COMPARATIVE Verdier, Breen / EUROPEANIZATION POLITICAL STUDIES AND / April GLOBALIZATION 2001

The authors attempt to sort out three exogenous factors affecting the domestic societies of European Union (EU) member countries: market globalization, the European single market, and European supranational institutions. They offer a research design to separate the respective manifestations of each factor and apply it to four domestic dimensions: labor market, capital market, electoral competition, and center-local government relations. Although they find systematic evidence in the cases of the labor and capital markets supporting the widely shared claim that the EU is an agent of globalization, the results also point to the importance of the voluntarist component in the electoral and subgovernmental domains.

EUROPEANIZATION AND GLOBALIZATION Politics Against Markets in the European Union DANIEL VERDIER RICHARD BREEN European University Institute

T

he academic debate about European integration no longer bears on whether there is integration, as it used to until 15 years ago. It also does not seem to bear on whether the actual agent of this integration is the council, the commission, or the court. The debate bears, instead, on the mechanism that is responsible for that integration: Is it the market or the will to build a polity?

AUTHORS’ NOTE: We thank Geoffrey Garrett, Dennis Quinn, and Duane Swank for supplying us with some of their data. We thank our colleagues at the European University Institute, Jim Caporaso, three anonymous reviewers, and Brian McCormack for useful comments. A draft of this article was presented at the 1999 Meeting of the International Studies Association, Omni Shoreham, Washington, DC, in February 2000. Correspondence concerning this article should be addressed to Daniel Verdier, European University Institute, Via dei Roccettini 9, 50016 San Domenico di Fiesole (FI), Italy. COMPARATIVE POLITICAL STUDIES, Vol. 34 No. 3, April 2001 227-262 © 2001 Sage Publications, Inc.

227

228

COMPARATIVE POLITICAL STUDIES / April 2001

The debate is complicated by global trends. In Europe, as elsewhere, factor markets are being deregulated, partisan identification is eroding, local governments are becoming more active, and so forth. Yet from a European Union (EU) resident’s perspective, the origins of these ubiquitous changes are unclear. It is unclear whether they reflect global trends, referred to as globalization, or the play of forces that are directly attributable to European integration, Europeanization. When trying to explain changes in EU countries, therefore, observers are faced with a trinity of plausible factors: global markets, the single market, and the political union. Sorting out their respective effects is a daunting task, to which we give a first crack. We first try to theorize the general implications of each effect and then empirically test for their actual occurrence. Although we find evidence supporting the widely shared claim that the EU is an agent of globalization, our results also point to the importance of the voluntarist component.

THE QUESTION European integration can theoretically proceed in two ways. One is the construction of a single market without a political union. The other way is to build a political union with a wide range of centralized policies. Although reality falls somewhere in between these two ideals, political scientists are of the opinion that European integration as of late has leaned toward market integration more so than political voluntarism. Apologists praise what Majone (1996) calls the gradual “depoliticization of the Common Market” (p. 330), which has taken common market countries away from planning, corporatist self-regulation, and the public ownership of natural monopolies toward the regulation by experts and regulatory commissions of private and privatized monopolies. Consistent with this line of argument is the creation of a single currency managed by an independent central bank. Critics alike lament what Scharpf (1996) and Streeck and Schmitter (1991) call negative integration—the practice of striking down national regulation without replacing it with supranational regulation. The policy of merely purging markets from barriers to competition worked because, as Lange (1992) put it with respect to social policies, “There is no ‘compelling need’ to harmonize social policies in order for the single market to operate effectively” (p. 253). The outcome is even deemed by some as biased toward “big business”: Grahl and Teague (1989) equate European integration with “the gradual erosion of social constraints on the normless self-definition of economic objectives by the strongest enterprises themselves” (p. 50). The only dissenting voice

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

229

comes from two American economists. Alesina and Wacziarg (1999) write that Europe is going too far on many issues that would be better dealt with in a decentralized fashion, while it is not going far enough on policies that guarantee the free operation of market both across and within the countries of the Union. (p. 3)

Except for these two, the consensus among Europeanists is that EU institutions have promoted an exclusively economic form of integration, doing little to construct a centralized system of interest representation and decision making in a broad range of issues—a polity. If Europeanization is de facto synonymous with deregulation (or market-conforming “re-regulation,” as Majone, 1996, puts it), then its effects should be similar to the effects of globalization. The globalization of national markets is indeed a case of market integration by deregulation, involving the enlargement of the national market to the global market and requiring no (inter-)governmental action other than the deregulation and opening of markets.1 The deregulatory effects of globalization, presently the object of a voluminous literature, are commonly said to be four-pronged. We quickly survey these results, temporarily suspending judgment on their empirical validity. First, the openness of product markets intensifies competition between firms, forcing the path of innovation (see Castells, 1996; M. E. Porter, 1990) and increasing the instability of input markets, in particular, the labor market (see Streeck, 1987). The capacity of firms to locate new investments in wage havens makes domestic investment sensitive to domestic wage levels (see Chase, 1998; Thurow, 1996). Labor instability creates a demand for government to insure workers against market risk through unemployment benefits, government employment, and the provision of assorted social services. Second, and simultaneously, cross-border capital mobility undermines the capacity of the government to deliver this much-needed insurance. Capi1. The identity between globalization and market competition is not absolute. Global markets do not exist in an institutional vacuum but are embedded in NATO, the OECD, the GATT, the World Trade Organization, and the belief, held by the most advanced countries, in the desirability of trade and financial openness. The difference between these global regimes and the European Union (EU) is one of degree—the latter has stronger coordination mechanisms than the former. Because all EU countries are also members of all global regimes, the potentially spurious impact of global regimes on markets is automatically controlled for, allowing the analysis to focus on the added impact of EU regional institutions. Furthermore, although some non-EU countries are also involved in some form of regional organization (European Free Trade Agreement, North American Free Trade Agreement, Mercosur, and so forth), none of these schemes have reached a level comparable to the EU.

230

COMPARATIVE POLITICAL STUDIES / April 2001

tal mobility increases the elasticity of the domestic tax base with respect to the tax rate (see Bates & Da-Hsiang, 1985; Rodrik, 1997; Steinmo, 1993). Capital mobility also deters governments from financing budget deficits by printing money or over borrowing (see Kurzer, 1993; Strange, 1986). Which of these two forces—the rising demand for public insurance and the sinking capacity to meet that demand—prevails is a priori indeterminate and a matter for empirical research. Both Garrett (1995, 1998) and Swank (1998) argue that some of the alleged effects, far from being general, are mediated by domestic institutions. Third, as markets are becoming more important, governments are becoming less so. Loyalty to political parties declines, along with voter turnout and government stability. Politicians are losing the capacity to govern at the same time as government loses some of its prior relevance to market allocation. Fourth, greater factor mobility frees up latent economies of agglomeration, causing a territorial relocation of mobile factors away from poor or declining peripheries to wealthier and upcoming centers (see Fujita, Krugman, & Venables, 1999; Krugman, 1991). Increasing territorial inequalities between local jurisdictions raises the demand for offsetting territorial transfers administered by the central government. The capacity of the central government to supply these transfers, however, is on the decline. Districts that win from relocation oppose these transfers (see Bartolini, 1998). Lower dependence on the national market makes secession a credible threat (see Alesina & Spolaore, 1997; Bolton, Roland, & Spolaore, 1996). Which of the two forces—the increasing demand for offsetting territorial transfers or the declining supply of such transfers—wins, here again, seems to be an empirical matter. All the presumed effects of globalization—the emphasis on labor market flexibility, bank privatization, financial deregulation, low voter turnout, increasing electoral volatility, the threat of secession—are not unfamiliar to EU countries. These similarities fuel the claims of the above-mentioned literature that Europeanization is globalization by another name. The EU is seen as a simple agent of globalization or an irrelevant intervening factor. European economies would have reached a qualitatively similar state of market deregulation by simply exposing themselves to the global winds without engaging in the costly and painstaking construction of Europe. The commission is taking the praise (or alternatively, the blame) for an outcome over which it has limited control. What would be an alternative to globalization? Assuming for a moment that Europeanization is not reducible to globalization but that the political component is as active and as much developed as the market component, what would Europeanization look like? We venture that the strengthening of

