Politics matter: changes in unionisation rates in rich countries, 1960–2010

Politics matter: changes in unionisationrates in rich countries, 1960–2010John Schmitt and Alexandra Mitukiewicz

ABSTRACTirj_675 260..280

We use the Institutional Characteristics of Trade Unions, Wage Setting, State Inter-vention and Social Pacts database to examine the changes in collective bargainingcoverage and union density among 21 OECD countries over the past 50 years. Theobserved patterns suggest that national politics are a more important determinant ofrecent trends in unionisation than globalisation or technological change.

There is a reason for the decline in union membership and clout. The old model of big unions facingdown big companies has become obsolete in the face of globalisation and technological advancement. . .1

—David G. Tuerck, ‘Union campaign against union jobs’, Boston Business Journal, 19 August 2011

[Past union successes] changed in the 1970s and 1980s. Imports and ‘transplant’ factories creatednew competition in steel and autos. Airlines, trucking and communications . . . were deregulated,allowing new low-cost rivals into the market. Digital technology and the Internet transformedcommunications and threatened many industries . . .2

—Robert Samuelson ‘Is Organised Labor Obsolete?’ Washington Post, 28 February 2011

1 INTRODUCTION

The unionised share of the US workforce has been declining from the early 1960sthrough the present. The pace of that decline was slow at first but accelerated atthe end of the 1970s, leaving only about 12 per cent of all workers and 7 per centof private sector workers in unions by 2010 (see Figure 1). The public debate inthe United States has generally focused on the interrelated roles of ‘globalisation’and ‘technology’ in explaining the considerable drop in unionisation. Unions raiselabour costs and reduce employers’ flexibility in ways that are incompatible, theargument goes, with economic success in a ‘weightless’, information-based, globa-lised economy.

❒ John Schmitt is Senior Economist at the Center for Economic and Policy Research. Alexandra Mituk-iewicz is Research Intern at the Center for Economic and Policy Research. Correspondence should beaddressed to John Schmitt, Center for Economic and Policy Research, 1611 Connecticut Avenue, NW,Fourth Floor, Washington, DC 20009, USA; email: [email protected] http://www.bizjournals.com/boston/print-edition/2011/08/19/union-campaign-against-union-jobs.html?page=all2 http://www.washingtonpost.com/wp-dyn/content/article/2011/02/27/AR2011022702873.html

Industrial Relations Journal 43:3, 260–280ISSN 0019-8692

© 2012 The Author(s)Industrial Relations Journal © 2012 Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and350 Main Street, Malden, MA 02148, USA.

The academic debate on dwindling unionisation, however, has offered additionaland generally more nuanced explanations for the collapse in unionisation.3 Farberand Krueger (1993), for example, argue that the most important factor was a dropin workers’ demand for union representation, though Lipset and Meltz (2004) andFreeman and Rogers’ (2006) analysis of more recent survey data on worker attitudessuggests a substantial unmet demand for unionisation among US workers remains.4

Most researchers have instead maintained that, for various reasons, it is the fallingsupply of union jobs that has been the main factor behind the waning trade unionpresence. A rise in employer opposition to unions (sometimes posited to coincide withstrategic failures within trade unions themselves) is one possible cause of the decreasein the availability of union jobs (Bronfenbrenner, 2009; Freeman, 2007; Logan, 2006;and Schmitt and Zipperer, 2009).5 The bulk of the research on the reduction in thesupply of potential union jobs, however, has stressed structural shifts in the economyaway from industries and occupations with higher levels of union organisation.6

Researchers in this vein have typically debated the correct way to apportion thesechanges in industrial and occupational structure between ‘technology’ and ‘trade’.7

3 For relatively recent discussions of the possible causes of union decline in the United States and interna-tionally, see Baldwin (2003), Blanchflower (2007), Farber (2005), Farber and Weston (2001), Flanagan(2007), Pencavel (2005) and Wallerstein and Western (2000).4 Moreover, as Baldwin (2003) observes, distinguishing between workers’ tastes and exogenous economicforces may be difficult: ‘many of the same basic economic forces affecting employers’ profit-orienteddecisions could also affect workers’ decisions about the desirability of union representation’. (p. 2).5 For Britain, several studies show that the failure of unions to organise new workers at new establishmentsexplains much of the decline in union membership there: Machin (2000), Disney et al. (1995) and Brysonand Gomez (2005).6 For an early example, see Dickens and Leonard (1985).7 The exact context of the debate is not always the effects of trade or technology on unionisation; it is oftenanalysed in terms of the effects of trade or technology on employment in the (previously) heavily unionisedmanufacturing sector or on the relative wages of manufacturing workers or workers with less than a highschool degree.

Figure 1: Union membership in the United States, 1948–2010

Source: Labor Research Association analysis updated using Bureau of Labor Statis-tics ‘Union Members’ reports, 2003–10.

Changes in unionisation 261

© 2012 The Author(s)Industrial Relations Journal © 2012 Blackwell Publishing Ltd

Researchers emphasising the primacy of technology have followed two lines ofthinking. The first stems from the simple observation that productivity growth ishistorically much faster in manufacturing than in services (see Baumol, 1967). As aresult, the share of total employment in manufacturing will fall continuously overtime. Given that manufacturing was traditionally much more heavily unionised thanthe services sector, the relative erosion of manufacturing implies, all else constant, along-term deterioration in union density.

