European economic integration and the labour compact, 1850-1913

European economic integration and thelabour compact, –MICHAEL HUBERMAN† AND WAYNE LEWCHUK‡†Département d’histoire, Université de Montréal, CIRANO, CIREQ, QuébecHC J, Canada‡ Department of Economics, Labour Studies, McMaster University

Globalisation was a fact of life in Europe before , but as trade sharesincreased, so did wage and employment instability. Faced by growingpressure from workers, national authorities established labour compacts –a packet of labour market regulations and social insurance programmes –that defended workers against the risks they faced in and outside thefactory. The labour compact provided workers with insurance because itcompressed wage structures. We construct an index of labour marketregulations and social insurance schemes for seventeen Europeancountries and find that the extent of the labour compact varied with thelevel of openness. We conclude that the labour compact gave workersreason to support free trade because it protected them from external risk.Contrary to the received view, globalisation before was compatiblewith state intervention. Our findings are consistent with Rodrik’s andAgell’s for the period after .

. Introduction

In Globalization and History, O’Rourke and Williamson () claimed thatthe winds of integration before carried the seeds of its own destruction.Their argument is that migration from the Old to New World and the trans-port revolution that brought cheap grains to Europe precipitated changes infactor prices and in the distribution of income on both sides of the Atlantic.Growing inequality in the United States led to controls on immigration;falling agricultural prices in Europe resulted in the reimposition of tariffs.By the interwar years, ‘globalization [had] killed itself (ibid., p. )’. Stateintervention had pushed aside markets. These were the lessons of history.But were these trends inevitable? Were alternative policies available to offsetgrowing inequality and to conserve the benefits of economic integration?

In this article we draw other lessons from the pre- years. The pointof departure for O’Rourke and Williamson is the theory of factor-priceequalisation developed by Eli Heckscher and Bertil Ohlin to help accountfor Scandinavian catch-up. But twenty years earlier, another Swedish econ-omist, Gustav Cassel, sought to understand the potential role for state

European Review of Economic History, , –. Printed in the United Kingdom © Cambridge University Press

policy in promoting economic growth. Writing at a time when the devel-opment of the Swedish welfare state was in its infancy, Cassel criticisedGerman tariff policy and Bismarckian social measures for their rigidity andargued that they were inappropriate for Sweden because of its dependenceon international trade. He believed that social policy had multiple goals.Social reform could be designed to provide workers with a degree of pro-tection that was compatible with open market policy.

Extending Cassel’s insights, we examine the association between socialpolicy and openness in Europe before . We define social policy toinclude the labour compact – the package of labour market regulations andsocial entitlements – national authorities established in the decades beforeWorld War I that were intended to defend workers against the risks theyfaced in and outside the factory. It is our contention that the extent of thelabour compact varied positively with the degree of openness. An expansivelabour compact gave workers reason to support free trade because it pro-tected them from the wage and employment instability that was associatedwith openness or external risk. The compact was a viable alternative to tariffprotection. We conclude that globalisation before was consistent withactivist social policy.

Rodrik (, ) and Agell (, ) – see also Cameron ()and Krueger () – provide theoretical support for the relationshipbetween openness and social policy. These authors have argued that in theperiod after , increased exposure to international trade led to greaterwage and employment instability, which in turn forced national authoritiesto extend their safety nets. While Rodrik stressed the relationship betweeneconomic integration and the instability of labour-market outcomes, Agellemphasised the role of what he called ‘labour market institutions’ in com-pressing wage structures, thereby providing workers with a degree of socialinsurance. Summarising these views and echoing Cassel, Rodrik (, p.) remarked that markets and governments are not substitutes, but thatthere exists a ‘degree of complementarity’ between the two.

Our contribution to this line of reasoning is mainly empirical. For European countries, we construct a Labour Compact Index that incorporates different types of labour market policies, from the date of the first factory actto the creation of unemployment insurance. We then examine the relationshipbetween openness and the LCI. The empirical correlation we find is robustbecause: () it is unaffected by the inclusion of other control variables; () itholds for various measures of government intervention; () it exists for measuresof government intervention drawn from other data sets; () it prevails for bothhigh and low income countries; and () it is not an artifact created by outliers.

This article is organised as follows. Section surveys contemporary opin-ions and more recent treatments of the relationship between openness and

European Review of Economic History

We thank Jonas Agell for sharing with us his knowledge of Cassel.

social policy. There is much recent research on openness and the origins ofthe welfare state, but there is no attempt to draw these two phenomenatogether. Section summarises the intuitive reasons why we would expectto see a link between openness and social policy, and we introduce otherapproaches to trade policy, mainly from political science, that we draw uponin our statistical analysis. In Section , we construct the LCI and provide apreliminary examination of the data. Section is devoted to multivariateanalysis of the relationship between openness and social policy. A conclud-ing section situates our findings in the larger debates on globalisation.

. European economic integration, standards of living andsocial policy

The first great wave of global economic integration and the first stepstoward the creation of the welfare state coincided in the thirty years or sobefore World War I, but despite their simultaneity these two phenomena aremost often treated separately. In fact, economists and historians who havestudied these two phenomena take somewhat opposing positions.According to one school, globalisation led to improvements in Europeanworkers’ living standards. This view ignores the demand for social policy.But others, giving emphasis to domestic factors, have traced the origins ofsocial policy to the deterioration of living and working conditions during thesame period. This view ignores the role of globalisation. In many ways,these two approaches seem to be talking past each other. Contemporariesthought differently. They saw a direct link between openness and thedemand for social policy. The objective of this section is to survey the dis-tinct approaches to globalisation and the rise of the welfare state and, fol-lowing up on the insights of contemporaries, suggest ways in which thesetwo phenomena may in fact be connected.

.. European economic integration and the demand for social reform

Exploiting a newly constructed real wage series, O’Rourke and Williamsonhave provided new benchmark estimates of the effect of openness on stan-dards of living and the distribution of income around the world between and . The transport and grain revolutions that forged the inte-gration of Old and New Worlds affected European countries unequallybecause of different employment and demographic structures and trade

European economic integration, –

There are exceptions, notably the work of political scientists. We discuss theircontributions in the next section.

The wage series is from Williamson (). For other contributions, see Williamson(), Williamson (), Taylor and Williamson (), Lindert and Williamson(), and O’Rourke ().

policies. O’Rourke and Williamson observed that workers in countries thatremained open fared better than elsewhere. Relative to Great Britain, wagegrowth was particularly rapid in Scandinavia, moderate in France, Belgiumand Germany, and slow in Spain and Portugal. The distribution of incomealso narrowed in northern Europe. This analysis gives little place to thedemands for social policy because it presumes that the market provided foreverything. Government intervention in the Old World was restricted totariff policy. Although tariffs may have been positively correlated withgrowth (O’Rourke , Clemens and Williamson ), they acted tolimit real wage gains and to dampen changes in the distribution of income.

O’Rourke and Williamson were mainly concerned with Old and NewWorld exchanges and the resulting movement in factor prices across conti-nents; they (O’Rourke and Williamson , p. ) wrote that although‘[c]onvergence did take place in Europe, it was a more modest affair’. Butif we tilt the focus to forces promoting convergence within Europe – thatwent hand-in-hand with globalisation – a different perspective on workerwelfare emerges. Consider first the levelling effect of migration on wages.Population movements within Europe were considerable: from to about per cent of Italian migrants chose European destinations (Rosoli, Hatton and Williamson ); elsewhere, Belgians travelled to Franceto work, and Poles to Germany. To be sure, for some countries like theNetherlands cross-border population movements were negligible, and onthe whole Belgium recorded a net inflow of migrants (Hatton andWilliamson , p. ). For these countries as elsewhere, because tradecan operate as a substitute for factor mobility, intra-European exchangesmay have impacted on wages and working conditions in a way that chal-lenges O’Rourke and Williamson’s rather benign view of European econ-omic integration.

Table gives estimates of the value of exports and imports as a share ofGDP for a sample of European countries. The average trade share increasedby nearly per cent in the period, with most of the change occurring after. Trade between countries with similar factor endowments and tech-nologies dominated exchanges. In , about per cent of Belgium’strade (exports plus imports) was with France and Germany; per cent ofDenmark’s trade was with Norway and Sweden; and for Britain, trade withFrance, Germany and the Netherlands was about per cent of the valueof its trade with the US and India (Mitchell ). Exchanges of this typemade the demand for labour very responsive to changes in its price (Rodrik, pp. –). As continental markets became more integrated, employ-


In discussing social spending before , Lindert and Williamson (, p. ) wrotethat ‘safety nets were not yet in place’.

Intra-European exchanges maintained their share in the world trading system from to (Kenwood and Lougheed ).

ers and final consumers substituted foreign European workers for domesticworkers (this had the same effect as increasing the domestic labour supply),either by investing abroad or by importing the products made abroad – andwages and working conditions for comparable groups of workers across thecontinent would have been driven down. The established trading networksexacerbated the degree of external risk. Certain economies had becomedependent on a restricted number of trading partners, and this meant thatmany of the continent’s workers were highly susceptible to changes in com-mercial policy and faced uncertainty because of it. The Netherlands’ tradeshare failed to increase over the period because its most important tradingpartner, Germany, had brought in tariffs. This was also the case forSwitzerland. Denmark got off easier; the UK was the main destination of itsexports.

The upshot is that movements of people and goods within Europe on theone hand, and exchanges across continents on the other, may have had off-setting effects on worker welfare. The net result may have spurred workersto seek new forms of social protection. Recent research has pointed to theseconflicting trends. While real wages increased rapidly in certain countries,working conditions may have not improved to the same extent. In fact,


In , Dutch exports to Germany were nine times the value of its trade with Belgiumand three times that with the UK (Mitchell ).

Table . Imports and exports as a share of GDP.

