEIN

notion of a crisis in compliance, Alexander notes that, ‘‘The NRA never took a decisive stand on enforcement . . . as soon as low-cost firms realized they were able, they reverted back to competition on the basis of low costs’’ (p. 327). Complementary to our general conclusions, this suggests that the NIRA enforcement mechanism worked initially, but became largely inept after firms no longer feared punishment. While our paper complements the existing literature, it adds a detailed examination of the NIRA cartel enforcement mechanism, particularly the role played by consumer and producer beliefs in the power of the Blue Eagle emblem, as well as an examination of specific enforcement strategies employed by the NRA Compliance and Litigation Divisions. Furthermore, our study provides the most detailed analysis of the nature and causes of one of the defining moments of the NIRA – the compliance crisis of 1934.

CONCLUSION Grossman (2004, p. 7) notes that ‘‘in any discussion of cartel behavior, it is easier to ask the questions than to find clear-cut answers.’’ While the so-called ‘‘compliance crisis’’ has been a key feature of recent empirical analyses of the National Industrial Recovery Act, scholars have done little in the pursuit of answers behind its nature and causes. To gain general insight into the mechanics of cartel success and failure we explore the compliance crisis that befell the government-sponsored cartels of the 1930s, when over 700 industries had their own specific cartel code passed into law. The game-theoretic model developed here demonstrates the importance of expectations of punishment, rather than the objective probabilities, for the success and failure the NIRA cartels. Clearly, as long as firms expected to be punished for violating their cartel codes and expected to lose customers upon being stripped of the Blue Eagle emblem, firms’ best responses were generally to comply with the NIRA cartel codes. However, as consumers lost enthusiasm for the Blue Eagle, firms realized that the NIRA compliance mechanism – including the Blue Eagle emblem – was largely innocuous, and firms began to defect from the cartel. When these defections went unpunished, other firms lowered their evaluations of punishment, leading to further defections. By the time the NRA Litigation Division began referring violators for prosecution in earnest, the compliance crisis was too far underway. The NRA Compliance and Litigation Divisions attempted to compensate with what we have called the ‘‘smoke-and-mirrors’’ and

265

Anatomy of a Cartel

‘‘pick-your-battles-wisely’’ strategies. Specifically, the NRA attempted to prosecute a few high-profile firms in jurisdictions with NIRA-friendly judges. However, with thousands of complaints flooding field offices each month, these strategies were not enough to make the threat of prosecution credible in the eyes of most cartel participants. The lethal combination of declining enthusiasm for the Blue Eagle emblem and declining expectations of punishment ultimately doomed the NIRA cartel experiment. Our model and empirical work illustrate the main features of this argument, but they are of course not definitive. The model can be extended in several ways. For example, following Hinloopen (2003), we could model belief formation explicitly, rather than treat beliefs as exogenous shift parameters.35 More generally, we can try to model additional ways in which the legal and competitive environments changed during the course of the NIRA. As Levenstein and Suslow (2006) point out in their recent survey of empirical research on cartels, the key to cartel stability over time is flexibility – cartels tend to stay together when firms can adjust cartel agreements in light of changing economic conditions. Our model highlights the most important of these changes during the NIRA period – consumer beliefs about the Blue Eagle emblem and firms’ beliefs about government enforcement – but other changes could be important as well. Additional empirical work on the spatial distribution of cartel defections under the NIRA would be valuable to test the extent of a geographic ‘‘contagion effect’’ in cartel breakdown. We look forward to pursuing these and other extensions in future work.

NOTES 1. Bittlingmayer (1995) argues, by contrast, that cartelization led to cost reductions that mitigated the price increases and output restrictions traditionally associated with cartels. More generally, cartelization may provide efficiency gains by reducing uncertainty about rivals’ actions, pooling promotion and distribution costs, establishing standards, and so on (High, 1984–85). Kinghorn (1996) and Troesken (1989) examine late twentieth-century German cartels in this context, while Sjostrom (1989) and Pirrong (1992) similiarly note such cartel efficiencies in ocean shipping. 2. See Krepps (1997) for a discussion of the importance of open price filing provisions under the NIRA codes. 3. U.S. Committee of Industrial Analysis (1937, p. 74). 4. This was particularly true of the early codes. Brand (1988, pp. 106 and 119) notes that with over 700 code applications coming into the government offices within the first six weeks, NIRA administrators could not review every provision of every code, and therefore the early codes most closely reflect those of an unrestricted cartel.

266

JASON E. TAYLOR AND PETER G. KLEIN

5. The high-wage doctrine thus differs from modern efficiency-wage theories, which argue that higher wage rates benefit firms by increasing worker productivity. Here the emphasis is strictly on the perceived boost to aggregate demand. 6. Hourly wage rates paid to 25 manufacturing industries – the best data source on monthly wage rates in the manufacturing sector, the primary sector covered by the NIRA – rose from 45 cents to 58.8 cents, an increase of nearly 31 percent, during the same period (National Industry Conference Board). 7. Document 158, 75th Congress, 1st Session, p. 48. 8. In fact, few business owners were imprisoned for cartel violations. However, this is less important than potential violators’ expectations of punishment. 9. Although the ability to display the Blue Eagle was implicit in all codes, some industry codes included more detailed provisions on how the emblem could be displayed. The Men’s Clothing Industry code, for example, said that the Blue Eagle insignia had to be attached to all garments. ‘‘The privilege of using such labels and the issuance thereof may be withdrawn [from] any such manufacture whose operations, after appropriate hearings . . . shall be found to be in substantial violation’’ of the cartel code. 10. The Lansing State Journal, October 11, 1933, p. 1. 11. As noted above, the NIRA codes sometimes allowed, or even mandated, firms to adjust prices in response to changes in cost; however, unless consumer demand is perfectly inelastic, the firms will be unable to pass the entire cost increase on to consumers. 12. For simplicity we assume that the defecting firm can purchase inputs at market prices from non-compliant suppliers. Otherwise the defector still bears some franction of g. 13. See Collie (2003) for a contrary view, however. 14. For simplicity, we assume a constant per-period probability of punishment, regardless of the previous-period outcome. In other words, the probability of being penalized for defection in a given period is the same as the probability that the penalty will be continued in subsequent periods. Alternatively, we could assume that F and BE are imposed only in the period immediately following defection, and not afterward (i.e., firms play a grim-trigger strategy while the government and consumers play tit-for-tat), or that F and BE are imposed in every period, or that F and BE are imposed with probabilities that are a function of time. None of these alternative modeling strategies change the basic insights described below. 15. As in Hinloopen (2003), for the trigger strategy to be effective in sustaining a noncooperative equilibrium discount rate, the government fine, and the detection probabilities must be sufficiently low. 16. Consistent with Stigler (1964), in the repeated game, as in the one-shot game, collusion becomes harder to sustain as n increases, suggesting that the entry of new firms into the cartelized industries could lead to a reduction in compliance. To see this, assume as before that pni ¼ ð1=4nÞðða  cÞ2 =bÞ and pdi ¼ ð1=16Þððc  aÞðn þ 1Þ ða þ c  2g  ða þ c þ 2gÞnÞÞ=n2 bÞ: The Cournot profit pci can be expressed similarly as (1/(nþ1)2((a  c)2/b). When we substitute these expressions into inequality 5 we get that the right-hand side is increasing in n; that is, an increase in the number of firms in an industry raises the critical value of dNIRA needed to sustain collusion in the infinitely repeated game, ceteris paribus. (This applies

Anatomy of a Cartel

267

regardless of our assumption about firms’ willingness to retaliate.) Even when cartelization is enforced through the twin mechanisms of BE and F, defections are more likely when there are many firms in a given industry. 17. Incidentally, the price level rose around 7 percent during this period, corresponding to a real interest rate of approximately –2.5 percent. Ceteris paribus, such low real interest rates should have made collusion easier to sustain. 18. Recall that BEi is the subjective value of the loss in business that firm i feels it will endure if it loses its Blue Eagle. Since we are assuming firms are symmetric, the firm will estimate that the loss in business to any firm j for defecting will be the same as its own, so BEj ¼ BEi. Therefore GPi, will consist of the extra business the n–m complying firms will divide equally from the mBEi losses. 19. Note we are assuming here that industry demand is not fixed; i.e., if all firms in an industry defect, are caught, and lose their Blue Eagles, consumers boycott the industry entirely. Otherwise the Cournot payoff for each firm would simply be pc  y2Fpc  y2F, as BE would equal zero. The results below are not sensitive to this assumption. 20. Taylor and Selgin (1999) show that this desire to force free riders to help boost aggregate demand, rather than the desire to help workers per se, was an important force behind the push for minimum-wage laws during the 1920s and 1930s in the U.S. 21. Of course there is no reliable measure of market wages during the fixed-wage NIRA period. Furthermore, the moderately high inflation rate likely meant that market nominal wages rose, increasing the likelihood of cartel compliance. 22. Lansing State Journal, August 7, 1933, p. 1. 23. U.S. Law Week 17, September 12, 1933. 24. Lansing State Journal, August 11, 1933, p. 9. 25. The first two and final two observations reported are actual numbers rather than the five-week central moving average. 26. The results are not sensitive to the inclusion of the time trend. 27. U.S. Committee of Industrial Analysis, 1937, p. 70. 28. Blackwell Smith to Averell Harriman, April 9, 1934, memo file, January–April 1934, Box 45, Richberg Papers, LC. 29. Both quotes are from The Washington Post, March 1, 1934, p. 1. 30. The Washington Post, March 2, 1934, p. 1. 31. U.S. Committee on Industrial Analysis, 1937, p. 72. 32. Washington Post, December 2, 1933, p. 8. 33. Washington Post, December 12, 1933, p. 1. 34. Defection, Alexander notes, was likely more attractive to a low-cost firm in a heterogeneous-cost industry than for an average firm in a homogeneous-cost industry. While admitting that her analysis is a ‘‘very rough’’ attempt to test empirically the effects of heterogeneity on cartelization, Alexander notes her analysis is primarily meant to offer some empirical support to her macaroni case study and her model on the effects of cost heterogeneity. 35. Hinloopen (2003) studies price-fixing cartels and models the probability of detection and government fine as a function of cartel characteristics, the resources available to the enforcement authority, and the number of periods in which the cartel has been in operation. The qualitative results of his analysis are similar to ours.

268

JASON E. TAYLOR AND PETER G. KLEIN

36. Note that g ¼ (cu  c)x; i.e., the cost of compliance is computed assuming the firm is producing its cartel-specified level of output.

ACKNOWLEDGMENTS We thank Chris Bailey, Fred Bateman, Juan Carlos Bisso, Harvey James, and George Selgin for helpful comments and Per Bylund, Mario Mondelli and Mark Reffit for valuable research assistance. The usual caveat applies.

REFERENCES Alexander, B. (1994). The impact of the national industrial recovery act on cartel formation and maintenance costs. Review of Economics and Statistics, 76(2), 245–254. Alexander, B. (1997). Failed cooperation in heterogeneous industries under national recovery administration. Journal of Economic History, 57(2), 322–344. Bittlingmayer, G. (1995). Output and stock prices when antitrust is suspended: The effects of the NIRA. In: F. S. McChesney & W. F. Shughart, II. (Eds), The causes and consequences of antitrust: A public choice perspective. Chicago: University of Chicago Press. Brand, D. (1988). Corporatism and the rule of law: A study of the national recovery administration. Ithaca: Cornell University Press. Collie, D. R. (2003). Sustaining Collusion with Asymmetric Costs. Working paper, Cardiff Business School, Cardiff University. Dick, A. R. (1996). When are cartels stable contracts? Journal of Law and Economics, 39(1), 241–283. Eckbo, P. L. (1976). The future of world oil. Cambridge: Ballinger Publishing. Flynn, J. T. (1948). The Roosevelt myth. New York: Devin-Adair. Reprint, San Francisco: Fox & Wilkes, 1998. Friedman, J. W. (1971). A non-cooperative equilibrium for supergames. Review of Economic Studies, 38(1), 1–12. Genovese, D., & Mullin, W. P. (2001). Rules, communication, and collusion: Narrative evidence from the sugar institute case. American Economic Review, 91(3), 379–398. Green, E. J., & Porter, R. H. (1984). Noncooperative collusion under imperfect price information. Econometrica, 52, 87–100. Griffin, J. M. (1989). Previous cartel experience: Any lesson for OPEC? In: L. R. Klein & J. Marquez (Eds), Economics in theory and practice: An eclectic approach (pp. 179–206). Dordrecht: Kluwer Academic Publishers. Grossman, P. Z. (2004). Introduction: What do we mean by cartel success? In: P. Z. Grossman (Ed.), How cartels endure and how they fail: Studies of industrial collusion (pp. 1–8). Northhampton, MA: Edward Elgar. Hawley, E. W. (1966). The new deal and the problem of monopoly. Princeton: Princeton University Press.