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

231

the political dimension, had it occurred, would translate into the preservation of the existing (and/or the development of an alternative) interventionist capability. The reasoning runs like this: Political voluntarism, irrespective of its substantive goals, is ineffective without a centralized decision-making process, easing coordination among all interested parties—not only the national governments but also a comprehensive system of interest representation, including the political parties that are temporarily in opposition, trade associations and trade unions, employers’ and employees’ peak associations, consumers’ associations, and so forth. The superiority of such a centralized system of interest representation is its unique capacity to produce policies that are not enforceable without the consent of the interested parties. The existence of such a mechanism was found to be essential to the stabilization of European economies in the wake of the oil shocks. In the context of the EU, such a centralized and comprehensive decision-making process can take one of two forms: intergovernmental or federal. The intergovernmental mode of centralized decision making, the one that is more prevalent in the present state of bounded federalism, relies on the maintenance of separate centralized interest representation mechanisms in each member country and their coordination at the supranational level through government delegates (the existing Council of Ministers and derived committees). It is a case in which interest groups and political parties remain linked to national governments for two reasons. First, these national governments enjoy veto power within the EU legislative process. Second, the absence of an EU budget makes any compensatory policy the province of the national government. Among the four issue areas that we survey below, this intergovernmental decision-making process prevails in three—labor market, financial market, and electoral volatility. The second type of centralized interest representation is federal. It would involve the creation of regional forms of interest representation in factor markets and the polity—that is, social corporatism at the union’s level in the labor market, EU-regulated credit allocation in the capital market, disciplined parties in the European Parliament, and centralized bureaus in Brussels. This form of interest representation is unknown to Brussels (see Streeck & Schmitter, 1991; Turner, 1996). Its closest, yet still far remote, approximation is the structural funds policy. This is a rare case in which the EU has a budget and the commission enjoys some real spending power. Certainly, governments decide on country quotas; but mutual suspicion, backed by experience, that funds might be wasted on consumption by national recipients forced governments to establish strict standards on spending, relinquishing monitoring to the commission. Governments also agreed to delegate spending authority to subnational jurisdictions and allow direct

232

COMPARATIVE POLITICAL STUDIES / April 2001

bargaining with the commission. The upshot has been the growth of a centralized system of subnational interest representation in Brussels (see Marks, Nielsen, Ray, & Salk, 1997; Smyrl, 1998). Although these two modes are usually represented as theoretically antithetical (viz. the debate between so-called “intergovernmentalists” and the manifold heirs to Haas’s “neofunctionalism”), they are both distinct from integration via market deregulation. Either one could supply European political elites with a voluntarist capability, enabling them to pursue outcomes in concurrence with markets or beyond what markets can deliver. There is no necessary one-to-one correspondence between political voluntarism and federalism or between market liberalism and intergovernmentalism. The absence of EU-wide collective bargaining is no evidence that European integration is irrelevant to labor policy in Europe. We should also note that the intergovernmental and federal facets of political voluntarism are not empirically incompatible. A regional system of interest representation, if any, would have to be built in a first stage on the shoulders of the existing national ones. European peak associations and European parties will be conglomerations of national units for an indefinite amount of time. We recapitulate the argument. Europeanization, unlike globalization, walks on two legs—market efficiency and political voluntarism. The market has decentralizing and deregulating effects, making Europeanization synonymous with globalization. In contrast, the polity has centralizing effects, distinguishing Europeanization from globalization. Equipped with these definitions, we are now in a position to sort out the relative impact of the single market and the political union while controlling for the impact of globalization. We consider four areas of interest representation—the labor market, the capital market, and the political system, with the latter subdivided into political parties and subnational governments. In each case, we ask two questions: Are any of the changes that are observable in interest representation attributable to Europeanization as opposed to globalization? If so, are any of these changes associated with the voluntarist component of Europeanization as opposed to the market component? We answer both questions in the affirmative with respect to the party system and subnational governments only. In factor markets, we find Europeanization to be globalization by another name. We first present the research design followed by a quantitative survey of the four points of impact of globalization and Europeanization. The article ends with some general conclusions.

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

233

RESEARCH DESIGN Our choice of method should enable us to separate three hypothetical effects: market globalization, market Europeanization, and voluntarist Europeanization. Either of the last two effects, moreover, may either be triggered by globalization or be sui generis.2 This makes for five hypothetical situations: 1. The first, termed globalization-plus, implies that Europeanization is a regional case of market broadening and deepening—a regional instance of globalization. According to this hypothesis, EU-member countries are subject to two cumulative forms of market broadening—global and regional. As a result, European countries should evince a stronger case of globalization than non-European countries. The globalization-plus effect implies that Europeanization is driven by the market. 2. The opposite state, dubbed globalization-minus for reasons soon to be apparent, implies that Europeanization is an insurance against external market vicissitudes. According to this hypothesis, European countries are subject to two opposite demands—globalization and antiglobalization—which force them to choose a lower level of globalization than non-European countries. Cases of globalization-minus directly reveal the impact of the voluntarist component of Europeanization. 3. The sui generis market effect implies that Europeanization has effects that are unique to EU countries (that is, an effect associated with Europe that is not caused by globalization). Two cases are possible depending on whether this sui generis effect is generated by the market or by voluntarist policies. 4. The sui generis voluntarist effect is the second case of sui generis effects. 5. Last, the null effect corresponds to the situation in which Europeanization has no discernible impact one way or another. According to the null hypothesis, European countries should exhibit the same level of globalization as non-European countries.

We will test for the first, second, third, and fourth hypotheses combined and the fifth by means of one multiple regression equation. We will then separate the third and fourth hypotheses by resorting to two different specifications of the Europeanization variable. Our generic equation examines the impact of globalization and Europeanization together and interactively. Yt = β1 + β2(Yt – 1) + β3(Globt) + β4(Eurot) + β5(Globt*Eurot) + β6(Xt) + εt. 2. We assume that globalization is exogenous to Europeanization.

(1)

234

COMPARATIVE POLITICAL STUDIES / April 2001

Our dependent variable is one of the four areas of interest representation (Yt) already referred to—labor, finance, elections, and local governments. Our model specification says that the domestic dimension at time t can depend on its prior value (Yt – 1), globalization (Globt), Europeanization (Eurot), the interaction of globalization and Europeanization (Globt*Eurot), and a set of control variables (Xt) to be specified in each case. The coefficients β3, β4, and β5 play the central role in tests of our hypotheses. The presence of the interaction term between Europeanization and globalization means that we can interpret β3 as the effect of globalization where Europeanization is absent (see Friedrich, 1982). This is most easily seen where Europeanization (Eurot) is simply captured by a dummy variable distinguishing the EU countries (coded 1) from the other countries (coded 0). In the case in which Eurot is equal to zero—that is, for the non-EU countries—Equation 1 reduces to: Yt = β1 + β2(Yt – 1) + β3(Globt) + β6(Xt) + εt.

(2)

In Equation 2, the coefficients β4 and β5 do not appear because non-EU countries have a zero score on the variable to which these coefficients apply. β3 is the effect of globalization in non-EU countries. In the case in which Eurot is equal to 1—that is, for EU countries—the corresponding equation is ∆Yt = β1 + β4 + β2(Yt – 1) + (β3 + β5 )(Globt) + β6(Xt) + εt.

(3)

In this case, the coefficient for the EU dummy variable (β4) can be thought of as an extra term added to the intercept of the regression model, and the effect of globalization on EU members is given by the sum β3 + β5. In sum, estimating Equation 1 yields the impact of globalization on EU countries (β3 + β5) and non-EU countries (β3). The globalization-plus hypothesis implies that the effect of globalization in the EU countries—namely, the sum β3 + β5—has the same sign as its effect on non-EU countries—β3—and is significantly larger than it (significantly and significant are used throughout to mean statistically different from zero at the 5% level).3 In this case, globalization is operating in the same direction within and outside the EU, but its effect is stronger within the EU. A special case occurs when β3 is not significant but β3 + β5 is. In this case, globalization would be having an effect within the EU but not outside it. European mem3. The standard error of β3 + β5 is given by the square root of the sum of the variance of β3 plus the variance of β5 plus twice the covariance between these two parameters. These quantities can all be obtained from the variance-covariance matrix of the parameter estimates. Note that β3 and β3 + β5 are significantly different if β5 is significantly different from zero because β5 is the difference between β3 and β3 + β5.

