Corporatism and Unemployment in the 1980s and 1990s Author(s): Lane Kenworthy Source: American Sociological Review, Vol. 67, No. 3 (Jun., 2002), pp. 367-388 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/3088962 Accessed: 16-11-2015 12:19 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/ info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Sage Publications, Inc. and American Sociological Association are collaborating with JSTOR to digitize, preserve and extend access to American Sociological Review. http://www.jstor.org
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions CORPORATISM AND UNEMPLOYMENT IN THE 1980S AND 1990S
LANE KENWORTHY Emory University
A number of studies have found an association between corporatist institutions and low unemployment in the 1970s and/or 1980s. Three gaps in our understanding of corporatism's labor market effects are addressed here: (1) Which of the two princi- pal forms of corporatism-corporatist wage-setting or union participation in eco- nomic policymaking, or both-generates these effects? (2) What are the causal mechanisms? (3) Did these effects continue in the 1990s in the face of globalization, restrictive monetary policy, growing dissension within labor movements, and related developments? The impact of corporatism across 16 affluent OECD countries in the 1980s and 1990s is assessed using pooled time-series cross-section analysis. The results suggest that wage coordination was conducive to low unemployment in the 1980s because it fostered moderation in labor costs, spurred faster economic growth, and encouraged governments to more aggressively pursue policies to reduce unemployment. In the 1990s, this effect disappeared, largely because unemployment outcomes in low wage-coordination countries improved rather than because unem- ployment outcomes in high wage-coordination countries deteriorated. Union partici- pation in economic policymaking was associated with low unemployment throughout the two decades, conditional on the presence of leftist government. Union participa- tion appears to have had this effect mainly via government policy.
C ORPORATISM refers to various force are bargained by national union and types of institutional arrangements employer confederations. A second impor- whereby important political-economic deci- tant form is participationby such confedera- sions are reached via negotiation between, or tions in the formation of government deci- in consultation with, peak-level representa- sions about nonwage issues such as fiscal tives of employees and/or employers, some- policy, monetary policy, active labor market times involving other interest groups and the policy, and so on. Corporatist arrangements state. One of the best-known forms of cor- have been among the key political-economic poratism is centralized wage-setting, in institutions in a number of affluent capitalist which wages for a large portion of the work nations since the 1960s. They have thus been of substantial interest to macro-comparative Direct all correspondence to Lane Kenworthy, sociologists, political scientists, and econo- Department of Sociology, Emory University, At- mists. lanta, GA 30322 ([email protected]). For A number of studies have found an asso- helpful comments I thank Alex Hicks, Bernhard ciation between the prominence of corporat- Kittel, and the ASR reviewers, as well as Art ist institutions and low unemployment in the Alderson, Anke Hassel, Fritz Scharpf, Michael 1970s and/or 1980s (Bruno and Sachs 1985; Smith, Wolfgang Streeck, and Michael Waller- Calmfors and Driffill 1988; Cameron 1984; stein. I am also grateful to Stephen Nickell for Compston 1997; Garrett 1998; Hall and allowing me access to data from the Centre for Hicks Kenworthy 1998; Economic Performance. Portions of this paper Franzese 1998; and were written while I was a visiting scholar at the Iversen 1999; Janoski, McGill, and Tinsley Max Planck Institute for the Study of Societies 1997; Layard, Nickell, and Jackman 1991; in Cologne, Germany. Scharpf [1987] 1991; Soskice 1990). Unem-
AMERICAN SOCIOLOGICAL REVIEw, 2002, VOL.67 (JUNE:367-388) 367
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions 368 AMERICAN SOCIOLOGICAL REVIEW ployment is one of the most important eco- existing research has mistakenly emphasized nomic outcomes in affluent countries. It is a the wrong form of corporatism. major contributor to individual unhappiness (Clark and Oswald 1994) and to societal ills WHAT ARE THE CAUSAL MECHANISMS? such as poverty, neighborhood decay, and crime (Chiricos 1987; Wilson 1996); it also Among researchers interested in wage deter- impedes economic growth (Okun 1973) and mination, the predominant notion has been is a potential threat to the welfare state that centralized or coordinated wage-setting (Esping-Andersen 1999). Large-scale unem- yields low unemployment by engendering ployment has arguably been the prime eco- real wage restraint. (I use the term "wages" nomic, social, and political issue in much of throughout as shorthandfor wages plus ben- Western Europe over the past two decades. efits.) The general logic is relatively simple, I attempt to answer three questions that although specific applications of it can be reflect notable gaps in previous research complex (Franzese 1999; Organization for linking corporatism with low unemploy- Economic Cooperation and Development ment: (1) Are corporatism's effects on unem- [OECD] 1997). If employees bargain ag- ployment generated by both corporatist gressively for high wage increases and em- wage-setting and union participation in eco- ployers acquiesce, the latter can do five main nomic policymaking, or by only one of these things in response: raise productivity, raise forms of corporatism? (2) What are the prices, reduce profits paid out to investors, causal mechanisms through which such ef- reduce investment, and/or reduce the num- fects occur? (3) Did these effects continue ber of employees. Where wages are bar- in the 1990s in the face of globalization, re- gained at a large number of separate firms strictive monetary policy, growing dissen- or plants, each individual union may reason- sion within labor movements, and related ably hope that its employer's response to a developments? pay increase will consist predominantly of some combination of the first four options. None of these responses will necessarily THREE QUESTIONS have an adverse short-term effect on em- ployment or inflation-adjustedwages, which WHICH FORM OF CORPORATISM? are the principal concerns of union negotia- The focus of most research on corporatism's tors. Even if the firm chooses to reduce em- macroeconomic effects has been on wage- ployment, those laid off should be able to setting arrangements. Yet some scholars find work elsewhere as long as the pattern view corporatism as participation by orga- of wage increases and layoffs is not general- nized interest groups in various types of ized throughout the economy. Thus, where public policymaking. This conception, origi- bargaining is decentralized and uncoordi- nally highlighted by Lehmbruch (1984), has nated, there is an incentive for unions to pur- received relatively little attention in the em- sue a strategy of wage militancy. pirical literatureduring the past two decades. By contrast, if the wage negotiations cover Indeed, only three studies have examined the a relatively large share of the work force, effect of union participation in policymak- union bargainerscan be reasonably sure that ing on unemployment (Boreham and a large wage increase will have an adverse Compston 1992; Compston 1997; Traxler, impact on their members. For instance, when Blaschke, and Kittel 2001:227-31). Each an individual firm raises prices, this is likely found a beneficial effect. to have little or no effect on the living stan- No prior study has examined the macro- dard of that firm's own employees (unless economic impact of these two principal the company's goods or services happen to forms of corporatism in conjunction. Analy- account for a large share of what those em- ses of corporatist wage-setting have not con- ployees consume). But when firms repre- trolled for union participation in policy- senting a sizable share of the economy raise making, and vice-versa. These two forms are prices, the resulting inflation offsets or nul- likely to be correlated with one another,rais- lifies the wage gains of most workers. Simi- ing the possibility that at least some of the larly, a reduction in employment at a single
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions CORPORATISM AND UNEMPLOYMENT 369 firm does not necessarily reduce the job ful sector, as in Germany, or by a group of prospects for workers who are laid off; but influential firms, as in Japan. Pattern-setting if wage increases and ensuing layoffs are exists where bargaining is formally decen- economy-wide, employment opportunities tralized, but in practice one or a few wage will diminish. Centralized or coordinated settlements (e.g., the metalworkers' settle- wage-setting thus generates an incentive for ment in Germany) are seen by all or most wage moderation. other bargaining parties as determining the Wage moderation, in turn, is presumed to outcome that they will follow. contribute to lower unemployment. The as- Centralization has received the bulk of at- sumption is that slower growth of real labor tention in the corporatist literature. But to costs (often referred to as "real unit labor understandthe effects of wage-setting insti- costs") leads to lower unemployment. "La- tutions on labor cost developments and un- bor costs" refer to wages and benefits ad- employment, it makes more sense to focus justed for productivity, and "real" denotes on wage coordination (Flanagan 1999:1172; adjustment for inflation. If wage increases Soskice 1990; Traxler and Kittel 2000). The are offset by increases in productivity and/ key is that major players have an incentive or prices, then real labor costs have not in- to moderate labor costs and minor players creased. Consequently, there is no need for tend to adhere to what the major players de- employers to reduce payments to investors, cide. Pattern-settingand informal bargaining investment, or employment. However, if real centralization are likely to be just as effec- labor costs do increase, one result-perhaps tive as formal centralization at generating the main result-will be a higher unemploy- this set of incentives and behaviors. The lim- ment rate as employers respond by reducing ited empirical evidence accumulated thus far the size of their work force. has generally been supportive (Hall and Some researchers have focused on the ef- Franzese 1998; Nickell and Layard 1999; fects of wage-bargaining centralization, Soskice 1990; Traxler and Kittel 2000). while others have emphasized wage coordi- If labor cost restraint has been the princi- nation. Centralizationrefers to the level(s) pal presumed link between wage-setting ar- peak, sectoral, company/plant-at which rangements and unemployment, a second wages are determined. Coordination refers link is economic growth. One of the out- to the degree to which minor players in the comes of wage coordination, achieved either wage-setting process (e.g., firm- or plant- informally or explicitly in corporatist pacts, level bargainers) intentionally follow along may be greater investment, which in turn with what the major players decide. Bargain- tends to