Anatomy of a Cartel

269

High, J. (1984–1985). Bork’s Paradox: Static vs. dynamic efficiency in antitrust analysis. Contemporary Policy Issues, 3(2), 21–34. Hinloopen, J. (2003) Cartel stability with time-dependent detection probabilities. Working Paper, University of Amsterdam and ENCORE, November. Irons, P. H. (1982). The new deal lawyers. Princeton: Princeton University Press. Johnson, H. S. (1935). The blue eagle, from egg to earth. Garden City, NY: Doubleday, Doran & Co. Kinghorn, J. R. (1996). Kartells and cartel theory: Evidence from early twentieth century German coal, iron and steel industries. Essays in Economic and Business History, 14, 339–363. Krepps, M. B. (1997). Another look at the impact of the national industrial recovery act on cartel formation and maintenance costs. Review of Economics and Statistics, 79(1), 151–154. Leuchtenburg, W. E. and the Editors of LIFE. (1964). The LIFE history of the United States volume 11: New deal and global war. New York: Time Inc. Levenstein, M. C. (1997). Price wars and the stability of collusion: A study of the pre-World War I bromine industry. Journal of Industrial Economics, 45(2), 117–137. Levenstein, M. C., & Suslow, V. Y. (2004). Studies of cartel stability: A comparison of methodological approaches. In: P. Z. Grossman (Ed.), How cartels endure and how they fail: Studies of industrial collusion (pp. 9–52). Northhampton, MA: Edward Elgar. Levenstein, M. C., & Suslow, V. Y. (2006). What determines cartel success? Journal of Economic Literature, 44(1), 43–95. Marquez, J. (1994). Life expectancy of international cartels: An empirical analysis. Review of Industrial Organization, 9(3), 331–341. Mason, D. T. (1935). The lumber code. Yale University, School of Forestry. O’Brien, A. P. (1989). A behavioral explanation for nominal wage rigidity during the great depression. Quarterly Journal of Economics, 104(4), 719–735. Pirrong, S. C. (1992). An application of core theory to the study of ocean shipping markets. Journal of Law and Economics, 35, 89–131. Rotemberg, J. J., & Saloner, G. (1986). A supergame-theoretic model of price wars during booms. American Economic Review, 76, 390–407. Rothbard, M. (1984). The federal reserve as a cartelization device: The early years, 1913–1930. In: B. Siegel (Ed.), Money in crisis (pp. 89–136). San Francisco, MA: Pacific Institute for Public Policy Research and Ballinger Publishing Company. Rothschild, R. (1999). Cartel stability when costs are heterogeneous. International Journal of Industrial Organization, 17, 717–734. Schmalensee, R. (1987). Competitive advantage and collusive optima. International Journal of Industrial Organization, 5, 351–367. Sjostrom, W. (1989). Collusion in ocean shipping: A test of monopoly and empty core models. Journal of Political Economy, 97, 1160–1179. Stigler, G. J. (1964). A theory of oligopoly. Journal of Political Economy, 72, 44–61. Suslow, V. (2005). Cartel contract duration: Empirical evidence from inter-war international cartels. Industrial and Corporate Change, 15(5), 705–744. Taylor, J., & Selgin, G. (1999). By our bootstraps: Origins and effects of the high-wage doctrine and the minimum wage. Journal of Labor Research, 20, 447–462. Taylor, J. E. (2002). The output effects of government sponsored cartels during the new deal. Journal of Industrial Economics, 50(1), 1–10.

270

JASON E. TAYLOR AND PETER G. KLEIN

Taylor, J. E. (2007). Cartel code attributes and cartel performance: An industry-level analysis of the national industrial recovery act. The Journal of Law and Economics, 50, 597–624. Troesken, W. (1989). A note on the efficacy of German steel and coal syndicates. Explorations in Economic History, 18, 595–600. U.S. Committee of Industrial Analysis. (1937). The national recovery administration: Report of the president’s committee of industrial analysis. Washington, DC: Department of Commerce. U.S. National Recovery Administration. (1933). A handbook of NRA laws, regulations, and codes. Washington: Federal Codes Inc. Vedder, R. K., & Gallaway, L. E. (1993). Out of work: Unemployment and government in twentieth-century America. New York: Holmes and Meier. Weinstein, M. M. (1980). Recovery and redistribution under the NIRA. Amsterdam: NorthHolland. Wolvin, A.D. (1968). The 1933 blue eagle campaign: A study in persuasion and coercion. Ph.D. dissertation, Purdue University.

APPENDIX. COMPARATIVE-STATIC ANALYSIS OF ONE-PERIOD MODEL Consider a one-period, n-firm symmetric Cournot model with linear demand given by p ¼ a  bX where p is price, X the total industry output, and a and b parameters. Firms produce output x at a constant marginal cost c. When complying with the NIRA, the firm’s costs increase from c to cu (where (g ¼ (cu  c)x). Under perfect collusion firm i produces output xni ¼ ð1=nÞðða  c0 Þ=2bÞ, representing one nth of the monopoly output. The market price p ¼ (1/2)(a þ cu) (the monopoly price) and firm i’s profit pni ¼ ð1=4nÞðða  c0 Þ2 =bÞ. If firm i deviates by maximizing its own profit, assuming that other firms continue to produce their shares of the collusive output, then firm i chooses xi to maximize ða  bðxi þ X ni ÞÞxi  cxi , where X ni represents the (cartellevel) output of all firms j6¼i. We assume that the defecting firm avoids the costs of compliance g, thus facing per-unit costs of c rather than c0 . The first-order condition is a  2bxi  bX ni  c ¼ 0, which implies that xdi ¼ ða  bX ni  cÞ=2b: Noting that X ni can be expressed as ððn  1Þ=nÞ ðða  c0 Þ=2bÞ, letting g ¼ c0  c, and making the appropriate substitutions, we obtain the following solution: xdi ¼

1 a  c þ an  nc 4 nb

(A1)

271

Anatomy of a Cartel

1 a  c  2g þ ða þ 3c þ 2gÞn 4 n

(A2)

1 ðc  aÞðn þ 1Þðc  a þ 2g  ða þ c  2gÞnÞ 16 n2 b

(A3)

pd ¼ pdi ¼

Note that as expected, firm i’s profit from defecting, pdi , is greater than its profit from cooperating, pni . We can now do some comparative statics on the effectiveness of the NIRA under various market conditions. For instance, the right-hand side of inequality (2) in the main text above can be expressed as 1 ðc  aÞðn þ 1Þðc  a þ 2g  ða þ c  2gÞnÞ 1 ða  cÞ2  þg b 16 n2 b 4n which is an increasing function of n.36 A decreasing function of a, and a (generally) decreasing function of c. If we model changes in industry demand as vertical, parallel shifts in the demand curve, then changes in demand are captured by changes in a, just as changes in the industry’s cost structure are captured by changes in c.

GOVERNOR EUGENE MEYER AND THE GREAT CONTRACTION James L. Butkiewicz ABSTRACT Eugene Meyer governed the Federal Reserve Board during most of the Great Contraction. Yet his role and import are almost unknown. He was not misguided by incorrect policy indicators or the real bills doctrine; the usual explanations for the failure of monetary policy. Meyer urged the adoption of expansionary policies and created the Reconstruction Finance Corporation to assist banks, especially nonmembers. However, the diffusion of power enabled the district bank Governors to stifle his efforts, although an expansionary policy was finally adopted in 1932. His unquestioning commitment to gold and lack of operational authority are the reasons policy failed.

INTRODUCTION One of the most debated questions in American economic history is why the Federal Reserve failed to arrest the monetary collapse that caused the Great Contraction of 1929–1933. In their classic analysis of the Great Contraction, Milton Friedman and Anna J. Schwartz (1963, chapter 7) attribute the deep depression to a collapse of the quantity of money, a collapse the Federal

Research in Economic History, Volume 26, 273–307 Copyright r 2008 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0363-3268/doi:10.1016/S0363-3268(08)26006-4

273

274

JAMES L. BUTKIEWICZ

Reserve could have prevented, but failed to because of a lack of leadership following the death of New York Federal Reserve Bank Governor Benjamin Strong in 1928,1 a shift of power from the Federal Reserve Bank of New York to the newly created Open Market Policy Conference (OMPC), and an incorrect theory of policy, the real bills doctrine. Allan Meltzer (2003) blames the failure of policy on the failure to distinguish between nominal and real interest rates, an incorrect policy framework, and failure to act as a lender of last resort. Paul Trescott (1982) estimates a model of open market operations and finds that Federal Reserve policy changed after 1929, which he attributes to the creation of the OMPC. In contrast, David Wheelock (1989) estimates econometric models of the determinants of Federal Reserve open market policy, finding that Federal Reserve policy during the contraction was consistent with its 1920s policy.2 Elmus Wicker (1965, 1966) also argues that Federal Reserve policy was consistent before and after Strong’s death, with international considerations dominating policy decisions in both periods. Wicker further emphasizes the diffusion of power, confusion over excess reserves,3 and a lack of understanding of the role of a central bank. The importance of domestic versus international factors in the formulation of Federal Reserve policy remains an issue. The Friedman and Schwartz position is that the contraction began in the United States and was spread to the rest of the world by the gold standard. Barry Eichengreen (1992) and Peter Temin (1989) argue that policy was constrained by the gold standard and Temin states that the gold standard rules mandated deflation. Wicker (1966) also argues that international considerations dominated monetary policy.4 Recently, Hsieh and Romer (2006) challenge the ‘‘gold standard view,’’ arguing that Federal Reserve policy was not limited by the gold standard. These varying accounts raise six issues: (1) Where was the leadership within the Federal Reserve System? (2) Did the diffused power structure hinder policy? (3) What determined open market policy during this period? (4) Did Federal Reserve officials accept the real bills belief that low nominal rates and accumulating excess reserve indicated monetary ease? (5) Why did not the Fed act as a lender of last resort? (6) Was the gold standard the ultimate monetary policy constraint? The numerous accounts of monetary policy during the Great Contraction have focused primarily on New York Federal Reserve Bank Governor George L. Harrison’s role and are based on the archival records of the Federal Reserve Board and the papers of Harrison, supplemented by other sources, especially the diaries of Board member Charles S. Hamlin.5

Eugene Meyer and the Great Contraction

275

Friedman and Schwartz claim that Harrison and the Federal Reserve Bank of New York attempted to provide policy leadership. Far less attention has been paid to the role of Federal Reserve Board Governor Eugene Meyer. However, the Library of Congress holds an extensive collection of Meyer’s papers, which has largely been ignored by economic historians. These papers, along with other archival sources, provide a new and different view into the events of this crucial period. Meyer was the dominant member of Hoover’s policy-making triumvirate that also included Harrison and Treasury Undersecretary/Secretary Ogden Mills.6 He played a significant role in the development of Federal Reserve policies during this period, and many of its success and failures are the result of Meyer’s efforts, decisions, and understanding of economics. In September 1930, President Hoover appointed Eugene Meyer as Governor of the Federal Reserve Board. Even before his appointment, Meyer wrote to Hoover signaling his support for open market purchases and lower discount rates. However, Meyer’s commitment to the gold standard delayed his efforts to engage in open market purchases until the level of foreign deposits, which he felt was dangerously high, was reduced. In the spring and summer of 1931, Meyer advocated making significant open market purchases, but his plans were frustrated by opposition within the Federal Reserve System. Following Britain’s departure from the gold standard in September 1931, Meyer defended the gold standard by increasing the discount rate, likely in response to French pressure. In early 1932, Meyer devised a complex strategy of expansion that combined the 1932 Glass–Steagall Act facilitating open market purchases with the Reconstruction Finance Corporation (RFC) that was designed to end bank panics. To Meyer, the RFC was essential to the success of open market purchases. Large-scale purchases ensued, but were ultimately terminated because bank lending failed to respond as expected and gold outflows resumed. Meyer was unable to counter the internal opposition’s arguments that the policy had failed. After this Meyer felt the election campaign and ensuing interregnum harmed his best efforts and an incipient recovery. During the final banking crisis of February–March 1933, Meyer resisted calls for open market purchases, following only the gold standard rule of increasing interest rates to stop an external gold drain. For several days prior to the inauguration he unsuccessfully urged Hoover to declare a bank holiday, his preferred solution. Frustrated with politics and suffering poor health, he tendered his resignation three weeks after FDR’s inauguration.