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

235

bership would thus be revealing the otherwise latent effect of globalization. Although it is a subcategory of the globalization-plus hypothesis, we will refer to this case as one of revealed globalization.4 The globalization-minus hypothesis implies (a) that β3 + β5 has the same sign as β3 and that β3 + β5 is significantly closer to zero than β3 or (b) that β3 and β3 + β5 have different signs and are significantly different. In the first case, the overall effect of globalization is weaker in the EU than outside it; in the second, the effect runs in different directions in each.5 If β3 is statistically significant but β5 is not, then the impact of globalization is the same within the EU as in the rest of the OECD. In this case, neither the globalization-plus nor the globalization-minus hypotheses would be supported. Our third hypothesis—the sui generis European effect—is tested using the β4 coefficient. If this is statistically significant, it means that there is a difference between the EU and the other countries in our sample that does not arise as a result of the differential effects of globalization (because these are captured in β3 and β5). Specifically, the change in the dependent variable would, on average, be either larger (β4 positive and significant) or smaller (β4 negative and significant) in the EU than outside it. Note, however, that if β4 were significant and the conditions for the globalization-plus hypothesis, given above, were met, this would be a case in which there were two sorts of Europeanization taking place—catalyzed by globalization and sui generis. Similarly, we could also find a globalization-minus effect operating together with the sui generis effect. Finally, our fourth hypothesis—the null effect—would be supported if in Equation 1, given β3 is significant, neither β4 nor β5 was significant. In such a case, the impact of globalization would not be distinctive in Europe and there 4. The revealed-globalization effect is not to be confused with a sui generis effect of Europeanization soon to be presented because the latter does not require the presence of globalization in order to occur. 5. A (fictitious) example may help visualize the difference between the two cases. Were greater trade openness to have the overall effect of decentralizing wage bargaining between employers and unions from the national to the plant level, European governments could weaken that impact by raising trade barriers. In such a case, globalization would decentralize wage bargaining outside the EU (β3 is negative and significant) and would have a different impact among EU countries (β5 is positive and significant), that is, no impact whatsoever (β3 + β5 not significantly different from zero). Alternatively, European employers could respond to the trade challenge by increasing coordination with the unions at the central level, as in a planned economy. In this second case, globalization would still decentralize wage bargaining among non-EU countries (β3 is negative and significant) and would still have a different effect in EU countries (β5 is positive and significant), that is, to centralize wage bargaining (β3 + β5 is positive and significant).

236

COMPARATIVE POLITICAL STUDIES / April 2001

would not be any sui generis European effect. A subcategory of the null effect is no effect anywhere, inside and outside the EU. The same strategy applies in those cases in which Europeanization is measured as a continuous rather than as a dummy variable. The only minor complication is that the effect of globalization will be given by β3 + β5*Eurot for all countries and, because the coefficient for globalization is then a function not only of parameters but also of the value of the Europeanization variable, its statistical significance will likewise depend on the value of Europeanization. In summary, four of our five hypotheses are tested by focusing on the coefficients of Equation 1. It remains to explain how we differentiate between Hypotheses 3 (sui generis market effect) and 4 (sui generis voluntarist effect). We need two distinct measures of Europeanization, differing in terms of their relative sensitivity to each effect. The simplest measure of Europeanization—a dummy variable taking a value of 1 for countries that are part of the EU and 0 for countries that are not—is a reasonable measure of the voluntarist component of European integration. But it is not a good measure of the market component, for it presumes that all member countries have the same exposure to European market forces when in fact they do not—the large countries are much less exposed than the small ones. Conversely, a measure of the trade dependence of a country on the EU countries is a good measure of the market component of European integration but a poor measure of the voluntarist component of that same integration. It ranks Switzerland, which is not a member of the EU although almost totally dependent on it for its trade, above Germany, which although one of the original members of the EU, is a large country whose economy is comparatively less exposed to external market forces in general. Hence, we will choose between Hypotheses 3 and 4 according to whether the sui generis effect is stronger with the European transaction dependence variable or with the EU dummy. We will pattern the European transaction dependence variable after the globalization variables. A commonly used measure of globalization is trade dependence, calculated as the sum of a country’s imports and exports divided by that country’s gross domestic product. The equivalent measure of the market component of Europeanization can be calculated as the sum of a country’s imports and exports with members of the common market divided by that country’s gross domestic product. Another common measure of globalization is dependence on capital flows. Several variations of it are conceivable. Quinn (1997) built a yearly index of legal openness that summarizes each country’s exchange restrictions during the 1950-1993 period. One may also use differentials in interest-covered parity (Shepherd, 1994) on the grounds that the absence of flows does not constitute a priori evidence of

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

237

market segmentation. Or more simply, one may use the International Monetary Fund and OECD measures of actual capital flows—direct, portfolio, loans, or total. However, there is a problem with all these measures, for they do not allow for the calculation of an equivalent Europeanization index, the first two because a country’s exchange controls and interest rates are undifferentiated by countries of destination, the last one (actual flows) because of data limitations. Country-by-country breakdowns of capital flows exist for direct investment only, and flow data are not available before the mid-1970s. Because it takes about 10 years of flow data to calculate a reasonably accurate measure of stock, stock data are not available before the mid-1980s or even later for many countries.6 The models are tested on the population of OECD countries. The OECD is a club of rich countries, and this makes for a homogeneous sample of cases. The fact that all EU countries are OECD members but not all OECD countries are EU members provides us with the requisite control group, without which it would be impossible to distinguish between the effects of the two external mechanisms. Moreover, the fact that not all European countries are EU members is also important in separating the effects of EU membership from the effects that derive from a common political history and geographic proximity. All findings reported below are robust to the inclusion of a European geopolitical dummy, coded 1 for the countries located in the geographic region of Europe and 0 for others. The dependent variables are several dimensions of domestic societies that are affected by globalization and/or Europeanization. We cover the two factor markets (labor and finance) and, within the political arena, national parties and local governments. The estimation method varies with the type of data and their availability. In the presence of time-series cross-sectional data—our standard—we use generalized least squares with panel-corrected standard errors (see Beck & Katz, 1996). We include dummy variables for each country but one to eliminate idiosyncratic differences in scale between countries (“fixed effects”). Our standard specification is Equation 1, including the lagged dependent variable. We will refer to it as the lagged dependent variable model. We will use it as default, except in three cases. First, if the coefficient on the lagged dependent variable exceeds 0.9, we test for cointegration. A coefficient close to one indicates that the dependent variable has a long memory or is path dependent. We then look for variables 6. Flow (unlike stock) data are also notoriously volatile, making their use hazardous other than in the form of 10-year averages. Data were extracted from the annual issues of OECD’s International Direct Investment Statistics Yearbook and National Accounts over a 30-year period.

238

COMPARATIVE POLITICAL STUDIES / April 2001

that maintain a predictable relationship (are cointegrated) with the dependent variable in the long run and are weakly exogenous to it. The cointegration model we use is identical to the one suggested by Beck and Katz (1996, p. 11): ∆Yi, t = β1 + β3∆Globi, t + β4∆Euroi, t + β5[∆Globi, t*∆Euroi, t] + β6∆Xi, t + β2(Yt – 1 – γ3Globi, t – 1 – γ4Euroi, t – 1 – γ5[Globi, t – 1*Euroi, t – 1] – γ6Xi,t – 1) + εi, t,

(4)

with, on the right-hand side, coefficients for the change variables measuring short-term effects (β3, β4, β5, and β6) and coefficients for the lagged variables measuring long-term effects (the β2γs). Cointegration between the dependent variable and an independent variable is verified when both β2 and the product between β2 and the γ coefficient for the lagged independent variable are significant. If Euro is a dummy variable, Equation 4, like Equation 1, reduces to two simpler equations, with β2γ3 measuring the long-term impact of globalization on non-EU countries, β2(γ3 + γ5) measuring that on EU countries, β2γ5 measuring the long-term difference in the way globalization affects EU and non-EU countries, and β2γ4 measuring the sui generis effect. Second, if the coefficient does not exceed 0.9, we use the specification given by Equation 1. But following Beck and Katz (1996), we test this model against the more general model in which it is nested. We refer to this as the all-lagged model: Yi, t = β1 + β2Yt – 1 + β3Globi, t + β4Euroi, t + β5 [Globi, t*Euroi, t] + β6Xi, t + γ3Globi, t – 1 + γ4Euroi, t – 1 + γ5[Globi, t – 1*Euroi, t – 1] + γ6Xi, t – 1 + εi, t.

(5)

Finally, if pooling time and cross-sectional series is impractical, we regress the change in domestic dimension against its value at the beginning of the period and corresponding changes in the other right-hand-side variables. We will refer to this specification as the cross-sectional change model.7 Although quite intelligible, this method presents two drawbacks. First, it misses nonlinear changes (that is, changes between the initial and terminal value of the dependent variable that bounce around the trend spanning these two values). This is a minor problem, however, in the presence of data exhibiting trends. A second drawback of the cross-sectional design is the small number of observations that we can feed into it, requiring that we be watchful for potential outliers. We want to guard against reporting as finding results 7. It is similar to Equation 1 except that all variables on both sides (except for the lagged dependent variable) appear as first differences, and t-1 refers to the starting value of a multiyear period and t to its terminal value.