Austria-Hungary . () . () . ()Belgium . () . () . ()Bulgaria . () . () . ()Denmark . () . () . ()Finland . () . () . ()France . () . () . ()Germany . () . () . ()Italy . () . () . ()Netherlands . () . () . ()Norway . () . () . ()Portugal . () . () . ()Russia . () . () . ()Spain . () . () . ()Sweden . () . () . ()Switzerland . () . () . ()United Kingdom . () . () . ()European average . . .

Note: Rankings in parentheses.Sources: See Appendix.

Huberman and Lewchuk () found that in Denmark and Sweden until work hours remained long, between eleven and a half and twelve hoursper day on average. Huberman and Lewchuk also reported that across thecontinent work hours were longer than average in sectors, like textiles,which faced considerable international (that is, European) competition.

Other indicators give credence to the view that wage gains may not haveoffset the poor working conditions and the wage and employment instabilityof the new economic order. For Belgium, which according to O’Rourke andWilliamson saw moderate wage growth, Scholliers () found onlymeagre improvements in family budgets and fluctuations in the price ofbread were a source of great distress in the mid s; Alter () reportedthat the proportion of children at work in Belgium was the same in the earlys as it was at mid-century. For France and parts of Germany, new evi-dence on heights suggests that living standards and working conditions werenot improving at any great rate (Weir , Twarog ).

.. The rise of the welfare state in an open economy framework: somepossible links

For the most part, historians of the welfare state have not made much of theconnection between increased exposure to international trade and the risein workers’ demands for improved working conditions and some basicamount of insurance. Comparative histories of the rise of social policy tendto begin with Bismarck who first introduced compulsory accident andhealth insurance and old-age pensions in the s. The UK is believed tohave followed the German model after Lloyd George’s visit there in (Hay and Harris ). Like a contagion effect, other European coun-tries adopted similar measures (Kohler and Zacher ). A different pic-ture surfaces, however, if we broaden the definition of social policy toinclude both social entitlements and labour market reform – the labourcompact nations provided workers.

We have assembled a data set for countries in order to trace the devel-opment of social reform broadly defined and to assess its uniformity across


For work hours in , across countries and by sectors, see the British consular reports:UK Parliamentary Papers (PP) LXVI, PP LXVIII, PP LXII. Thereports were superseded by the Abstract of Foreign Labour Statistics publishedoccasionally before World War I and the Commission on Labour, Foreign Reports. See,PP XCII, PP LXXIII, PP CXIII, PP – XXXIX. For , thereports indicate that in Belgium the average work week was hours; Italy, hours;France and Germany, ; Spain, ; Denmark and the Netherlands, ; UK, . Thereis other evidence of long work hours in Scandinavia after . Denmark – UK PP– XXXIX, Part II, p. ; Finland – Heikkinen (, pp. –); Sweden –Tegle (, pp. –). We owe the latter reference to Lars Svennson. We leave a moredetailed examination of work hours for a future paper.

Europe. We have divided reforms into two main groups: labour marketregulations that influenced the supply of labour and regulations that affec-ted the conditions under which work was performed (Table ); and socialinsurance programmes that protected workers during periods when theywere unable to work (Table ). In constructing Table , wherever possiblewe used the standards of the Final Protocol of the International Conferenceon Labour in Factories and Mines held in Berlin in , which outlined amodel labour code that was to be the basis of a late nineteenth centuryEuropean social charter. We do not set any standards for factory inspectionor the introduction of factory acts reported in the last two columns of thetable. By , all countries had some form of factory inspection and mostcountries had put in place the years old standard for children’s work.Women were left unprotected in more than half of our sample.

Table gives the dates for the introduction of social insurance and acci-dent compensation schemes. Less than half of our countries had intro-duced some form of old age and unemployment insurance before . Wehave not included the Poor Law in our tabulation. In Western Europe, PoorLaw expenditures as a share of national product were declining from on – by they represented as little as . to . per cent of GNP. Table does not fix standards for social expenditure because of the wide variabil-ity of these measures and how they were funded. In Germany, sicknessinsurance coverage was compulsory with penalties falling on employers whofailed to make sure their workers were covered. Workers and employersshared the payment premiums. Denmark had a voluntary system, with high


Our first step was to track down when these reforms were introduced. This was notalways straightforward. In some cases our sources provided contradictory information.Where these conflicts could not be resolved, we gave priority to information found inofficial publications; when official reports were themselves in conflict, we assumed thatchange occurred mid-way between the last two dates identified. Sources consulted: factoryacts, hours of labour for young workers and women, age of admission to factory work, ageat which night work permitted, night work for women – Brooke (), Fraser (),IALL (), UK PP LXXIII, p. . Factory inspection – IALL (), ILO ().

The Final Protocol recommended that children under years of age not be allowed toperform factory work; women should be not permitted to work at night; and that thelength of the working day for children under years be set at hours and that forwomen, hours. Table provides the dates when national legislation embodied thesevalues. UK Commercial (), ILO (), Shotwell (), Troclet ().

Sources consulted: old age pensions, sickness insurance, unemployment insurance –Association Internationale (), Epstein (), Industrial Relations Counsellors(), Commons and Andrew (), Rimlinger (), Blaustein and Craig (),Fraser (), Ambrosius and Hubbard (). Accident compensation – Droz (),Bellom (), Villard (), US Department of Labor (, ), Ambrosius andHubbard ().

This trend reflected, according to Lindert (), the shift in the balance of politicalpower from labour-hiring rural landlords to urban employers who wished to reduce theirtax burden.

E

uropean Review

of Econom

ic History

Table . Labour market regulation in Europe until .

Minimum Ten hour Night work Night work Eleven hour Introduction Introduction of age working day children women working day of factory first factory acts

youths prohibited prohibited women inspection

Austria Belgium * e * Bulgaria * – e –Denmark * * Finland * – * * –France Germany Hungary e * Italy * * Netherlands Norway e * Portugal * * e * Russia * * * Spain * * e * Sweden e * Switzerland * United *

Kingdom

Notes: *Indicates did not enact such a regulation. – Indicates information not available. e Indicates mid-point estimate.Sources: See text.

levels of government subsidies and relatively widespread participation.Belgium’s voluntary system lay at the other extreme; it was actuariallyunsound and, with little government subsidy, membership stagnated rela-tive to the other funds. The French system fell somewhere in between interms of subsidy and membership. Unemployment insurance schemeswere equally varied. In all countries, coverage was limited to selected groupsof workers, and only in the UK was it compulsory.

Tables and suggest divergence as opposed to convergence in thespread of the labour compact. The UK and France sought to regulate fac-tory conditions and limit the work time of women and children; the north-ern European countries favoured social entitlements. Germany seems to bethe exception and not the rule. It introduced some social entitlements –excluding unemployment insurance – at an early date and its regulation offactory conditions was lax compared to the UK and France (Hentschel, Kemp ). How have historians explained these variations? Theyhave given preference to domestic factors, including the extension of the


Murray () reviews these different schemes.

Table . Social insurance in Europe until .

Accident Unemployment Sickness Old age compensation insurance insurance insurance

Austria * *Belgium /(V) (V)Bulgaria * * *Denmark (V) (V) Finland * * *France (V) (V) (V)Germany * Hungary * *Italy * (V) (V)Netherlands * Norway (V) *Portugal * * *Russia – * – *Spain * * *Sweden * (V) Switzerland * (V) *United

Kingdom

Notes: *Indicates did not enact such a regulation. – Indicates information not available. V stands for a voluntary scheme.Sources: See text.

franchise, levels of unionisation, and the demographic and occupational dis-tributions of populations. Exploiting an international data set covering theperiod to , Lindert () undertook a comprehensive study ofthe determinants of social spending, defined as expenditure on unemploy-ment relief, old age pensions, health care and housing. He found that thepercentage of the adult population eligible to vote, Protestantism, and anageing population were determinants of spending, while income levels andthe level of development were poor indicators.

Lindert’s own research suggests that other factors may have also been atwork. Except for age, Lindert does not specify the changes in the riskenvironment faced by workers. As for the differences in programmesbetween countries, it would be difficult to attribute these variations solely tothe extent of the male franchise, which by was roughly similar fornorthern and western European countries. Indeed, the UK which had per-haps the most extensive labour compact also had a relatively low proportionof adult males eligible to vote. We speculate that increased exposure tointernational trade – as Table demonstrates – was a fundamental changein the risk environment in the period, and that differences in the observedpattern of social protection were the result of different degrees of openness.The northern European countries may have offered more social entitle-ments because they were more open, and not only because of their domes-tic politics or social and population dynamics.

That said, there were other ways in which political representation mayhave mattered for the relationship between openness and social policy.While Germany had granted universal male suffrage, it would be difficult tocall it a democracy because its executive power was independent of the leg-islative assembly. Lindert does correct for this. But along with the extensionof the franchise, certain European states introduced new systems of rep-resentation, most famously electoral systems based on proportional rep-resentation (PR). Political scientists like Rogowski () have observedthe correlation between PR and trade shares. Thus, new forms of rep-


As Lindert () noted, the positive effect of an ageing adult population on socialtransfers may simply have reflected increased out-migration in the period before theGreat War.

The percentage of the adult male population eligible to vote in in Belgium was.; Denmark, .; Finland, .; France, .; Germany, .; Italy, .;Netherlands, .; Norway, .; Sweden, .; Switzerland, .; UK, .. Source:see Appendix.

The countries and dates of the introduction of proportional representation are found inthe Appendix.

The point here is not that countries chose PR to increase their trade shares, but that theintroduction of PR had the unintended consequence of increasing trade. Theintroduction of PR in Belgium was the result of political negotiation between Catholicand Liberal parties; it had the result of keeping Belgium open to trade because workerscould participate in policy formation. See Section ..

resentation may have increased external risk, but we also know that PRregimes permit workers and other large constituencies who may not bemembers of the ruling party to have their voices heard in the formulation ofsocial policy. Thus, in an open economy, social reform may be as muchrelated to the type of representation as to the extension of the franchise. Weexplore this possibility further below.