spur more rapid growth of economic ing centralization is one means of achieving output (Alvarez, Garrett, and Lange 1991; coordination. A second is state-imposed cen- Hicks 1988). Faster economic growth gener- tralization, as in Belgium, the Netherlands, ally helps to reduce the unemployment rate. Denmark in some years, and even Canada A third view holds that the mechanism and the United States in a few years. A third linking corporatist wage-setting with low is guidance of industry or firm-level bargain- unemployment is government policy. Policy ing by peak union and/or employer confed- orientations, in this conception, are a key erations, as in Switzerland, Austria (prior to determinant of cross-country differences in the mid-1980s), Norway (in a number of unemployment rates (Hibbs 1977; Korpi years), the Netherlands (since 1983), and 1991; Soskice 2000; Therborn 1986). Italy (since 1993). This might be termed "in- Policymakers in countries with coordinated formal centralization." In informal central- wage-setting are likely to feel more confi- ization, union and/or employer confedera- dent than their counterparts in countries tions tacitly centralize the bargaining pro- with fragmented bargaining that labor cost cess by suggesting and/or approving wage increases will be moderate. Thus, they agreements that are formally negotiated at should tend to worry less about wage-push lower levels, or sometimes simply by shar- inflation. This may increase their willing- ing information and fostering a climate of ness to pursue stimulative monetary and/or consensus. A fourth means of achieving co- fiscal policy, active labor market policy, ordination is pattern-setting led by a power- public employment, or other programs that
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions 370 AMERICAN SOCIOLOGICAL REVIEW reduce unemployment. By contrast, policy- such policies are effective, the result should makers in nations with less coordinated be lower rates of joblessness. wage arrangements may feel compelled to I empirically explore these hypothesized resort to higher levels of unemployment in causal links between wage coordination, order to keep inflation in check (Hall and union participationin policymaking, and un- Franzese 1998; Kenworthy 1996; Soskice employment. Only a few cross-country stud- 1990). Note that this hypothesis does not ies of corporatism have actually examined imply that the rate of labor cost increase the relationship between wage-setting and necessarily will be lower in high-coordina- labor cost developments. Most have looked tion countries than in low-coordination only at the relationshipbetween wage-setting ones, only that policymakers in high-coor- and unemployment and have simply pre- dination countries will tend to have greater sumed that the link between wage-setting and confidence that labor cost increases will be labor cost restraint (and also between labor moderate. cost restraint and unemployment) occurs as Among the nations with relatively coordi- hypothesized. The few prior studies to assess nated wage-setting, Sweden and Denmark this link yielded inconclusive results and have relied heavily on active labor market were somewhat limited in scope. Bruno and policy and public employment, Norway and Sachs (1985), Layard et al. (1991, chap. 9), Austria on investment by the publicly owned and Western and Healy (1999) all found an business sector and government subsidies to association between corporatistwage-setting business, Germany on education and job and real wage restraint. Because of limited training, Japan on government support for data availability, however, these studies ex- the practice of "lifetime employment" in amined wages only in the manufacturingsec- large firms, Switzerland on displacement of tor. Although manufacturinghistorically has foreigners from the labor market, and Ire- been a pacesetter for wage developments in land on stimulative fiscal policy. The evi- the rest of the economy, it has accounted for dence suggests that these various policy only one-fifth of total employment in most strategies have been effective (Boix 1998, affluent countries in recent decades. Thus, it chap. 3; Layard et al. 1991; Martin 2000- is important to examine wage patterns in 2001; Nickell and Bell 1996; Therborn other sectors as well. A recent OECD (1997) 1986). And there is reason to believe that study used wage data for the entire economy coordinated wage-setting has played a causal and found no association between wage cen- role in the development and sustained use of tralization or coordination and changes in many of these strategies. In some instances, real wages. That analysis, however, was wage coordination is one element in an ar- based on a highly underspecified model. In ray of interlinked institutions that reinforce addition, each of these previous studies fo- one another (Dore 1987; Soskice 1999; cused on wages, despite the fact that the Streeck 1997). In the absence of wage coor- theory specifies a link with labor costs.1 dination, other elements of the institutional I am not aware of any empirical assess- configuration, such as a particularpolicy ap- ments of the hypothesis that corporatism- proach toward joblessness, might be weak- in the form of wage-setting and/or union par- ened. In other cases, anti-unemployment ticipation in economic policymaking-af- policies are introduced or continued as part fects unemployment via government policy. of an explicit "political exchange" for union wage restraint (Pochet and Fajertag 1997; 1 Western and Healy's (1999) study was not Visser and Hemerijck 1997). aimed at examining the mechanism linking cor- How might union participation in eco- poratism with low unemployment and so had no nomic policymaking affect unemployment? reason to consider labor costs. Kenworthy (1996) Unions typically desire low unemployment. and Traxler and Kittel (2000) each found links Thus, the more input unions have into eco- between wage-setting and nominal labor cost re- straint and between nominal labor cost restraint nomic policy decisions, the more likely it is and low inflation. But neither study examined that government policies will prioritize low real labor costs, which are presumed to be the unemployment (Boreham and Compston mechanism through which corporatism generates 1992; Compston 1997). To the extent that low unemployment.
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions CORPORATISM AND UNEMPLOYMENT 371
Boreham and Compston (1992), Compston 2000). In some countries, this reflected a (1997), and Traxler et al. (2001:227-31) desire to comply with the low-inflation re- found union participation to be associated quirement for European monetary integra- with low unemployment, but they did not tion. In all nations it is partly a product of test the purportedintervening mechanismss. heightened financial globalization-in par- ticular, the growing importance of currency speculators in determining currency values. DID CORPORATIST EFFECTS CONTINUE By the mid-1990s, more than $1 trillion in IN THE 1990S? currencies was traded each day, much of it Existing research has been confined to the by speculators using low inflation as their 1970s and 1980s. Did corporatist effects principal criterion for selecting currencies continue in the 1990S?2 There are five plau- in which to invest (Woodall 1995:10). Mas- sible scenarios. sive currency sell-offs provoked by even (1) Corporatist effects weakened or disap- moderate inflation-or expected inflation, peared-because corporatist institutions be- as gauged by low domestic interest rates came less effective at generating low unem- and/or sizable government deficits-are ployment than in previous decades.3 First, embarrassing to government officials and coordinated wage-setting may have become potentially highly destabilizing (Krugman less able to secure labor cost moderation, as 1998). Convergence in monetary policy re- wage drift at the firm- or plant-level increas- moved one important tool that governments ingly offset the restraint generated in coor- in corporatist countries may have formerly dinated negotiations. The chief cause of this used to influence the unemployment rate. In wage drift is thought to be dissension among addition, the heightened ease of capital exit different segments of the work force, par- may have pressured policymakers in corpo- ticularly over the issue of relative pay levels ratist countries to cut back on other pro- (Ahlen 1989; Hernes 1991). One important grams for reducing unemployment, such as division is among private-sector blue-collar, education, active labor market policy, and private-sector white-collar, and public-sector public employment. employees; another is among highly skilled (2) Corporatist effects weakened or disap- versus semiskilled and unskilled employees. peared-because noncorporatist institutions A second reason why corporatist coun- became more effective at generating low un- tries may have been less able to secure low employmentthan in previous decades. Grow- unemployment in the 1990s is declining ing international trade and heightened inte- policy autonomy. Loose monetary policy is gration of financial markets have increased one way to stimulate the economy and the competitive pressures faced by many thereby hold down the rate of joblessness, firms and also enhanced companies' ability but various aspects of globalization encour- to move to other countries. These develop- aged policymakers in most affluent nations ments have increased employers' motivation to pursue a restrictive monetary policy in to demand labor cost moderation as well as the 1990s (Boix 2000; Scharpf and Schmidt their leverage vis-a-vis employees. Hence, labor cost restraint may have become more 2 A few studies, such as Iversen (1999), common in countries with fragmented wage- Nickell and Layard (1999), and Western (2001), setting institutions, eroding the difference included the early 1990s but did not examine the between low-coordination and high-coordi- possibility of changes in effects during these nation countries (Streeck 1984; Windolf years. 1989). In this view, corporatism'seffects dis- 3 A decline in wage coordination's prominence appear in the 1990s because low-coordina- can be misinterpretedas a decline in its effective- tion countries are increasingly able to re- ness. Wage coordination decreased in Sweden strain labor costs and hence can achieve the and Australia in the 1990s. If these countries low levels of unemployment formerly con- were less successful at securing wage restraint and low unemployment in the 1990s than they fined to high-coordination countries. formerly had been, that might be due to the re- (3) Corporatist effects persisted. Many duction in wage coordination's prominence observers have expressed skepticism about rather than to a reduction in its effectiveness. the degree to which globalization has altered