The second line of thinking focuses on the idea that technological change is biasedagainst less skilled workers (often operationalised as non-supervisory workers orthose with a high school degree or less). Acemoglu et al. (2000), for example, arguethat skill-biased technical change (SBTC) is ‘at the root of deunionisation’. SBTC‘undermines the coalition between skilled and unskilled workers underlying unions’by increasing the relative competitive-market pay of skilled workers, which ‘weakenstheir incentives to join the unionised sector’ (p. 231). More recently, David Autor, incombination with co-authors, has revised the simple more- versus less-skilled SBTCaccount to emphasise that technological developments since the 1970s have tended toundermine employment opportunities for workers in the broad middle of the wagedistribution.8 Autor et al. (2003) argue that the automation and offshoring of ‘routine’middle-skilled tasks has been the main driver of this process of ‘job polarisation’. Thisresulting ‘hollowing-out’ of work disproportionately leaves the economy with twokinds of non-routine jobs: those based on less skilled, face-to-face, ‘manual tasks’ andothers based on more skilled, ‘abstract tasks’, both of which are difficult to automateor offshore. Job polarisation, which is partly technological and partly related toglobalisation (through offshoring), has obvious implications for unionisation becausethe portion of the economy most susceptible to hollowing out is the one where unionrepresentation was traditionally highest. This discussion of SBTC and job polarisa-tion is, in turn, embedded in a broader debate over the changing nature of work(Castells, 2000; Giddens, 2007; Nolan and Slater, 2010; Rifkin, 1996), which haspotentially profound implications for the future of unions.

The other factor competing to explain the decline in the supply of jobs in unionindustries and union occupations is ‘globalisation’. Globalisation can affect uniondensity through trade, because union workers in rich countries must now competedirectly with low-wage workers in countries that previously did not participate inglobal markets (Baldwin, 2003; Bluestone and Harrison, 1982; Lee, 2005; Magnaniand Prentice, 2003; Wood, 1994). But, many recent researchers have emphasised therole of heightened capital mobility, which undermines union bargaining power byallowing employers to relocate production to other parts of the world or simplythreaten (credibly) to do so (Brady and Wallace, 2000; Bronfenbrenner, 2000;Slaughter, 2007).

As with the discussion of the connection between technology, the changing natureof work, and unionisation, the debate around the impact of globalisation on unionstakes place against the backdrop of a larger debate around the impact of globalisationon the welfare state. Much of this broader debate pits the ‘efficiency’ hypothesis—thatglobalisation has set in motion a ‘race to the bottom’ with respect to taxation andpublic expenditures (Scharpf, 1991; Kurzer, 1991; and Scharpf and Schmidt, 2000;cited in Busmeyer, 2009)—against a ‘compensation’ hypothesis—that welfare states

8 For evidence of job polarisation in Britain, see Goos and Manning (2007); for Europe, see Goos et al.(2009).

John Schmitt and Alexandra Mitukiewicz262


expand in the face of globalisation precisely in response to the heightened econo-mic insecurity facing national populations (Cameron, 1978; Katzenstein, 1985; andRodrik, 1997).

In the context of these larger debates, this article starts from a simple premise: if thedecline in unionisation in the United States is the inevitable response to the twin forcesof globalisation and technology, then we would expect unionisation rates to follow asimilar path in other countries subjected to roughly similar levels of globalisation andtechnology.

We review unionisation data for the last five decades for 21 rich economies. We findthat trends in unionisation have varied substantially across these economies. Unioncoverage (the share of workers whose terms of employment are covered by a collectivebargaining agreement) changed little and even rose slightly in a substantial number ofcountries, including the period since 1980. Union membership (the share of workerswho are members of a union) fell in most of the rich economies, with the United Statesexperiencing losses (from a low initial level of unionisation) near the middle of thedistribution. These differences across countries exposed to broadly similar levelsof globalisation and technological change suggest that neither factor mechanicallydetermines national levels of unionisation. These findings are broadly consistent withother research on long-term trends in unionisation in the world’s rich democracies,including Blanchflower (2007); Brady (2007); Bryson et al. (2011); Ebbinghaus andVisser (1999); and Scruggs and Lange (2002).

One simple factor, however, does appear to explain much of the observed variationin unionisation trends: the broad national political environment.9 Countries that havebeen strongly identified during the post-war period with social democratic parties—Sweden, Denmark, Norway and Finland—have generally seen small increases inunion coverage and only small decreases in union membership since 1980. Over thesame period, countries that are more typically identified as ‘liberal marketeconomies’—the United States, the UK, Australia, New Zealand, Ireland, Canadaand Japan—have generally seen sharp drops in union coverage and membership.Countries in the broad Christian democratic tradition, sometimes referred to as‘coordinated market economies’ or ‘continental market economies’—Germany,Austria, Italy, the Netherlands, Belgium, France and Switzerland—typically have hadoutcomes somewhere in between, with small drops in union coverage and moderatedeclines in union membership. These patterns are consistent with the view thatnational politics are a more important determinant of recent trends in unionisationthan globalisation or technological change.10

2 THE INSTITUTIONAL CHARACTERISTICS OF TRADE UNIONS, WAGESETTING, STATE INTERVENTION AND SOCIAL PACTS (ICTWSS) DATA

We use the ICTWSS database compiled by Jelle Visser at the Amsterdam Institute forAdvanced Labor Studies at the University of Amsterdam.11 The database providesannual information for the years 1960 through 2010 for Organisation for Economic

9 We follow the political categories established by Huber and Stephens (2001a; 2001b), as modified inNavarro et al. (2004).10 We also examine three ex-dictatorships, Greece, Portugal and Spain, where the pattern is not as clear cut.11 The complete database can be downloaded at http://www.uva-aias.net/208. This article uses the May2011 version of the database.