.. Contemporary insights into the relationship between openness andsocial reform

For contemporaries globalisation was a fact of life. Consider the case ofBelgian workers who around the time of the social upheavals of the sestablished boulangeries socialistes, bread co-operatives, that tied their welfaredirectly to imports of American wheat (Witte and Craeybeckx , p. ).The co-operatives they founded were intended to stabilise expenditures onbread, clearly a large part in workers’ household budgets (Scholliers ).Although they actively encouraged trade, leaders of the co-operative move-ment and the Parti ouvrier belge recognised that working class institutionsoffered only partial and incomplete insurance against the risk of increasedinternational exposure (Leboutte et al. ). In the case of the textileindustry, Belgian work hours were the longest in Europe and reformerslinked poor working conditions to international competition. Despite thewidespread belief that long hours were unhealthy, individual firms could notbe counted on to reduce collectively the length of the workday. Workerscould not negotiate these conditions themselves; state intervention, theyargued, was their best recourse. But the degree of Belgium’s internationalexposure – it was the greatest in Europe in – also affected the natureof workers’ demands. Alongside restrictions on work hours, there was apush for unemployment insurance to support workers faced with theprospects of job loss due to sudden changes in foreign markets. In the end,the Parti ouvrier belge backed free trade (Bairoch ), because it was ableto use its support for trade to gain a larger influence in designing social pro-tection (Kossmann ).

Like workers, late nineteenth century employers recognised the connec-tion between openness, the standard of living and the demand for socialreform, but in their eyes the three dimensions represented a type of‘trilemma’. Rising wages and access to the international markets wenttogether; and along another dimension, reform in a closed economy was


This argument was common among European reformers (Rist ) into the interwarperiod. Hicks () gave their arguments theoretical support in his classic study.

Trade unions and social democrats in Germany, Italy and the UK supported free trade(O’Rourke ). The German case is interesting because it points to the fact thatworkers were not satisfied with Bismarck’s social policy which was seen as acomplement of his tariff policy.

also possible. But in an open economy, wage gains and reform were irrec-oncilable. If employers could not offload their contributions to social enti-tlements and other programmes, the labour compact would lead to a loss ofcompetitive advantage, a balance of payments crisis and inevitably to down-ward pressure on wages.

National authorities had to balance employers’ concerns against workers’demands for reform. One option they tried was the creation of a widerEuropean social union that fixed standards for working conditions acrossthe continent. The alternative, they feared, was a race to the bottom, that isto the lowest standards of social insurance and factory regulation prevailingin Europe. The Swiss and then the Germans spearheaded the early move-ments to set international standards, but at the Berlin conference of even the developed countries could not find common ground (Rolin-Jaequemyns , Delevigne , Alcock ). There was only agreementon the content of phosphorous in matches. After , it was left to unof-ficial and independent organisations of reformers and workers, often at anindustry level, to exert pressure on national authorities in order to avoid theinevitable ‘race to the bottom’. Although multilateral agreements were dif-ficult to negotiate, states did sign bilateral arrangements. In , France,concerned about the guarantees it had made to its workers owing to aninflux of Italian migrants, sought to establish a labour treaty with its neigh-bour. It gave Italian workers in the hexagon the same level of benefits thatFrench workers received, and in exchange the Italian government intro-duced in the peninsula some measures of the more advanced French labourlegislation (Follows , p. ).

In sum, despite their simultaneity, the rise of social policy and the trendtoward international market integration have been treated as distinct devel-opments. Historians have analysed the rise of the welfare state in a closedeconomy framework; economic historians of globalisation have tended tosee tariff policy as the sole option available to governments. Contemporariesbelieved otherwise. Workers demanded that openness be met with socialpolicy. Faced by the failure of international accords, national authoritieshad to devise policies that reconciled the interests of workers and employ-ers. Where this was achieved, to what extent and how is the subject of theremainder of this article.

. The labour compact and openness: some formalhypotheses

There are two aspects to the link between openness and social policy. Thefirst is the relationship between the demand for social protection and


Yet, according to Follows (), Berlin did influence the Danish and Austriangovernments’ decision to forbid Sunday work.

exposure to trade, and the second is the role of the labour compact, broadlydefined, in providing this protection.

Rodrik (, ) treats the first aspect. His starting point is the obser-vation that the current debate about globalisation has tended to focus onNorth-South flows, giving less attention to exchanges among countries withsimilar factor endowments. But in the period after – as was the case inthe period before – trade between countries that look alike hasremained important. These types of exchanges have the effect of flatteninglabour demand curves and lead to a greater instability of labour-market out-comes. ‘Shocks in labor demand – caused, for instance, by a suddenincrease or decrease in labor productivity – result in much greater volatilityin both earnings and hours worked’ (Rodrik , pp. –). Increasedvolatility gave rise after to the demand for a safety net, or what Rodrikcalled ‘government consumption’ to smooth income flows.

Agell (, ) describes how the labour compact provides protectionagainst uncertainty. He defines labour market institutions to include theprovision of social insurance, and the enactment of regulations like mini-mum wage and factory legislation that affect the supply and demand forlabour. These institutions act as an insurance mechanism because they tendto compress wage structures. In the absence of perfect capital markets, therepresentative worker is willing to exchange a lower expected wage for awage structure that offers insurance against uncertainty concerning ‘who-one-will-be in the wage distribution’ (Agell ). Unions through collec-tive bargaining can negotiate this type of protection, but Agell shows thatgovernment intervention is preferred because it can deliver the sameamount of protection with smaller amounts of unemployment.

Social protection can take many forms. Social entitlements give workersdirect support against the vicissitudes of trade. Labour market regulationsalso provide insurance. In the context of the nineteenth century, lawsrestricting the labour supply of women and children and limiting the lengthof the workday raised the wages of unskilled male workers relative to aver-age wages. Compressed wage structures assured male workers that therewas some minimum level below which their standards of living would notfall. Factory regulation also provided insurance because it was universal.Before national authorities intervened, the provision of above-average work-ing conditions was restricted to paternalist employers who were scatteredover the countryside and industrial regions. National regulations gaveworkers assurance that their ‘rights’ to these conditions were transferable,and hence their income opportunities and mobility increased.Notwithstanding these features, even modest amounts of government intervention have redistributive effects and reduce the variance of living


In Agell’s model, unions’ interference with relative factor prices results in greater levelsof unemployment than that produced by redistributive government policy.

standards. Sinn () argued that, like the provision of public goods, theentire system of redistributive taxation involves some element of risk sharingand social insurance. Thus even nineteenth-century factory inspection pro-vided for income smoothing because it was paid for by taxes that deviatedfrom the benefit taxation criterion and made the rich the net contributorsand the poor the net receivers of economic resources.

Holding levels of openness constant, not all national authorities are likelyto react equally and provide the same degree or type of social protection.Economists analysing the determinants of growth (Alesina et al. ) haveobserved that more polarised countries tend to invest less in public goodssuch as schools. To Katzenstein (, p. ) and also Gourevitch (),country or population size as opposed to its linguistic or ethnic fragmenta-tion is the primary determinant of government spending. In historical con-text, small European states were best suited to provide social protection,because of their ‘economic flexibility and political stability’ (Katzenstein, p. ). Small states promoted democratic corporatism and its institu-tions – including labour market institutions – furnished them with an infra-structure to weather the ‘open and vulnerable’ nature of their economies.

Many of the small European states also have forms of proportional rep-resentation whose origins, we have noted, can be traced to the years beforeand immediately after World War I. For the period after , Rogowski() – and see also Mansfield and Busch () – proposed that it is pol-itical institutions and not size that matters in the relationship between open-ness and social policy. Polities with PR are most likely to support opennessbecause representatives of large electoral districts are more insulated fromregional and sectoral pressure. After reviewing a number of key episodes –and contrary to the received wisdom – Rogowski found that in PR systemspolitical parties tend to be ‘more disciplined, more coherent and more ideo-logical’ and more secure against sudden shifts in voter sentiments. PR


For most of the countries in our data set, public finances before conformed to themodel of redistributive taxation, even in the UK, France and Germany (Schremmer).

The idea here is that it is difficult to agree on public expenditures in fragmentedsocieties.

Rogowski (, p. ) summarised the main idea: ‘When automakers or dairy farmersentirely dominate twenty small constituencies and are a powerful minority in fifty more,their voice will be heard in a nation’s councils. When they constitute but one or two percent of an enormous district’s electorate, representatives may defy them more freely . . .The smaller the size of each constituency, the more homogeneous is each district, thefewer is the number of special interests that are likely to exist per constituency, and thegreater will be the political influence of each pressure group in that district. Smallelectoral districts encourage patronage and pork-barrel politics . . . [and] yield tradepolicy that covers more types of goods and services than would be the case in a countrycharacterized by large electoral districts and less influential interest groups.’ Largeconstituencies make PR more exact.

builds trust among political parties of diverse stripes. Excluding extremistfactions, various combinations of political parties can come together at anytime to set the policy agenda, but to do so the parties need to demonstratetheir willingness and ability to be reliable, if not trustworthy, partners.Where workers participate in PR, they can form coalitions with liberal orcentrist parties that support free trade. Because of the long-run stability ofPR, workers in these circumstances can exchange their conditional supportof trade for improved social protection.

These insights from economics and political science suggest a number ofhypotheses that we will draw upon to study the link between globalisationand the labour compact before .

() For all countries it is expected that government interventionincreased with the degree and risk of openness. This correlationshould be robust across income levels and hold for differentmeasures of labour market intervention. The type of interven-tion would have varied. Where degrees of openness weregreater, we would expect to see more direct forms of protection.Labour market regulations would be more prevalent in coun-tries where the extent of openness was smaller. We can also testthe related hypothesis that social policy, because it stabilisedwages and employment, was an alternative to tariff protection.If the labour compact varied with openness, and because tariffprotection moved inversely with trade, we would expect to finda negative relationship between the labour compact and tariffs.