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions 372 AMERICAN SOCIOLOGICAL REVIEW institutional structures and policy choices in dependent variables are available through the rich OECD countries (Berger and Dore 1999, but because of data limitations for the 1996; Campbell 2000; Garrett 1998; union participation variable the analysis Kenworthy 1997; Kitschelt et al. 1999). Fur- covers only the period through 1997. All thermore, both globalization and growing variables are described in Table 1. work force divisions were already present to Some scholars suggest that the employed a significant degree in the 1980s. Perhaps, share of the working-age population is a then, whatever impact corporatistinstitutions more telling indicator of labor market per- had in the 1980s continued in the 1990s. formance than the unemployment rate (4) Corporatist effects increased-because (Scharpf 1999:125-26; Smith 1999). Unem- corporatist institutions became even more ployment, after all, can be "hidden"by low effective at generating low unemployment levels of labor force participation (as in than in previous decades. With monetary Italy, Belgium, and Ireland) or by various policy lost as a policy tool, union wage ne- active labor market, early retirement, and gotiators interested in political exchange or other social-welfare policies (as in Sweden). union representatives directly involved in Yet using the employment rate as an indica- economic policymaking may have turned tor of labor market outcomes is not without their attention to active labor market policy, problems, particularlybecause nations differ government employment, or other programs in the frequency with which women, espe- aimed at reducing joblessness. To the extent cially marriedwomen, participatein the paid that these programs succeeded, the unem- labor force. This variation, which is a result ployment gap between corporatist and of differing cultural traditions and govern- noncorporatist countries might have wid- ment policies, yields sizable cross-country ened in the 1990s. differences in employment rates that have (5) Corporatist effects increased-because little or nothing to do with the health of the noncorporatist institutions became even less labor market. Therefore, despite its limita- effective at generating low unemployment tions, unemployment remains a preferable than in previous decades. With all nations indicator (also see Nickell 1997). As it turns committed to a tight money orientation, "ex- out, the causal patterns are similar for these cessive" increases in labor costs may have two measures of labor market performance. been even more likely in the 1990s to result (These and other results not shown here are in higher unemployment because govern- available from the author on request.) ments were increasingly unable to "mon- The wage-coordination variable is my own etize" such labor cost increases (i.e., to al- (Kenworthy 2001a). It is an index ranging low higher inflation in order to prevent a rise from 1 to 5. This is the only existing coordi- in unemployment). Thus, if fragmented nation variable that is measured annually wage-setting continued to promote wage and provides scores through the late 1990s. militancy, the unemployment performance Unlike other coordination indicators- of noncorporatist countries may have wors- Soskice (1990), Layard et al. (1991), OECD ened in the 1990s. (1997), Hall and Franzese (1998)-it does not attempt to capture the degree of actual coordination in each METHOD, DATA, AND VARIABLES wage country. It is ex- tremely difficult to measure the degree to I use pooled time-series cross-section re- which the various actors involved in the gressions to explore the effects of wage co- wage-setting process deliberately harmonize ordination and union participation in eco- their bargaining. To do so, the researcher nomic policymaking on unemployment over must factor in both the share of the work the period 1980 to 1997. Sixteen affluent force whose wages are deliberately pegged OECD countries are included: Austria, Bel- to the agreements) reached by the major gium, Canada, Denmark, Finland, France, player(s) and the degree to which minor Germany, Ireland, Italy, Japan, the Nether- players follow along (how closely they ad- lands, Norway, Sweden, Switzerland, the here). This information is not easily discern- United Kingdom, and the United States. ible, much less readily available. Existing Data for unemployment and many of the in- measures thus tend to be highly impression-
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions CORPORATISM AND UNEMPLOYMENT 373 istic (see Kenworthy 2001b). My scores are participation in policymaking should have based instead on a set of expectations about greater influence on policy decisions, and which institutional features of wage-setting thereby generate lower unemployment, in a arrangementsare likely to generate more or context of sympathetic government. I thus less coordination. This avoids the problem include a union participation x leftist gov- of limited information about the intentions ernment interaction term in the regressions. of various actors in the wage-setting process A negative coefficient is expected for this and the difficulty in ranking those intentions variable. even if such information were available. The I include a variety of control variables. rationale for the scoring is explained in Each has been found in previous studies to greater detail in Kenworthy (2001a, 2001b). affect employment performance and may be Despite the difference in measurement strat- correlated with wage coordination and/or egy, this index is closely correlated (r = .88 union participation in policymaking (Hall or better) with the Soskice, OECD, and Hall- and Franzese 1998; Kenworthy 2002; Korpi Franzese coordination measures.4 1991; Nickell and Layard 1999; Scarpetta The union participation in economic poli- 1996; Western 2001). (1) Trade: Unemploy- cymaking variable is from Compston (1997). ment in nations more heavily dependent on It is an index ranging from 0 to 10, with an- trade may be influenced to a greater extent nual scores. The index aims to capture the by trends in the international economy. (2) degree of union input into the economic pol- Union density: Controlling for wage coordi- icymaking process, ranging from no repre- nation, more extensive unionization is ex- sentation at all to broad agreement with pected to generate greater labor cost in- policymakers. Compston's scores go only creases and therefore produce higher unem- through 1992, but I have extended them ployment. (3) Employmentregulations: This through 1996 based on information in a re- index gauges the strictness of legislation on cent article by Compston (1998) and, for two working time, fixed-term contracts, employ- countries, on extrapolation. Compston ment protection, minimum wages, and em- (1997) suggests that the effect on unemploy- ployees' representation rights on works ment of union participation in policymaking councils and company boards. These types is likely to lag one to two years, and in his of regulations are commonly believed to statistical analysis he uses an average of the weaken employers' willingness to hire addi- scores for the previous two years. I do the tional employees. (4) Tax rate on a typical same here. Given that his (extended) scores employee: This is a measure of the tax go through 1996, I can examine unemploy- wedge between labor costs for firms and ment outcomes through 1997. Unfortunately, take-home pay for workers. A higher wedge the Compston measure is available for only is expected to reduce employer demand for 13 countries. Among the 18 OECD nations labor. (5) Duration of eligibility for unem- commonly included in comparative research ployment benefits: The longer one can re- on corporatist effects, Australia, Canada, Ja- ceive unemployment compensation, the pan, New Zealand, and the United States are weaker the incentive to get a new job. omitted. I have added scores for Canada, Ja- The causal links between corporatism and pan, and the United States. These three macroeconomic performance are explored countries are almost certain to be on the low using seven variables. Wage restraintis mea- end, except that in the early 1990s Japanese sured with an index of real labor costs. The unions began to participate somewhat exten- index is set at 1979 = 1 for each country to sively in peak-level tripartitebodies.' Union control for preexisting cross-national differ- ences in labor costs and thereby focus on developments in the 1980s and 1990s. Eco- 4 These correlations are based on a 1974-1989 period average, as the other wage-coordination measures are time-invariant (see Kenworthy not use those measures because Lehmbruch's is 2001b). based on the late 1970s and early 1980s and the 5 This is consistent with the scores for the only Traxler et al. scores suggest very little variation other measures of this kind, those of Lehmbruch over time in union participation in the poli- (1984:66) and Traxler et al. (2001:76, 312). I do cymaking process.
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions 374 AMERICAN SOCIOLOGICAL REVIEW
00 cn 0 00 r- I- C, cj,2 cn r- = 0 Z cn r- j=., m cn 73 O' cl 0 0 cn 0 -t:3 cn cn 0 4, M =3 00 r t:3 Cl
-t:3 >-. C13 73 .- " " " U > 73 m 0 0 0 cl 4-4 40. U 2. 73 0 -t:3-t:3 Q 73 C, 0
O
cli
4-4 73 tb 0 cn 0 m tm. 00 73 Z3 tt C m - C's tb ., >, -, -, tt m >, " r- r .0 tt tt tt tt tt 00 73 CLI M tb N Q C13 cn ,::3 ab 0 t; cl C's m = 94 = tb W - '.., tb UB.- C, r Cl >'73 tt cn = 0 0 It4 r 0 --4 C's 73 4) C,3 = 73 I 2 0 > tb C) = 4 0 m C4 '6 73.- -t:3 0 .0 , Q 0 " 00 -5 -, z 0 C, m C3 -t:3 - > 4-4 >-- r- 03 73 0 4 0 m 'n 0 " 0 .0 $:L, Q t,b 00 73 C, E tt m -t:3--4 00 73 m 0 4:3 N , 0 tb,.C,. 0 cli - 4-4 -t:3 0 4-4 0c)06 = Z Q 0 o cl cl N CA cn I I -, =It !:,, cn m m - 0 r- cn = cr : - 73 = 0 tb cn 73 m C's 73 m --4 0 tt C,3Q r 73 $. - >, cn $:L, --4 0 tt 0 E qj %-4 00 4-4 = 0 E .0 I- Q m : 00 O' m 0 - >, m =3 m -U- C13 - - 0 0 = 0 o cn 00 >, 0 C,3 0 > cl =3 " m cn =3 x - So ob 0 - -1 11:3 0 m m 0, CIN 73 r- 0-4 0 00 73 E 0 0 --, N 0 : E : --, -p -p C",$ 0 cl a m 7:3 r Q 73 0 0 0 o r 0 00 cn73 tt oc) 4. r- cl z 0 > > 0 0 0 0 Q s. cl cl 4. Q - cl too a, , 0 = 0 0 > -, = --4 - cl cl -I -$. tb 0 .4 .- tb ( C,3 73 to,=cn -r (U -, 40-4 "tb r- 40. ':L' M 0 M >1 0 tb0 = 0 00 cl 0 (1; Q C,3 : C13 " cl m CIN e Q C's r- 0 : K, cl Q Z 0 $. C's o C, ,is C, 4-4 --4 > 4-4 0 0 m uo cl . - tb --4 0 -t:3 , o 4) CA -- 73 o z cl 00 1* = > cl 1-1 = 0 - = Q Q cd Cl cl 0 cl a) 0 0 cl cl 73 tb C,4 cl 94 r --4 73 -4) q, cl : Q 0 cl 0 73 4-, O' C4 = 0>, -, 0 Q a) 0 ON = Z Cl cn c 73 1 0 00 cl cl 0 0 cl 7:3 4) cl r- --4 -1 - tb 4) 0 _0 tb t:3 0 q4 C4- ON ON r.. -- .- 'U 00 00 C,3 Cl 0 >-. 0 73 t4 a, 0 .- cI - 0 0 Q cl r- - cl 0 Q 0 r .- 123 cl >-. tb tb CZs m ,is to 73 U > CL, U ,is 44 Cl C,3 to cl to 10 m > cl cl U cl cl ct3 t) Z cl ., cl 0 -0 > ,be C4 C's C, cA cl cl cn >, C'3to 4.) CZ ci Co 00 r. 4.4 to cl cl U 4.) cj cl 0 - to cl I:,,- cl C) -.4 - k r r. cl 4--4 r.1:1, r. 0 10 C,3 C,3 cn ; 5 r. cl C,3 ci r. _ cl 0 r. 0 0X r. -0 ":: (:)- , = (7, = 'n cl r cl U W) cl '. 0 = = Q r . Ic; : .- 00 C) ;G Z cl cl z - m >, >, = cl 0 4. $. 00 cl 4. C13 m b 0 cl 0 N U E i. d 0. E m m m cl 4.) cl -0 C'3 7E. r.00en cA cl "o x C,3 0 Q cl 0 2 0 cn 40, cl 0 .0 0 'r. > > en I'D 00
04 cl ci ci ci to to -,t >, r-- cl
cl to
cl cl cl r.
z 0 Q) to CZ cl r cj =.- 64 & z CZ t > Q j
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions CORPORATISM AND UNEMPLOYMENT 375
co co co 0 cl Q cl cl 0co 0 R, R, 0 co co 00
4--4 4--4 C,4 co C-4 iZN cl --4 -4 17, co con CA
C14 0 cl 0 cl C14 4 C,4 C,4 C,4 4--4 4.4 4.4 E ,_, ,_, 1-1 cl cl cl cl cl 1.4 1.4 c cl cj 42 40, 40, U U U U U r. U cl = C $. cl .0 0 0 .- -cl w -cl cl cl cl cl C14 0 J C) U z U z U z U O U O U O U O U O
C14 --4 c (71, C14 00 tf) CT! r cli 0 --4 r- r- r- C14 C-i
CZ 00 c C14 00 C14 tf) C14 00 C14 tf) 0 4) C14 00 m OR C14 0 (1) 0 00 06 tt cli C-i
C-i U. +1 co cl +-- cj $.