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

239

that are driven by outlying values and against discarding correlations that are hidden by outliers.

ORGANIZATION OF THE LABOR MARKET There exists a consensus in the labor literature that globalization of markets is forcing employer-employee relations to become less corporatist and more fragmented. The internationalization of capital is rendering national labor confederations irrelevant because these national organizations are unable to coordinate across national boundaries and together establish an international labor confederation. As a result, wage bargaining is decentralized to lower levels, rates of unionization are dropping, strike activity is declining, and government employment is shrinking. On the rise are unemployment, part-time employment, and performance-related pay (see Crouch, 1993; Ferner & Hyman, 1992; Streeck, 1987). Is this trend equally felt among EU countries? Two hypotheses are plausible a priori. On one hand, European labor markets are not immune to market shocks but, in fact, are even more affected by them in light of the more advanced level of product and financial market integration achieved in Europe. On the other hand, prospects for coordination among national labor confederations are brighter inside than outside the EU. The existence of intergovernmental institutions, along with the prodding of the commission and the court, makes it at least conceivable that enough regulatory coordination could be achieved to offset the worst effects of capital internationalization. The “social pillar” of the 1992 Maastricht Treaty, however fledgling it may look, signals a different choice for Europe. In sum, EU membership may offer trade unions and their governments additional political options besides market adjustment, justifying the preservation of social corporatism at the national level. We test these two hypotheses against each other and the null hypothesis on two labor market dimensions: trade union density and wage-bargaining level. Our findings offer support for the market integration hypothesis. We find that Europeanization has no bearing on trade union density. With respect to bargaining level, we find that financial openness has a globalization-plus effect in which the EU significantly outperforms the rest of the OECD. We start with union density. We control for the presence of left parties in government. Because the literature is unanimous about the idea that left government is positively correlated with corporatism, the test gains in accuracy if that effect is held constant. We also control for the four cases in which the payment of unemployment benefits is administered by the unions—

240

COMPARATIVE POLITICAL STUDIES / April 2001

Belgium, Denmark, Finland, and Sweden. In these countries, unemployment increases the incentives for workers to join a union. We finally control for geographic presence in Europe.8 We use two alternative measures of globalization: nominal capital openness and trade dependence. Data availability allows us to pool data for 16 countries during 30-plus years of observations. The coefficient of the lagged dependent variable is very close to one, requiring the use of the cointegration method presented in Equation 4. Two series of results are worth reporting (see Table 1). First, two of the control variables confirm existing claims. Left government is positively related with trade union density in the long run; it tests significant in the financial globalization regression (Equation 1), and marginally so (at the 5.6% level) in the trade dependence regression (Equation 2). Long-run upward shifts to the left of the ideological spectrum generate long-run upward growth in trade union density. Also, countries in which unemployment benefits are distributed by the unions show a long-term level of unionization higher than average. The European geopolitical dummy fails to test significant, however, suggesting that there is nothing particular about European geography or history with respect to trade union membership. Second, we find three types of effects of Europeanization—a globalizationplus effect adding to the demobilizing effect of globalization, a sui generis effect pointing to mobilization, and a null effect. The first two effects are found in Regression 1 (the financial globalization regression). The globalizationplus effect can be read from the coefficients β2γ3, β2γ5, β2(γ3 + γ5); they are all negative and significant. Nominal capital openness has a significantly greater negative long-term impact on trade union membership in EU than in non-EU countries. This globalization-plus effect is supplemented with an equally long-term sui generis effect.9 That effect works in the opposite direction of globalization (β2γ4 is positive and significant), canceling the latter at low values of globalization but being canceled at higher values. The tipping point, before which the sui generis effect prevails and past which the globalization-plus effect predominates, is equal to 8.11.10 Every EU member had already passed that threshold by the time it joined the common market, sug8. Our fixed-effects specification also controls for all factors that are country specific and time invariant. 9. The sui generis effect disappears if one substitutes the variable European trade dependence for the EU membership dummy, suggesting a political rather than a market origin. 10. The calculation of the tipping point runs as follows. The two effects on EU countries add up to β2γ4 + β2(γ3 + γ5)*Globi, t – 1. When the net effect is equal to 0, Globi, t – 1 is equal to –β2γ4 / β2(γ3 + γ5)—that is, 8.11 in Regression 1.

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

241

Table 1 Trade Union Density (cointegration model with fixed effects, generalized least squares estimates, and panel-corrected standard errors) Dependent Variable: ∆Trade Union Density 1 Trade union densityi, t – 1 ∆Nominal financial opennessi, t ∆Trade dependencei, t ∆EU membershipi, t ∆Nominal financial opennessi, t *EU membershipi, t ∆Trade dependencei, t *EU membershipi, t ∆Left party cabinet portfolioi, t Nominal financial opennessi, t – 1 Trade dependencei, t – 1 EU membershipi, t – 1 Nominal financial opennessi, t – 1 *EU membershipi, t – 1 Trade dependencei, t – 1 *EU membershipi, t – 1 Left party cabinet portfolioi, t – 1 Unemployment benefits paid by unions (dummy for Belgium, Denmark, Finland, and Sweden)i Geopolitical Europei (dummy) Intercept

Number of observations Number of groups Number of time periods Log likelihood Probability (chi-square)

2

β2 β3 β3 β4

–0.02 –0.12

(–2.51) (–1.18)

0.17

(0.25)

β5

0.37

(0.48)

a

β5 β2γ3 β2γ3 β2γ4 β2γ5

0.004 (1.42) a –0.10 (–2.45) 2.19

(2.31)

a

–0.17

(–2.08)

a

β2γ5 0.004 (2.10)

a

1.13 (1.67) 0.35 (1.39) a 1.39 (2.41) –0.12 (–1.18) β 3 + β5 b –0.27 (–3.76) β2(γ3 + γ5) 592 c 16 d 37 –685.7757 0.0000

–0.02 (–2.50)

a

0.79 (0.98) –4.98 (–1.49)

30.02 (1.49) 0.003 (0.93) –0.81 (–2.03)a –0.33 (–0.73)

–0.52 (–0.89) 0.003 (1.90)

b

2.46 (3.41) 0.77 (0.66) –1.52 (–1.85) 0.79 (0.98) b –1.32 (–2.94) 512 c 16 e 32 –535.6337 0.0000

Note: EU = European Union. The dependent variable is the first difference in trade union density adjusted for missing data by Golden, Lange, and Wallerstein (1997). Nominal financial openness is the level of nominal financial openness coded on a 0 to 12 scale by Quinn and Toyoda (1997). Left power is the percentage of all cabinet portfolios held by left parties; the source is Swank (1998). Trade dependence is the ratio (imports + exports)/gross domestic product; the source is OECD (National Accounts; see note 6). EU membership is a dummy variable coded 1 for member countries at year t, 0 for all others. Geopolitical Europe is a dummy coded 1 for the countries located in the geographic region of Europe and 0 for others. All the multiplicative variables are the product of their unstandardized components. Values of z statis-

(continued)

242

COMPARATIVE POLITICAL STUDIES / April 2001

Table 1 Continued tics are given in parentheses. All Greek symbols refer to Equation 4 in the text. The unit of observation is the country year. a. z is significant at the 5% level. b. z is significant at the 1% level. c. Australia, Austria, Belgium-Luxembourg, Canada, Denmark, Finland, France, Germany, Italy, Japan, the Netherlands, Norway, Switzerland, Sweden, the United Kingdom, and the United States. d. 1955 to 1992. e. 1960 to 1992.

gesting that the net effect of EU membership on trade union membership was always negative. The impact of trade dependence on trade union membership is also generally negative. Regression 2 suggests that trade dependence is cointegrated with the decline in trade union density within and outside the EU (β2γ3 and β2[γ3 + γ5] are negative and significant). However, there is no difference between EU and non-EU members (β2γ5 is not significant). The findings conform with the null hypothesis, according to which Europeanization makes no difference one way or another. We now turn to the bargaining level (see Table 2). The coefficient on the lagged dependent variable (β2) is well below 0.90, allowing us to bypass the cumbersome cointegration technique. Moreover, tests for residual serial correlation suggest that the lagged dependent variable takes out most of the serial correlation from the data.11 We also estimated both the full “all-lagged” model (Equation 5), in which the lagged values for both the dependent and the independent variables are included on the right-hand side of the regression, and the simpler model (Equation 1), including only the lagged dependent variable. We then performed a likelihood-ratio test. The hypothesis that the two equations are identical could not be rejected at the 5% confidence level. These results allow us to use the simple lagged-dependent-variable model of Equation 1. The left government variable tests positive and significant in Regression 1 and marginally significant in Regression 2. The dummy for European geography and history tests positive and highly significant in the second regression. The story about financial globalization is much the same as for trade union density. Although its impact on non-EU countries (β3) is indeterminate, that on EU members is significantly different (β5) and negative (β3 + β5)—a typical revealed-globalization effect, that is, a subcategory of globalization-plus. 11. We regressed the residuals against their lagged value and found no significant correlation.