276

JAMES L. BUTKIEWICZ

The following account of Meyer’s role provides answers to the six questions listed above. Meyer consistently and forcefully attempted to exert his leadership. He did work effectively with New York Federal Reserve Bank Governor George Harrison. However, his best efforts where frustrated by the diffusion of power within the System. The Reserve banks jealously guarded their authority and resisted any attempt by the Board to provide leadership. Meyer’s desire to pursue expansionary open market purchases was sincere; not motivated by external (Congressional) pressures. He understood the necessity of driving nominal rates as low as possible and anticipated the accumulation of excess reserves by banks desiring to be as liquid as possible. When Federal Reserve Banks did little to assist nonmembers, he created the RFC to do the job the Fed would not. However, above all else he felt that preservation of the gold standard was his responsibility. It proved to be the ultimate constraint. The next section summarizes what others have said about Meyer and his role with the Fed and briefly summarizes Meyer’s views about key issues. Then Meyer’s efforts and major decisions as Federal Reserve Board Governor are recounted. In chronological order the major events are Meyer’s efforts to reduce foreign, especially French, deposits in the US; the failed attempt to make open market purchases in 1931; Britain’s departure from gold and raising the discount rate; creation of the RFC, the 1932 Glass–Steagall Act, and the open market program of 1932; poor health and political turmoil surrounding the election; and the final crisis.

MEYER’S ECONOMICS In spite of his prominent position as Governor of the Federal Reserve Board from 1930 to 1933, Meyer is a relatively obscure figure.7 Temin categorizes Meyer as ‘‘an orthodox Wall Street financier with a strong international orientation’’ but says nothing else about him (Temin, 1989, p. 97). Eichengreen’s only reference to Meyer cites a statement in Congressional testimony where Meyer says that open market purchases were not made following Britain’s departure from gold in 1931 due to fears of further gold losses (Eichengreen, 1992, pp. 297–298). Wicker argues that Meyer’s inexperience weakened the board and its ability to provide leadership (Wicker, 1966, p. 159). In their discussion of why monetary policy was inept, Friedman and Schwartz say of Meyer: ‘‘Perhaps, if he had more time to develop his leadership of the System,

Eugene Meyer and the Great Contraction

277

he might have been able to lead the System along a different route’’ (Friedman & Schwartz, 1963, p. 417). This statement suggests that Meyer favored more expansionary policy but the diffusion of power within the System confounded his expansionary intentions. Allan Meltzer also notes that Meyer and fellow Board member Adolph Miller favored expansionary policies but ‘‘they did little to promote their views against the dominant view in the System’’ (Meltzer, 2003, p. 281). Meltzer’s discussion of the contraction notes that Meyer frequently advocated open market purchases (Meltzer, 2003, pp. 330, 335, 337, 339–340, 341, 363, 371, 376–377, 407, 410). None of these accounts provides much insight about Meyer or his activities as Board Governor. In 1901, Meyer purchased a seat on the New York Stock Exchange and by 1915 his fortune was estimated to be $40–60 million. In 1917, he closed his financial firm to aid in the war effort. His significant governmental positions were director and later managing director of the War Finance Corporation from 1918 to 1925 and commissioner of the Farm Loan Board from 1927 to 1929. He served as Governor of the Federal Reserve Board from 1930 to 1933.8 Upon leaving the Fed Meyer purchased the Washington Post. With the aid of his daughter and son-in-law, Katharine and Philip Graham, they reversed the fortunes of the failing newspaper. He also served briefly as the first president of the World Bank. While a life-long Republican, Meyer was not traditional. He stated that he opposed deflation and resigned from his government position in 1920 in protest of what he correctly believed to be deflationary policies.9 Meyer’s statement is best understood as separating himself from the ‘‘industrial equilibriumists’’ who advocated deflation as a cure for expansionary excesses. Neither was he a devotee laissez-faire economics.10 Rather, Meyer consistently advocated federal assistance, albeit on a temporary basis, in periods of distress. When he served as managing director of the War Finance Corporation, he convinced Congress in 1920 to extend the agency’s life with a new objective of making loans to support prices of agricultural products, even obtaining an override of a Presidential veto. Meyer also rejected the real bills doctrine saying that the real bills ideas of Senator Carter Glass and his economic advisor H. Parker Willis ‘‘belonged in a museum.’’ He understood that real bills rules focusing on specific types of credit were useless, as the asset used as collateral for Federal Reserve discounts had no relationship to the use of the borrowed funds. He knew that ‘‘the country was nearly dead from deflation’’ and that ‘‘reflation’’ was

278

JAMES L. BUTKIEWICZ

in order while real bills advocates feared that open market purchases would generate inflation.11 Meyer’s belief in government intervention accorded well with the Progressive views of President Hoover. When appointed to the Federal Reserve Board, Meyer was expected to pursue expansionary policies. However, he deferred his plans due to concerns about the gold standard. This conflict between defense of the gold standard and expansionary monetary policy plagued Meyer’s tenure at the Board.

WITH ONE EYE ALWAYS ON FRANCE Even before his appointment as Governor, Meyer was wary of French monetary policy. On July 21, 1930, prior to his appointment, Meyer wrote to his brother-in-law, who was in Paris, asking if he could learn anything about the intentions of the Bank of France and the French Government. Meyer wrote that while the French claimed that they were not interested in attracting gold, he did not believe them.12 Meyer felt that his term on the Federal Reserve Board was dominated by foreign debts and domestic bank failures. He was very worried about foreign short-term liabilities, particularly liabilities to France. He understood that the United States was a net creditor, but more of its international assets were long-term, while its liabilities were more short-term. A failure of the United States to meet any demands made on it would create world chaos in an interdependent global economy. Upon learning of the potential problem when he joined the Board, he apprised Hoover of the danger of the situation, urging him to tackle the war debts-reparations conundrum, as Meyer felt this would provide some relief to the global credit strain. But Hoover did nothing at that time.13 Depicted in Fig. 1 is the ratio of short-term foreign claims to United States monetary gold. The ratio in September 1930, when Meyer’s term began, was 59 percent. By April of 1931 the ratio had fallen to 45 percent, and fell below 40 percent by July. Meyer accomplished this reduction by urging the Federal Reserve Banks to lower their discount rates. Meyer discussed his efforts to lower Federal Reserve Banks’ discount rates in a letter to his son: ‘‘One of the reasons I urged the bank rate reductions here in the spring of 1931 was to increase the differential to reduce the balances over here, which were dangerously large.’’14 Meyer’s concern reflected his unquestioning commitment to the gold standard. During Senate hearings on a bill to establish a price index as the

279

Eugene Meyer and the Great Contraction 0.7 0.6

Percent

0.5 0.4 0.3 0.2 0.1

Mar

1933 Jan

Nov

Sep

Jul

Mar

May

1932 Jan

Nov

Sep

Jul

Mar

May

1931 Jan

Nov

Sep

Jul

Mar

May

1930 Jan

0

Date

Fig. 1. Ratio of Foreign Claims to U.S. Monetary Gold. Sources: Foreign Claims, Banking and monetary statistics, 1914–1941. Monetary Gold, Federal Reserve Annual report, 1933.

Federal Reserve’s policy target, Meyer was asked if there was a chance that the United States would leave the gold standard. Meyer replied ‘‘Decidedly not.’’ He continued that United States law ‘‘prescribes a gold standard’’ and that countries leave the gold standard by force, not choice, and are anxious to return to a gold standard as soon as possible.15 Gold standard concerns delayed his efforts to pursue an expansionary open market policy.

OPEN MARKET POLICY IN 1931: TO STRIKE A BOLD STROKE Prior to his appointment as Governor, Meyer wrote to President Hoover on July 2, 1930 advocating open market operations. He noted that the Federal Reserve Bank of New York wanted to purchase securities, but was opposed by the other governors and by the Board. He noted that ‘‘(T)he policy of the Federal Reserve Bank of New York is based on the theory that government

280

JAMES L. BUTKIEWICZ

purchases at this time would do some good, and might do a great deal, and that, in any event, they could do no harm.’’16 Meyer’s first meeting with the OMPC was on September 25, 1930, shortly after he joined the Board.17 At the meeting between the Board and Reserve bank Governors, Board member Adolph Miller asked whether a program of aggressive purchases had been considered. Governor Harrison replied that purchases would be ‘‘fraught with dangers . . . attendant to a policy of inflation.’’ Meyer replied that any easing from purchases would be offset by gold exports.18 Fearful of losing gold, Meyer delayed pursuing purchases. Before he would suggest purchases, the failures of the Caldwell group of banks and the Bank of United States created the first banking crisis.19 Now a nervous public began hoarding currency and cautious bankers began accumulating excess reserves and making only the most secure loans, raising obstacles to expansion which were never surmounted. In April 1931, with foreign balances falling and gold reserves remaining strong, Meyer made his first attempt to gain support for open market purchases. Philadelphia Governor George Norris told Board member Charles S. Hamlin that Meyer worked with Harrison to endorse a policy of purchases.20 Meltzer (2003, p. 330) notes that Meyer and Miller advocated purchases at the April New York Federal Reserve Bank Directors’ meeting, and that Meyer urged reducing rates regardless of how low they seemed. At the April 29–30, 1931, meeting of the OMPC, there was a lengthy discussion of the fact that gold inflows had been sterilized. Harrison advocated reducing rates and making purchases. Norris spoke next and, with reservations, endorsed the policy. Meyer was invited to join the conference in progress and attempted to meet objections to the proposed purchases.21 At this meeting, purchases of up to $100 million were authorized, but by the June meeting, no purchases had been made.22 On June 16, Hamlin reported that Meyer felt that currency hoarding was sterilizing gold inflows and that he wanted to purchase securities.23 Meyer discussed his concern with the Federal Reserve Board staff economists Emmanuel A. Goldenweiser and Winfield Riefler.24 Both disagreed with Meyer’s plan, but Meyer was not deterred.25 On July 1, Meyer met with the Philadelphia Federal Reserve Bank directors: Governor Meyer spoke at some length regarding the reason for the purchase of $50,000,000 of Government securities at this time, placing particular emphasis on the fact that $350,000,000 of currency is being hoarded, and that we have received from abroad about $300,000,000 of gold since the first of the year which has been practically

Eugene Meyer and the Great Contraction

281

sterilized. He also stated that the purchases of these securities will not have any material effect on the money market.26