Cooperation and Development (OECD) countries, European Union Member States,and selected emerging economies including Brazil, China and India. From theICTWSS database, we collected annual union coverage and membership data for 21rich countries, all of which are members of the OECD; have roughly comparablestandards of living; and have nearly complete unionisation data for the entire periodfrom 1960 to the late 2000s. For reasons of economic comparability (as well asincomplete data availability), we have not included the ICTWSS data from develop-ing or eastern European countries; due to their small size, we have also excludedIceland and Luxembourg.

We focus on two ICTWSS variables: bargaining coverage (union coverage) and theunion density rate (union membership). The ICTWSS database defines its adjusted‘bargaining (or union) coverage’ as the employees ‘covered by wage bargaining agree-ments as a proportion of all wage and salary earners in employment with the right tobargaining, expressed as percentage, adjusted for the possibility that some sectors oroccupations are excluded from the right to bargain (removing such groups from theemployment count before dividing the number of covered employees over the totalnumber of dependent workers in wage and salary employment)’. The ICTWSS definesthe union density rate as ‘net union membership as a proportion of wage and salaryearners in employment’.12

Union coverage and union membership rates may, and frequently do, differ in thesame country in the same year. In the liberal market economies, including the UnitedStates, workers who are covered by a union contract at their workplace generally maychoose not to join the union that represents them. As a result, union membership ratesin the United States and other liberal market economies are typically lower thanunion coverage rates. In many of the countries we refer to here as continental marketeconomies, ‘[m]ulti-employer bargaining and public policies extend . . . the negotiatedcontract to nonorganised firms guarantee[ing] very high coverage rates . . . far inexcess of union density rates’.13 By contrast, union membership and coverage rates aresimilar (and high) in the social democratic countries. The most important reason is theuse, in three of these countries (Denmark, Finland and Sweden), of the ‘Ghentsystem’, which gives unions a major role in the delivery of social benefits, includingunemployment insurance benefits, therefore providing strong incentives for workersto join unions.14

The distinction between membership and coverage rates is an important one. AsVisser explains: ‘Whereas union density is closer to measuring potential union bar-gaining pressure . . . bargaining coverage [is] closer to measuring the effectiveness ofunions in providing and defending minimum standards of income and employmentprotection in labor markets’.15

The data used here have three other characteristics relevant to our analysis. First,while the union membership data are generally available annually, the adjusted unioncoverage data are usually not reported annually, often only every five years. Second,post-war unionisation data for the ex-dictatorships—Greece, Portugal and Spain—areeither unavailable or only sporadically recorded until the 1970s. During much of the

12 Visser (2011), p. 18.13 See Visser (2006), p. 47; his remarks quoted here refer to ‘most European countries’.14 For recent discussions of the Ghent system and challenges facing it, see Scruggs (2002), Bockerman andUusitalo (2006) and Van Rie et al. (2011).15 Visser (2006), p. 39.



period before the mid-1970s, unions in these countries were either illegal orunder direct government control.16 We show all ICTWSS data available for theex-dictatorships, including data covering periods before democracy, but we limit ouranalysis of trends in these countries only to the period since 1980. Finally, the data forGermany refer to West Germany until unification, and unified Germany from then on.

3 THE TRENDS

Figure 2 displays the ICTWSS data for both union coverage (solid bars) and unionmembership (marked with an X) in 2007, the most recent year for which completedata are available for coverage and membership across the full sample. Coverage ratesin just under half the countries were 80 per cent or higher. Almost two-thirds of thecountries had coverage rates above 50 per cent. At 13.3 per cent, the United States hadthe lowest coverage rate, followed closely by Japan (16.1 per cent) and New Zealand(17.0 per cent).

For almost every country, membership rates were lower than coverage rates. Inmany countries, this gap between coverage and membership rates was large. In Austria,for example, about 99 per cent of eligible workers were covered by a collectivebargaining agreement, but only 30 per cent were members of a union. In France, about90 per cent of workers were covered, but fewer than 10 per cent were union members.Other countries with large gaps between coverage and membership include Belgium,Spain, the Netherlands, Italy, Greece, Portugal, Germany, Switzerland and Australia.

In the United States, coverage rates were only slightly higher than membership rates.This small difference generally reflects decisions by some workers who are representedby unions at their place of work not to join the union that represents them.

In two countries, Japan and New Zealand, membership rates were higher thancoverage rates. This pattern appears to arise from a feature of ICTWSS adjustedcoverage rate data that we use here. As mentioned in the review of the ICTWSSdata in the preceding section, the adjusted coverage rate is calculated after excludingfrom the calculation workers who are not legally entitled to collective bargaining(e.g. some public sector workers in some countries). As a result, the workforce used inthe coverage calculation can, in some cases, be smaller than the workforce used in themembership calculation (which includes all wage and salary employees).