> C,4 O cl 4--4
4--4 O cl CZ O,, j=., cl
CZ CZ 1.4 co
> cl cl co +-- = co C-i Q O CZ cl cl 0 Q 1.4 $. cl 42 *1 0 = 42 $. cl 0 1.4 co
cl 4-4 4-4 1.4 O +, ;0 0 0 u, I .-" CZ U4 cl 4. 7: CZ CZ CZ cl 42 .- co $. $. 4--4 cl cl 4-4 > cl co cl
CZ CZ 4-4 1.4 O,, r- 4--4 CZ oll 0 Q 4--4 s.4 -4 cl s. cl
cl CZ cl = = CZ CA CZ cl 1.4 1.4 CA CZ cl co cl cl 01.4 +1 cl co >-. CZ 0 cl .0 +-- 0 C-i 0 cl 42 co > ll:: Q. r. I = >1 = cl 4--4 cl co 4--4 cl 4--4 0 = bo Q CZ > co cl 1.4 CZ cl >. 0 4--4 0 s.., 0 0 cl > cl 4--4 cl 0 0 4.4 4--4 1.4 4. CZ 1-4
cl co 1-4 CZ 44 O cl cl cl I-, 0 0 cl > O., CZ cl cl CZ CZ > .,..4 cl 0 s. CZ 0 0 CZU-4 >
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions 376 AMERICAN SOCIOLOGICAL REVIEW nomic growth is operationalized with an in- fine the subperiods. This allows the data to dex of real gross domestic product (GDP), determine if and when a structuralbreak in also with 1979 = 1. Five indicators assess the effects has occurred (Western and Healy notion that corporatism affects unemploy- 1999). The year 1990 marks the start of a ment via government policy choices: real new business cycle, and it begins a period in long-term interest rates, total government which policymakers in a number of coun- expenditures, government expenditures on tries became increasingly worried about education, government expenditures on ac- lowering inflation in order to meet the crite- tive labor market policy, and government ria for European unification and to avoid employment. currency sell-offs. Thus, there is reason to The regressions use annual data in a suspect that if a break occurred, it took place pooled time-series cross-section design. Fol- sometime around the turn of the decade. Be- lowing common practice in recent compara- cause these regressions include a period tive political economy research, I estimate dummy variable, I drop the dummy variables the models using ordinary least squares for individual years.7 (OLS) with panel-corrected standard errors (Beck and Katz 1995). Similar results were obtained using "HC3" standarderrors (Long FINDINGS and Ervin 2000) as well as random-effects WAGE COORDINATION generalized least squares (GLS) with robust standard errors. Unemployment rates in af- Model 1 in Table 2 shows the regression re- fluent countries tend to be sticky over time; sults for the two corporatism indicators plus one of the best predictors of a country's un- the control variables and the lagged unem- employment rate in a given year is its unem- ployment rate. The results suggest a nega- ployment rate in the previous year. I there- tive effect of wage coordination on unem- fore include the lagged (one-year) unem- ployment.8 ployment rate as a regressor. This also helps Model 2 adds the wage coordination x pe- avert autocorrelation. A country fixed-ef- riod interaction term (and a similar interac- fects model cannot be used here because tion for union participation in policymaking several of the key independent variables, in- x period, discussed below). In regressions cluding the wage-coordination variable, do searching for a break in effects using all pos- not vary over time for a number of countries. sible periodizations, the t-value for the in- In this circumstance, the best course is to fo- teraction term and the R-squares for the cus on the cross-sectional variation by in- equation are largest in the regression using cluding year dummy variables without coun- 1992 as the break point. I therefore use this try dummy variables (Beck and Katz 2001; periodization in the Model 2 regression. The Traxler et al. 2001:27-28). This is consistent results indicate that wage coordination re- with the cross-country focus of most prior duced unemployment between 1980 and research on this issue. A test for unit roots 1991, but not thereafter. In the 1992-1997 revealed no problem with nonstationarity in period there is no effect, as the coefficient the dependent variable.6 for the wage-coordination variable (which To test the possibility of a structuralbreak estimates the effect for 1980-1991) almost in effects, I use a dummy variable that di- vides the 1980-1997 period into two sub- 7 Leaving the year dummy variables in the re- periods. It is coded 0 for each observation gression requires dropping the period dummy prior to a given year and coded 1 for obser- variable, although interpretation of the interac- vations from that year forward. Each of the tion term remains the same. Doing this yields two corporatism variables is interacted with similar results. this dummy variable. I examine 18 regres- 8 To check to see if this effect is genuinely lin- sions, each with a different year used to de- ear, I tried entering wage coordination as a set of dummy variables representing the various cat- egories. This allows comparison of effects for 6 The null hypothesis of nonstationarity for the each specific level of wage coordination. The re- Im-Pesaran-Shin test (see Im, Pesaran, and Shin sults suggest that the relationship is indeed lin- 1997) was rejected at the p < .001 level. ear.
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions CORPORATISM AND UNEMPLOYMENT 377
Table 2. Unstandardized OLS Coefficients from Regressions of Unemployment on Corporatism Variables and Control Variables: 1980s and 1990s
1980-1997 1980-1991 1992-1997 IndependentVariable Model 1 Model 2 Model 3 Model 4 Wage coordination -.11 -.23** -.19** -.03 (.05) (.06) (.07) (.12)
Wage coordination x 1992-1997 period dummy .24** (.09) Union participationin economic policymaking -.18** -.20** -.10 -.31** x leftist government (.06) (.06) (.06) (.07)
Union participation in economic policymaking -.12* x leftist governmentx 1992-1997 period dummy (.05) Union participationin economic policymaking .06* .02 .01 .05 (.03) (.03) (.04) (.05) Leftist government 1.19** 1.56** 1.13** 1.35* (.37) (.44) (.45) (.59) 1992-1997 period dummy -.32 (.50) Trade -.43 -.15 .17 -.72* (.28) (.31) (.40) (.28) Union density .50 .82* .32 .97 (.39) (.39) (.38) (.66) Employmentregulations .06 .08 .04 .16 (.06) (.06) (.07) (.12) Tax rate on workers -.01 -.01 -.02 .01 (.01) (.01) (.01) (.02) Unemploymentbenefit duration .04 .06 .06 .10* (.04) (.05) (.05) (.06) Lagged unemployment(t- 1) .96** .91** .94** .88** (.03) (.03) (.04) (.05)
R2 .95 .93 .96 .95 Durbin's h 1.51 1.49 1.33 1.25 Numberof observations 288 288 192 96
Wage coordination -.19a .09a -.25 to -.job -.11 to .13b -.20 to -.13c -1 1 to .09c
Union participationin economic policymaking _.11a -.29a x leftist government -.16 to -.05b -.54 to -.24b -.14 to-.08c -.53 to -.20c Note: Numbersin parenthesesare panel-correctedstandard errors. Results for year dummies(Models 1, 3, and 4) are not shown. a Coefficient in a regressionwith the other corporatismvariable omitted. b Range of coefficients in regressionswith countriesomitted one at a time ("jackknife"). c Range of coefficients in regressions with all possible combinationsof the control variables ("extreme bounds"). *n < .05 n < .01 (one-tailed tests)
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions 378 AMERICAN SOCIOLOGICAL REVIEW exactly equals the oppositely signed coeffi- due to the sudden collapse of the Soviet mar- cient for the interaction term (which esti- ket, Germany took on the burden of unifica- mates the change in effect between 1980- tion, Norway benefited from substantial oil 1991 and 1992-1997).9 Why 1992? One revenues, the United States lowered its un- possible reason is that, although 1990 is employment rate in part by incarcerating a commonly identified as the beginning of the larger share of its unskilled males, and so on 1990s business cycle (see OECD various (Smith 1999; Western and Beckett 1999). years), that is largely based on the fact that The second test reestimates the regressions the U.S. economy went into recession in that with all possible combinations of the control year. Most of Europe, by contrast, did not variables ("extreme bounds"). Both tests enter the downturn until around 1992. Of suggest that the results for the wage-coordi- course, any such break in effects is unlikely nation variable are reasonably robust. to have occurred simultaneously across all What accounts for this shift in the rela- countries or even to have taken place sharply tionship between wage coordination and un- and decisively in a single year within any employment? Figure 1 indicates that, aside given individual country. Models 3 and 4 from Finland and Sweden, countries with co- show separate regressions for 1980-1991 ordinatedwage-setting generally did not per- and 1992-1997. They reinforce the results in form worse in the 1990s than in the 1980s. Model 2: Wage coordination was associated Indeed, several high-coordination countries, with superior unemployment performance in such as the Netherlands, Belgium, and Den- the 1980s, but in the 1990s that association mark, experienced markedly better unem- disappeared. ployment performance in the 1990s than in Note that these regressions assess change the previous decade. Instead, the difference in unemployment outcomes, as they include between the two periods is attributable the lagged unemployment rate as a regres- mainly to the fact that low-coordination sor. If the lagged dependent variable is countries-Canada, France, the United dropped, the wage-coordination variable is Kingdom, and the United States-improved negatively signed and easily statistically sig- their performance.10 nificant in the 1992-1997 regression (not To understandwhy the impact of wage co- shown). In other words, controlling for other ordination disappeared in the 1990s, we relevant determinants,countries with greater need to know the mechanisms) through wage coordination continued to enjoy lower which wage coordination affected jobless- unemployment rates in the 1990s. However, ness in the 1980s. If the effect of wage coor- the Table 2 results, with the lagged unem- dination on unemployment worked through ployment rate included, suggest that this was labor cost moderation, economic growth, a legacy of superior performance in earlier and/or government policies, entering these decades. variables into the 1980s regression should Table 2 includes results of two sets of ro- reduce the size of the coefficient for the bustness tests for the 1980-1991 and 1992- wage-coordination variable. Table 3 shows 1997 regressions. The first test drops each the results of doing this. They suggest some country one at a time ("jackknife"). Outliers support for the conventional notion that the may be of particular concern for the 1990s: Finland experienced a deep economic shock 10 It is worth emphasizing the substantial cross-country variation in unemployment perfor- 9 I also exploredthis issue using periodaver- mance evident in Figure 1. In the 1980s, there ages for the two decades instead of annual obser- was virtually no increase in joblessness relative vations. The data were pooled (yielding 32 obser- to 1974-1979 in Japan, Sweden, Switzerland, vations) and a wage coordination x 1990s period Finland, and the United States, whereas in the interaction term entered in the regression. As with Netherlands, Belgium, France, Ireland, and the the regressions using yearly data, the wage-coor- United Kingdom the unemployment rate rose by dinationvariable is negativelysigned and statisti- more than four percentage points. In the 1990s, cally significant,while the wage coordinationx the spread was even more pronounced, with sev- 1990s interactionterm is positively signed and eral countries achieving reductions of two per- also significant. Results for the other variables are centage points or better while others experienced generally consistent with those in Table 2. increases of four percentage points or more.