() At a given level of openness, the level and nature of protectionwould have varied with social and political factors. Populationsize and its fragmentation would be negatively correlated withgovernment intervention. Katzenstein believed that the distinc-tion between small and large countries took shape in the tumultof the interwar years. But many of the institutional and histori-cal conditions that he thought relevant for the s were wellin place in the years before . Does the relationship betweensmall economies and labour market institutions hold for theearlier period?


We benefited from discussions with Peter Katzenstein and Bo Rothstein on this point. In this article, we limit ourselves to the relationship between openness and labour

market intervention. We leave aside the ancillary hypothesis that labour market policycompressed wage structures. In passing, we note that there is some support for thishypothesis. For the Netherlands, Smits and van Zanden () observed that in theperiod before , the gap between skilled and unskilled wages closed. Heikkinen() reported a similar finding for Finland. O’Rourke and Williamson’s evidence ofcompressed distributions of income in continental countries that remained open is alsoconsistent with the hypothesis advanced here that state intervention in the labour marketincreased in response to openness.

() Most European countries had expanded the male franchise by. In certain histories, the rise in the percentage of eligiblevoters was a sufficient condition for the expansion of the welfarestate. But not all states were alike in their political represen-tation. Taking our lead from Rogowski, we consider whetherworkers in PR systems could use their newly won political cloutto gain larger safety nets.

. The Labour Compact Index: preliminary analysis

Our data set is composed of published and unpublished sources. TheAppendix gives the published sources for the macroeconomic, social,political, demographic and geographic variables in the data set.

Wherever possible we have used the most widely available sources. TheLabour Compact Index posed a different set of problems. Lindert() provides estimates of social expenditure for a sample of Europeancountries for to . As a share of national product, these expen-ditures were small; Denmark was the most munificent country inEurope, spending . per cent of its national product on social trans-fers. These figures underestimate the degree of intervention in thelabour market before . Following Agell, labour market legislationenforcing who could work and when also provided workers with somedegree of insurance.

.. Building the Labour Compact Index

Our first step was to combine information we had on the origins and natureof labour market reforms to allow for comparisons with trade shares acrosstime and space. We summarise the data in a Labour Compact Index (LCI)given in Table . The LCI exploits the same methodology underlying theUnited Nation’s Human Development Index. Each country’s LCI is acomposite of the equally weighted labour market regulations and social


The ethnic variable is the most worrisome. Because of data availability, we wereforced to use figures on ‘ethnolinguistic fractionalisation’ for . This data series isfrequently used in studies of growth after . Its applicability to the period before may be dubious. The assumption here is that the language mix of particularcountries in our data set did not change much or that all countries changeduniformly. Consider Belgium whose linguistic divisions were about the same in as ; or the UK and France where a considerable portion of migrants after came respectively from Commonwealth and North African countries speaking Englishand French. Having said that, in statistical analysis the ethnic variable was notsignificant.

Peter Lindert graciously provided us with his international data set on education andsocial spending. Wherever possible, we checked our figures against Lindert’s.

Crafts () gives a description of the methodology.

insurance entitlements. For continuous variables in the LCI, benchmarkminimums and maximums were defined. The minimum represented stan-dards of working conditions and social entitlements around . Themaximum represented conditions found in under the most progressivelegislation or those set by the Berlin Conference on Labour.Admittedly the choice of endpoints is arbitrary (thus an argument could bemade that the minimum age of child workers was actually below eightyears). Having said that, this approach allows us to establish the distancecountries had travelled from a state of unregulated working conditions tothe creation of a national labour compact. For discrete variables, such as theadoption of factory legislation, signifies that the regulation was in place; not in place. No judgement was made on the effectiveness of factory inspec-tion or on coverage. Thus Britain is given a value of in for its FactoryAct of , even though it covered only children in textiles. With regard tosocial entitlements, no attempt was made to give precise monetary values tonational programmes; signifies the programme was not in place, and represents the maximum. We say more about our valuation of these pro-grammes below.

Minima Maxima

Labour market regulationsand work conditions() Minimum age children allowed to

work in manufacturing establishments() Maximum length of working day

of young workers() Age at which night work allowed () Women excluded from night work () Length of working day for women () Factory Act adopted () Factory inspection in place

Social insurance entitlements() Accident compensation () Unemployment insurance

() Sickness insurance () Old-age insurance


We recognise the arbitrary nature of our choice of components and weights. Amongother measures, we have ignored minimum wage and trade union laws, as well as lawsgoverning sweatshops. Our choice of components was ultimately dictated by theiravailability for as large a data set as possible. We chose equal weights because, in theabsence of ascribing monetary values, we had no basis on which to give preference toone component of the labour compact over another.

We then calculated an index measuring in percentage terms the progresseach country had made between the minimum and the maximum. TheAppendix gives values for the continuous variables in the key years we usedin our calculations. To illustrate, in the Swiss established as theminimum age for children to work at night. Using the minimum and maxi-mum values we established, the Swiss earned a score of ( – )/( – ) �. for this component in .

In calculating the indices for the provision of entitlements, we assigned avalue of . where authorities introduced partial and/or voluntary plans forsick, unemployment and pension benefits, and accident compensation, as aprecursor to more comprehensive entitlements. A number of countriesincreased employer liabilities in the case of accidents before introducingstate administered compensation schemes. These intermediate plans werealso valued at .. Some examples will help here. The British compulsoryunemployment insurance plan of was given a value of even thoughits coverage was narrow, because contemporaries believed that it was themost comprehensive and generous system in Europe (Harris , p. ).The story in France is more complicated. Unemployment relief was apatchwork of private, religious and local charity, with financial aid from thecommunes, departments and the state; it got a index value of . ((. – )/( – )) for this social entitlement in . In , France put inplace a compulsory pension scheme for workers earning below a certainamount (its original scheme of was voluntary). With three and halfmillion workers covered in (Kemp , p. ), France received ascore of for old age pensions in that year.

Table indicates that the introduction of the labour compact wasuneven. In only two countries, Switzerland and Austria, had any typeof regulation in place. By France, Germany, Hungary and the UK hadjoined them. Thereafter, the UK took the leadership. The period from to saw little change in terms of the number of countries regulating theirlabour markets and the extent of regulation in each country. Switzerlandinitiated the next wave of regulation in the s; the mantle of changepassed to Germany which by had the most comprehensive body oflabour market regulations and social insurance entitlements. The UK didnot match these improvements and by ranked only third. The exten-sion of the labour compact was rapid in Scandinavia after , and by, Denmark, Norway and Sweden provided more expansive labourmarket protection than did Belgium. The latter did catch up in the periodafter . For all of Europe, our measures of convergence (bottom panelof Table ) suggest that some uniformity in the regulatory environmentoccurred only at the end of the period. Between and the gapbetween the most and least regulated nation grew and the interquartile


This meant that we dropped the standards underlying Table .

European econom

ic integration, –

Table . Labour compact index.

Austria . () . () . () . () . () . () . () . () . ()Belgium . () . () . ()Bulgaria . ()Denmark . () . () . () . ()Finland . () . () . ()France . () . () . () . () . () . () . ()Germany . () . () . () . () . () . () . () . ()Hungary . () . () . () . () . () . () . () . ()Italy . () . () . ()Netherlands . () . () . () . ()Norway . () . ()Portugal . () . ()Russia . () . () . ()Spain . () . () . () . ()Sweden . () . () . ()Switzerland . () . () . () . () . () . () . () . () . ()United . () . () . () . () . () . () . () . ()

KingdomMax-min . . . . . . . . .St. dev. . . . . . . . . .Mean . . . . . . . . .Coef. var. . . . . . . . . .Interquartile . . . . . . . .

Note: Rankings in parentheses.Sources: See text.

spread increased as well; the coefficient of variation only began fallingtowards the end of the century.

There was variation across the type of insurance. Table decomposes theLCI into two sub-components: a labour regulation index (LRI, categories to above) and a social insurance index (SII, categories to ). In termsof labour regulations, Switzerland was the leader in ; the UK’s relativeposition steadily eroded over the years, that of France improved, andGermany’s was stable. While the northern European countries had rela-tively poor labour market regulations, their social insurance systems weremore advanced. This was clearly the case for Belgium, Denmark and theNetherlands, whose systems of social entitlements were as good as any-where else. Was this a difference between small and large countries, PRversus parliamentary systems, or open versus closed economies?

.. Preliminary analysis of the LCI and its components

To fix ideas, consider again the histories of social reform in Britain andGermany on the one hand, and the smaller northern European states on theother. Inspired by liberalism, the UK was anxious to provide a frameworkfor stable labour markets, but was reluctant to restrict the right of individ-uals to enter into contracts. Starting in the s and s, individualswere given new rights to protect themselves through collective bargaining;those deemed unable to defend their self-interest, including women andchildren, were offered special protection (Low , Ogus ). The adultmale worker was expected to fend for himself and to make his own provisionfor unemployment, old age and accident insurance. The UK moved earlierthan almost any other European nation to regulate the employment ofwomen and children, but during the s and s British officialsrejected the Swiss proposal to set limits to the working day for men acrossEurope. The UK was also slow to adopt advances in social insurance pion-eered elsewhere in the s. In , the replacement of accident com-pensation by a compulsory insurance scheme marked the first meaningfulregulation that infringed on the rights of male adults to make their owndeals with their employers.

German protective legislation was shaped in the course of a politicalstruggle that pitted the conservative land-owning class against the mostpowerful socialist party in Europe. In an attempt to find middle groundand to consolidate the power of the state, Bismarck introduced a mix of


Our measure of the interquartile spread is the value of the th centile minus the valueof the th centile, divided by two.