Our main interest, however, is in trends in unionisation rates over time. Figure 3shows the coverage and membership rates for the 21 rich economies in our sample from1960 (or the earliest year available) to the late 2000s (typically 2007 or 2008). Thedashed line represents union coverage, and the solid line represents union membership.

Trends in unionisation rates differ substantially across the sample. Initially, weidentify five distinct patterns over the five decades. First, in just over half the coun-tries, union coverage rates were roughly constant or even rising (Austria, Belgium,Canada, Denmark, Finland, France, the Netherlands, Norway, Spain, Sweden andSwitzerland). Second, in a smaller share of countries (7 of the 21), union membershiprates remained roughly constant or rose slightly (Belgium, Canada, Denmark,Finland, Norway, Sweden and Spain after 1980). Third, in 10 of the 21 countries,

16 Figure 3 below reproduces the data exactly as they appear in the ICTWSS data, showing somewhaterratic behaviour before the early 1980s. For details on trade unions in Greece, Portugal and Spain in thefirst three decades after the end of the Second World War, see Ioannou (2000), Robolis (2008) and Teixeira(2001).



Figure 2: Union coverage and union membership, 21 rich countries, 2007.*2008 data used

Source: 1960–2010 Institutional Characteristics of Trade Unions, Wage Setting,State Intervention and Social Pacts data from Amsterdam Institute for AdvancedLabour Studies.



Figure 3: Union coverage and union membership, 21 rich countries, 1960–2010




union coverage was falling (Australia, Germany, Greece, Ireland, Italy, Japan, NewZealand, Portugal, UK and the United States). Fourth, in two-thirds of the sample,union membership was falling, frequently sharply (Australia, Austria, France,Germany, Greece, Ireland, Italy, Japan, the Netherlands, New Zealand, Portugal,Switzerland, UK and United States). Finally, while in most countries union coverageand membership trends tracked each other (both rising or both falling) or were not farout of sync, coverage and membership trends did diverge in five countries (Austria,France, the Netherlands, Norway and Switzerland). In all five cases, coverage rateswere flat or rising, while membership rates dropped.

This initial review of the data raises serious doubts about the inevitability of uniondecline. While union membership rates fell in two-thirds of the sample, membershiprates were flat or constant in the other third. While union coverage rates fell in justunder half of the 21 countries, coverage rates were flat or rising in the other half.

At least at face value, globalisation and technological progress do not appearcapable of explaining these observed differences, primarily because all of these econo-mies have been subjected to similar sets of forces in both regards. Figure 4, forexample, shows that exports and imports are a large share of all 21 economies,ranging from just under 30 per cent of total gross domestic product in the case of theUnited States to just over 160 per cent in the case of Belgium.17 If anything, countrieswith a higher level of globalisation (by this measure) appear to have a higher level ofunion coverage.18

17 The trade measure is the sum of the absolute value of exports and imports, both expressed as a share ofGDP.18 Rodrik (1997) argues that higher levels of social protection, typically associated with higher levels ofunion coverage, facilitate trade openness by reducing the adjustment costs for workers whose interests are

Figure 4: Union coverage and trade as share of gross domestic product (GDP),21 rich countries, 2007

Source: Authors’ analysis of Institutional Characteristics of Trade Unions, WageSetting, State Intervention and Social Pacts and World Bank data.



While there are differences in the technological level, the 21 countries generallyhave similar levels of access to and use of an important range of business technologies.Figure 5, for example, shows each country’s score (on a scale from zero to six) on theWorld Economic Forum (WEF)’s Networked Readiness Index in 2009–10. TheUnited States ranked fourth (after Sweden, Denmark and Switzerland),19 with all butGreece, Italy, Portugal and Spain clustered fairly closer together. The figure alsographs the WEF technology index against union coverage and finds no significantrelationship across our sample.

Given that all of these countries were at approximately the same level of economicdevelopment and all were subjected to the same kinds of pressures from globalisationand technological progress, the different trends apparent across the rich economiesstrongly suggest that more globalisation and better technology do not inexorably leadto lower unionisation rates.

4 UNIONISATION AND NATIONAL POLITICS

We argue, instead, that national politics, not globalisation or technology, are themost important determinants of unionisation trends over time. We draw on theextensive literature, initiated by Gosta Esping-Andersen,20 on the ‘worlds of welfarecapitalism’ in rich countries. This research has typically organised the better-off

threatened by increased foreign competition. Dreher and Gaston (2005) argue that a more sophisticatedindex of globalisation suggests that globalisation has had an effect on unionisation.19 In the full WEF list, the United States also placed behind Singapore, which is not included in our sample.20 Gosta Esping-Andersen (1990).

Figure 5: Union coverage and WEF (World Economic Forum) index of networkreadiness, 21 rich nations

Source: Authors’ analysis of Institutional Characteristics of Trade Unions, WageSetting, State Intervention and Social Pacts and WEF data (2009–10).