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions CORPORATISM AND UNEMPLOYMENT 379
10o 10- m e*Finland
o 0 4) Ireland * Sweden UnitedKingdom * 5 .c 0 France Netherlands* C 0 0 Italy 0 4) 0 e Belgium -0 *Switzerland *Italy * Germany Q France e Germany X E *Canada Denmarke AustriaV Norway r E Norway. 0 0 o C Sweden Austria* Japan. e o0 *United States Finlande Japan 0 *Canada Switzerland @ Denmarke Belgium. o o ~~~~~~~~~~~~~~~*UnitedStatesIrln ~ ~ Netherlands*
-5 - | | | a -5 1 2 3 4 5 1 2 3 4 5 Wage Coordination, 1980-1991 Average Wage Coordination, 1992-1999 Average Figure 1. Wage Coordination and Unemployment Performance in the 1980s and 1990s a 1980-1991 average percentunemployed minus 1974-1979 average. b 1992-1999 average percentunemployed minus 1980-1991 average. mechanism linking wage coordination with costs (Layard et al. 1991; O'Connell 1994; superior unemployment performance in the Volgy, Schwarz, and Imwalle 1996; Western 1980s was labor cost restraint.When the real and Healy 1999). These two processes-slow labor cost index is added, the wage-coordi- growth of labor costs helping to reduce un- nation coefficient decreases (Model 2). Yet employment, along with low unemployment the magnitude of the decrease is small.1" causing faster growth of labor costs-will However, simply entering labor costs as an partially or fully offset each other. intervening variable may not accurately cap- It is therefore useful to separately exam- ture its role as a causal mechanism. The hy- ine the effect of wage coordination on labor pothesized causal sequence links slower costs and the effect of labor costs on unem- growth of real labor costs with lower unem- ployment. This also makes it possible to ex- ployment, but low unemployment enhances plore whether coordination's lack of asso- workers' bargainingpower and thereby likely ciation with low unemployment in the leads to faster increases in wages and labor 1990s was due to wage coordination no longer being associated with labor cost I This could stem from the time frame used moderation or to labor cost moderation no here. If wage coordination affects unemployment longer being associated with low unem- via its impact on real labor costs, that effect may ployment (or both). Table 4 shows the re- not be observable in the very short run. Regres- sults of two regressions. In Model 1, the sions that examine a longer period, such as a de- real labor cost index is regressed on lagged cade, might be better able to detect it. In analy- ses not shown here, I estimated an OLS regres- labor costs, wage coordination, a wage co- sion using decade-average data for the 1980s ordination x period dummy interaction, un- with change in unemployment (1980-1991 aver- employment (average of t, t- 1, and t-2), age minus 1974-1979 average) as the dependent and three additional variables that have variable. With only 16 countries, a regression us- proved relevant in prior research on deter- ing decade averages for a single period is hin- minants of wage developments: inflation, dered by limited degrees of freedom. I therefore trade, and union density (O'Connell 1994; included various combinations of just two con- Volgy et al. 1996; Western and Healy trol variables in addition to the wage-coordina- 1999). The results indicate that wage coor- variable is tion variable. The wage-coordination dination was associated with labor cost re- negative and statistically significant, as in the in Table 2 regression using annual data. Again, straint in the 1980s but not the 1990s. however, adding the labor cost index reduces the Here the regressions suggest a break point size of the wage-coordination coefficient by a in either 1991, 1992, or 1993, with slightly relatively small amount. stronger support for 1993. In Model 2, un-
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions 380 AMERICAN SOCIOLOGICAL REVIEW
Table 3. Unstandardized OLS Coefficients from Regressions Testing Various Links between Corporatism and Unemployment
1980-1991 1992-1997 IndependentVariable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Wage coordination -.19** -.16* -.17* -.14* (.07) (.07) (.08) (.08) Union participation in economic -.31** -.15* policymakingx leftist government (.07) (.07) Note: Numbersin parenthesesare panel-correctedstandard errors. Model 1: Same variables as in Table 2, Model 3. Model 2: Same variables as in Table 2, Model 3 plus real labor costs variable. Model 3: Same variables as in Table 2, Model 3 plus real GDP variable. Model 4: Same variables as in Table 2, Model 3 plus governmentexpenditures on education variable, governmentexpenditures on active labor marketpolicy variable,and governmentemployment variable. Model 5: Same variables as in Table 2, Model 4. Model 6: Same variablesas in Table 2, Model 4 plus total governmentexpenditures variable, government expenditures on education variable, governmentexpenditures on active labor marketpolicy variable, and governmentemployment variable. * ** p < .05 p < .01 (one-tailed tests) employment is regressed on the same set of curate to reverse the direction of the causal variables as in Table 2, except that wage co- arrows. ordination is replaced by the labor cost in- Figure 2 provides some case study insight dex. Faster growth of labor costs is associ- into this issue. It shows year-by-year devel- ated with higher unemployment throughout opments in wage coordination, real labor both decades; there is no evidence of a costs, and unemployment in six relevant break in effects. (Hence no period dummy countries in the 1970s, 1980s, and 1990s. interaction term is included in Model 2.) Canada, France, the United Kingdom, and Thus, consistent with the assumption under- the United States have consistently had the lying much of the prior research on corpo- least coordinated wage-setting arrangements ratist wage-setting, the evidence suggests among the 16 nations, and Denmark and that in the 1980s wage coordination helped Sweden shifted from high to intermediate to reduce unemployment by fostering real coordination during the 1980s. The charts labor cost restraint. However, although la- suggest that Canada, France, and the United bor cost moderation continued to contribute States almost certainly are not countries for to low unemployment in the 1990s, coun- which reverse causality is plausible. None of tries with coordinated wage-setting were no these three experienced any noteworthy longer better at securing such moderation. change in wage coordination over time, Can we safely presume, based on these aside from brief periods of wage-price con- findings, that the causal path in the 1980s trols in Canada and the United States during truly was wage coordination -* real labor the early 1970s. Nor was there reverse cau- cost moderation -* low unemployment? Per- sality in Sweden, as its unemployment rate haps some countries with rapid growth in la- remained quite low throughout the 1980s bor costs and/or high unemployment in the even as wage bargaining became less coor- late 1970s or 1980s shifted to fragmented dinated. wage bargaining in the hope that reducing The United Kingdom and Denmark are institutional "rigidities" would solve their more likely candidates. Both suffered a labor market problems. If such nations were sharp rise in unemployment in the early unable to turn things around, their poor per- 1980s, just as they shifted toward less coor- formance in the 1980s would suggest that dinated wage-setting. But did wage mili- low coordination yields poor labor market tancy generated by wage coordination in ear- outcomes when in fact it would be more ac- lier years cause these increases in jobless-
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions CORPORATISM AND UNEMPLOYMENT 381
Table 4. Unstandardized OLS Coefficients ness? Britain certainly suffered from rapid from Additional Regressions Testing increases in real labor costs in the mid- Wage Restraint as a Link between 1970s, but arguably despite rather than be- Corporatism and Unemployment cause of its temporary attempt at a national DependentVariable incomes policy during that period (Flanagan, Real Unemploy- Soskice, and Ulman 1983; Scharpf [1987] LaborCosts, ment, 1991). Moreover, the sharp rise in unem- Independent 1980-1997 1980-1997 ployment beginning in 1980 was largely an Variable Model 1 Model 2 explicit aim of the Thatcher government's restrictive monetary policy, a centerpiece of Wage coordination -.26** its efforts to curtail the power of organized (.09) labor (Rhodes 2000). Denmark, too, experi- Wage coordination .21* enced above average increases in real labor x 1993-1997 period (.11) costs in the mid-1970s, which contributed to dummy growing unemployment in those years. But 1993-1997 period -1.43** the very high levels of joblessness in the dummy (.48) 1980s were at least as much, if not more, a Unemployment -. 14** product of Denmark's hard currency ar- (.04) rangement, which forced the country to ad- Inflation -.11 ** here to the German Bundesbank's policy of (.04) tight money (Benner and Vad 2000). Over- Real labor cost index 3.58** all, then, the conclusion that wage coordina- (1.29) tion was conducive to low unemployment in Union participationin - -.10 the 1980s may be somewhat affected by re- economic policymakingx (.06) verse causality, but the magnitude of that leftist government bias probably is not large. Union participationin - -.00 What about reverse causality in the 1990s? economic policymaking (.04) The regression results suggest that the asso- Leftist government -.95* ciation between wage coordination and low (.37) unemployment eroded in that decade, but Trade -.49 -.03 might that be because high or rising levels (.38) (.29) of joblessness spurred an increase in wage Union density .89 .19 coordination in some countries? Two na- (.65) (.37) tions-Ireland and Italy-shifted from low Employment .04 to high coordination in the 1990s (see Fig- regulations (.06) ure 1), and in both countries the aim was to Tax rate on workers -.00 restrain wages and thereby reduce unem- (.01) ployment. However, Ireland was quite suc- Unemployment -.02 cessful at lowering unemployment in 1990s, benefit duration (.05) while Italy did not fare particularly well. Lagged real labor costs .93** Hence, these two cases probably offset one (t- 1) (.03) another, so that the regression results were Lagged unemployment .98-* unaffected. The jackknife analyses indicate (t- 1) (.02) that removing either or both does not alter the findings. R2 .93 .95 Is the finding of no association between Durbin's h 1.31 1.47 wage coordination and unemployment per- Numberof observations 288 288 formance in the 1990s a result of a mis- specification of the functional form of the Note: Numbersin parenthesesare panel-corrected relationship? Calmfors and Driffill (1988) standarderrors. In Model 1, all coefficients, except rela- that for lagged real labor costs, are multiplied by and others have posited a hump-shaped 100. Results for year dummies (Model 2) are not tionship between wage-bargaining central- shown. ization and labor cost changes, whereby low * p < .05 ** p < .01 (one-tailed tests) and high levels of centralization are best at