To construct the sub-indexes in Table , we follow the same procedures as for the LCIin Table .

This paragraph is based on Canning (), Machlan (), Rimlinger (), andZollner ().

labour market regulations and social insurance programmes whose evol-ution was almost the mirror image of that found in the UK. In the s,Bismarck focused on social insurance schemes because he saw them as themost effective way to build and maintain the loyalty of the working class tothe monarchy. Bismarck believed that legislation regulating working hoursand work conditions would only increase the visibility of trade unions.Germany would only move towards labour market regulations in the lastdecade of the century as labour strife and socialist agitation continued tochallenge the government. In , Wilhelm II, prompted by a major strikein the coal mining regions the previous year, announced a ‘new course’ thatincluded comprehensive labour market regulations on the British model, ahalf century after they became the norm in the UK.

By , the UK and Germany had converged, despite their different tra-jectories over most of the previous century. Still, throughout the period,they did share a strong predisposition to limit the use of tax revenues to payfor reform. In Bismarckian social policy, outside of old age pensions, thestate’s contributions were minimal. True to his Junker origins and smallemployer background (Rimlinger , p. ), Bismarck was reluctant to


Table . Indices of labour market regulation and social insurance.

Labour regulation index Social insurance index

Austria . () . () . () . () . ()Belgium . () . () . () . ()Bulgaria . () . ()Denmark . () . () . () . ()Finland . () . () . () . ()France . () . () . () . () . ()Germany . () . () . () . () . ()Hungary . () . () . () . ()Italy . () . () . () . ()Nether- . () . () . () . ()

landsNorway . () . () . () . ()Portugal . () . () . ()Russia . () . () . ()Spain . () . () . ()Sweden . () . () . () . ()Switzer- . () . () . () . () . ()

landUnited . () . () . () . () ()

Kingdom

Notes: Rankings in parentheses.Sources: See text.

use state funds, an inclination that underlay his decision not to introduceunemployment insurance at an early date. Accident compensation was runby the state as an insurance system rather than an assistance programme(Rimlinger , p. ). In the UK, there was also a strong movement tolimit the redistribution of resources (Lindert ). In the period leading upto the introduction of unemployment insurance, the degree of the centralauthority’s contribution was a source of much debate. The Conservativescriticised a tripartite system, and in the end the Liberals agreed to limit thestate’s contributions to one-third of the total of employers’ and workers’shares (Fraser , p. ; Hay , p. ). With regard to the actualburden of the insurance schemes, we cannot say with any certainty whetherlabour demand was less elastic than labour supply, but it was the strongbelief that employers in both countries – in Germany this was facilitated bytariffs – could pass on price increases to consumers in their large domesticmarkets.

The situation was different in small, northern European countries.Because of their greater exposure to international exchanges (see Table ),firms did not have the margin for manoeuvre found in larger countries.Recognising that labour market regulations would raise costs in the exportsector, national authorities in these countries opted for social insuranceschemes that were paid for out of general tax revenue. In Denmark, thestate actively participated in subsidising trade union sponsored unemploy-ment insurance initiatives. Employers did not contribute to the system until (Blaustein ). As elsewhere in northern Europe, the Danish planwas voluntary. Still, as one commentator wrote: ‘it is probably true todescribe the Danish system as the most comprehensive attempt [in Europe]to bring unorganised workers into a primary voluntary plan’ (Wolfenden, p. ). In Sweden, state financing of a voluntary programme was inplace by (Clark ). In Belgium and the Netherlands, in contrast,national initiatives were slower to take hold and municipalities retainedfinancial responsibility.

The division between the Low Countries and Scandinavia meshes withthe view that political fragmentation was a factor in the development ofsocial spending. Scandinavia was celebrated, to borrow a phrase fromKindleberger (), for its ‘group behaviour’. Gourevitch (, p. )and also Rustow () and Rogowski (, p. ) attributed Swedish col-laboration to the nature of its industrial and social development. Based ontheir initial coalition in support of free trade, groups of farmers, commercialand financial interests, industrialists and workers coalesced around issues ofsocial reform. Elsewhere, the Danish and Norwegian co-operative move-ments were among the strongest in Europe (Kindleberger ). In these


See Beveridge’s statements on wage offsetting cited in Harris (, p. ) and Hay(, p. ).

states, redistributive policy was an easier sell than in the Low Countries,and in Britain, France and Germany.

Table presents a preliminary analysis of these issues. We are primarilyinterested in the relationship between LCI and trade, but we also includethe (Pearson) correlation coefficients for the other continuous independentvariables in our data set. For all variables the values in this exercise are for, and . Consider, first, the trade variable. As expected, it isinversely related to tariffs, area and population. Small countries needed totrade, while larger countries (measured by area) had higher tariffs. As forthe labour compact, there were differences among the LCI and its two sub-components, the labour market regulation, LRI, and social insurance, SII,indices. Consistent with Rodrik and Agell, the LCI and SII are positivelyrelated to the trade share; however the relation between LRI and trade,while positive, is not significant. Those countries that opted for labourmarket regulations seem to have done so for domestic reasons, as in thestandard story of the rise of the welfare state. Size (area and population) didnot have much effect on the delivery of social insurance either, but note thatthe negative coefficient for SII-AREA is five times as large as that for LRI-AREA. This gives only small comfort to the Katzenstein hypothesis for theperiod before . Finally, SII is negatively correlated with tariffs. Socialinsurance was an effective substitute for tariff policy. Open economiesneeded to support their workers – they could have turned to tariffs or theycould have provided a safety net. In the next section, we test whether theserelations hold when we control for other variables.

. The LCI and trade: another look

Our objective in this section is to expand on the relationship between thelabour compact and openness, taking into account macroeconomic, social,political and geographic variables. We first examine the association betweenthe LCI and trade, and then examine the relationship between the sub-com-ponents of the index (LRI and SII) and trade. Finally, we turn our atten-tion to the effect of political institutions on this relationship. In all stages inour argument, we rely on tobit regressions. Our dependent variables, LCI,


We use values for , and because of data limitations for some of ourvariables, and because the average country in our sample had reached less than percent of Berlin and other ‘best practice’ standards prior to .

In his empirical work, Rodrik used two measures of risk caused by openness, the degreeof openness (the trade share), and openness multiplied by the standard deviation of theterms of trade. Because of data limitations, we will use the former measure. Williamson() has assembled the most comprehensive data set on terms of trade before .For Europe, he reported that the terms of trade were declining almost everywhere,which implies that, holding the trade share constant, European countries were facing thesame change in risk.

E

uropean Review

of Econom

ic History

Table . Correlation coefficients of dependent and independent variables.

Labour market compact Macroeconomic variables Size variables Social and political variables

LCI LRI SII TRADE TARIFF GDP GDP% POP AREA NONAG VOTE UNION ETHNIC

LCI –LRI .** –SII .** .** –TRADE .* . .* –TARIFF �. �. �.* �.** –GDP .** .** .** .** �.** –GDP% .** .** .** .* �.* �. –POP . . . �.** . �. �. –AREA �. �. �. �.** .* �.* �. .** –NONAG .** .** .** .** �.** .** .* .* �.** –VOTE .** .** .** .** �. .** .** . �. .** –UNION .** .** .** .** �.* .** . �. �. .** .** –ETHNIC �. . �. .** �. .* . �. �. .* . . –

Notes: **Significant at the . level. *Significant at the . level.Sources: See Appendix.

LRI and SII, combine zero and continuous values. For this type of ‘cen-sored’ data, which Greene (; see also Wooldridge ) compares tocount data, tobit estimation is an appropriate procedure. Unless indicatedotherwise, we use values for , and .

.. Trade and other influences on the labour compact

Columns – of Table give the results of regressing LCI on trade and otherindependent variables. Because of the possible endogeneity between LCIand trade, in the sense that greater government expenditures raise incomeand hence imports, we have lagged the trade variable by ten years. Wereport the regression coefficients and the (absolute) ‘t’ values in parentheses;for the trade value we also report the ‘slopes’ or marginal effects in squarebrackets. Our baseline regression, column , is our story. The coefficient onthe lnTRADE variable translates into an elasticity of ., evaluated at thesample mean of LCI. This equates to an increase in the index of ., againat the mean value of LCI. A visual representation of the positive and sig-nificant relationship between the LCI and trade is found in Figure . (Thisis an OLS estimation – there are no values for LCI in ).

All regressions control for population. This captures the Katzensteinhypothesis, but it is also possible that the relationship between the LCI andtrade would be spurious without its inclusion or that of country area(Rodrik ). Assume that the provision of public services is subject to sig-nificant indivisibilities. Every country requires a Ministry of Labour. Thengovernment social expenditure would be negatively correlated with countrysize, but since openness is negatively correlated with country size as well,the observed link in column between LCI and trade could be spurious andcould be due to an omitted variable. That said, we do not find this.Population and area enter with significant and positive signs in columns–.

From regression on, we have included time dummies for and. They are significant in columns –, suggesting the effects of omitted


One problem is that our dependent variable is bounded between and . However, we haveonly one value in our data set equal to (the SII value for the UK in ). When we dropthe UK from our data set, the resulting estimates do not change. As a check on ourprocedures, we multiplied the LCI by a scalar (we chose the log odds correction); this meantgiving small values to the zero values in the index. We then used OLS to estimate therelationship between the new values for LCI and the independent variables. Again, our mainresult that there is a positive relationship between LCI and trade was confirmed. Wepreferred the tobit method because of the arbitrary nature of assigning small values. Webenefited from discussion with Éric Renault on selecting the appropriate estimation strategy.

N � . We have combined Austria and Hungary because of data limitations. To calculate marginal effects, estimated coefficients in tobit regressions need to be

corrected with an adjustment factor. We follow Woolridge (, pp. –).

E

uropean Review

of Econom

ic History

Table . Some influences on the labour compact.