John Maynard Keynes made a five-week visit to the United States in June and July 1931. Keynes spent time with Federal Reserve authorities, ‘‘especially Mr. Eugene Meyer.’’ During his voyage home in July he wrote a report that provides considerable insight and understanding about Meyer, Harrison, and their policies: It would not be true to say that the heads of the Reserve System in New York and Washington are as confident and whole-hearted adherents of the so-called ‘monetary’ school as, for example, I am myself. But they lean very decidedly to that side of the fence, and their qualifications would relate, not to the objectives to be aimed at, but to the likelihood of attaining them by monetary expedients. That is to say, they are, at any rate, definitely not of the party of the ‘equilibriumist’ (whom I discuss below). They want to raise commodity prices as much and as soon as they can; they want to reduce the longterm rate of interest by any means in their power; and they believe that a revival of the construction industries is probably a necessary condition of the recovery of industry in general. Consequently they are quite in favor of trying the experiment of cheap money and abundant credit carried a` outrance, convinced that at worst it can do no harm, and that it must be of the right tendency in present conditions, and only doubtful (and I share their doubts) as to how substantial the practical results will prove to be in the near future.27

Keynes’ report clearly states, as do Meyer’s comments cited above, that low nominal interest rates were not viewed as a sign of monetary ease by Meyer and Harrison. They wanted to further reduce interest rates to stimulate construction. Meyer’s efforts to make open market purchases were intended to provide ‘‘cheap money and abundant credit.’’ Next came the pivotal August 11, 1931 meeting, which was initiated by the Board to consider a program of purchases.28 The minutes reveal little of what actually transpired. In testimony four years later, Hamlin recalled: Governor Meyer . . . went before the committee for 2 hours explaining that . . . nothing but a major stroke would help the situation . . . that the System should make a bold stroke and buy, say 300 millions or 400 millions of Government securities, hoping that that might turn the tide.29

During the meeting Governor Calkins of San Francisco, who consistently opposed purchases, asked each of the governors to state the position of his board of directors. Only Governor Black of Atlanta supported purchases on the scale Meyer advocated. Governor Fancher of Cleveland recommended purchases of $100 million. Nine governors opposed with four explicitly citing a lack of free gold and two others saying their banks had gone as far

282

JAMES L. BUTKIEWICZ

as possible, suggesting a free gold problem.30 Harrison said his directors did not favor purchases but motioned that the Executive Committee be authorized to purchase up to $300 million at its discretion. Calkins, with Fancher seconding, motioned for authority to purchase or to sell $100 million, plus the remaining $20 million of authority from the April meeting, which was approved.31 When the OMPC members met with the Board, Board members expressed their disappointment with the small quantity of purchases recommended.32 During this meeting Goldenweiser stated that there was enough free gold to support purchases.33 While purchases had been authorized, the OMPC executive committee had to initiate any action. When the OMPC met again on November 30, 1931, no purchases had been made.34 Meyer faced several obstacles dealing with the OMPC. The Board was not highly regarded, and the 12 district banks jealously guarded their independence and authority to initiate policy. Also, the district bank Governors were typically subordinate to powerful directors, whose interests conflicted with monetary policy goals. Chandler (1971, pp. 40–42, 67, 465) recounts that Ben Strong had sought to minimize the influence of the Federal Reserve Board, and felt that the position of (Federal Reserve district bank) governor should be the top position in the System. Strong alternately created opposition to the Board and capitalized on public and Congressional opposition to centralization of power with the Board. After Strong’s death, the district banks remained reluctant to follow the Board, preferring instead to take their lead from their boards of directors. The Federal Reserve Board suffered from a bad reputation. Discussing Meyer, Sidney Hyman (1976, p. 113) writes: the board he had inherited was weak, divided and gave the appearance of serving more as a place of dignified retirement for its aging members than of being a command post for the management of an infinitely complex financial crisis. Its members had no standing in the eyes of the governors of the Federal Reserve Banks.

Another problem was the concentration of power in the hands of a few members of several district banks’ boards of directors. While district bank directors were appointed for three-year terms, many had been reappointed and served continuously since 1914. While the original intent had been for the chair of each district bank board of directors to be the executive officer, the position of Governor was created as district bank executive officer. The Governor was chosen by the district bank directors. The Federal Reserve

Eugene Meyer and the Great Contraction

283

Board approved each Governor’s salary but had no other approval authority. Thus, many Governors felt greater responsibility to their board than to the Federal Reserve Board.35 In testimony for the Banking Act of 1935, Marriner Eccles argued that while open market operations were the most important instrument of monetary control, existing open market policy was governed by local interests, not national interests, since many of the boards of directors were dominated by bankers more interested in their banks’ profits than in national or international economic considerations.36 Meyer was frustrated by the subservient and indifferent attitudes of at least some of the reserve bank Governors and by his lack of operational authority. Following the August 11, 1931 meeting of the OMPC, Meyer expressed his frustrations, complaining about the procedure, requesting that the OMPC first meet with the Board. He asked the Governors to come prepared to have a free discussion with the Board about open market policy, not with predetermined instructions from their directors, and that they not depart before meeting with the Board.37

BRITAIN’S DEPARTURE FROM GOLD Britain’s departure from the gold standard on September 21, 1931 renewed bank panics and sparked a run on U.S. gold. Meyer urged the classic response when he recommended that Harrison increase the New York Federal Reserve’s discount rate. During the two weeks following Meyer’s recommendation, the discount rate was raised in two steps from 1-1/2% to 3-1/2%. When a New York Federal Reserve Bank official expressed concern about a large amount of bills that a member bank wanted to sell to the Bank, Meyer reminded him that the Federal Reserve should lend freely. Meyer felt he was following Bagehot’s dictum of lending freely at a penalty rate.38 Friedman and Schwartz (1963, pp. 395–396) criticize the increased discount rate, as they feel Bagehot’s rule was only partly followed. They argue that the discount rate should have been increased to meet the external drain, but the internal drain called for not only lending freely, but also open market purchases. Meyer encouraged discounting and buying all bills offered, but refused to make open market purchases. When testifying before Representative Goldsborough’s committee in 1932, both Harrison and Meyer defended their decision not to make purchases, saying that purchases would have been counter to the increased discount rate, which was needed to stop the gold drain. Recalling the events,

284

JAMES L. BUTKIEWICZ

Meyer said that open market purchases were not made because ‘‘We could not undertake anything of that character in October without increasing the loss of gold.’’39 It is clear that the discount rate increases were made either in response to French pressure or due to Meyer’s concerns about French intentions. Kenneth Moure´ (1991. pp. 74–75) argues, based on archival records of the Bank of France, that Bank of France Deputy Governor Charles Farnier and Director of Research Robert Lacour-Gayet who were in the United States at the time pressured Harrison to raise the discount rate because the French wanted to raise their discount rate, and did follow the first New York increase with an identical increase the next day.40 Wicker (1966, pp. 167–168) cites a letter from Harrison to Meyer in which Harrison denies increasing the discount rate in response to French pressure. However, in the same letter Harrison states that Meyer had extended conversations with Farnier and Lacour-Gayet. Meyer’s account does not mention his conversations with the Bank of France officials, but does state that he urged the increases in the discount rate to signal to the French that the United States was strong, and could withstand the gold drain.41 A short time later, Charles Rist, French economist and previously assistant to Bank of France Governor Moreau, confirmed Meyer’s instinct that raising the discount rate was normal and signaled confidence to the French, and that the French were watching to see what the Federal Reserve did. The increases in the discount rate in 1931 were either in response to French pressure or with an eye toward French intentions. The increased discount rate increased the severity of the contraction.42 However, Federal Reserve data do not support the belief that the banking panics were the direct result of the discount rate increases.43 Daily data from Federal Reserve preliminary reports on bank suspensions for September and October 1931, displayed in Fig. 2, indicate that failures increased immediately upon the September 21 announcement of Britain’s suspension of gold convertibility, and peaked around the time of the first discount rate increase on October 9.44 When gold losses began again in January 1932, the problem originated in France. This time French fears were aroused by an article by H. Parker Willis45 that criticized Federal Reserve policy. Meyer asked Rist to help stop publication of further Willis articles in major French outlets.46 Governor Harrison also struggled with Willis’s attempts to fan French fears. On January 14, 1932, Lacour-Gayet called Harrison, primarily about a cable by Willis appearing in a French publication stating that ‘‘inflation is now the order of the day in the United States.’’47 Lacour-Gayet reported

285

Eugene Meyer and the Great Contraction

40 35 30

Number

25 20 15 10 5

1-Sep 3 5 7 9 11 13 15 17 19 21 23 25 27 29 1-Oct 3 5 7 9-Oct 11 13 15 17 19 21 23 25 27 29 31

0

Date

Fig. 2.

Bank Suspensions: September and October 1931. Source: Federal Reserve System, Papers, box 1690.

that this article was creating concern among the public and even the Bank of France.48 Lacour-Gayet questioned recent open market purchases, which Harrison assured were seasonal, and that the Federal Reserve’s goal was to stop the deflation.49 Meyer’s dealings with Rist and Harrison’s conversations with LacourGayet reveal American sensitivity to French concerns. Defending open market purchases in April 1932, Harrison attempted to assure LacourGayet that the European conception of inflation was an increase in currency, while the Federal Reserve’s goal was to stop the deflation of credit. Lacour-Gayet doubted that the increase in excess reserves created by the open market purchases would be used to expand credit.50 In January 1932, Harrison asked Lacour-Gayet’s opinion about cutting the New York discount rate. Lacour-Gayet recommended against a cut. Again in April, Harrison went to great length to explain that the open market policy had reduced market rates below the discount rate, and an adjustment of the latter was necessary.51 Lacour-Gayet protested when Harrison said that some in America felt the French policy was deflationary. When asked if the French planned to cut their discount rate, Lacour-Gayet

286

JAMES L. BUTKIEWICZ

replied that no cuts were contemplated and that French bankers were pressing for increases.52 Years after he left the Federal Reserve, Harrison’s life and careers were featured in the Saturday Evening Post. There it was reported ‘‘One of the most dramatic episodes of his career was the silent ‘battle of the dollar’ in 1931 and 1932; Harrison stoutly defended the gold standard when all the world seemed to be drawing gold on the Federal Reserve here.’’ To both Harrison and Meyer, defending the dollar was an unquestioned duty.53

THE RFC AND OPEN MARKET PURCHASES Meyer wanted an agency similar to the War Finance Corporation that he had directed to address banking problems even before Britain left gold.54 However, Hoover had plans for a National Credit Corporation (NCC) organized privately to make loans to troubled banks. A secret meeting was arranged with a group of prominent bankers on October 4, 1931. Hoover, Treasury Secretary Andrew Mellon and Under Secretary Mills all made their pitches to the bankers, without much success. Meyer felt that the bankers had agreed prior to the meeting to do nothing. Left alone with the bankers, Meyer pleaded with them to form the NCC, promising that should (or when) it prove inadequate, he would do everything in his power to create the RFC. In testimony for the RFC legislation, Meyer said legislation for the RFC was promised when the NCC was formed and that the NCC was intended to be temporary.55 Meyer felt that the Federal Reserve paid little attention to nonmember banks (Committee on the History of the Federal Reserve System, 1954–1955, Eugene Meyer, p. 6). When asked in the Senate hearing for the proposed RFC legislation why the Fed could not take care of the situation, Meyer replied: ‘‘ . . . there is a large area in banking which they do not touch among nonmember banks’’ (United States Senate Committee on Banking and Currency, 1931, Creation of a Reconstruction Finance Corporation, p. 34). The RFC was authorized to loan to all banks, including nonmembers. The RFC was an essential component of the expansionary credit policy. Meyer hoped that RFC loans to banks would improve public psychology, since Meyer felt restoring confidence was essential for a recovery. If RFC loans stabilized the banking situation, improved confidence in the banks would result in a return flow of currency. If they were no longer worried about runs, bankers who were holding only the most liquid assets could resume lending on a normal basis, using the funds provided through open