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions 382 AMERICAN SOCIOLOGICAL REVIEW
Canada France
4 -4-
10 3- 3-
5 95 18 98 90 19 155 95 18 18 90 19 159 2 -2- .r35 A
0 ~~~~~~~~~00 V 1975 1980 1985 1990 1995 1999 1975 1980 1985 1990 1995 1999
United Kingdom United States
4 ~~~~~~~~~~~~~~~4-
-10 ------~~~~~~~~~~~10 3 - 3999
2-~~~~~~~~~~~~~~~
1975 1980 1985 1990 1995 1999 1975 1980 1985 1990 1995 1999
5 95 18 95 19 95 19 1555 95 18 95 19 159 Denmark Sweden 5 ~~~~~~~~~~155-1
I1'M 10 F, 3 ':A~~~ H~~~~~~~~~~~~~~~~~
4AN
1 ~~~~~~ ~ ~ ~ ~ 1'4
0 -~~~~~~~~~~~00 - 1975 1980 1985 1990 1995 1999 1975 1980 1985 1990 1995 1999
Figure 2. Wage Coordination, Real Labor Costs, and Unemployment by Year: Selected Countries, 1973 to 1999 Note: Bars represent wage coordinationscores (left axis applies). Dashed lines representthe real labor cost index, with 1973 = 10 (right axis applies). Solid lines representpercentage unemployed (right axis applies). generating labor cost restraint and thus nent view holds that the moderating effect lower unemployment. However, several fol- of corporatist wage-setting institutions on low-up studies of the relationship between labor costs is contingent on the presence of wage centralization and unemployment find a leftist government (Alvarez et al. 1991; little or no support for this view (OECD Garrett 1998). Still another view suggests 1997; Traxler and Kittel 2000), and the logic that wage-setting arrangementsmay interact is not compelling when applied to wage co- with central bank independence in determin- ordination (Soskice 1990). Another promi- ing labor cost developments (Franzese 1999;
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions CORPORATISM AND UNEMPLOYMENT 383
Hall and Franzese 1998; Iversen 1999). In It appears, then, that wage coordination auxiliary regressions not shown here I tested contributed to low unemployment in the each of these alternative specifications. 1980s via labor cost moderation, economic None is supported. growth, and government policies (and per- Do the regression analyses suggest no ef- haps other as-yet-undiscovered mecha- fect of wage coordination in the 1990s be- nisms). Does this explain the lack of effect cause of multicollinearity? The two corpor- in the 1990s? The scatterplots in Figure 1 in- atism variables correlate at .66 in that de- dicate that the key change in the 1990s was cade, compared with .41 in the 1980s. How- that low-coordination countries improved ever, the findings for wage coordination are their unemployment performance, not that robust to exclusion of the union participation high-coordination countries began to per- variable, as is indicated by the figures at the form worse. Figure 2 suggests that labor cost bottom of Model 4 in Table 2. moderation was importantin Canada and the Unemployment dropped sharply in Swe- United States, as real labor costs decreased den and several other countries in the late by nearly 10 percent in these two countries 1990s. Perhaps the fact that the analyses here during the 1990s. Labor cost restraint may extend only to 1997 is responsible for the also have played a role in Britain's decline apparent lack of association between wage in joblessness (Nickell and Van Ours 2000), coordination and unemployment perfor- though there was little actual decline in real mance in the 1990s. However, a reestimation labor costs in the 1990s. In fact, between of the regression in Model 4 of Table 2 (not 1980 and 1999, labor costs fell just 5 per- shown) adding the years 1998 and 1999 and cent in the United Kingdom, compared with dropping the union participation variable 15 percent in both Canada and the United (due to lack of data) suggests not. The wage- States. Other institutional changes may have coordination variable in this regression is been more important in Britain, such as the positive, although not statistically signifi- loosening of restrictions on firing (Deakin cant. and Reed 2000). In France, another country The second hypothesized causal link be- with relatively uncoordinated wage-setting, tween wage coordination and unemployment expansion of public employment seems to is economic growth. The regressions indi- have played an important role. Indeed, cate some supportfor this mechanism: When France was one of only four affluent OECD the real GDP index is added to the 1980s re- countries to increase the proportion of its gression, the wage-coordination coefficient working-age population employed by the decreases somewhat, though not a great deal government during the 1990s (OECD 2001). (Model 3 in Table 3). In Denmark, which has had an intermediate What about the possibility that wage coor- level of wage coordination, a sizable expan- dination lowers unemployment via govern- sion in active labor marketpolicy in the mid- ment policy? The hypothesis here is that 1990s contributed to a dramatic fall in un- policymakers presume wage coordination employment in the last half of that decade will engender labor cost moderation, and (Auer 2000; Benner and Vad 2000; Bjork- hence that inflationary pressures will be lund 2000). minimal. This encourages governments to pursue more vigorous anti-unemployment UNION PARTICIPATION IN ECONOMIC programs. Entering the five policy variables POLICYMAKING one by one into the 1980s regression sug- gests that only education expenditures, ac- For union participation in economic policy- tive labor market policy, and government making, the second form of corporatism ex- employment are likely channels linking amined here, the story is different. The re- wage coordination with low unemployment. gression results in Table 2 suggest that, con- Model 4 in Table 3 shows the result for the ditional on the presence of a leftist govern- wage coordination coefficient when these ment, union participation tended to be asso- three policy variables are added to the Table ciated with lower unemployment in both the 2 regression. The wage-coordination coeffi- 1980s and the 1990s. The estimates in Model cient decreases from -.19 to -.14. 2 indicate that this effect grew stronger be-
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions 384 AMERICAN SOCIOLOGICAL REVIEW ginning in the early 1990s. This is indicated CONCLUSION by the statistically significant negative coef- ficient for the union participation x leftist I have attempted to enhance our understand- government interaction term (which, in ing of corporatism's macroeconomic effects Model 2, estimates the effect in the period by answering three questions: (1) Which of 1980-1991) and the significant negative co- the two principal forms of corporatism- efficient for the three-way union participa- corporatist wage-setting or union participa- tion x leftist government x 1992-1997 pe- tion in economic policymaking, or both- riod dummy interaction term (which esti- has helped to reduce unemployment? (2) mates the change in effect between 1980- What are the causal mechanisms? (3) Did 1991 and 1992-1997). The data suggest these effects continue in the 1990s in the roughly equal support for a break in the ef- face of globalization, restrictive monetary fect of union participation in 1991, 1992, or policy, growing dissension within labor 1993. I show the results using 1992 as the movements, and related developments? break-point simply for symmetry with the It is important to be cautious in interpret- wage coordination break. The findings for ing my findings. The two corporatism vari- the separate 1980-1991 and 1992-1997 re- ables are based on subjective judgment and gressions in Models 3 and 4 are consistent thus are susceptible to measurement error. with those in Model 2. (In Model 3, the In addition, these analyses are highly aggre- union participation x leftist government in- gated; they may or may not shed light on teraction term is just shy of significance: p = the experiences of any particular nation. .06.) And the results of the jackknife and ex- Broad statistical analyses such as this need treme bounds tests, shown at the bottom of to be supplemented by in-depth country the table, indicate that these findings are ro- case studies. bust to the group of countries included and With these caveats noted, the findings here to model specification. Regressions in which suggest that coordinated wage-setting was the union participation variable was not in- associated with low unemployment in the teracted with leftist government indicate no 1980s. There are compelling theoretical rea- support for a direct effect in either decade sons to suspect that this effect worked (not shown here). through real labor cost restraint, economic Models 5 and 6 in Table 3 show that add- growth, and government anti-unemployment ing four of the policy variables-total gov- policies, and each of these mechanisms re- ernment expenditures, spending on educa- ceives empirical support in the analyses. In tion, spending on active labor market policy, the 1990s the association between wage co- and public employment-reduces the size of ordination and superior unemployment per- the union participation x leftist government formance disappeared, mainly because low- interaction coefficient from -.31 to -.15. coordination countries performed better in This suggests that participation by unions in the 1990s than they did in the 1980s. By the the economic policymaking process reduced early 1990s, the balance of power between unemployment mainly, though perhaps not workers and employers in at least some low- exclusively, via its influence on government coordination countries appears to have policies and programs. shifted to a point where unions were consid- I examined the sensitivity of results for my erably less able to disrupt market processes. (expanded) Compston measure by substitut- Students of corporatism have long recog- ing the only other existing time-varying nized the potential for substantial modera- measure of union participation in economic tion of labor costs in countries with weak la- policymaking: that of Traxler et al. (2001: bor (Alvarez et al. 1991; Calmfors and 76). Over the 1980-1997 period, these two Driffill 1988). Fragmentedbargaining struc- measures correlate at .72. The findings (not tures create an incentive for union militancy, shown) turn out to be reasonably robust, al- but this incentive is immaterial if unions are though with the Traxler-Blaschke-Kittel sufficiently weak. In the 1970s and 1980s, measure there is no indication of a strength- unions in low-coordination countries evi- ening of union participation's unemploy- dently were strong enough that the incentive ment-reducing effect in the 1990s. did matter, but that was no longer the case