OECD economies into three or four categories by broad features of their politicalsystems and then explored the impact of these different systems on social and eco-nomic outcomes. The terminology and exact composition of the countries variesacross the hundreds of published papers,21 but three frequently recurring groupingsare, loosely: ‘social democratic’; ‘Christian democratic’; and ‘liberal market’. Thework of Hall and Soskice22 on ‘varieties of capitalism’ has also been influential, butthey divide the same set of countries into only two categories ‘coordinated marketeconomies’ and ‘liberal market economies’.

We follow, with one minor modification and a slight change of terminology,the political typology used in Navarro, Schmitt and Astudillo,23 which assigns 21OECD countries to one of four broad political regimes: ‘social democratic’(Denmark, Finland, Norway and Sweden); ‘continental market’ (Austria, Belgium,France, Germany, Italy, the Netherlands and Switzerland); ‘liberal market’, by the19th century and contemporary European usage of the term liberal (Australia,Canada, Ireland, Japan, New Zealand, UK and United States); and ‘ex-dictatorships’(Greece, Portugal and Spain).24

The four traditions are based on an analysis of the political parties in power in thesecountries during the years 1946 through 1980.25 The minor modification of theNavarro et al. framework is that here we have moved Austria from the ‘socialdemocratic’ to the ‘continental market’ group. While Austria had social democraticgovernments during much of the initial post-war period (1946–80), the Austrian socialdemocrats often ruled in coalition with Christian democrats. Our findings below arenot sensitive to where we place Austria, nor to other common variations in countryclassification.

Figures 6–9 present the same unionisation trends as in Figure 3 but now groupingthe rich countries by their political system. This simple exercise reveals strong simi-larities in unionisation trends within—and strong differences across—politicalgroupings. In reviewing these trends by country type, we focus on the period from1980, when globalisation and technology accelerated in most accounts.

The social democratic countries (Figure 6) have consistently high levels of unioncoverage (dashed line) and union membership (solid line). Coverage rates were flat oreven rising in all four countries. Membership rates, which are nearly as high ascoverage rates, changed little or fell slightly.

The continental market economies (Figure 7), except arguably Switzerland, havehigh levels of union coverage. Since 1980, these coverage rates remained roughly flator even increased in six of the seven countries. The exception is Germany, wherecoverage fell 25 percentage points between 1980 and the late 2000s. By contrast, unionmembership rates fell in all seven countries (though the decline was small in Belgium).All the continental market economies show a large gap between coverage and mem-bership rates. The collective bargaining systems in these countries extend negotiatedwages and employment conditions to many non-union workers.

21 See Scruggs and Allan (2006) and Huber and Stephens (2005).22 Hall and Soskice (2001).23 Navarro et al. (2004).24 We have also changed our category labels slightly. Navarro et al. (2004) use ‘social democratic’,‘Christian/conservative’, ‘liberal’, and ‘ex-dictatorships’.25 See Huber and Stephens (2001a; 2001b) and Navarro et al. (2004) for further discussion and a descriptionof the criteria used for assigning countries to categories.



The liberal market economies (Figure 8) generally have relatively low unionisationrates and these rates have declined in every one of these countries. In the late 2000s,relative to the rest of the liberal market economies, coverage rates were higher inAustralia and Ireland and membership rates were higher in Canada, Ireland and UK.Between 1980 and 2007, union coverage and membership rates fell only about fivepercentage points in Canada. In the rest of these countries, however, unionisationrates fell substantially. In all of the liberal market economies except Australia, thereis little difference between coverage and membership rates because the collectivebargaining systems generally do not extend contract terms to non-union workers.

Results are less clear cut for the ex-dictatorships of Greece, Portugal and Spain(Figure 9). Up until the mid-1970s, when all three were ruled by authoritarian right-wing governments, union membership was either mandatory (in Portugal, in state-controlled unions) or essentially illegal (Spain and Greece).26 As a result, through

26 For a discussion of unions in Portugal and Spain during the transition to democracy, see Teixeira (2001);for Greece, see Ioannou (2000) and Robolis (2008).

Figure 6: Union coverage and union membership, social democraticeconomies, 1960–2010




about 1980, the ICTWSS coverage and membership data show either erratic patterns(Portugal and Spain) or contain no data (Greece). Focusing on the period from 1980to the end of the 2000s, union coverage rates rose in Spain and declined slightly inGreece and substantially in Portugal. Membership rates, meanwhile, fell in all threecountries (sharply in Portugal and only slightly in Spain).

Figure 10 summarises the changes in union coverage between 1980 and the late2000s. The 21 countries in the sample are organised by the four political categories,from top to bottom: social democratic, continental market, liberal market andex-dictatorships. While there is some variation within each political grouping, thechange in coverage rates is strongly influenced by national political tradition. Between1980 and 2007, the social democratic countries saw coverage rates increase, onaverage, five percentage points. Over the same period, the continental market econo-mies experienced a small decline in coverage that averaged four percentage points.

Figure 7: Union coverage and union membership, continental marketeconomies, 1960–2010




Meanwhile, coverage rates fell in all of the liberal market economies, with an averagedecline of 26 percentage points (20 percentage points at the median). In theex-dictatorships, the average change over the period in coverage was close tozero.