Eugene Meyer and the Great Contraction

287

market purchases. A reduction in public currency holdings would also increase free gold. Meyer initially chaired the RFC’s board and worked to begin RFC operations as quickly as possible, Meyer located RFC loan offices in Federal Reserve facilities, and many Federal Reserve officials did double duty working with the RFC.56 Free gold in early 1932 was less than in August 1931, when Meyer wanted to make purchases. Meyer wrote that when purchases of securities began in 1932, free gold was $400 million, and foreign (French) withdrawals of almost twice this amount were expected.57 Richmond Governor Seay expressed his concern in a January 1932 letter to Harrison: ‘‘free gold held by the various Federal (R)eserve banks is rather widely observed, I would rather not reduce our free gold at present.’’ In February New York Deputy Governor W. Randolph Burgess wrote to Harrison saying that the Federal Reserve had done nothing due to ‘‘the limited amount of free gold in the face of European gold withdrawals.’’58 While working on the RFC, Meyer and others began to address the free gold issue. Meyer asked Harrison to organize a meeting of New York bankers on January 3, 1932. At this meeting, J. P. Morgan partner Russell Leffingwell suggested the use of government securities as collateral for notes to overcome the free gold issue. They next considered the difficult problem of getting Carter Glass, who dominated financial legislation in the Senate and who Meyer felt was opposed to the RFC and the use of government bonds as collateral for notes, to approve this change, so Burgess, Harrison, Leffingwell, and Meyer worked to persuade Glass. Leffingwell, who had been Assistant Treasury Secretary when Glass was Treasury Secretary from 1918 through 1920, was very close to Glass. He had extended conversations with Glass and wrote lengthy letters urging support for the use of government bonds as collateral. Glass agreed to support the bill that he later renounced.59 With the Glass–Steagall legislation in place, Meyer had overcome earlier ‘‘free-gold’’ objections to a program of open market purchases. At the February 25, 1932 OMPC meeting, weekly purchases of $25 million were approved in anticipation of the passage of the Glass–Steagall Act two days later. It is important to reemphasize the connection Meyer made between the RFC and the open market purchases. Keynes (1981, p. 565) implies this connection when he reported Meyer’s and the New York Federal Reserve Bank officials’ expectations: There are also two other reasons why open-market purchases may have to take place on an inconveniently large scale before they are effective. Bank nervousness in the

288

JAMES L. BUTKIEWICZ

United States is causing a most dangerous tendency to hoard currency . . . For the same reason, namely the extreme nervousness of the member banks, they may, for a time, actually refuse to make use of the additional reserves put at their disposal by the Reserve System, i.e. they may maintain idle reserves in excess of their legal requirements, a thing which is almost unthinkable in London and has never previously occurred in New York.

An accumulation of excess reserves was expected, and they were not expected to be used until the bank panics ended and confidence restored.60 To Meyer, the RFC was needed to eliminate bank runs, thus making the purchases effective. Hamlin records Meyer saying at this time ‘‘that if damned hoarders would cease hoarding and the damned banks begin loaning, all would be well.’’ By the April 12 OMPC meeting Hamlin felt that Meyer had lost the optimism he displayed a month earlier, and said that Meyer favored a bold policy of purchases. At this meeting, purchases of $500 million, at a rate of $100 million per week were approved.61 The open market purchases faced numerous internal difficulties. A number of the Governors, especially Roy Young of Boston and James McDougal of Chicago were opposed to the idea, and their banks held large amounts of gold. Because government securities were not used as collateral until May, the Dallas bank in February and the Atlanta and Richmond banks in April all wrote saying that they wanted to cooperate, but did not have enough free gold to take their apportioned share of the securities.62 There was general agreement at the time that the RFC had reduced bank failures, but the expected return flow of currency was short lived. Fig. 3 shows Federal Reserve note circulation for 1932. After reductions in March and April, currency demand increased in May and again in June and July due to the banking crisis in Chicago. The high level of hoarding and Chicago banking panic in June did little to relieve bankers’ fears, so the increase in lending that was anticipated never occurred. Fig. 4 depicts member bank loans and investments in 101 leading cities. While the open market purchases did result in an increase in investments, bank lending continued to follow a relentless downward trend.63 At the May meeting, the resolution authorized purchases of $500 million, as deemed appropriate, but some objections to continued expansion were raised. The Executive Committee decided to purchase $80 million during the next week.64 The May 30 issue of Time (1932, p. 11) magazine reported that while $725 million of government securities had been purchased, and further purchases had been approved at the May OMPC meeting, the program was not working, as banks accumulated excess reserves rather than lending as expected.

289

Eugene Meyer and the Great Contraction 2900000 2850000

Thousands of $

2800000 2750000 2700000 2650000 2600000 2550000 2500000 2450000 December

November

October

September

August

July

June

May

April

March

February

January

2400000

1932

Fig. 3.

Federal Reserve note circulation. Source: Banking and monetary statistics, 1914–1941.

The fact that purchases of a large quantity of government bonds would have little immediate impact, resulting in criticism, was foreseen. Keynes (1981, p. 565) wrote: the Reserve System may have to purchase a very large volume of Government securities to make its immediate purpose effective, and in the course of doing this it may lose all its bill holdings-both of which developments may tend to evoke hampering criticism and give a handle to the deflationist party.

Proponents of the real bills doctrine would point to the accumulation of excess reserves as the result of a lack of demand.65 That was not Meyer’s opinion. He told the Governors in July that while bank presidents were saying that credit was available, Meyer and everyone knew that people were experiencing difficulty getting credit.66 At the May meeting an attempt was made to use moral suasion to increase bank lending. Miller suggested the formation of Banking and Industrial Committees in each Federal Reserve district. The stated purpose of the committees was ‘‘to enlist the cooperation of bankers and businessmen in an effort to develop ways and means of making effective use of the funds which were being made available by the open market operations of the System.’’ Meyer observed years later that he

290

JAMES L. BUTKIEWICZ 25000

Millions of $

20000

15000

10000

5000

Total Loans and Investments Loans

Jan-31 Feb-31 Mar-31 Apr-31 May-31 Jun-31 Jul-31 Aug-31 Sep-31 Oct-31 Nov-31 Dec-31 Jan-32 Feb-32 Mar-32 Apr-32 May-32 Jun-32 Jul-32 Aug-32 Sep-32 Oct-32 Nov-32 Dec-32

0

Year and Month

Fig. 4.

Member Bank Loans and Investments 101 Leading Cities. Source: Banking and monetary statistics, 1914–1941.

was not sure if these committees did any good, but at least an attempt was made.67 In May and June there was another gold outflow. Russell Leffingwell, who was a key figure in the development of the 1932 Glass–Steagall Act, wrote his close friend Carter Glass about the effect of the drain on Federal Reserve officials: ‘‘The Glass-Steagall Bill and the open-market policy under it started things going again in 1932. But another run on our gold probably destroyed the nerve of the Federal Reserve authorities.’’68 There were extensive concerns about the gold drain within the System. A staff member at the Minneapolis bank wrote Governor Geery on May 6, 1932 arguing that extensive gold exports were expected and the Federal Reserve should not exhaust its ability to buy securities. Richmond Governor Seay wrote Burgess on May 24 stating his support for the policy, but he was unable to continue his bank’s participation as his reserve ratio was bordering on 50%. Burgess replied that Richmond and Dallas had low reserve ratios and could not take their full allotments, and

Eugene Meyer and the Great Contraction

291

that St. Louis and San Francisco may adjust; they were waiting for decisions by their boards.69 Such was the concern over the gold drain that Harrison wrote a letter on June 8 to the other governors. His cover memo to Meyer stated: ‘‘I have received so many telephone calls from the other Federal reserve banks about different aspects of the problem that I thought it advisable to write a more or less comprehensive letter to each of them.’’ In the letter Harrison stated his expectation that pressure on the dollar would end very soon, as the Bank of France was completing its conversion of its dollar balance to gold.70 Burgess, who favored continuation of large scale purchases, wrote to Harrison on June 30: ‘‘Our partners in the enterprise are altogether too ready to vacillate, and every time we show anything of the same spirit ourselves it strengthens their position.’’ Governor Martin of St. Louis wrote Burgess on June 30 saying that the public is paying attention to the reserve percentages of the reserve banks. On July 2, Governor McKinney of Dallas wrote Burgess saying that the narrow margin of reserves is a difficulty confronting most Federal Reserve Banks. At the July 14 OMPC meeting a number of Federal Reserve Bank governors stated that their gold reserve percentages had dropped to close to 50%, and that their directors were reluctant to continue, ‘‘ . . . unless the operation were a united system undertaking.’’71 At the July meeting Meyer reported defensively that while the purchase program had not resulted in an extension of credit to business, he did not feel the program could be judged a failure. Harrison spoke in support of Meyer. Governors Young and McDougal noted that their banks had both assisted New York by taking more securities than their allocated share,72 but that both were opposed to the program and McDougal emphasized his directors’ desire to stop. In the preliminary memorandum for the July meeting, the effectiveness of the purchases was questioned: ‘‘It, therefore, seems desirable at this time to review the general business and industrial situation as a basis for determining the extent to which further purchases of government securities would likely prove effective.’’ The policy agreed upon was an excess reserve target of $200 million. Over the next few weeks, $30 million of purchases were made. By August, purchases ended.73 Data for the Federal Reserve Banks’ and system reserve ratios for February, June and July of 1932 are reported in Table 1. All of the banks except Boston and Chicago had reserve ratios falling toward or below 50%, which appears to be the minimum ratio the banks’ directors were comfortable with. Prior to the meeting Chicago Governor McDougal wrote

292

JAMES L. BUTKIEWICZ

Reserve Bank Gold Reserve Ratios. Reserve Bank

Boston New York Philadelphia Cleveland Richmond Atlanta Chicago St. Louis Minneapolis Kansas City Dallas San Francisco System

Reserve Ratio – End of Month February 1932

June 1932

July 1932

72.0 73.8 58.3 61.6 65.8 64.6 78.7 70.8 63.5 60.6 56.6 51.8 68.4

65.0 52.2 54.4 54.9 48.2 52.1 71.2 55.8 50.4 53.9 59.0 52.7 57.6

64.2 52.6 51.5 55.3 49.3 45.5 72.0 53.7 46.6 51.8 47.9 46.3 56.5

Note: The system reserve percentages differ slightly from the percentages reported in Banking and Monetary Statistics (1943, p. 348). The contemporaneous data are used here to best reflect the information available at the time the decisions were being made. Source: Federal Reserve Board, Federal Reserve Bulletin, March, July and August 1932.

to Burgess and to Harrison arguing that due to its large amount of outstanding currency, it needed to maintain its gold surplus, which it could lose to the other banks, should the currency return.74 McDougal also expressed concern about low nominal rates.75 In House hearings in April, Harrison had anticipated the outcome. Congressman Goldsborough asked if announcing a policy of stabilizing a price index or making a specific amount of open market purchases wouldn’t do more to improve confidence.76 Harrison replied that if a specific policy were announced, and there was a gold drain, the policy would have to be terminated, to the embarrassment of the Federal Reserve.77 Friedman and Schwartz (1963, pp. 384–385) argue that the 1932 purchases were carried out only in response to Congressional pressure. However, Meyer favored open market purchases even before he became Governor. As noted above, Meyer, Harrison, and Leffingwell worked very hard to obtain Carter Glass’s reluctant support for the RFC and Glass–Steagall bills. In House hearings Congressman James Strong observed: ‘‘as the result of an Executive proposal we passed the Reconstruction Finance bill and the Glass-Steagall bill . . . ’’ At the same hearings Meyer said: ‘‘I came here some weeks ago and asked you to

Eugene Meyer and the Great Contraction

293

authorize, and you did, a reconstruction finance agency and amendments to the Federal Reserve Act . . . ’’ referring to the RFC and the 1932 Glass–Steagall Act.78 At the April 1932 OMPC meetings, Meyer did make statements suggesting that he was motivated by Congressional pressure. Meyer’s statements are better understood to be part of a continuing effort to obtain support of the Reserve Bank governors whom he felt that he always had to exhort to obtain any unified action.79 Hsieh and Romer (2006) argue that forward exchange rates did not exhibit devaluation expectations or fear of speculative attacks on the dollar. Internal disagreement and success of the policy in increasing excess reserves were the reasons they argue the program ended. However, as noted above, Meyer and contemporary observers judged the purchases a failure when bank lending did not revive as expected. The decline in gold reserve ratios due to gold outflows and internal opposition were the reasons the purchases ended. Meyer could not gain support for a program that was deemed a failure. Meltzer (2003, p. 357) cites the fact that the authority to pledge government bonds as collateral for notes was not used until May as evidence that free gold was not a problem. However, Meyer told the governors in April that a strong program might end hoarding and make the total program unnecessary.80 If hoarding had ended, the return of currency would have increased free gold, and the Glass–Steagall authority would never have to be invoked. To obtain Glass’s support for the bill, Meyer appears to have told Glass that he needed the bill only to deal with opposition within the System, and that the powers would not be used. During the Senate hearings for the Banking Act of 1935 (US Senate Committee on Banking and Currency, 1935, p. 686), Glass expressed his opposition to the bill he had sponsored: The so-called ‘‘Glass-Steagall bill’’ that practically returned us for the time to a bondsecured currency. And I want to say again for the record that from my individual records I never would have agreed to have reported that bill but for the fact that we were assured over and over again by men in authority that they did not expect to use it; that they wanted it for psychological purposes.