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions CORPORATISM AND UNEMPLOYMENT 385 in the 1990s. Hence, wage coordination was tries." American Journal of Sociology. no longer a prerequisite for labor cost re- Alvarez, R. Michael, Geoffrey Garrett, and Peter straint. Lange. 1991. "Government Partisanship, La- Has corporatism, then, become irrelevant bor Organization, and Macroeconomic Perfor- Science Review for labor market outcomes? I answer no, for mance." American Political 85:539-56. two reasons. First, beneficial effects of cor- Auer, Peter. 2000. Employment Revival in Eu- poratist institutions on unemployment per- rope: Labor Market Success in Austria, Den- formance were still observable in the 1990s, mark, Ireland, and the Netherlands. Geneva, but they were generated by union participa- Switzerland: International Labour Organiza- tion in economic policymaking rather than tion. by wage coordination. The causal link ap- Beck, Nathaniel and Jonathan N. Katz. 1995. pears to be anti-unemployment policies "What To Do (and Not To Do) with Time-Se- such as general government spending, edu- ries Cross-Section Data." American Political cation, active labor market policy, and pub- Science Review 89:634-47. . 2001. "Throwing Out the Baby with the lic employment. Although union participa- Bath Water: A Comment on Green, Kim, and tion in policymaking was a central concern Yoon." International Organization 55:487-95. in the early-1980s corporatist literature Benner, Mats and Torben Bundgaard Vad. 2000. (Lehmbruch 1984), it has received scant at- "Sweden and Denmark:Defending the Welfare tention in empirical research on corpor- State." Pp. 399-466 in Welfare and Work in atism's macroeconomic effects. The find- the Open Economy, vol. 2, Diverse Responses ings here seem to warrantrenewed interest. to Common Challenges, edited by F. W. Second, other researchers find an associa- Scharpf and V. A. Schmidt. New York: Oxford tion between corporatism and low levels of University Press. Berger, Suzanne and Ronald Dore, eds. 1996. earnings inequality, and the structure of National Diversity and Global Capitalism. wage-setting arrangementsis responsible for Ithaca, NY: Cornell University Press. this association (Alderson and Nielsen forth- Bjorklund, Anders. 2000. "Going Different coming; Lucifora 2000; Rueda and Pon- Ways: Labour Market Policy in Denmark and tusson 2000; Wallerstein 1999). Coordinated Sweden." Pp. 148-80 in Why Deregulate and/or centralized wage arrangements may Labour Markets?, edited by G. Esping- no longer be advantageous for securing low Andersen and M. Regini. New York: Oxford unemployment, but as long as they are no University Press. worse in this respect than fragmented wage Boix, Carles. 1998. Political Parties, Growth, and New York: Univer- structures and better at limiting pay inequal- Equality. Cambridge sity Press. ity, they are likely to remain prominent in . 2000. "Partisan Governments, the Inter- the institutional landscapes of affluent capi- national Economy, and Macroeconomic Poli- talist nations. cies in Advanced Nations, 1960-93." World Politics 53:38-73. Lane Kenworthyis AssistantProfessor of Soci- Boreham, Paul and Hugh Compston. 1992. ology at EmoryUniversity. His researchfocuses "LabourMovement Organization and Political on the effectsof institutionsand government poli- Intervention: The Politics of Unemployment in cies on socioeconomicperformance in affluent the OECD Countries, 1974-1986." European countries.His publicationsinclude In Searchof Journal of Political Research 22:143-70. NationalEconomic Success: BalancingCompe- Bruno, Michael and Jeffrey D. Sachs. 1985. Eco- tition and Cooperation (Sage, 1995) and recent nomics of Worldwide Stagflation. Cambridge, articles in the AmericanJournal of Sociology, MA: HarvardUniversity Press. Social Forces,and WorldPolitics. Calmfors, Lars and John Driffill. 1988. "Bargain- ing Structure, Corporatism, and Macroeco- REFERENCES nomic Performance." Economic Policy 6:14- 61. Ahlen, Kristina. 1989. "Swedish Collective Bar- Cameron, David R. 1984. "Social Democracy, gaining under Pressure: Inter-Union Rivalry Corporatism,Labour Quiescence, and the Rep- and Incomes Policies." British Journal of In- resentation of Economic Interest in Advanced dustrial Relations 27:330-46. Capitalist Society." Pp. 143-78 in Order and Alderson, Arthur S. and Frangois Nielsen. Forth- Conflict in Contemporary Capitalism, edited coming. "Globalization and the Great U-Turn: by J. H. Goldthorpe. Oxford, England: Income Inequality Trends in 16 OECD Coun- Clarendon.
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions 386 AMERICAN SOCIOLOGICAL REVIEW
Campbell, John L. 2000. "States, Politics, and Hall, Peter A. and Robert J. Franzese, Jr. 1998. Globalization: Why Institutions Still Matter." "Mixed Signals: Central Bank Independence, Department of Sociology, Dartmouth College, Coordinated Wage Bargaining, and European Hanover, NH. Unpublished manuscript. Monetary Union." International Organization Centre for Economic Performance (CEP), Lon- 52:505-35. don School of Economics and Political Sci- Hernes, Gudmund. 1991. "The Dilemmas of So- ence. N.d. "Centre for Economic Performance cial Democracies: The Case of Norway and OECD Dataset." London, England: CEP. Sweden." Acta Sociologica 34:239-60. Chiricos, Theodore G. 1987. "Rates of Crime and Hibbs, Douglas A., Jr. 1977. "Political Parties Unemployment: An Analysis of Aggregate Re- and Macroeconomic Policy." American Politi- search Evidence." Social Problems 34:187- cal Science Review 71:1467-87. 212. Hicks, Alexander. 1988. "Social Democratic Clark, Andrew and Andrew Oswald. 1994. "Un- Corporatism and Economic Growth." Journal happiness and Unemployment." Economic of Politics 50:677-704. Journal 104:1025-43. Hicks, Alexander and Lane Kenworthy. 1998. Compston, Hugh. 1997. "Union Power, Policy "Cooperation and Political Economic Perfor- Making, and Unemployment in Western Eu- mance in Affluent Democratic Capitalism." rope, 1972-1993." Comparative Political American Journal of Sociology 103:1631-72. Studies 30:732-51. Im, Kyung So, M. Hashem Pesaran, and Yong- . 1998. "The End of National Policy Con- cheol Shin. 1997. "Testing for Unit Roots in certation? Western Europe since the Single Heterogeneous Panels." Department of Ap- European Act." Journal of European Public plied Economics, University of Cambridge, Policy 5:507-26. Cambridge, England. Unpublished manuscript Deakin, Simon and Hannah Reed. 2000. "River (available at: http://www.econ.cam.ac.uk/fac- Crossing or Cold Bath? Deregulation and Em- ulty/pesaran/lm.pdf). ployment in Britain in the 1980s and 1990s." Iversen, Torben. 1999. Contested Economic In- Pp. 115-47 in Why Deregulate Labour Mar- stitutions. New York: Cambridge University kets?, edited by G. Esping-Andersen and M. Press. Regini. New York: Oxford University Press. Janoski, Thomas, Christa McGill, and Vanessa Dore, Ronald. 1987. Taking Japan Seriously. Tinsley. 1997. "Making Institutions Dynamic Stanford, CA: Stanford University Press. in Cross-National Research: Time-Space Dis- Ebbinghaus, Bernhard and Jelle Visser. 2000. tancing in Explaining Unemployment." Com- Trade Unions in Western Europe since 1945. parative Social Research 16:227-68. London, England: Macmillan. Kenworthy, Lane. 1996. "Unions, Wages, and Esping-Andersen, G0sta. 1999. Social Founda- the Common Interest." Comparative Political tions of Postindustrial Economies. New York: Studies 28:491-524. Oxford University Press. . 1997. "Globalization and Economic Flanagan, Robert J. 1999. "Macroeconomic Per- Convergence." Competition and Change 2:1- formance and Collective Bargaining: An Inter- 64. national Perspective." Journal of Economic 2001 a. "Wage-Setting Coordination Literature 37:1150-75. Scores." Version dated June 17, 2001. Data Flanagan, Robert J., David Soskice, and Lloyd set. Retrieved June 17, 2001 (http://www. Ulman. 1983. Unionism, Economic Stabiliza- emory.edu/SOC/lkenworthy). tion, and Incomes Policies. Washington, DC: . 2001b. "Wage-Setting Measures: A Sur- Brookings Institution. vey and Assessment." World Politics 54:57- Franzese, Robert J., Jr. 1999. "The Interaction of 98. Credibly Conservative Monetary Policy with . 2002. "Do Affluent Countries Face an Labor- and Goods-Market Institutions: A Re- Incomes-Jobs Tradeoff?" Discussion Paper view of an Emerging Literature." Department No. 01/10, Max Planck Institute for the Study of Political Science, University of Michigan, of Societies, Cologne, Germany (available at: Ann Arbor, MI. Unpublished manuscript. http://www.mpi-fg-koeln.mpg.de). Garrett, Geoffrey. 1998. Partisan Politics in the Kitschelt, Herbert, Peter Lange, Gary Marks, and Global Economy. New York: Cambridge Uni- John D. Stephens, eds. 1999. Continuity and versity Press. Change in Contemporary Capitalism. New Golden, Miriam, Peter Lange, and Michael York: Cambridge University Press. Wallerstein. 1997. "Union Centralization Korpi, Walter. 1991. "Political and Economic among Advanced Industrial Societies: An Em- Explanations for Unemployment: A Cross-Na- pirical Study." Data set. Retrieved May 5, tional and Long-Term Analysis." British Jour- 2001 (http://www.shelley.polisci.ucla.edu/ nal of Political Science 21:315-48. data). Krugman, Paul. 1998. "The Confidence Game."