Corresponding data for union membership rates appear in Figure 11. Thechange in membership rates also clusters by political category. Membership ratesfell across all four types, but the average changes differ substantially acrossthe groups. On average, membership rates declined 5 percentage points for socialdemocratic countries; 14 percentage points for continental market economies; 23percentage points for liberal market economies; and 18 percentage points forex-dictatorships.

Taken together, the data on changes in union coverage and membership in Fig-ures 10 and 11 suggest that national political traditions (as embodied in labour law,industrial relations practices, political party structures and other factors) are a major

Figure 8: Union coverage and union membership, liberal marketeconomies, 1960–2010




determinant of trends in unionisation over the last three decades.27 These findings arerobust to the most common reclassifications of political traditions.28

5 CONCLUSION

National political traditions established in the period 1946 through 1980 have a strongcapacity to predict changes in unionisation rates from 1980 to the present. Thechanging nature of work—driven by technological change and globalisation—undoubtedly weigh on unions in the United States and other rich democracies. But,trends in unionisation vary substantially and systematically across countries by broadnational political types. Unions in countries in the social democratic tradition have

27 Our findings are consistent with the more detailed analysis of union membership trends by Sano andWilliamson (2008), who conclude ‘concerns over globalisation are not as crucial to union densities as manyanalysts have suggested’ (p. 496).28 For example, dividing the sample according to Hall and Soskice’s (2001) classifications, between 1980and 2007, union coverage fell an average of 29 percentage points in their ‘liberal market’ countries and anaverage of only 2 percentage points in their ‘coordinated market’ economies.

Figure 9: Union coverage and union membership, ex-dictatorships, 1960–2010




managed to maintain and even expand collective bargaining coverage while experi-encing little or no decline in trade union density. Countries, such as the United States,in the liberal market tradition, meanwhile, have generally seen collective bargainingcoverage and trade union density plummet in the era of SBTC and globalisation. Andcountries identified with the continental market (sometimes referred to as the Chris-tian Democratic) tradition have fallen somewhere in between, with small to moderatedeclines in collective bargaining and somewhat larger declines in union density.

The differences within countries in outcomes for collective bargaining andunion density are suggestive that the last three decades have seen a struggle between

Figure 10: Change in union coverage, 1980–2007




economic fundamentals and political power. In most of the countries analyzed here,collective bargaining coverage and union density have moved in the same direction. Inthe large majority of cases, however, union density, which is arguably more directly afunction of economic fundamentals because it is directly tied to the ‘supply’ of unionjobs, has performed worse than collective bargaining coverage, which is more directlya function of political institutions and therefore political power.

Our approach has foregone the specificity offered by formal econometric analysisbuilt around country-level panel data sets, which has been successfully employed inearlier research seeking to explain differences in international trends in unionisation

Figure 11: Change in union membership, 1980–2007




(see, e.g. Lee, 2005). That research has contributed in important ways to our under-standing of changes and differences in unionisation across countries and over time.But, by its nature, regression analysis attempts to quantify the effects of individualvariables (in this context, features of national labour market and welfare-stateinstitutions)—holding other variables constant. The ‘varieties of capitalism’ andrelated literatures, however, have emphasised the importance of the ways in whichinstitutions combine, reinforce, and sometimes undermine each other, followingbroadly similar patterns across national political regimes. ‘Interaction’ terms inregression analysis can only poorly capture these institutional dynamics andinterplay. In the end, the four simple political groupings used here show a remarkablecapacity to explain patterns in unionisation across the main OECD countries over thelast three decades.

Our findings suggest the primacy of politics over technology and globalisation butshed little light on the particular features of the social democratic or continentalmarket economies that have allowed them to resist the decline in collective bargainingcoverage that appears to have so strongly affected the liberal market countries. Ourfindings also demonstrate that, across political regimes, collective bargaining has heldup better than union density, but our methods do not allow us to explore the under-lying causes for this frequent divergence. Given the evident differences in unionisationtrends across the four political regimes, future research would do well to focus on thespecific political features that have produced these diverging outcomes. Panel datatechniques may help, but the complexity of the interactions involved suggest thatcareful historical case studies are likely to provide more nuanced and convincingexplanations.

Acknowledgements

The authors would like to thank Eileen Appelbaum, Dean Baker, Kris Warner andtwo anonymous referees for helpful comments and the Ford Foundation and thePublic Welfare Foundation for generous financial support.

References

Acemoglu, D., P. Aghion and G. L. Violante (2000), ‘Deunionization, Technical Change, andInequality’, Carnegie-Rochester Conference Series on Public Policy, 55, 229–264.

Autor, D., F. Levy and R. J. Murnane (2003), ‘The Skill Content of Recent TechnologicalChange: An Empirical Exploration’, Quarterly Journal of Economics, 118, 4, 1279–1334.

Baldwin, R. E. (2003), The Decline of US Labor Unions and the Role of Trade (Washington, DC,Peterson Institute for International Economics).

Baumol, W. J. (1967), ‘Macroeconomics of Unbalanced Growth: The Anatomy of UrbanCrises’, American Economic Review, 57, 3, 415–426.

Blanchflower, D. G. (2007), ‘International Patterns of Union Membership’, British Journal ofIndustrial Relations, 45, 1, 1–28.

Bluestone, B. and B. Harrison (1982), The Deindustrialization of America (New York, BasicBooks).