From March 15 through May 4, 1932 free gold fell from $349 million to $169 million. As the expected return flow of currency ended, Meyer could delay no longer. On May 5, government securities were pledged as collateral for Federal Reserve notes, ending the free gold issue.81

294

JAMES L. BUTKIEWICZ

Health and Politics Meyer had not intended simultaneously to head both the Federal Reserve Board and the RFC, but did so to gain Democratic support for the RFC. The stress of both jobs took its toll on his health. He suffered from neuritis and in August returned to his New York home to recover his health, being mostly absent from the Board through the end of October. While his health improved, he did not fully recover until he left the Board in May 1933. He felt the RFC and open market purchases improved the economy,82 but politics undermined his efforts to promote recovery. John Nance Garner, Speaker of the House, forced publication of the names of all banks receiving RFC loans.83 Meyer (1961, pp. 621–622) said that he ‘‘ . . . regarded that as sabotage of the gravest kind of the national welfare.’’ He continued, ‘‘I think the Democratic policy was to make the depression as bad as possible for fear they wouldn’t get the Democratic party in.’’ To Meyer (1961, pp. 688–689) the interregnum polarized policy: From February 1 to March 4, 1933 was a period of going along as best you could from day to day, knowing that you weren’t getting anything except black looks from the White House. There was an inability to function . . . My relationship with Hoover hadn’t deteriorated completely . . . He had all these problems and was trying to get Roosevelt to cooperate in his own way – the Hoover way.

At that time the Board Governor served at the pleasure of the President and the archival evidence suggests that Meyer anticipated the election outcome and his own departure from the Board.84 He submitted his resignation on March 24, and it was officially accepted effective May 10, 1933.

THE FINAL CRISIS AND COLLAPSE The final banking crisis in 1933 sparked another gold drain and culminated in a nation wide bank holiday in March. There are two views of the final crisis, the ‘‘gathering storm’’ view and the ‘‘Michigan shock’’ view. Friedman and Schwartz (1963, pp. 324–325) argue that the final banking crisis began in the Midwest and Far West in the final months of 1932, spreading east. They cite the Michigan bank holiday as but one in a series of holidays. Meltzer (2003, p. 379) concurs, saying the final crisis developed over a period of months as it spread from state to state. While recognizing that the lengthy depression had seriously weakened many banks throughout the country, Wicker (1996) and James L. Butkiewicz (1999)

295

Eugene Meyer and the Great Contraction

dispute the gathering storm view. They argue that the final crisis was the direct result of a single shock, the declaration of a bank holiday in Michigan on February 14, 1933. Following the Michigan crisis, banking difficulties and holidays quickly spread to surrounding states, and ultimately throughout the country. Meyer also felt that the Michigan crisis was a unique shock. Discussing the occurrence of local bank holidays and moratoria culminating in the Nevada bank holiday in November 1932, he concluded ‘‘there was certainly nothing about it to prepare the public for the proclamation, three months later, of a banking holiday in Michigan.’’85 Prior to the Michigan crisis, the dollar exchange rates for the major gold block countries: France, Holland, Belgium, and Switzerland, were at par or near the gold import point. By the end of the week of the Michigan crisis, all had jumped to or above the gold export point. Daily dollar–French franc exchange rate data are depicted in Fig. 5. From the beginning of the year through February 14 the rate is below par.86 The exchange rate data do not

3.98

Cents per French franc

3.96

3.94

3.92

3.9 Par value Daily closing price Gold Export Point

3.88

3-Jan 5 7 10 12 14 17 19 21 24 26 28 31 2-Feb 4 7 9 11 15 17 20 23 25 28 2-Mar

3.86

Month and Day

Fig. 5. Dollar-French Franc Exchange Rate. Sources: Exchange rates, New York Times, (1933). Par value and gold export point. Morgenstern, O. (1959) International financial transactions and business cycles.

296

JAMES L. BUTKIEWICZ

evidence concern about the dollar until after the Michigan crisis. On February 25, 1933, Charles Rist wrote to J. P. Morgan partner Russell Leffingwell, stating: The dollar is the chief topic in discussions. It is accepted as an obvious truth that the devaluation of the dollar will be the only remedy by which the American Government will be able to solve the farm debt problems. This state of mind which has been in evidence for several weeks has particularly gained ground as a result of the recent banking holiday in the State of Michigan. Since that measure was instituted not a single day has passed without my being asked by two or three parties at a time about my opinion as to the probability of a fall in the rate of the dollar.87

The Michigan crisis precipitated an attack on the dollar, and Meyer again followed what he felt were gold standard rules, as he had done in October 1931. When Ogden Mills pressed for open market purchases as RFC loans were going to drain Treasury balances from banks, increasing monetary tightness, Meyer objected, saying that higher interest rates were called for, as European rates had increased.88 While opposing purchases for the system, Meyer did press Governor McDougal of Chicago to provide relief by purchasing government securities from banks in his district, the focal point of the crisis. McDougal agreed to help in his characteristic unhelpful fashion, stating that member banks would learn of his bank’s willingness to purchase securities when they requested such assistance.89 Barry Wigmore (1987) argues that the bank holiday was declared due to the Federal Reserve’s loss of gold. He further argues that if the crisis resulted from a run on the dollar due to Roosevelt’s discussions of devaluations, the only remedy available to Hoover and the Federal Reserve would have been to declare a bank holiday. In the days preceding Roosevelt’s inauguration, Meyer pressed Hoover to declare a bank holiday.90 Throughout the archival records and Congressional testimony, Meyer repeatedly acknowledges the force of international factors upon his decisions. The interchange between Meyer and Mississippi Representative Jeff Busby in House hearings is illustrative: Busby: . . . in all of your discussions about the things which have to do with our domestic economical welfare, you have coupled your premise to European and foreign conditions. Meyer: . . . I am giving it some weight, because it affects our domestic situation.91

Eugene Meyer and the Great Contraction

297

CONCLUSION Understanding Eugene Meyer and his role within the Federal Reserve System affords better understanding of Federal Reserve policy during the Great Contraction.92 Meyer favored an expansionary open market policy and philosophically believed in government intervention in times of economic crisis. He attempted to lead the Federal Reserve System although the Board had no authority to initiate operations. He and Board colleague Adolph Miller traveled frequently to the district banks attempting to persuade the district bank Governors and directors to support expansionary policies, generally with little success. Meyer was responsible for two significant pieces of legislation, the 1932 Glass–Steagall Act and the RFC. These bills made possible the expansionary program of purchases in 1932. The Federal Reserve System during Meyer’s tenure was dysfunctional. Ben Strong provided effective leadership because he was a reserve bank Governor and thus readily acceptable to the Governors and directors of the other reserve banks. When Meyer, Governor of the Board, attempted to lead, he was consistently opposed and resisted by the leadership of most of the 12 district banks as they jealously guarded their power to determine monetary policy. The System reforms embodied in the Banking Act of 1935 were clearly needed. Beginning in April 1931, Meyer advocated open market purchases to offset sterilization of gold inflows. His failure to ‘‘strike a bold stroke’’ was due to a lack of operational authority. In 1932, Meyer did develop a significant expansionary policy resulting in over $1 billion of open market purchases. The purchases ended when Meyer’s quantitative target, bank lending, failed to increase as expected, and the reserve banks suffered gold drains, adding strength to those within the system who opposed the purchase program. By 1932 his efforts seemed, to him, unable to overcome currency hoarding and bankers’ demand for liquidity. Meyer disavowed the real bills doctrine. He felt easy money was called for and wanted to drive nominal rates as low as possible. Neither was Meyer confused by the accumulation of excess reserves. He recognized this to be the bankers’ response to their desire to remain as liquid as possible. Feeling that the Federal Reserve district banks would not assist nonmember banks that accounted for about two-thirds of all banks, Meyer was the force behind the creation of the RFC. While RFC loans provided some relief, it was not enough to reverse the currency hoarding. In this respect, the RFC failed to attain its intended objective.

298

JAMES L. BUTKIEWICZ

But above all he sought to maintain the gold standard. Meyer’s (and Harrison’s) commitment to gold never wavered. From his appointment in September 1930 through March 1931 he delayed pursuing open market purchases due to concerns about the high level of foreign deposits relative to United States monetary gold. Gold drains caused contractionary policies in October 1931 and February–March 1933.93 Most importantly, a gold drain ended the 1932 open market purchases. The only solution was abandoning the gold standard, a choice that was never considered.

NOTES 1. The heads of the 12 Federal Reserve Banks initially held the title of governor. The Board was simply the Federal Reserve Board with first five and then six appointed members and two ex officio members. Two appointed board members were designated executive officers with the titles ‘‘governor’’ and ‘‘vice-governor’’. The district bank heads’ titles and Board name and composition were changed to the present structure by the Banking Act of 1935. 2. The difference between Trescott and Wheelock is a difference of interpretation, as both of their models forecast much larger open market purchases than were made during the contraction period. Trescott assumes that open market operations would be made to offset gold and currency flows, while Wheelock assumes that open market purchases would end once member bank borrowing from the Federal Reserve approached zero. 3. Federal Reserve member banks were required to hold a percentage of deposits in reserve. Reserves above the required amount are excess reserves. 4. However, Friedman recently attributed increased responsibility for the severity of the contraction to failure of France and the United States to follow gold standard rules Moreau (1991). Friedman further argues that the gold standard might have worked had the rules of the game been followed Parker (2002). 5. Chandler (1971), Eichengreen, (1992), Friedman and Schwartz (1963), Meltzer (2003) and Wicker (1966). 6. Mills was appointed Under Secretary of the Treasury in 1927, and became the Secretary of the Treasury on February 13, 1932. 7. Meyer’s archival material was not publicly available until 1974, so it was unavailable to Friedman and Schwartz (1963) and Wicker (1966). The lack of references to Meyer’s archives suggests that it remains a relatively unknown resource. 8. While Hoover had offered Meyer several other positions in his new administration, Meyer preferred to remain at leisure from the time he left the Farm Loan Board until he joined the Federal Reserve Board. 9. Meyer, Papers, box 181. Meyer did resign from the War Finance Corporation but was soon reappointed by President Harding. 10. Meyer studied economics at Yale with the champion of laissez-faire economics, William Graham Sumner. However, after college, Meyer worked and studied in Europe, where his economic views were heavily influenced from a course