Dependent () () () () () () ()variables LCI LCI LCI LCI LCI LINDERT LCI

all countries all countries all countries all countries poor countries all countries all countries

Constant �. �. �. �. �. . �.(.) (.) (.) (.) (.) (.) (.)

lnTRADE .** .** .* .* .* .**(.) (.) (.) (.) (.) (.)[.] [.] [.] [.] [.] [.]

lnPOP .** .** .** . �.** .*(.) (.) (.) (.) (.) (.)

lnAREA .*(.)

lnGDP .(.)

GDP% . -. -. -.(.) (.) (.) (.)

VOTE .** .** . .**(.) (.) (.) (.)

lnETHNIC �. �.(.) (.)

lnTARIFF �.*(.)

T .** �.** .** . �. .**(.) (.) (.) (.) (.) (.)

T .** .** .** . �. .**(.) (.) (.) (.) (.) (.)

N

SE of est/‘R’ ./. ./. ./. ./. ./. ./. ./.

Notes: **Significant at the . level. *Significant at the . level. Tobit regressions. Unnormalised coefficients (t statistics in parentheses). ‘R’ is the squared correlation betweenobserved and expected values of the dependent variable. GDP and TRADE lagged years. GDP is an instrument predicted from a regression described in text. Means: LCI � .; LRI � .; SII � .; LINDERT � .; TRADE � .; POP � ,; AREA � .; GDP � ,; GDP% � .; VOTE � .; ETHNIC � .; TARIFF � ..Sources: See Appendix.

variables. In subsequent regressions we try to uncover these factors.Regression includes political, social and income variables. Obviously tradecan affect income. We have experimented with a number of alternatives tocontrol for this problem. We have lagged income years and we have alsoincluded the average rate of growth of income in the ten years before thedate of the corresponding LCI value. The lagged income variable incolumn is an instrumental variable. In all regressions neither income norits growth rate was significant. Lindert found a similar result in his analysisof social spending. As in Lindert’s study, the percentage of the adult malepopulation eligible to vote is significant, and although the linguistic frag-mentation coefficient is negative as expected it is not significant.

Regression provides a check on the robustness of our estimates and


The rate of growth variable can be taken as an indicator of rapid expansion in theeconomy, resulting from technical innovation, structural change, or urbanisation, thatgave rise to the demand for social protection. The poor performance of the incomevariable might indicate that richer countries had alternative private means of insurance.

Exogenous variables in the regression were: lagged income, non-agricultural share of thelabour force and all the other independent variables in Table . The R (F statistic) ofthe estimated equation was . (.).

We have experimented with including the non-agricultural labour force and the uniondensity rate as exogenous variables. The former presented us with problems ofmulticollinearity; the latter was insignificant.

Figure . Openness and the labour compact in .

examines the relationship between LCI and trade for only poor countries,that is the bottom half of our sample. Note that the time variables do notshow up as significant here. Something is going on in rich countries that wehave yet to explain.

Some clues as to what may be happening can be found in column whichuses Lindert’s social spending estimates (measured as a percentage of nationalincome) as the dependent variable. The trade variable is significant which sitswell with our basic line of reasoning; and its elasticity, calculated at the meanvalue of social spending, has a value of .. Lindert’s finding of a relation-ship between democracy and spending is also confirmed. The time variablesare not important and, again in contrast with our previous results, the popu-lation variable is negative and significant. Our reading of these results is that,holding constant the level of openness, small countries do seem to spend moreon social spending and this predisposition is stable over time. This mesheswith the Katzenstein hypothesis. The last column in the table asks whetherextending the labour compact was an alternative to tariff policy. Our answeris affirmative. The coefficient on tariffs means that a reduction in protectionby per cent was associated with a . per cent increase in the LCI.

.. Trade and other influences on labour regulation and socialinsurance

The relationship between trade and the labour compact holds equally for thesub-components of the LCI (columns and of Table ). There are differ-ences in some of the other influences, however. The extent of the male fran-chise is not a significant determinant of labour market regulation. Allcountries, even Bulgaria, Russia and Spain, introduced some minimal stan-dards of regulation. This is the contagion effect in the standard histories of therise of the welfare state. As for social insurance, the time dummies and thepopulation coefficients are not significant, results that closely mimic those ofregression (Lindert) in Table . All told, the effect of trade was greater onsocial insurance than on labour market regulation. The calculated elasticity ofthe trade variable (.) for column is almost twice as large as that of thecorresponding value for column (.), when both are measured at meanvalues. The relationship between tariffs and the labour compact also shows upfor its sub-components, and the trade-off between protection and social insur-ance is greater than that between protection and labour market regulation.


We have performed other checks: in one set of regressions we excluded the UK; inanother we looked at rich countries only. The results are similar to those reported incolumn of Table .

After adjustment, the slope of lnTARIFF is �.. Evaluated at the mean level of LCI,the elasticity is .. We do not make any conjecture on the relation between tariffsand growth. These issues are better dealt with in Clemens and Williamson () whouse a much larger data set.

.. Trade, the labour compact and political institutions

Recall that Rogowski hypothesised that PR systems are, contrary to expec-tations, more stable and that they provide an institutional framework forworker participation in decision-making. Workers’ parties can use theirsupport for free trade as leverage to meet their goal of more expansive socialprotection. In Table we test for this proposition. Our baseline regression isreported in column . Holding constant the trade share, PR is a significantfactor in explaining the extent of the labour compact; we have calculated a


Recall, as well, that Rogowski found that after countries having assemblies orparliaments with large constituencies (fewer representatives) had greater trade shares.Using data from Mackie and Rose (), we did not find that this relation held for ourperiod. There was a connection between trade and PR, however.

We have included the Netherlands as a PR country. It went to PR in but, lookingover its shoulder at the Belgium experience, it was well on the way to introducing itbefore the War (Kossmann ).

Table . Some influences on regulation and insurance.

Dependent Labour market regulation (LRI) Social insurance (SII)variables

() () () () () ()

Constant �. �. �. �. �. �.(.) (.) (.) (.) (.) (.)

lnTRADE .** .* .** .*(.) (.) (.) (.)[.] [.] [.] [.]

lnPOP .** .** .* .* . .(.) (.) (.) (.) (.) (.)

GDP% . �.(.) (.)

VOTE . .**(.) (.)

lnTARIFF �. �.*(.) (.)

T .** .** . .(.) (.) (.) (.)

T .** .** . .(.) (.) (.) (.)

N SE of est/ ./. ./. ./. ./. ./. ./.

‘R’

Notes: See Table . **Significant at the . level. *Significant at the . level.

marginal effect of a change from no PR to PR of .. PR was mainlyintroduced in Europe after and it may represent one of the omittedvariables we noted above. The inclusion of PR in regression is in somesense a substitute for the time dummy for we reported in Tables and. There may be an identification problem here because the PR states in oursample were all ‘small countries’. Still, not all small countries had PR sys-tems by World War I. We looked next at the relationship between theextension of the franchise and PR. VOTE in column is significant at the per cent level, its importance being overwhelmed by the interaction ofVOTE and PR. For given levels of male franchise and trade, there was a dif-ferent and positive dynamic in the relationship between the labour compactand PR regimes. When we examined this relationship for the sub-indices,the relationship held up for social insurance but not for labour market regu-lation. The latter regression () performs poorly. The suffrage rate is notimportant; note as well that the trade variable is significant only at the percent level.

Belgium’s experience appears to conform to these findings. Thecountry saw rapid economic growth and its trade share rose, but late intothe century it had few measures of social protection, and despite sizeablewage increases, work hours remained very long. It was very hard hit by thelong depression from to and there was social unrest. The majorpolitical aim of the trade union movement, and its ally, the Parti ouvrier belge(POB) formed in , was universal franchise in order to promote greatersocial protection. Up until the granting of the right to vote, the dominantCatholic party remained indifferent to these demands. In men receivedthe right to vote. The Catholic majority remained entrenched, but the oppo-sition was able to push through proportional representation in .Thereafter, social reforms multiplied: introduction of old age pensions in, accident compensation in , and unemployment insurance in. Militancy waned and, after the POB formed a coalition with the rad-ical liberals, workers began to actively participate in setting the policyagenda. Throughout this period, workers’ movements and their politicalpartners supported free trade, but only with PR were workers able to usetheir newly gained voting rights to achieve some of their demands for socialprotection.


This amounted to calculating E[y|x] at the means of the other variables, first when PR� and second when PR � , and then taking the difference between the two. Wethank Denise Young for providing us with the SHAZAM program to make thiscalculation.

We have run regression in Table without a PR dummy, but with a ‘small country’dummy. The results are similar. We postpone sorting out these two effects until a futurepaper.

This paragraph is based on Kossmann ().

. Conclusion: more lessons from history

The decades before saw the expansion of international trade and therise of social protection. This was not a mere coincidence. With globalisa-tion, European workers faced increasing uncertainty and their conditions ofwork suffered. They demanded social protection and, in response, nationalauthorities introduced labour compacts. Even in the absence of generoussocial entitlements, the labour compact provided workers with insurancebecause it compressed wage structures. We have found that the labour com-pact was most extensive where the exposure to trade was greater. Ourresults are robust. They hold for different subsets of our sample and for dif-ferent measures of the labour compact. Moreover, we find the same associ-ation between trade and social protection using Lindert’s estimates of socialspending.

Holding trade shares constant, the proportion of adult male voters in thepopulation was a significant determinant of the extent of the labour com-pact. But the nature of political institutions may have mattered more.


Table . The labour compact and proportional representation.

Dependent () () () () ()variables LCI LCI LCI LRI SII

Constant �. �. �. �. �.(.) (.) (.) (.) (.)

lnTRADE .* .* .* . .*(.) (.) (.) (.) (.)[.] [.] [.] [.] [.]

lnPOP .** .** .** .** .(.) (.) (.) (.) (.)

GDP% . . . . .(.) (.) (.) (.) (.)