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

243

Table 2 Bargaining Level (lagged-dependent variable model with fixed effects, generalized least squares estimates, and panel-corrected standard errors) Dependent Variable: Level of Wage Bargaining 1 Bargaining leveli, t – 1 Nominal capital opennessi, t Trade dependencei, t EU membershipi, t Nominal capital opennessi, t *EU membershipi, t Trade dependencei, t *EU membershipi, t Left party cabinet portfolioi, t Geopolitical Europei (dummy) Intercept

b

β2 β3 β3 β4

0.56 (16.94) –0.009 (–0.60) 0.75

(2.22)

β5

–0.08

(–2.86)b

β5

β3 + β5 Number of observations Number of groups Number of time periods Log likelihood Probability (chi-square)

2

a

a

0.001 (2.06) 0.15 (1.78) 1.49 (8.45)b b –0.09 (–3.57) 608 c 16 d 38 –132.2239 0.0000

0.56 (15.58)

b

–0.34 (–1.95) –0.08 (–0.39)

0.54 (2.05)a 0.001 (1.86) b 1.23 (3.92) –0.13 (–0.36) 0.20 (0.97) 528 c 16 e 33 –138.5172 0.0000

Note: EU = European Union. The dependent variable is the level of wage bargaining coded on a 1 to 4 scale by Golden, Lange, and Wallerstein (1997); the higher the number, the more centralized the bargaining. Values of z statistics are given in parentheses. All Greek symbols refer to Equation 1 in the text. The unit of observation is the country year. a. z is significant at the 5% level. b. z is significant at the 1% level. c. Australia, Austria, Belgium-Luxembourg, Canada, Denmark, Finland, France, Germany, Italy, Japan, the Netherlands, Norway, Switzerland, Sweden, the United Kingdom, and the United States. d. 1955 to 1992. e. 1960 to 1992.

Simultaneously, the EU dummy exhibits a positive sui generis effect (β4), prevailing over the former effect for values of globalization inferior or equal to 8.33—a threshold barely higher than the one we calculated for trade union density (see Regression 1, Table 1). Once again, because all EU members had more or less passed that threshold by the time they joined the common market, the net effect of EU membership on trade union membership was always negative. The effect of trade dependence on bargaining level in Regression 2 is unclear, hesitating between a weak case of globalization-minus and the null

244

COMPARATIVE POLITICAL STUDIES / April 2001

effect. Trade dependence has a marginally negative effect on non-EU countries (β3 is positive but not significant at the 5% level); it affects EU members differently (β5 is positive and significant)—that is, not at all (β3 + β5 is not different from zero). Most of the action is stolen by the geopolitical dummy, of which the positive coefficient suggests that bargaining levels in Europe are generally higher than elsewhere when controlling for trade dependence. Taking all results into consideration, trade does not seem to cause much of a difference between EU members and nonmembers. Financial openness, in contrast, exhibits a prevailing globalization-plus effect on trade union density and wage bargaining. In sum, our analysis lends plausibility to the detractors of European social policy and the claim that European integration is essentially market driven. Corporatism in the labor market, along with the possibility of market-correcting intervention, is declining fast among EU countries, faster than elsewhere. We found no evidence that labor confederations in the EU are any more capable of coordination across national boundaries within the EU ambit than outside it. Employees desert national labor unions everywhere, indeed more so within than outside the EU. European integration does not offer trade unions and their government political options justifying the preservation of social corporatism at the national level.

ORGANIZATION OF THE CAPITAL MARKET Recent research has shown that capital markets, like labor markets, can be more or less corporatist (Deeg, 1998; Verdier, 2000b). A corporatist capital market is one in which the interests of various borrowers (savers are never organized) are organized and articulated by various centralized institutions. Two types of borrowers compete for cash in capital markets—small- and medium-sized enterprises, which do not enjoy sufficient visibility to be traded on the market, and large enterprises, which do possess that visibility. These groups are not directly organized into confederations, but their respective bankers are. The bankers for large firms are the large center banks, headquartered in the national financial center and engaged in fierce competition with each other. The bankers for the small- and medium-sized companies, in contrast, are typically sheltered from the competition of the large banks. Their identities vary according to country: In Germany, Austria, Italy, Switzerland, and the Scandinavian countries especially, small firms bank with the nonprofit sector, which is composed of the savings banks, cooperative societies, and local banks controlled by local governments. In the United States, the only country that still allows local governments to charter for-profit banks, small firms also bank with locally chartered for-profit country banks. In

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

245

France, Belgium, New Zealand, and the Netherlands, the small borrowers’ main banker is (or was) the state credit sector, which includes all the specialized credit facilities that enjoy state borrowing privileges.12 The relative market share of each sector varies greatly across countries: In 1990, the competitive sector represented 92% of all banking assets in Australia but only 27 % in Germany; the state sector captured 25% of French banking assets but was nonexistent in Ireland, Sweden, and Switzerland. Each banking sector in each country is organized in a national trade association, whose role is to ensure that the rules of coexistence between sectors laid down by the central government agencies are not disadvantageous to its own sector. Unsurprisingly, these sectors hold conflicting regulatory preferences: The large banks favor market mechanisms, whereas nonprofit, local, and state banks favor, indeed need, regulatory protection from the center banks. Capital markets have been greatly affected by financial globalization. Deregulation has caused an increase in competition, forcing banks to concentrate and shift away from lending toward market activities.13 The Basle agreement on capital-assets ratios, one of the few instances of voluntarism in the domain of financial globalization, forced banks throughout the world to strengthen their solvability. Although the impact of financial globalization is much discussed, that of Europeanization seems nonexistent. Banking and financial regulation has received a lot of attention from the council and the commission, especially in the early 1980s.14 Yet, of the few studies that have looked for a European specificity, none has found any. Firms and banks offer as much diversity in Europe as outside Europe (Cerasi, Chizzolini, & Ivaldi, 1998). We argue that the effect of Europeanization is a priori indeterminate; it depends on which of the market or voluntarist components prevail. Integration through market deregulation is likely to favor the market-oriented, for-profit sector. Although it may lead to greater concentration in that sector, it would also yield a decentralization of interest representation. This is because the interests of each class of borrower would no longer be allocated through summit negotiations, involving government regulators along with representatives of the various banking sectors but would be decided by competition among a handful of oligopolies. Voluntarist integration, in contrast, is likely to keep intact existing mechanisms for the regulation of credit (or reproduce them at the supranational level in the European Ecofin committee), and thus freeze the existing allocation of market shares between sectors. 12. On the state sector, see Verdier (2000a). 13. On concentration, see T. Porter (1993) and Cerasi (1996); on securitization, see Thompson (1995). 14. See contributions by and to Underhill (1997).

246

COMPARATIVE POLITICAL STUDIES / April 2001

Therefore, if financial globalization favors concentration and market orientation, it should correlate with a redistribution of market shares away from the sheltered sectors (the local nonprofit, country, and state banks) toward the unsheltered sector (the center banks). If membership in the EU makes a difference, the impact should be significantly distinct in the case of the single-market countries. We ran the tests on two sectors, center and state, using respective market shares calculated in total assets. The coefficient on the lagged dependent variable being close to one, we use the cointegration model. We used only two measures of globalization—nominal capital openness and trade dependence— because various measures of cross-border capital flows variables yielded no results whatsoever.15 The findings support the two sides of the null hypothesis (see Table 3). First, long-term increases in globalization correlate with long-term increases in center banks’ market share and long-term decline in state banks’ market share. Second, European integration makes no difference whatsoever. Regressions 1, 2, and 3 have a common structure. The long-term impact of the global variable on non-EU countries, which can be read from β2γ3, is correctly signed and significant—positive for center banks, negative for state banks. However, the coefficient on the long-term interaction term (β2γ5) is insignificant, suggesting an identical impact within and outside the EU. Regression 4 also fails to show a difference between the two groups, detecting no impact of globalization anywhere. These results clearly suggest that the European capital market is an integral component of the global market. Market reform took place among EU countries at the same speed as among non-EU countries. We found no trace of Europeanization, let alone voluntarism, in financial markets. Furthermore, aside from the isolated negative result of Regression 4, globalization does have a long-term deregulatory impact on interest organization among financial institutions.