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions CORPORATISM AND UNEMPLOYMENT 387
The New Republic, October 5, pp. 23-25. 01#2001 [MRDF]. Paris, France: OECD [pro- Layard, Richard, Stephen Nickell, and Richard ducer and distributor]. Jackman. 1991. Unemployment: Macroeco- . Various years. OECD Historical Statis- nomic Performance and the Labor Market. tics. Paris, France: OECD. New York: Oxford University Press. . N.d. "OECD Education Database." Data Lehmbruch, Gerhard. 1984. "Concertation and set. Retrieved May 5, 2001 (http://www.oecd. the Structure of Corporatist Networks." Pp. org/els/stats/edudb/edu-db.htm). 60-80 in Order and Conflict in Contemporary Pochet, Philippe and Guiseppe Fajertag, eds. Capitalism, edited by J. H. Goldthorpe. Ox- 1997. Social Pacts in Europe. Brussels, Bel- ford, England: Oxford University Press. gium: European Trade Union Institute. Long, J. Scott and Laurie H. Ervin. 2000. "Using Rhodes, Martin. 2000. "Restructuringthe British Heteroscedasticity Consistent Standard Errors Welfare State: Between Domestic Constraints in the Linear Regression Model." American and Global Imperatives."Pp. 13-68 in Welfare Statistician 54:217-24. and Work in the Open Economy, vol. 2, Di- Lucifora, Claudio. 2000. "Wage Inequalites and verse Responses to Common Challenges, ed- Low Pay: The Role of Labour Market Institu- ited by F. W. Scharpf and V. A. Schmidt. New tions." Pp. 9-34 in Labour Market Inequali- York: Oxford University Press. ties, edited by M. Gregory, W. Salverda, and Rueda, David and Jonas Pontusson. 2000. "Wage S. Bazen. New York: Oxford University Press. Inequality and Varieties of Capitalism." World Martin, John P. 2000-2001. "WhatWorks among Politics 52:350-83. Active Labour Market Policies: Evidence from Scarpetta, Stefano. 1996. "Assessing the Role of OECD Countries' Experiences." OECD Eco- Labour Market Policies and Institutional Set- nomic Studies 30:79-113. tings on Unemployment: A Cross-Country Nickell, Stephen. 1997. "Unemployment and La- Study." OECD Economic Studies 26:43-98. bor Market Rigidities: Europe versus North Scharpf, Fritz W. [ 1987] 1991. Crisis and Choice America." Journal of Economic Perspectives in European Social Democracy. Translated by 11(3):55-74. R. Crowley and F. Thompson. Ithaca, NY: Nickell, Stephen and Brian Bell. 1996. "Changes Cornell University Press. in the Distribution of Wages and Unemploy- . 1999. Governing in Europe: Effective ment in OECD Countries." American Eco- and Democratic? New York: Oxford Univer- nomic Review (AEA Papers and Proceedings) sity Press. 86:302-08. Scharpf, Fritz W. and Vivien A. Schmidt, eds. Nickell, Stephen and Richard Layard. 1999. "La- 2000. Welfare and Workin the Open Economy, bor Market Institutions and Economic Perfor- vol. 1, From Vulnerability to Competitiveness. mance." Pp. 3029-84 in Handbook of Labor New York: Oxford University Press. Economics, vol. 3C, edited by 0. Ashenfelter Smith, Michael R. 1999. "Doing Well by Doing and D. Card. Amsterdam, Netherlands: Good? What Happened to Corporatist Coun- Elsevier. tries in the 1990s, and Why?" Presented at the Nickell, Stephen and Jan Van Ours. 2000. "Mi- annual meeting of the Society for the Advance- rage or Miracle? Labour Market Performance ment of Socio-Economics, June, Madison, WI. in Britain and the Netherlands." CentrePiece, Soskice, David. 1990. "Wage Determination: spring. Retrieved June 6, 2000 (http:Ilwww. The Changing Role of Institutions in Ad- centrepiece-magazine.com). vanced Industrialized Countries." Oxford Re- O'Connell, Philip J. 1994. "National Variation in view of Economic Policy 6(4):36-61. the Fortunes of Labor: A Pooled and Cross- . 1999. "Divergent Production Regimes: Sectional Analysis of the Impact of Economic Coordinated and Uncoordinated Market Crisis in the Advanced Capitalist Nations." Pp. Economies in the 1980s and 1990s." Pp. 101- 218-42 in The Comparative Political Economy 34 in Continuity and Change in Contemporary of the Welfare State, edited by T. Janoski and Capitalism, edited by H. Kitschelt, P. Lange, A. Hicks. New York: Cambridge University G. Marks, and J. D. Stephens. New York: Press. Cambridge University Press. Okun, Arthur M. 1973. "Upward Mobility in a . 2000. "Macroeconomic Analysis and the High-Pressure Economy." Brookings Papers Political Economy of Unemployment." Pp. 38- on Economic Activity 1:207-52. 74 in Unions, Employers, and Central Banks, Organization for Economic Cooperation and De- edited by T. Iversen, J. Pontusson, and D. velopment (OECD). 1997. "Economic Perfor- Soskice. New York: Cambridge University mance and the Structureof Collective Bargain- Press. ing." Pp. 63-93 in OECD Employment Out- Streeck, Wolfgang. 1984. "Neo-Corporatist In- look. Paris, France: OECD. dustrial Relations and the Economic Crisis in . 2001. OECD Statistical Compendium West Germany." Pp. 291-314 in Order and
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions 388 AMERICAN SOCIOLOGICAL REVIEW
Conflict in Contemporary Capitalism, edited Lawrence E. Imwalle. 1996. "In Search of by J. H. Goldthorpe. Oxford, England: Economic Well-Being: Worker Power and the Clarendon. Effects of Productivity, Inflation, Unemploy- 1997. "German Capitalism: Does It Ex- ment, and Global Trade on Wages in Ten ist? Can it Survive?" New Political Economy Wealthy Countries." American Journal of Po- 2:237-56. litical Science 40:1233-52. Swank, Duane. N.d. "18-Nation Pooled Time-Se- Wallerstein, Michael. 1999. "Wage-Setting Insti- ries Data Set: Strength of Political Parties by tutions and Pay Inequality in Advanced Indus- Ideological Group in Advanced Capitalist trial Societies." American Journal of Political Countries." Data set. Retrieved October 10, Science 43:649-80. 2001 (http://www.marquette.edu/polisci/ Western, Bruce. 2001. "Institutions, Investment, Swank.htm). and the Rise in Unemployment." Pp. 71-93 in Therborn, Gbran. 1986. Why Some Peoples Are The Rise of Neoliberalism and Institutional More Unemployed than Others. London, En- Analysis, edited by J. L. Campbell and 0. K. gland: Verso. Pedersen. Princeton, NJ: Princeton University Traxler, Franz, Sabine Blaschke, and Bernhard Press. Kittel. 2001. National Labour Relations in In- Western, Bruce and Katherine Beckett. 1999. ternationalized Markets. New York: Oxford "How Unregulated Is the U.S. Labor Market? University Press. The Penal System as a Labor Market Institu- Traxler, Franz and Bernhard Kittel. 2000. "The tion." American Journal of Sociology 104: Bargaining System and Performance: A Com- 1030-60. parison of 18 OECD Countries." Comparative Western, Bruce and Kieren Healy. 1999. "Ex- Political Studies 33:1154-90. plaining the OECD Wage Slowdown: Reces- U.S. Bureau of Labor Statistics. N.d. "Compara- sion or Labour Decline?" European Sociologi- tive Civilian Labor Force Statistics, Ten Coun- cal Review 15:233-49. tries, 1959-2000." Data set. Retrieved May 5, Wilson, William Julius. 1996. When Work Dis- 2001 (http://stats.bls.gov/fls/home.htm). appears. New York: Vintage. Visser, Jelle and Anton Hemerijck. 1997. "A Windolf, Paul. 1989. "Productivity Coalitions Dutch Miracle": Job Growth, Welfare Reform, and the Future of European Corporatism."In- and Corporatism in the Netherlands. Am- dustrial Relations 28:1-20. sterdam, Netherlands: Amsterdam University Woodall, Pam. 1995. "A Survey of the World Press. Economy." The Economist, October 7, pp. 1- Volgy, Thomas J., John E. Schwarz, and 38.
This content downloaded from 192.124.250.5 on Mon, 16 Nov 2015 12:19:08 UTC All use subject to JSTOR Terms and Conditions