VOTE o. o. o.(.) (.) (.)

PR .** .**(.) (.)

PRVOTE o.* o. o.*(.) (.) (.)

T .* .*(.) (.)

N SE of est/ ./. ./. ./. ./. ./.

‘R’

Notes: See Table . The mean of PR � .. **Significant at the . level. *Significant atthe . level.

Proportional representation appears to have been a vehicle for workers toexchange their support of trade for concessions from the state. The upshotis that democracy was good for growth because it was good for trade andworkers. As for its components, the labour compact varied across Europe.Larger states with sizeable domestic markets tended to use labour marketand factory regulations; countries with more exposure to internationalmarkets, many of which were small, opted for social insurance type pro-grammes. Many of these findings are echoed in research on current trendsin European economic integration (Ehrenberg , Jacoby ). Overall,however, it is difficult to separate the influence of PR and small states in theformation of the labour compact. This is a subject for future research.

The received view is that national authorities in Europe before ,faced by falling world grain prices and strong international competition, hadtwo choices: laissez-faire policies or tariff protection. We have argued thatthere was an alternative. Economic integration involved dislocation and wascostly to all types of workers, in agriculture as in industry. The labour com-pact was an alternative to tariffs because it provided workers with protectionagainst external risk. Thus, in countries with extensive labour compacts, likeBelgium and Denmark, industrial workers were enthusiastic supporters offree trade. Herein lies another lesson of history. Globalisation expanded inthe period before because it was associated with increased worker pro-tection. As today, free trade was less well received where workers wereforced to go it alone.

Acknowledgements

We benefited from the comments of seminar participants and discussants at theLaboratoire d’Économie Appliquée (INRA), Paris; the London School of Economics;and the Universities of California (Davis), Copenhagen, Göteborg, Toronto, andZurich. We owe special special thanks to Jonas Agell, Steven Broadberry, ThomasDavid, Peter Lindert, Peter Scholliers, and two anonymous referees for their com-ments and suggestions. The usual disclaimer applies. Huberman’s research wasfunded by the Social Sciences and Humanities Research Council of Canada.

Appendix

AREA: Boundaries and area in millions of square kms. Values for , and. Source: Foreman-Peck and Lains ().

ETHNIC: Average value of five different indices of ethnolinguisticfractionalisation. Its value ranges from to . Components are: () index ofethnolinguistic fractionalisation which measures the probability that tworandomly chosen individuals in a given country are not a member of the samelinguistic group; () probability that two randomly selected individuals do notspeak the same language; () probability of two randomly selected individualsspeaking different languages; () percentage of population not speaking the


official language; () percentage of population not speaking the most widelyused language. All values for . Source: La Porta, et al. ().

GDP: GDP per capita in Geary-Khamis Dollars. Values for , and. Source: Madison ().

GDP%: Average annual rate of growth of GDP per capita in the period prior tothe year selected. For , average annual growth for the period –;for , average annual growth for –; for , –. Source:Madison ().

NONAG: Share of labour force in non-agricultural sectors. Values for , and . Source: Lindert (data sheet), Mitchell ().

POP: Population. Values for , and closest date to . Source: Mitchell().

PR: A binary variable ( � proportional representation; � none). All countriesthat introduced PR up until . Belgium, ; Denmark, ; Finland,; Netherlands, ; Sweden, . Source: Mackie and Rose ().

TARIFF: Average (manufacturing and agriculture) tariff rates. Values for , and . Source: Foreman-Peck and Lains ().

TRADE: Value of exports and imports as a share of GDP. Values for , and , unless otherwise indicated. Sources: Bairoch (), Grassman(), Mitchell (), Maddison ().

UNION: Percentage of labour force unionised. Values for the closest years to, and . Unreported countries received a value of zero. Source:Crouch ().

VOTE: Percentage of male population over years of age having right to vote.Values for the closest years to , and . Unreported countriesreceived a value of zero. Sources: Flora (), Crouch ().

References

AGELL, J. (). On the benefits from rigid labour markets: norms, marketfailures and social insurance. Economic Journal , pp. F–F.

AGELL, J. (). On the determinants of labour market institutions: rent seekingvs. social insurance. German Economic Review , pp. –.

ALCOCK, A. E. (). History of the ILO. New York: Octagon Books.ALESINA, A., BAQIR, R., and EASTERLY, W. (). Public goods and ethnic

divisions. Quarterly Journal of Economics , pp. –.ALTER, G. (). Work and income in the family economy of Belgium. Journal

of Interdisciplinary History , pp. –.AMBROSIUS, G. and HUBBARD, W. H. (). A Social and Economic History of

th Century Europe. Cambridge, MA: Harvard University Press.ASSOCIATION INTERNATIONALE POUR LA LUTTE CONTRE LE CHÔMAGE. ().

State Unemployment Insurance in Various Countries. Reports to the reAssemblée Générale. Geneva.

BAIROCH, P. (). Commerce extérieur et développement économique de l’Europe auXIXe siècle. Paris: Mouton.

BAIROCH, P. (). European Trade Policy, –. In P. Mathias and S.Pollard (eds), Cambridge Economic History of Europe, vol. VIII, The Industrial



Economies: The Development of Economic and Social Policies. Cambridge:Cambridge University Press.

BELLOM, M. (). État actuel de la question des accidents du travail dans lesdifférents pays. Congrès International des Accidents du Travail, Rapports.Brussels: P. Weissenbruch.

BLAUSTEIN, S. J. (). Unemployment Insurance in the United States: The FirstHalf Century. Kalamazoo: Upjohn Institute.

BLAUSTEIN, S. J. and CRAIG, I. (). An International Review of UnemploymentInsurance Schemes. Kalamazoo: Upjohn Institute.

BROOKE, E. (). Factory Laws of European Countries. London: Grant Richards.CAMERON, D. R. (). The expansion of the public economy: a comparative

analysis. American Political Science Review , pp. –.CANNING, K. (). Languages of Labor and Gender: Female Factory Work in

Germany, –. Ithaca, NY: Cornell University Press.CLARK, H. (). Swedish Unemployment Policy – to . Washington:

American Council on Public Affairs.CLEMENS, M. and WILLIAMSON, J. G. (). A tariff-growth paradox?

Protection’s impact the world around, –. Working Paper no. ,National Bureau of Economic Research.

COMMONS, J. R. and ANDREW, J. B. (). The Principles of Labor Legislation. thedn. New York: Harper and Bros.

CRAFTS, N. F. R. (). The Human Development Index and changes instandards of living: some historical comparisons. European Review of EconomicHistory , pp. –.

CROUCH, C. (). Industrial Relations and European State Traditions. Oxford:Clarendon.

DELEVINGNE, M. (). The pre-war history of international labor legislation. InJ. T. Shotwell (ed.), The Origins of the International Labor Organization, vol. .New York: Columbia University Press.

DROZ, N. (). État de la question des accidents du travail. Congrès Internationaldes Accidents du Travail, Rapports, vol. . Paris.

EHRENBERG, R. G. (). Labor Markets and Integrating National Economies.Washington: Brookings Institution.

EPSTEIN, A. (). Insecurity: A Challenge to America: A Study of Social Insurancein the United States and Abroad. New York: Harrison Smith.

FLORA, P. (). State, Economy and Society in Western Europe, –.Frankfurt: Campus Verlag.

FOLLOWS, J. W. (). Antecedents of the International Labour Organization.Oxford: Oxford University Press.

FOREMAN-PECK, J. and LAINS, P. (). European economic development: thecore and the southern periphery, –. In S. Pamuk and J. G.Williamson (eds), The Mediterranean Response to Globalization before .London: Routledge.

FRASER, D. (). The Evolution of the British Welfare State. nd edn. London:Macmillan.

GOUREVITCH, P. (). Politics in Hard Times: Comparative Responses toInternational Economic Crises. Ithaca, NY: Cornell University Press.

GRASSMAN, S. (). Long-term trends in openness of national economies.Oxford Economic Papers , pp. –.

GREENE, W. (). Econometric Analysis. New York: Macmillan.HARRIS, J. (). Unemployment and Politics: A Study in English Social Policy,

–. New York: Oxford University Press.HATTON, T. J. and WILLIAMSON, J. G. (). The Age of Migration: Causes and

Economic Impact. New York: Oxford University Press.HAY, J. R. (). The Origins of the Liberal Welfare Reforms, –. London:

Macmillan.HEIKKINEN, S. (). Labour and the Market: Workers, Wages and Living

Standards in Finland, –. Helsinki: Finnish Society of Sciences andLetters.

HENTSCHEL, V. (). German economic and social policy, –. In P.Mathias and S. Pollard (eds), Cambridge Economic History of Europe, vol. VIII,The Industrial Economies: The Development of Economic and Social Policies.Cambridge: Cambridge University Press.

HICKS, J. R. (). A Theory of Wages. New York: St. Martin’s Press.HUBERMAN, M. and LEWCHUK, W. (). Glory days? Work hours, labour

market regulations and convergence in late nineteenth century Europe. In G.Federico et al. (eds.), Integration of Commodity Markets in History. Proceedingsof the th International Economic History Congress. Madrid: FundacionFomento de la Historia Economica.

INDUSTRIAL RELATIONS COUNSELORS (). An Historical Basis for UnemploymentInsurance. Minneapolis: University of Minnesota Press.

INTERNATIONAL ASSOCIATION FOR LABOUR LEGISLATION (IALL) (). Report ofthe th General Meeting of the Committee of the IALL. Geneva, – September,. London; Labour Representative Printing & Publishing.

IALL (). First Comparative Report on the Administration of Labour Laws:Inspection in Europe. London: P. S. King & Son.

INTERNATIONAL LABOUR OFFICE (ILO) (). Factory Inspection: HistoricalDevelopment and Present Organization in Certain Countries. Geneva: ILO.