Bockerman, P. and R. Uusitalo (2006), ‘Erosion of the Ghent System and Union MembershipDecline: Lessons from Finland’, British Journal of Industrial Relations, 44, 2, 283–303.

Brady, D. (2007), ‘Institutional, Economic or Solidaristic? Assessing Explanations for Union-ization across Affluent Democracies’, Work and Occupations, 34, 1, 67–101.



Brady, D. and M. Wallace (2000), ‘Spatialization, Foreign Direct Investment and LaborOutcomes in the American States, 1976–1996’, Social Forces, 79, 1, 67–105.

Bronfenbrenner, K. (2000), ‘Uneasy Terrain: The Impact of Capital Mobility on Workers,Wages, and Union Organizing’, statement submitted to the U.S. Trade Deficit ReviewCommission.

Bronfenbrenner, K. (2009), No Holds Barred: The Intensification of Employer Opposition toOrganizing (Washington, DC, Economic Policy Institute).

Bryson, A. and R. Gomez (2005), ‘Why Have Workers Stopped Joining Unions?: Accountingfor the Rise in Never-Membership in Britain’, British Journal of Industrial Relations, 43, 1,67–92.

Bryson, A., B. Ebbinghaus and J. Visser (2011), ‘Introduction: Causes, Consequences, andCures of Union Decline’, European Journal of Industrial Relations, 17, 2, 97–105.

Busmeyer, M. R. (2009), ‘From Myth to Reality: Globalisation and Public Spending in OECDCountries Revisited’, European Journal of Political Research, 48, 4, 455–482.

Cameron, D. R. (1978), ‘The Expansion of the Public Economy: A Comparative Analysis’,American Political Science Review, 72, 4, 1243–1261.

Castells, M. (2000), The Rise of the Network Society (Oxford, Blackwell Publishers).Dickens, W. and J. S. Leonard (1985), ‘Structural Changes in Unionization: 1973–1981’,

National Bureau of Economic Research Working Paper No. 1882.Disney, R., A. Gosling and S. Machin (1995), ‘British Unions in Decline: Determinants of

the 1980s Fall in Union Recognition’, Industrial and Labor Relations Review, 48, 3, 403–419.

Dreher, A. and N. Gaston (2005), ‘Has Globalisation Really Had No Effect on Unions?’, KOFSwiss Economic Institute Working Paper No. 110 (Zurich).

Ebbinghaus, B. and J. Visser (1999), ‘When Institutions Matter: Union Growth and Decline inWestern Europe, 1950–1995’, European Sociological Review, 15, 2, 135–158.

Esping-Andersen, G. (1990), The Three Worlds of Welfare Capitalism (Princeton, NJ, PrincetonUniversity Press).

Farber, H. S. (2005), ‘Union Membership in the United States: The Divergence between thePublic and Private Sectors’, Industrial Relations Section Working Paper No. 503 (PrincetonUniversity).

Farber, H. and A. Krueger (1993), ‘Union Membership in the United States: The DeclineContinues’, in B. Kaufman and M. Kleiner (eds), Employee Representation: Alternatives andFuture Directions (Ithaca, New York, Cornell University Press) pp. 105–134.

Farber, H. S. and B. Western (2001), ‘Accounting for the Decline of Unions in the PrivateSector, 1973–1988’, Journal of Labor Research, 22, 3, 459–485.

Flanagan, R. J. (2007), ‘Has Management Strangled U.S. Unions?’, in B. E. Kaufman andJ. T. Bennett (eds), What Do Unions Do? A Twenty-Year Perspective (Edison, NJ, Transac-tion Publishers) pp. 459–491.

Freeman, R. B. (2007), ‘What Do Unions Do? The 2004 M-Brane Stringtwister Edition’,in J. T. Bennett and B. E. Kaufman (eds), What Do Unions Do? A Twenty Year Perspective(Edison, NJ, Transaction Publishers) pp. 607–636.

Freeman, R. B. and J. Rogers (2006), What Workers Want (Ithaca, NY, Cornell UniversityPress).

Giddens, A. (2007), Europe in the Global Age (Cambridge, Polity Press).Goos, M. and A. Manning (2007), ‘Lousy and Lovely Jobs: The Rising Polarization of Work

in Britain’, Review of Economics and Statistics, 89, 1, 118–133.Goos, M., A. Manning and A. Salomons (2009), ‘The Polarization of the European Labor

Market’, American Economic Review Papers and Proceedings, 99, 2, 58–63.Hall, P. A. and D. Soskice (2001), Varieties of Capitalism: The Institutional Foundations of

Comparative Advantage (Oxford, Oxford University Press).Huber, E. and J. D. Stephens (2001a), Development and Crisis of the Welfare State: Parties and

Policies in Global Markets (Chicago, IL, University of Chicago Press).



Huber, E. and J. D. Stephens (2001b), ‘Welfare State and Production Regimes in the Era ofRetrenchment’, in P. Pierson (ed.), The New Politics of the Welfare State (Oxford, OxfordUniversity Press) pp. 107–145.

Huber, E. and J. D. Stephens (2005), ‘Welfare States and the Economy’, in N. J. Smelserand R. Swedberg (eds), The Handbook of Economic Sociology (Princeton, NJ, PrincetonUniversity Press) pp. 552–574.