ELECTORAL TURNOUT AND VOLATILITY As markets thus become more important in (re)distributing income, political parties should become relatively less so. One should expect rational individuals to reallocate some of their wealth-maximizing effort away from politics toward markets. Loyalty to political parties should decline, and the 15. The extreme volatility of annual cross-border flows data considerably reduce their efficiency as a measure of financial globalization. Stock data are steadier, but time series are too short to be used in a time-series cross section.

247

Center banks’ market sharei, t – 1 State banks’ market sharei, t – 1 ∆Nominal financial opennessi, t ∆Trade dependencei, t ∆EU membershipi, t ∆Nominal financial opennessi, t *EU membershipi, t ∆Trade dependencei, t*EU membershipi, t Nominal financial opennessi, t – 1 Trade dependencei, t – 1 EU membershipi, t – 1 Nominal financial opennessi, t – 1 *EU membershipi, t – 1 Trade dependencei, t – 1 *EU membershipi, t – 1 Geopolitical Europei (dummy) Data breaks (dummies) Intercept β3 + β5 β2(γ3 + γ5)

–0.01 (–1.42) –0.001 (–0.88) 0.0009 (1.20)

b

–0.0008 (–0.92)

β2γ5

(3.09)

(0.87)

0.007

0.02

(3.46)b

0.002

β2γ5

(–0.91)

–0.002

(1.29)

–0.02

b

(3.48) (1.25)

(1.93)

0.065 0.02 0.01

c b

(4.12) (1.73) a (2.16)

–0.008 (–1.39)

0.02 –0.006

0.33

(1.73) (–0.88)

(0.27)

0.0004 0.02 –0.02

(–3.63)b

–0.59

(–2.30)a

–0.03

β5 β5 β2γ3 β2γ3 β2γ4

β2 β2 β3 β3 β4

2

1

∆Center Banks’ Market Sharei, t

(0.86)

(–0.98)

b

(–0.64)

(1.92)

(–1.95)

0.02 (2.69)a 0.002 (1.29) 0.0004 (0.60)

–0.003

0.001

–0.01

–0.0009 (–2.48)a

0.001

–0.015

–0.03 (–2.91) 0.0003 (0.22)

3

(–2.88)

(–0.80)

b

(continued)

–0.008 (–0.98) –0.003 (–0.26) –0.008 (–1.64)

c

–0.002 (–0.33)

–0.006 (–1.53) 0.001 (0.39)

–0.11

–0.003 (–0.26) 0.007 (0.56)

–0.04

4

∆State Banks’ Market Sharei, t

Dependent Variable

Table 3 Banking Sectors Market Shares (cointegration model with fixed effects, generalized least squares estimates, and panel-corrected standard errors)

248 532 d 13 41e 1719.698 0.0000

1 437 d 13 33f 1419.527 0.0000

2

∆Center Banks’ Market Sharei, t

532 d 13 41e 1843.257 0.0000

3

437 d 13 33f 1541.928 0.0000

4

∆State Banks’ Market Sharei, t

Dependent Variable

Note: EU = European Union. The dependent variables are the first differences of the respective market shares of center and state banks. Both typology and data are from Verdier (2000a). Values of z statistics are given in parentheses. All Greek symbols refer to Equation 4 in the text. The unit of observation is the country year. a. z value is significant at the 5% level. b. z value is significant at the 1% level. c. Dropped. d. Austria, Belgium-Luxembourg, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Norway, Sweden, the United Kingdom, and the United States. e. 1950 to 1995, with missing years. f. 1960 to 1995, with missing years.

Number of observations Number of groups Average number of time periods Log likelihood Probability (chi-square)

Table 3 Continued

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

249

floating vote should increase. Europeanization could have a similar effect on account of its deregulatory component. But it could also act in the opposite direction, mobilizing parties and electorates around competing projects of European integration, on account of its voluntarist component. The mobilization we have in mind need not involve the European Parliament, members of the European Parliament, or European elections. The electoral literature sees European elections as opinion polls, “second-order national elections,” or tests of the incumbent’s performance, characterized by low stakes and low turnout (see Eijk, Franklin, & Marsh, 1996; Franklin, Marsh, & McLaren, 1994, Reif & Schmitt, 1980). All that is required for the voluntarist leg to trip the trend toward voters’ disaffection is that European coordination be seen by national parties and electorates as a plausible alternative to global market allocation. We test two hypotheses: Europeanization preserves voter turnout from the market-engineered disaffection with politics, and Europeanization lowers electoral volatility. We begin with voter turnout, measured as the percentage of electorate casting valid votes at parliamentary elections (presidential in the United States). For years without elections, we use the score of the most recent election. Data are available for 18 countries over 38 years (1955-1993). We use the all-lagged model because the lagged dependent variable has a coefficient lower than 0.90 and the lagged independent variables make a statistically significant difference.16 The single regression of Table 4 reports the impact of nominal capital openness (the trade variable produced coefficients with insignificant coefficients, although with identical signs). The fixed effect dummies, which we use throughout, are particularly welcome here in light of the multiple national idiosyncrasies of a legal and cultural nature affecting voter turnout. We also include a measure of rising affluence in the regression to capture the common idea that embourgeoisement demobilizes working-class voters. The findings, reported in Table 4, point to a clear null effect: A rise in financial globalization is associated with a drop in voter turnout, but this effect is not significantly different within and outside the EU (γ3 is negative and significant but γ4, γ5, and γ3 + γ5 are not significant; none of the corresponding β’s are significant either). Note that wealth has the expected negative impact on voter turnout, as does being a European country, a result that can be explained by the fact that European countries used to have higher-thanaverage turnout rates and have been the most dramatically affected by the trend toward lower turnout. 16. The likelihood test ratio between the all-lagged and the lagged-dependent variable models allows us to reject the hypothesis that the two models yield similar results at the 1% confidence level.

250

COMPARATIVE POLITICAL STUDIES / April 2001

Table 4 Voter Turnout (all-lagged model with fixed effects, generalized least squares estimates, and panel-corrected standard errors) Dependent Variable: Percentage of Electorate Casting Valid Votes Turnouti, t – 1 Nominal financial opennessi, t EU membershipi, t Nominal financial opennessi, t*EU membershipi, t Gross domestic product per capitai, t Nominal financial opennessi, t – 1 EU membershipi, t – 1 Nominal financial opennessi, t – 1*EU membershipi, t – 1 Gross domestic product per capitai, t – 1 Geopolitical Europei (dummy) Intercept

β2 β3 β4 β5 γ3 γ4 γ5

β3 + β 5 γ3 + γ5 Number of observations Number of groups Number of time periods Log likelihood Probability (chi-square)

b

0.83 (36.20) 0.42 (1.85) 3.88 (1.32) –0.38 (–1.21) 1.21 (1.45) a –0.49 (2.19) –4.22 (1.42) 0.48 (1.51) a –1.68 (–2.00) b –1.94 (–2.59) 20.39 (7.74)b 0.04 (0.18) –0.02 (–0.07) 594 c 18 33d –938.9915 0.0000

Note: EU = European Union. The dependent variable is the percentage of the electorate casting valid votes. The source is Mackie and Rose (1991, 1997); the data were compiled by Swank (1998). All the multiplicative variables are the product of their unstandardized components. Values of z statistics are given in parentheses. All Greek symbols refer to Equation 5 in the text. The unit of observation is the country year. a. z value is significant at the 5% level. b. z value is significant at the 1% level. c. Australia, Austria, Belgium-Luxembourg, Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Switzerland, Sweden, the United Kingdom, and the United States. d. 1960-1993.