ILO (). The Organisation of Labour Inspection in Industrial and CommercialUndertakings. Geneva: ILO.

JACOBY, S. (ed.) (). The Workers of Nations: Industrial Relations in a GlobalEconomy. New York: Oxford University Press.

KATZENSTEIN, P. (). Small States in World Markets: Industrial Policy in Europe.Ithaca, NY: Cornell University Press.

KEMP, T. (). Economic and social policy in France. In P. Mathias and S.Pollard (eds), Cambridge Economic History of Europe, vol. VIII, The IndustrialEconomies: The Development of Economic and Social Policies. Cambridge:Cambridge University Press.

KENWOOD, A. G. and LOUGHEED, A. L. (). The Growth of the InternationalEconomy, –. London: George Allen & Unwin.

KINDLEBERGER. C. P. (). Group behavior and international trade. Journal ofPolitical Economy , pp. –.

KOHLER, P. A. and ZACHER, H. (eds) (). The Evolution of Social Insurance:–. London: St. Martin’s Press


KOSSMANN, E. H. (). The Low Countries, –. Oxford: Clarendon Press.KRUEGER, A. B. (). From Bismarck to Maastricht: the march to European

Union and the labor compact. Working Paper no. , National Bureau ofEconomic Research.

LA PORTA, R., LOPEZ-DE-SILANES, F., SHLEIFER, A., and VISHNY, R. (). Thequality of government. Journal of Law, Economics, & Organization , pp.–.

LEBOUTTE, R., PUISSANT, J., and SCUTO, D. (). Un siècle d’histoire industrielle(–). Belgique, Luxembourg, Pays-Bas: industrialisations et sociétés. Paris:SEDES.

LINDERT, P. H. (). The rise of social spending. Explorations in EconomicHistory , pp. –.

LINDERT, P. H. (). Poor relief before the welfare state: Britain versus thecontinent, –. European Review of Economic History , pp. –.

LINDERT, P. H. and WILLIAMSON, J. G. (). Does globalization make the worldmore unequal? Working Paper no. , National Bureau of Economic Research.

LOW, M. (). A Short History of Labor Legislation in Great Britain. Bulletin ofthe Bureau of Labor, Washington, no. , pp. –.

MACHLAN, L. (). Workers’ insurance versus protection of the workers: statesocial policy in Imperial Germany. In P. Weindling (ed.), Social History ofOccupational Health. London: Croom Helm.

MACKIE, T. T. and ROSE, R. (). The International Almanac of ElectoralHistory. London: Macmillan.

MADDISON, A. (). Monitoring the World Economy, –. Paris:Organization for Economic Cooperation and Development.

MANSFIELD, E. D. and BUSCH, M. L. (). The political economy of nontariffbarriers: a cross-national analysis. International Organization , pp. –.

MITCHELL, B. R. (). European Historical Statistics, –, nd edn. NewYork: Facts on File.

MURRAY, J. E. (). Government insurance, worker health, and labor supply inturn of the century Europe. University of Toledo, unpublished manuscript.

OGUS, A. I. (). Great Britain. In P. A. Kohler and H. Zacher (eds.), TheEvolution of Social Insurance, –. London: St. Martin’s Press.

O’ROURKE, K. H. (). The European grain invasion, –. Journal ofEconomic History , pp. –.

O’ROURKE, K. H. (). Tariffs and growth in the late th century. EconomicJournal , pp. –.

O’ROURKE, K. H. (). Globalization and inequality: historical trends. WorkingPaper no. , National Bureau of Economic Research.

O’ROURKE, K. H., and WILLIAMSON, J. G. (). Globalization and History: TheEvolution of a Nineteenth-Century Atlantic Economy. Cambridge: MIT Press.

RIMLINGER, G. V. (). Welfare Policy and Industrialization in Europe, Americaand Russia. New York: Wiley.

RIST, C. (). La réglementation légale de la journée de travail de l’ouvrier adulte.Paris: Larose.

RODRIK, D. (). Has Globalization Gone Too Far? Washington: Institute forInternational Economics.


RODRIK, D. (). Why do more open economies have bigger governments?Journal of Political Economy , pp. –.

ROGOWSKI, R. (). Trade and the variety of democratic institutions. IndustrialOrganization , pp. –.

ROGOWSKI, R. (). Commerce and Coalitions: How Trade Affects DomesticPolitical Alignments. Princeton, NJ: Princeton University Press.

ROLIN-JAEQUEMYNS, M. G. (). La conférence de Berlin sur la législation dutravail et le socialisme dans le droit international. Revue de droit international etde législation comparée , pp. –.

ROSOLI, G. (). Italian migration to European countries from politicalunification to World War I. In D. Hoerder (ed.), Labor Migration in the AtlanticEconomies. Westport, CN: Greenwood Press.

RUSTOW, D. A. (). The Politics of Compromise: A Study of the Parties andGovernment in Sweden. Princeton, NJ: Princeton University Press.

SCHOLLIERS, P. (). Family income, needs and mother’s wages: a criticalsurvey of working class budget inquiries in Belgium, –. In T.Pierenkemperer (ed.), Zur Ökonomik des privaten Haushalts. Frankfurt: Verlag.

SCHREMMER, D. E. (). Taxation and public finance: Britain, France andGermany. In P. Mathias and S. Pollard (eds), Cambridge Economic History ofEurope, vol. VIII, The Industrial Economies: The Development of Economic andSocial Policies. Cambridge: Cambridge University Press.

SHOTWELL, J. T. (ed.) (). Origins of the ILO, vol. . New York: ColumbiaUniversity Press.

SINN, H.-W. (). Social insurance, incentives and risk-taking. Working Paperno. , National Bureau of Economic Research.

SMITS, J. P. and VAN ZANDEN, J. L. (). Industrialization and incomeinequality in the Netherlands –. In L. Borodkin and P. Lindert, (eds),Trends in Income Inequality during Industrialization. Proceedings of the thInternational Economic History Congress. Madrid: Fundacion Fomento de laHistoria Economica.

TAYLOR, A. M. and WILLIAMSON, J. G. (). Convergence in the age of massmigration. European Review of Economic History , pp. –.

TEGLE, S. (). Den ordinarie veckoarsbetstiden i Sverige, –: enöversikt. Unpublished manuscript, University of Lund.

TROCLET, L. E. (). Législation sociale internationale. Les Cahiers de l’Institutede Sociologie Solvay, Cahier . Brussels.

TWAROG, S. (). Heights and living standards in Germany, –: thecase of Wurttemberg. In R. H. Steckel and R. Floud (eds.), Health and Welfareduring Industrialization. Chicago, IL: University of Chicago Press.

UNITED KINGDOM. Parliamentary Papers (PP) (). Reports from Her Majesty’sDiplomatic and Consular Agents, vol. LXVI.

UK: PP (). Reports from Her Majesty’s Diplomatic and Consular Agents, vol.LXVIII.

UK: PP (). Reports from Her Majesty’s Diplomatic and Consular Agents, vol.LXII.

UK: PP (–). Commission on Labour, Foreign Reports, vol. XXXIX.UK: PP (). First Annual Abstract of Foreign Labour Statistics, vol. XCII.


UK: PP (). Second Annual Abstract of Foreign Labour Statistics, vol. LXXIII.UK: PP (). International Conference on Labour, vol. LXXIII.UK: PP (). Third Annual Abstract of Foreign Labour Statistics, vol. CXIII.UK COMMERCIAL (). Further Correspondence Respecting the International

Labour Conference at Berlin, C–, no. , pp. –.UNITED STATES: DEPARTMENT OF LABOR (). Twenty-fourth Annual Report of

the Commissioner of Labour (). Workmen’s Insurance and CompensationSystems in Europe. Washington.

UNITED STATES: DEPARTMENT OF LABOR (). Workmen’s Compensation Lawsof the United States and Foreign Countries. Bulletin of the United States Bureauof Labor Statistics, Department of Labor, no. .

VILLARD. H. G. (). Workmen’s Accident Compensation and Insurance inBelgium, Norway, Sweden, Denmark and Italy. New York.

WEIR, D. R. (). Economic welfare and physical well-being in France,–. In R. H. Steckel and R. Floud (eds.), Health and Welfare duringIndustrialization. Chicago, IL: University of Chicago Press.

WILLIAMSON, J. G. (). The evolution of global labor markets since :background evidence and hypotheses. Explorations in Economic History , pp. –.

WILLIAMSON, J. G. (). Globalization, convergence, and history. Journal ofEconomic History , pp. –.

WILLIAMSON, J. G. (). Globalization and inequality, past and present. WorldBank Research Observer , pp. –.

WILLIAMSON, J. G. (). Land, labor and globalization in the pre-industrialThird World. Working Paper, no. , National Bureau of EconomicResearch.

WITTE, E. and CRAEYBECKX, J. (). La Belgique politique de à nos jours: lestensions d’une démocratie bourgeoise. Brussels: Éditions Labor.

WOLFENDEN, H. H. (). Unemployment Funds: A Survey and Proposal.Toronto: Macmillan,

WOOLDRIDGE, J. M. (). Introductory Econometrics: A Modern Approach. SanteFe: South-Western College Publishing.

ZOLLNER, D. (). Germany. In P. A. Kohler. and H. Zacher (eds), TheEvolution of Social Insurance: –. London: St. Martin’s Press.


European econom

ic integration, –

Appendix Table. The components of the labour compact index.

Minimum Maximum Minimum Maximum Minimum Maximum Minimum Maximumage of work length of age night length of age of work length of age night length of

working day work allowed working day working day work allowed working day (youth) (women) (youth) (women)

Austria Belgium Bulgaria Denmark Finland France Germany Hungary Italy Netherlands Norway Portugal Russia .Spain Sweden Switzerland United . . . .

Kingdom

European economic integration and the labour compact, 1850-1913

Documents

Transcript of European economic integration and the labour compact, 1850-1913