Eugene Meyer and the Great Contraction

299

he took with German state socialist Adolph Wagner. Wagner rejected laissez-faire and favored an increasing role for the state. Wagner’s philosophy was a conservative socialism. Ironically, Wagner was famously anti-Semitic; Meyer was a Jew. Evalyn Clark (1940, pp. 378–411) discusses Wager’s political and economic views. 11. Meyer, Papers, box 181; U.S. Senate Committee on Banking and Currency (1932, p. 217). While Meyer never distinguished real from nominal interest rates, he wrote in a 1931 letter ‘‘that business cannot be stimulated in times of depression by low rates alone.’’ Meyer, Papers, box 74. 12. Meyer, Papers, box 4. 13. Meyer, Papers, box 181. 14. Meyer, Papers, boxes 4 and 181. 15. U.S. Senate Committee on Banking and Currency (1932, pp. 218–219). In a 1952 letter Meyer remained critical of the decision to leave the gold standard. He wrote that Roosevelt ‘‘repudiated the gold standard and announced it was a forced move; it was not. During the bank holiday this country gained large amounts of gold.’’ Meyer, Papers, box 167. 16. Hoover, Papers, Presidential Subject File, Federal Reserve Board, Correspondence (1930). Meyer wrote a second letter suggesting lowering the discount rate at four Federal Reserve banks. 17. The 12 Federal Reserve district bank Governors comprised the OMPC. The OMPC determined system open market policy, subject to Board approval. The Executive Committee of the OMPC, consisting of five of the Governors, determined if and when the policy recommendations were conducted. 18. Federal Open Market Committee (1956, pp. 513–516). Meyer’s statement reflected his fear of losing gold, Hamlin, Papers, reel 19. 19. Meyer explored the feasibility of purchases at this time. A December 27, 1930 memo to Meyer from Edward L. Smead, Federal Reserve Board Chief of the Division of Bank Operations discussed the possibility of open market purchases. Smead estimated that purchases of between $500 and $800 million might be possible, Meyer, Papers, box 120. 20. Hamlin, Papers, reel 19. 21. The OMPC’s procedure was to meet and then consult with the Board after their meeting, and this was frequently a cursory consultation. Meyer repeatedly requested that the Governors meet with the Board both before and following the OMPC meeting. That Meyer was allowed to join the conference in progress was a small step in his continuing effort to have greater input and influence in open market policy decisions. 22. Open Market Policy Conference, Minutes, April 29, 1931 and June 19, 1931. 23. Hamlin, Papers, reel 19. At the June 22 meeting of the OMPC Executive Committee Harrison again argued that gold flows had been sterilized, and Meyer added that hoarding (increases in currency demand) was sterilizing gold inflows, and that if nothing were done, critics would charge the Federal Reserve System with sterilizing gold inflows. Meyer had strongly advocated purchases. Purchases of $50 million were authorized, but Meyer had wanted more. From June 22 through August 3, purchases totaled $80 million. Open Market Policy Conference, Minutes, June 19 and 22 and August 4, 1931. 24. Wicker (2002) cites the memos recording the meetings, noting that Meyer understood the effect of a currency drain on bank reserves, but did not understand

300

JAMES L. BUTKIEWICZ

the effect of an increased currency ratio on the money supply, as this latter effect was not yet understood. 25. Goldenweiser, Papers, box 1. Yohe (1982) notes that the Board economists espoused the real bills doctrine. 26. Federal Reserve Bank of Philadelphia, Minutes, July 1, 1931, p. 396. 27. Keynes (1981, p. 562). 28. Federal Reserve System, Papers, box 1437. 29. U.S. Senate Committee on Banking and Currency (1935, pp. 945–946). 30. The Federal Reserve System was subject to both reserve requirements and a collateral requirement. The banks were required to hold a 35% reserve in gold or lawful money (United States notes and silver certificates) against bank deposits, and 40% gold reserves against Federal Reserve notes in circulation. The reserve requirements could be suspended in emergencies. The second requirement was a collateral requirement against all Federal Reserve notes, both those in circulation and notes held by Federal Reserve banks for future distribution. The collateral requirement was 40% gold and 60% eligible paper (discounted loans and bankers’ acceptances) or gold. In addition, the Federal Reserve was required to deposit gold not less than 5% of eligible paper used as note collateral with the Treasury in a gold redemption fund. Free gold was the total gold reserve minus the amounts of gold required for deposit and note reserves, additional gold collateral beyond the 40% minimum when eligible paper was less than 60% of notes and gold required for the gold redemption fund. Collateral requirements could not be suspended (Federal Reserve Board, Federal Reserve Bulletin, March 1932, pp. 143–144). 31. Open Market Policy Conference, Minutes, August 11, 1931, pp. 6–10. 32. Meyer told Hamlin that Harrison had been unable to sell the program, but he felt the Board (i.e. Meyer) would have been successful, which Hamlin doubted. Charles S. Hamlin, Papers, reel 19. 33. Federal Reserve Board, Minutes, August 11, 1931, pp. 128–129. 34. Open Market Policy Conference, Minutes, November 30, 1931, p. 1. 35. F. Cyril James (1938, p. 876) writes that when it was decided that a Governor of the Federal Reserve Bank of Chicago was to be selected, the leading Chicago bankers determined that the Governor position had less standing than the presidency of a leading Chicago bank. 36. U.S. House of Representatives Committee on Banking and Currency (1935, pp. 188–189). 37. Federal Reserve Board, Minutes, August 11, 1931, pp. 128–129. Meyer continued to press the district banks’ boards to adopt an expansionary policy. Miller went to San Francisco to make the Board’s case. In a September 3, 1931 telegram to Meyer, Miller reported that in a discussion of open market policy, he did his best to argue the Board’s position but that ‘‘there is still much banking opposition to overcome.’’ Meyer, Papers, box 74. 38. Meyer, Papers, box 181. 39. United States House of Representatives Committee on Banking and Currency (1932, pp. 478 and 555). 40. Moure´ (1991, p. 75n) also writes that contemporary observers believed that France was exerting financial pressure on the United States to obtain a better war

Eugene Meyer and the Great Contraction

301

debts agreement, as Premier Laval was scheduled to visit Washington later in October 1931 to discuss the issue. 41. On October 6 George Harrison sent Meyer an article clipped from that day’s New York Evening Post, 1931. The article discussed rumors circulating in Europe, especially Paris, about U.S. commitment to the gold standard, resulting in extensive selling of American assets. One owner of a large amount of American assets was quoted as saying: ‘‘What Europe needs is a strong, convincing statement from Washington that the United States will hold firmly to the gold standard.’’ Meyer, Papers, box 74. The first of the two discount rate increases was approved on October 8. 42. Between August 1931 and February 1932 the Federal Reserve’s adjusted index of industrial production fell 11.5% and wholesale prices declined 8% (Federal Reserve Board, Bulletin, September 1932 and September 1933). 43. Bordo (2006, p. 651). 44. The data in the daily preliminary reports are incomplete in that the total number of failures for September and October are 298 and 512, while the final totals reported in the September 1937 Federal Reserve Bulletin are 305 and 522 for the same months. The small difference in the total number of failures would not change the evidence depicted in Fig. 2. Examination of Federal Reserve files reveals that reports of some small and private bank failures arrived with a lag of weeks or even months. 45. Willis had written much of the original Federal Reserve Act and was highly critical of the Federal Reserve during this period Meltzer (2003, p.70n). 46. Meyer, Papers, box 181. Friedman and Schwartz (1963, p. 407, note 162) also discuss the Willis episode. 47. To Willis, any deviation from his strict version of the real bills doctrine was inflationary. 48. Reading the Willis article to the Board, Meyer called it untrue and traitorous, Hamlin, Papers, reel 20. 49. Harrison, Papers, 3125.3. 50. Harrison, Papers, 3125.3. The French were critical of the U.S. budget deficit as well. 51. The New York discount rate was lowered from 3-1/2% to 3% on February 26, and lowered to 2-1/2% on June 24. Board of Governors of the Federal Reserve System (1943, p. 441) 52. Harrison, Papers, 3125.3. 53. Saturday Evening Post. February 26, 1949, p. 60, in Meyer, Papers, box 26. 54. Meyer served as the managing director of the War Finance Corporation during much of its existence. He transformed the agency’s functions from business lending and bond price support to lending to support agricultural exports and prices. It was this later version of the WFC that became his model for the Reconstruction Finance Corporation. 55. Epstein and Ferguson (1984) incorrectly attribute the formation of the NCC to the initiative of private bankers. Meyer wanted the RFC but accepted Hoover’s NCC expecting it would not succeed and Hoover would then agree to the RFC. The essential details about the formation of the NCC and RFC are in Meyer, Papers, box 181; Fortune, May 1940, pp. 46–47; Gerald D. Nash (1959); Merlo Pusey (1974); U.S. House of Representatives Committee on Banking and Currency (1931–1932, pp. 29–30).

302

JAMES L. BUTKIEWICZ

56. Meyer, Papers, box 181. In the proposed legislation there was a provision for the RFC to have a subsidiary to support commodity prices, as the RFC’s Commodity Credit Corporation later did under the New Deal. Senator Carter Glass struck this provision from the bill. Meyer continued his efforts to support commodity prices to aid farmers, going outside of the government. In August 1932, he organized the Commodities Finance Corporation, a $100 million bankers’ pool to lend for basic commodities. After RFC operations began, Meyer decided that the RFC should buy banks’ preferred stock and asked RFC Board member Jesse Jones to inquire whether Democratic House Speaker John Nance Garner would support this change. Jones reported back that Garner would not. A few months later, the Emergency Banking Act of 1933, approving President Roosevelt’s bank holiday, also authorized RFC purchases of bank preferred stock. 57. Meyer, Papers, box 85. Internal Federal Reserve documents report free gold of $416 million on February 24, 1932 (Federal Reserve System, Papers, box 1229). 58. W. Randolph Burgess, Papers, No. 410.5. 59. Meyer, Papers, box 181; Harrison, Papers, 2690.2; Glass, Papers, box 4; Leffingwell, Papers, box 3. 60. Keynes’ comments refute Wicker’s (1966, p. 195) argument that Federal Reserve officials misinterpreted the accumulation of excess reserves, at least in regard to Meyer and Harrison. 61. Hamlin, Papers, reel 20; Open Market Policy Conference, Minutes, April 12, 1932. 62. W. Randolph Burgess, Papers, No. 410.5. 63. The nominal value of bank loans continued to fall until the December 31, 1935 call report (Board of Governors of the Federal Reserve System, 1943, p. 18). 64. Open Market Policy Conference, Minutes, May 17, 1932. 65. Hsieh and Romer (2006) argue that the purchases ending because the objective of increasing excess reserves was accomplished. The following discussion in the text refutes this argument, showing that the purchases were deemed a failure and ended due to the loss of gold and opposition within the system. 66. Hamlin, Papers, reel 20. Years later, Meyer did attribute the failure of the open market policy to a lack of demand, Meyer, Papers, box 181, but that was not his belief at the time. 67. Federal Reserve Board, Annual Report, 1932, p. 22; Meyer, Papers, box 181; Open Market Policy Conference, Minutes, May 17, 1932; Hamlin, Papers, reel 20. The formation of the Banking and Industrial Committees reflects the belief at the time that the problem was on the supply side of lending, not a lack of demand which would be the ‘‘real bills’’ explanation. 68. Glass, Papers, box 4. 69. Burgess, Papers, 410.5. 70. Federal Reserve System, Papers, box 2448. 71. Open Market Policy Conference, Minutes, July 14, 1932, p. 3; Burgess, Papers, 410.5. 72. At this time a reallocation of shares of security purchases among the banks was proposed. It was recognized that Boston and Chicago, which had the highest gold