Voter turnout is only part of the story. Are EU voters, those who still vote, also forsaking past, or not developing new, partisan loyalties? Answering this question raises a difficult measuring problem. Standard measures of electoral volatility do not easily reveal trends. Aggregating variations from one election to the next in each party’s score, these measures are erratic—the score is high when the incumbent loses, small when it wins. The infrequency of elections (one every 5 years on average) makes the use of period averages

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

251

extremely sensitive to the choice of the cutting point and to the occurrence of new elections. Rather than measuring electoral volatility directly, we focus instead on its consequences for government stability. We use an index developed by Verdier (1995). For each government, each year, we calculate a position on a left-right continuum. This number is a weighted average of the relative position of the parties forming the government coalition. Each party’s relative position, in turn, is determined by how much it scored at the last election. This way of positioning parties requires no outsider’s judgment on the parties’ ideological orientation except for how parties are rank-ordered on a left-right continuum (we used Leonard & Natkiel, 1986, for the ranking). Consider, for instance, a simplified system of two parties, left and right, each polling half of the votes. The left party position on the 0-1 axis (0 for extreme left, 1 for extreme right) is its median voter—the middle of the space it occupies on the continuum—that is, 1/4. The right party being to the right of the left party, its position is 1/2 (the position of the right party’s most leftist voter) plus 1/4 (the distance between the party’s most leftist voter and its median voter), that is, 3/4. Were the two parties to govern in a grand coalition, the government position would be the simple average of their respective median voters (1/4 + 3/4 = 1/2) each year for the duration of the government or the legislature, whichever ends first. Were the two parties of different electoral size, we would weigh their presence in the government by the size of their respective electoral shares. Once we have generated a number summarizing the government position for each year, we calculate its standard deviation over two different periods (1950-1979 and 1980-1996).17 The difference between the two standard deviations measures the change in political volatility between the two periods. The technical presentation of the variable is left to a footnote.18 Because we have only two data points for the dependent variable, we use the cross-sectional change model. Globalization is measured by foreign direct investment stocks (the foreign direct investment–flow measure reached comparable yet outlier-sensitive results, whereas the trade and nominal financial openness variables displayed no association whatsoever). The Swiss observation is an outlier, reflecting its strong international financial 17. The two periods were chosen to fit data availability for the independent variable. 18. The dependent variable is the first-order difference between the average standard deviation of the government partisan orientation for the period 1950-1979 and that for the period 1980-1996. The index was calculated by first assigning to each party a positive integer i (i ∈ Z+) according to the ordinal pattern: 1 to the most leftist party, 2 to the next to the most leftist, and so forth until all parties are ordinally arranged on a left-right axis (we excluded unclassifiable other parties but made sure that the category “other” always remained below 5%). Each i was then

252

COMPARATIVE POLITICAL STUDIES / April 2001

specialization, out of line with the rest of the economy. Switzerland is also ruled by a stable coalition, making regular elections (as opposed to referenda) less prone to convey changes in electoral behavior than in other countries. As previously, we control for the initial value of the dependent variable and for being part of the European region at large. We also control for the growth in affluence, although the sign of the effect is a priori indeterminate—rising affluence lessens loyalty but also makes voters less likely to oust incumbents (the main source of instability). The results, presented in Table 5, first point to a case of conditional convergence: That is, countries with much political volatility in the past exhibit less today and vice versa. Second, there is no significant sui generis effect (β4 is not significant) but a clear case of globalization-minus. An increase in the financial globalization variable is associated with an increase in volatility in non-EU countries (β3 is significantly greater than zero) but not in EU countries (β5 is significant and of opposite sign) where there is no impact (β3 + β5 is insignificant). EU members were, on average, insulated from the effects of increasing financial exposure that elsewhere were associated with higher electoral volatility. The results thus unambiguously point to the impact of the voluntarist component of Europeanization. Financial liberalization is not associated with the dealignment of EU voters the way it is with non-EU voters. We do not observe a clear effect of the control variables—the European geopolitical dummy and growth in affluence. This is not a truly surprising result. It reflects the fact that irrespective of whether European integration is driven by Adam Smith’s “invisible hand” or by government, there is a political center that is there to take the praise (or alternatively, the blame) for most of what happens in relation to integration. It is less plausible for EU voters to sacrifice politics to markets in the face of greater market openness than it is for non-EU voters. The simultaneous inteassigned a positive real number pi (pi ∈ R: 0 < pi < 1) to reflect each party’s share of the electorate according to the formula i −1

r pi =  i  + ∑ rj ,  2  j=0 with ri party i’s percentage of votes and j an integer representing the parties on the left of i, with 0 ≤ j < i and rj = 0 if j = 0. Last, the government orientation index pg (pg ∈ R: 0 < pg < 1) was calculated by averaging the pis for those parties in government at any point during the given period: 1 pg =    ∑ ∑ ( pit * G it ) ,   T   t =1 i =1 T

N

with Git = [0,1], depending on whether party i at time t is part of the government (Git = 1) or not (Git = 0), T = the number of years, and N = the number of parties.

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

253

Table 5 Change in Electoral Volatility Between the 1950-1979 and 1980-1996 Periods (cross-sectional change model with ordinary least squares estimates) Dependent Variable: ∆Volatilityt Volatilityt – 1(1950-1979 average) EU12 ∆FDI stockst (1980-1995) ∆FDI stockst*EU12 Gross domestic product per capita growtht (1980-1995) Geopolitical Europe (dummy) Intercept

β2 β4 β3 β5

β3 + β5 Adjusted R-squared Number of observations

–1.12 0.08 0.005 –0.005 0.000 0.009 0.08 –0.0003 0.52 c 18

b

(–4.49) (1.42) a (2.96) a (–2.37) (1.43) (0.22) (0.12) (–0.19)

Note: The dependent variable is the difference between the volatility ratio defined in Note 18 calculated over the 1950-1979 period and that same ratio for the period 1980-1996. Sources for volatility include Mackie and Rose (1991, 1997) and Woldendorp, Keman, and Gudge (1993, 1998). The lagged dependent variable is the average for the first period. ∆FDI stocks is the first-order change over the 1980-1995 period of the ratio (total direct investment stocks in + total direct investment stocks abroad)/gross domestic product; source is United Nations (1995). EU12 is coded 1 for the 12 EC members of the late 1980s and 0 for all others. Values of t statistics are given in parentheses. All βs refer to Equations 1-3 in the text. a. t value is significant at the 5% level. b. t value is significant at the 1% level. c. Australia, Austria, Belgium-Luxembourg, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Sweden, the United Kingdom, and the United States.

gration of markets and governments in Europe maintains voters’ loyalty to their political parties. Because it is quite plausible that the lower turnout disproportionately affects voters without partisan loyalty, this would explain the discrepancy between turnout, indiscriminately lower, and volatility, lower outside the EU. Voters seem to expect something to come out of the voluntarist component of Europeanization.

STATE DECENTRALIZATION Local governments constitute another important level of interest articulation. How is their power affected by globalization and Europeanization? There is an emerging consensus across disciplines that modern production

254

COMPARATIVE POLITICAL STUDIES / April 2001

has a territorial, local dimension, reinforcing economic disparities between districts. M. E. Porter (1990) writes that “more open global competition makes the home base more, not less, important” (p. 58). Krugman (1991) shows that the decline in transportation costs makes firms want to agglomerate. Students of flexible specialization stress the importance of geographical concentration in attracting talented people and the role of proximity in the production of learning and innovation (Sabel, 1989; Saxenian, 1994; Storper, 1995). Amin and Thrift (1992) write that “the world economy may have become decentralized, but it is not necessarily becoming decentered” (p. 576). Economies of agglomeration have severe redistributional consequences for local districts. Those with an already dense industrial base may see it further reinforced, whereas those with a weak one risk losing what they have, and those without any might remain barren. The most favored areas are those located around large metropolitan regions (Tödtling, 1994). Less favored are the industrial districts located at the periphery. The upshot is an end to the postwar consensus. The losers of globalization want more territorial transfers and thus more tax revenues redirected to the central government, whereas the winners resist any further centralization, or even ask for less of it. Like globalization, market integration within the EU causes a need for territorial transfers aiming at a reduction of income disparities. An essential difference, however, is that EU members have set up a common budget to which they all contribute according to means and from which they draw according to needs, with the result that some members are net contributors whereas others are net beneficiaries. Another difference is that these transfers have a decentralizing impact on state structures. About half of the so-called structural funds that are distributed by Brussels are earmarked for local governments, thereby forcing centralized states to revitalize local government. France, Spain, and Britain, among others, have recently implemented decentralizing reforms. The rationale for such decentralization, which in the late 1980s went hand in hand with a shift of monitoring power toward the commission, hides no antigovernment motive but can be studied as a commitment mechanism that governments resorted to in order to allay mutual suspicions of waste and corruption (Tsoukalis, 1993). Therefore, whereas market integration should have the universal effect of increasing central government’s control over the flow of tax funds, political integration among EU countries should weaken it. One can assess the relative importance of the market and voluntarist components of European integration by tracking which of the centralizing or decentralizing effects is empirically more pronounced. Our measure of centralization is the ratio of central government receipts to general (central and local) government receipts. Although available on an