Ioannou, C. A. (2000), Trade Unions in Greece: Development, Structures and Prospects(Bonn, Friedrich Ebert Stiftung) Retrieved 14 October 2011 from http://www.fes.de/fulltext/bueros/athen/00740002.htm

Katzenstein, P. J. (1985), Small States in World Markets: Industrial Policy in Europe (Ithaca,NY, Cornell University Press).

Kurzer, P. (1991), ‘Unemployment in Open Economies: The Impact of Trade, Finance andEuropean Integration’, Comparative Political Studies, 24, 1, 3–30.

Lee, C.-S. (2005), ‘International Migration, Deindustrialization and Union Decline in 16Affluent OECD Countries, 1962–1997’, Social Forces, 64, 1, 71–88.

Lipset, S. M. and N. M. Meltz (2004), The Paradox of American Unionism: Why American LikeUnions More Than Canadians Do, But Join Much Less (Ithaca, NY, Cornell UniversityPress).

Logan, J. (2006), ‘The Union Avoidance Industry in the United States’, British Journal ofIndustrial Relations, 44, 4, 651–675.

Machin, S. (2000), ‘Union Decline in Britain’, British Journal of Industrial Relations, 38, 4,631–645.

Magnani, E. and D. Prentice (2003), ‘Did Globalization Reduce Unionization? Evidence fromUS Manufacturing’, Labour Economics, 10, 6, 705–726.

Navarro, V., J. Schmitt and J. Astudillo (2004), ‘Is Globalisation Undermining the WelfareState? ’, Cambridge Journal of Economics, 28, 1, 133–152.

Nolan, P. and G. Slater (2010), ‘Visions of the Future, the Legacy of the Past: Demystifying theWeightless Economy’, Labor History, 51, 1, 7–27.

Pencavel, J. (2005), ‘Unionism Viewed Internationally’, Journal of Labor Research, 26, 1,65–97.

Rifkin, J. (1996), The End of Work: The Decline of the Global Labor Force and the Dawn of thePost-Market Era (New York, Penguin Books).

Robolis, S. (2008), ‘The Trade Union Movement in Greece: Characteristics, Organisation,Prospects’, Manuscript (Panteion University) Retrieved 14 October 2011 from http://www.turi-network.eu/content/download/1147/5551/file/How%20trade%20union%20membership%20is%20understood%20in%20Greece.pdf

Rodrik, D. (1997), Has Globalization Gone Too Far? (Washington, DC, Institute for Interna-tional Economics).

Sano, J. and J. B. Williamson (2008), ‘Factors Affecting Union Decline in 18 OECD Countriesand Their Implications for Labor Movement Reform’, International Journal of ComparativeSociology, 49, 6, 479–500.

Scharpf, F. W. (1991), Crisis and Choice in European Social Democracy (Ithaca, NY, CornellUniversity Press).

Scharpf, F. W. and V. A. Schmidt (eds) (2000), Welfare and Work in the Open Economy: FromVulnerability to Competitiveness, Vol. I (Oxford, Oxford University Press).

Schmitt, J. and B. Zipperer (2009), Dropping the Ax: Illegal Firings During Union ElectionCampaigns, 1951–2007 (Washington, DC, Center for Economic and Policy Research).

Scruggs, L. (2002), ‘The Ghent System and Union Membership in Europe, 1970–1996’, Politi-cal Research Quarterly, 55, 2, 275–297.

Scruggs, L. and J. Allan (2006), ‘Welfare-State Decommodification in 18 OECD Countries:A Replication and Revision’, Journal of European Social Policy, 16, 1, 55–72.

Scruggs, L. and P. Lange (2002), ‘Where Have All the Members Gone? Globalization, Insti-tutions, and Union Density’, Journal of Politics, 64, 1, 126–153.



Slaughter, M. (2007), ‘Globalization and Declining Unionization in the United States’, Indus-trial Relations, 46, 2, 329–346.

Teixeira, P. (2001), ‘Labour Market Transition in Portugal, Spain, and Poland’, Manuscript(Economics Faculty, University of Coimbra) Retrieved 14 October 2011 from http://gemf.fe.uc.pt/workingpapers/pdf/2001/gemf01_05_newversion.pdf

Van Rie, T., I. Marx and J. Horemans (2011), ‘Ghent Revisited: Unemployment Insurance andUnion Membership in Belgium and the Nordic Countries’, European Journal of IndustrialRelations, 17, 2, 125–139.

Visser, J. (2006), ‘Union Membership Statistics in 24 Countries’, Monthly Labor Review,January, 38–49.

Visser, J. (2011), Data Base on Institutional Characteristics of Trade Unions, Wage Setting, StateIntervention and Social Pacts, 1960–2010 (ICTWSS), Version 3.0 (Amsterdam, AmsterdamInstitute for Advanced Labour Studies AIAS, University of Amsterdam).

Wallerstein, M. and B. Western (2000), ‘Unions in Decline: What Has Changed and Why? ’,Annual Review of Political Science, 3, 355–377.

Wood, A. (1994), North-South Trade, Employment and Inequality: Changing Fortunes in aSkill-Driven World (Oxford, Clarendon Press).



Documents

Politics matter: changes in unionisation rates in rich countries, 1960–2010