Eugene Meyer and the Great Contraction

303

reserve ratios, would have to take increased shares, and that both banks would strongly object. No change in allotments was made at this time (Burgess, Papers, 410.5). 73. Federal Reserve Board, Minutes, July 14, 1932, pp. 45–47; Open Market Policy Conference, Minutes, July 14, 1932, pp. 3–4, November 10, 1932, p. 1. The 1932 purchases exceeded $1 billion, by far the largest amount of purchases to that time. Meyer argued for further purchases at the December 1932 New York Federal Reserve Bank Directors’ meeting, without success Meltzer (2003, pp. 376–379). 74. Currency returned through other Federal Reserve Banks had to be promptly redeemed through the System’s gold settlement fund. Each reserve bank was required to maintain a minimum gold balance in this fund, so the return of Chicago’s notes through other banks would create a deficit position that the Chicago bank would have to cover with its surplus gold. For a description of the operation of the gold settlement fund, see Federal Reserve Board, Annual Report (1916, pp. 77–79). 75. Burgess, Papers, 410.5. Meyer believed that Chicago’s resistance came from director George Reynolds, Chairman of the Continental Illinois group of banks, based on Reynolds’ desire for higher nominal rates which he felt increased bank profits, Meyer, Papers, box 181. Keynes (1981, p. 563) commented about bankers’ desire for high rates, apparently reflecting Meyer’s view. Epstein and Ferguson (1984) argue that the purchases ended due to concerns that low nominal rates hurt member bank earnings. Meyer’s recognized that this was likely the real reason for the Chicago Fed’s opposition to purchases, but Meyer wanted to drive nominal rates as low as possible. Meyer succumbed to the internal opposition only when the purchases did not resulted in the expected increase in lending. 76. Goldsborough’s questioning reveals a sophisticated economic understanding. He felt that the problem was debt-deflation, that by the quantity theory monetary policy could be used to target a price index, and that transparent policies would be more effective than secrecy U.S. House of Representatives Committee on Banking and Currency (1932, pp. 466, 511). 77. U.S. House of Representatives Committee on Banking and Currency (1932, p. 471). Meyer made a similar statement, saying that a domestic policy may be hampered by international conditions (U.S. House of Representatives Committee on Banking and Currency, 1932, p. 534). 78. U.S. House of Representatives Committee on Banking and Currency (1932, pp. 533, 551–552) 79. Federal Reserve System, Papers, box 1438; Eugene Meyer, Papers, box 181. 80. Federal Reserve System, Papers, box 1438. 81. Federal Reserve Board, Bulletin, May 1932, p. 286; Meyer, Papers, box 120. 82. Meyer, Papers, box 181. 83. The first list was published in August 1932. Garner sought the Democratic Presidential nomination. He ultimately traded his support to run for Vice-President with Roosevelt, and served two terms in that position. 84. On October 31, 1932 Meyer wrote to one of his daughters saying ‘‘I shall be happy when the elections are over and when we can get down to doing something else.’’ Meyer, Papers, box 3. 85. Meyer, Papers, box 181.

304

JAMES L. BUTKIEWICZ

86. Daily data for the Dutch Guilder and weekly data for the Belgium belga and Swiss franc follow the same pattern of rates at or below par until the Michigan crisis, when they rise to and above the gold export point. 87. Leffingwell, Papers, box 7. 88. Federal Reserve Board, Minutes, February 27, 1933, pp. 256–258. 89. Federal Reserve Board, Minutes, February 23, 1933, pp. 231–232. 90. Meyer lamented his failed attempts to urge Hoover to declare a holiday, Meyer, Papers, box 94. 91. United States House of Representatives Committee on Banking and Currency (1932, p. 536). Although Meyer attempted to aid banks, especially nonmember banks through the RFC, he also felt the dual banking system was structurally unsound. This weakness and the collapse of real estate prices were the factors to which he attributed the banking collapse, Meyer, Papers, box 181. 92. His service with the Federal Reserve was a bitter memory for Meyer in an otherwise highly successful life. Sidney Hyman, who assisted Meyer with the writing of an unfinished autobiography, wrote that it was ‘‘painful’’ to obtain information about this period and that ‘‘EM was more sensitive about this period than any other one, and consistently baked away from it when questions were put to him.’’ Meyer, Papers, box 181. 93. While Meyer asserted that he followed Bagehot’s dictum in 1931, in the 1933 crisis Meyer ignored Bagehot’s advice of lending freely in a panic; he would only support higher interest rates to protect gold reserves. See Meltzer (2003, p. 282) for a further discussion on Bagehot and knowledge of his advice within the Federal Reserve System.

ACKNOWLEDGMENTS I thank Burt Abrams, Stacie Beck, Michael Bordo, Milton Friedman, Farley Grubb, Ken Lewis, Thomas Mayer, Allan Meltzer, Randy Parker, Scott Redenius, Larry Seidman, Evelyn Small, Frank Steindl, Paul Trescott, Eugene White, Elmus Wicker, Bill Yohe, the participants at seminars at the University of Delaware, Lehigh University and Rutgers University, and an anonymous referee for many helpful comments and suggestions. Responsibility for any errors is my own. Financial support was provided by the University of Delaware Research Foundation. Permission to quote from archival sources has been granted by Sarah Chilton, the Brookings Institution, for the Committee on the History of the Federal Reserve papers; Donald E. Graham, the Washington Post Company, for Eugene Meyer papers; Joseph Komljenovich, Federal Reserve Bank of New York, for the George L. Harrison papers and the W. Randolph Burgess papers; Courtney Smith, Butler Library, Columbia University, for the Reminiscences of Eugene Meyer; Herbert Taylor, Federal Reserve Bank of Philadelphia, for the Federal Reserve Bank of

Eugene Meyer and the Great Contraction

305

Philadelphia Board of Directors minutes; and Timothy Walch, Herbert Hoover Presidential Library, for the Herbert Hoover papers. Other archival sources are in the public domain.

REFERENCES Board of Governors of the Federal Reserve System. (1943). Banking and monetary statistics, 1914–1941. Washington, DC: Board of Governors of the Federal Reserve System. Bordo, M. D. (2006). Review of a history of the Federal Reserve. Volume I. Journal of Monetary Economics, 53, 633–657. Burgess, W. R. Papers, Federal Reserve Bank of New York. Butkiewicz, J. L. (1999). The Reconstruction Finance Corporation, the gold standard, and the banking panic of 1933. Southern Economic Journal, 66, 271–293. Chandler, L. V. (1971). American monetary policy, 1928–1941. New York: Harper & Row. Clark, E. (1940). Adolf Wagner: From national economist to national socialist. Political Science Quarterly, 55, 378–411. Committee on the History of the Federal Reserve System.(1954–1955). Eugene Meyer, Unpublished memoranda. Washington, DC: Brookings Institution. Eichengreen, B. (1992). Golden fetters. New York: Oxford University Press. Epstein, G., & Ferguson, T. (1984). Monetary policy, loan liquidation, and industrial conflict: The federal reserve and the open market operations of 1932. Journal of Economic History, 44, 957–983. Federal Open Market Committee. (1956). Open market investment policy: Excerpts – 1929 – Mid-1931. New York: Federal Reserve Bank of New York. Federal Reserve Bank of Philadelphia Board of Directors. (1931). Minutes. July 1. Federal Reserve Board. (1916, 1932). Annual report. Washington, DC: Federal Reserve Board. Federal Reserve Board. Federal Reserve Bulletin. Various issues as cited. Federal Reserve Board. Minutes. Washington, DC, National Archives II. Federal Reserve System. Papers. National Archives II. Fortune. The house of Jesse. May 1940. Friedman, M., & Schwartz, A. J. (1963). A monetary history of the United States, 1867–1960. Princeton: Princeton University Press. Glass, C. Papers. University of Virginia. Goldenweiser, E. A. Papers. Library of Congress. Hamlin, C. S. Papers. Library of Congress. Harrison, G. L. Papers. Federal Reserve Bank of New York. Hoover, H. Papers. Herbert Hoover Presidential Library-Museum. Hsieh, C.-T., & Romer, C. D. (2006). Was the Federal Reserve constrained by the gold standard during the Great Depression? Evidence from the 1932 open market purchase program. Journal of Economic History, 66, 140–176. Hyman, S. (1976). Marriner S. Eccles. Stanford: Stanford University Graduate School of Business. James, F. C. (1938). The growth of Chicago banks: Volume II. New York: Harper and Brothers.

306

JAMES L. BUTKIEWICZ

Keynes, J. M. (1981). The collected writings of John Maynard Keynes, Volume XX. In: D. Moggridge (Ed.), Activities 1929–1931: Rethinking employment and unemployment policies. New York: St. Martin’s Press. Leffingwell, R. C. Papers. Yale University Library. Meltzer, A. H. (2003). A history of the Federal Reserve. Volume I: 1913–1951. Chicago: University of Chicago Press. Meyer, E. (1961). The reminiscences of Eugene Meyer. Columbia University Oral History Research Office. Columbia University, New York, NY. Meyer, E. Papers. Library of Congress. Moreau, E. (1991). The golden franc: Memories of the governor of the bank of France: The stabilization of the franc (1926–1928). S. D. Stoller, & T. C. Roberts (Trans.). Boulder: Westview Press. Morgenstern, O. (1959). International financial transactions and business cycles. Princeton: Princeton University Press. Moure´, K. (1991). Managing the franc Poincare´: Economic understanding and political constraint in French monetary policy, 1928–1936. Cambridge: Cambridge University Press. Nash, G. R (1959). Herbert Hoover and the origins of the Reconstruction Finance Corporation. The Mississippi Valley Historical Review, 46, 455–468. New York Evening Post. (1931). Reading in Paris for laval talks. October 6, 1931. Open Market Policy Conference,Minutes, 1930–1933. Federal Reserve Bank of New York. Parker, R. E. (2002). Reflections on the Great Depression. Cheltenham, England: Edward Elger. Pusey, M. J (1974). Eugene Meyer. New York: Alfred A. Knopf. Saturday Evening Post. (1949). Money men are different now. February 26, 1949. Temin, P (1989). Lessons from the Great Depression. Cambridge: The MIT Press. Time. (1932). National affairs: Banks. May 30, 1932. Trescott, P. B. (1982). Federal Reserve policy in the great contraction: A counterfactual assessment. Explorations in Economic History, 19, 211–220. US House of Representatives Committee on Banking and Currency. (1931–1932). Reconstruction Finance Corporation. Hearings (72-1). US House of Representatives Committee on Banking and Currency. (1932). Stabilization of commodity prices. Part 2. Hearings (72-1). US House of Representatives Committee on Banking and Currency. (1935). Banking act of 1935. Hearings (74-1). US Senate Committee on Banking and Currency. (1931). Creation of a Reconstruction Finance Corporation. Hearings (72-1). US Senate Committee on Banking and Currency. (1932). Restoring and maintaining the average purchasing power of the dollar. Hearings (72-1). US Senate Committee on Banking and Currency. (1935). Banking act of 1935. Hearings, Parts 1 & 2 (74-1). Wheelock, D. C. (1989). The strategy, effectiveness, and consistency of Federal Reserve monetary policy, 1924–1933. Explorations in Economic History, 26, 453–476. Wicker, E. R. (1965). Federal Reserve monetary policy, 1922–1933: A reinterpretation. Journal of Political Economy, 63, 325–343. Wicker, E. R (1966). Federal Reserve monetary policy 1917–1933. New York: Random House. Wicker, E. R (1996). The banking panics of the Great Depression. Cambridge: Cambridge University Press.

Eugene Meyer and the Great Contraction

307

Wicker, E. R (2002). Money supply theory and the Great Depression: What did the fed know? History of Political Economy, 34, 31–53. Wigmore, B. A. (1987). Was the bank holiday of 1933 caused by a run on the dollar? Journal of Economic History, 47, 739–755. Yohe, W. P. (1982). The intellectual milieu at the Federal Reserve Board in the 1920s. History of Political Economy, 14, 583–607.

DATA SOURCES Board of Governors of the Federal Reserve System. (1943). Banking and monetary statistics, 1914–1941: Federal Reserve notes in circulation, p. 348; Foreign claims (short-term liabilities reported by New York City banks), pp. 574–575; Member bank loans and investments in 101 leading cities, pp. 144–145. Federal Reserve Board, Annual report, 1933: Monetary gold stock, p. 137 Federal Reserve Board, Federal Reserve Bulletin. (1932, 1933): Federal Reserve Bank gold reserve ratios – end of February, June and July, March, p. 183; July, p. 460; and August, p. 545 (1932); Wholesale Price Index, September 1932, p. 618; Index of Industrial Production, September 1933, p. 585. Federal Reserve System, Papers, National Archives II: Free gold, box 1229; Preliminary bank suspensions, September & October 1931, box 1690. Morgenstern, O. (1959). International financial transactions and business cycles. Dollar–franc gold export point and par exchange rate, p. 189. New York Times. (1933). Various issues. Daily dollar–franc exchange rate.