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

255

Table 6 Change in State Centralization, 1970-1994 (cross-sectional change model with ordinary least squares estimates) Dependent Variable: ∆State centralizationt 1 State centralization, 1970 ∆Nominal financial openness, 1973-1993 EU12 ∆Nominal financial openness*EU12 Structural funds, 1990 ∆Nominal financial openness *Structural funds Geopolitical Europe (dummy) Intercept

β2

–0.17

β3 β4 β5 β4

0.03 (2.18) 0.13 (1.72) –0.023 (–1.61)

(–1.02)

β5 β3 + β5

Adjusted R-squared Root MSE Number of observations

2

0.006 (0.15) –0.03 (–0.28) 0.009 (1.20) 0.25 0.05411 c 19

–0.14 a

(–1.14) a

0.0256

(2.37)

0.0028

(2.49)a

–0.00042 (–2.34)a 0.02 (0.56) –0.02 (–0.30) 0.31 0.04929 19c

Note: The dependent variable is the first-order change over the 1970-1994 period of the ratio central government receipts/(central and local government receipts – transfers from central to local governments); the source is OECD (National Accounts; see Note 6). Structural funds is a variable equal to per capita receipts in ECU of EC structural funds (regional, social, and agricultural) in 1990 for the 12 European Union countries and coded 0 for all others (Commission of the European Communities, 1994). Values of t statistics are given in parentheses. All βs refer to Equations 1-3 in the text. a. t value is significant at the 5% level. b. t value is significant at the 1% level. c. Australia, Austria, Belgium-Luxembourg, Canada, Finland, France, Germany, Greece, Ireland, Italy, Japan, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, and the United States.

annual basis, the time series are plagued with multiple breaks, considerably weakening efficiency. Our attempts to run pooled models bore no fruit, yielding contradicting and unintelligible results. We use instead the cross-sectional change model, a simpler and more manageable model. The dependent variable is the difference in state centralization between the 1994 and 1970 values. The base model includes the 1970 value of this variable and the European geopolitical dummy as control variables. Table 6 reports results for the nominal financial openness variable only—none of the other measures of globalization test significant. Regression 1 displays the distinctive signs of a null effect—β3 is positive and signifi-

256

COMPARATIVE POLITICAL STUDIES / April 2001

cant, whereas β5 and β3 + β5 are insignificant. We first consider the globalization effect. It has two dimensions. First, an increase in nominal financial openness is positively related with a centralizing impact on states (β3 is positive). This is a result in line with the mainstream economic literature on economies of agglomeration. Second, this centralizing effect seems absent in European countries (“seems” because although negative, β5 in Regression 1 is significant only at the 13% level). As suggested above, a possible cause might be the decentralizing effect of the structural funds policy. We examine this hypothesis in Regression 2. Rather than using the dichotomous “EU12” dummy (EU = 1, non-EU = 0), we try a semi-dummy variable that takes into account the varying importance of the structural funds for each EU member. This second variable takes, for non-EU countries, a zero value and, for EU countries, a positive, continuous value representing the per capita receipt of EC structural funds in 1990 (the value varies between ECU10 for the Netherlands and ECU209 for Ireland). Unsurprisingly, the results are similar to, although sharper than, those in the previous regression: β3 is positive and significant, and although we cannot calculate a fixed value for the globalization coefficient in the case of EU countries because that coefficient (β3 + β5 *Struct Funds) varies with the structural funds variable, β5 is negative and significant, suggesting that structural funds either cancel or reverse the overall effect of globalization. This is a case of globalization-minus. But the fact that the EU variable is now a real variable with respect to European countries allows us to say more about why European countries differ from other OECD countries. The coefficient for the financial globalization variable can be rewritten as (0.0256 – [0.00042 * Structural]).

From this, it is clear that the impact of financial globalization is positive for low values of structural funds, whereas it turns negative for sufficiently high values. The threshold, which is equal to 60.95 (=0.0256/0.00042), is crossed by four countries—Spain, Portugal, Greece, and Ireland. Put simply, an increase in financial openness has a centralizing impact on states except in European countries receiving substantial structural funds, where this effect is offset. The globalization-minus effect exists in concurrence with a positive sui generis effect—β4 is significantly greater than zero. This suggests that at zero or low values of the financial variable, the structural funds policy has a centralizing effect on EU countries. Indeed, the effect of structural funds on centralization can be rewritten as (0.0028 – [0.00042 * ∆Nom fin op]),

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

257

Figure 1. Structural funds and financial openness.

suggesting that structural funds have a centralizing impact for increases in nominal financial openness inferior to 6.67 (=0.0028/0.00042) but a decentralizing impact for values beyond it. Only two countries fall in the latter category—Spain and Portugal.19 These two seemingly contrary results—one says that a high value of structural funds cushions the otherwise centralizing impact of globalization, whereas the other says that a high value of globalization cushions the otherwise centralizing impact of structural funds—have one claim in common: Only when values of globalization and structural funds are high simultaneously do we observe decentralization. Both values are high in the case of four countries—Spain, Portugal, Greece, and Ireland. In fact, as shown in Figure 1, there is a general correlation between an increase in nominal financial openness and the importance of structural funds. Fitting a line between the log value of the structural funds variable and the nominal financial openness variable yields a t value that is significant at the 3% level. The reason is that the countries that opened last were also the most backward and thus the most eligible for structural funds. We conclude that the structural funds policy is associated with decentralization. Whereas globalization tends to be generally associated with greater centralization, the funds reduce centralization in accordance with the sums involved. This is a clear instance of market-correcting, redistributive policy. It testifies to the relocation in Brussels of a centralized decision-making pro19. The sui generis effect disappears if one substitutes the measure of European trade dependence for the structural funds variable, suggesting a political rather than market origin.

258

COMPARATIVE POLITICAL STUDIES / April 2001

cess in which governments no longer are the main channels of interest representation, as testified by their willingness to accept less centralization. Note, however, that the policy has little or no decentralizing impact on the policy-making process of the core countries’ institutions. It is only at very high levels of geographic redistribution, possible only for small and/or comparatively backward countries, that the structural funds policy seems to modify the relative distribution of resources between the central government and the local governments.

CONCLUSION There is a consensus among Europeanists that the integrative spurt of the last 20 years has been mostly due to the market and only secondarily, if at all, to political will. The first goal of this article was to propose a research design to help us separate the respective manifestations of the two mechanisms. To that effect, we hypothesized that a market-driven integration would have observable decentralizing effects, making Europeanization synonymous with globalization, whereas a voluntarist process would have centralizing effects, distinguishing Europeanization from globalization. We set up a research design allowing us to differentiate between the two hypotheses while holding constant the additive and interactive effects of globalization. We argued that the market-driven hypothesis would be supported in two cases: if Europeanization adds to globalization (globalization-plus and revealed globalization) or has effects of its own attributable to market variables. In contrast, the voluntarist hypothesis would be supported in two cases: if European integration cancels or reverses the impact of globalization (globalization-minus) or if it has effects of its own attributable to nonmarket variables. Although these tests are equally valid for institutional and policy variables, in this article, we limited our empirical foray to institutional variables—the representation of organized interests in labor markets, capital markets, and the polity. Equipped with these tools, we then looked for manifestations of the market and political facets of European integration. As in the rest of the literature, we found no significant instances of political voluntarism in the integration of factor markets. Our institutional hypothesis, that voluntarism, when any, works to strengthen the interventionist capability of European institutions, found no support in the labor market. Furthermore, trade unions also are not sheltered against the dilution of membership, and neither is the decentraliza-

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

259

tion of wage bargaining checked. In fact, in keeping with the faster growing financial openness of member economies, we found that corporatism was declining faster within than outside the EU. Likewise, we found no trace of state-correcting intervention in EU capital markets. The deregulation of internal capital markets proceeded at the same pace within and outside the European single market. This is a case in which European integration seems to make no difference. However, we found evidence that market-based integration has so far spared the more political arenas of interest representation—political parties and local governments. We found that globalization had a general dealigning impact on the electorate, except in Europe. Although voter turnout seems to have been negatively affected everywhere, electoral volatility, in contrast, is not declining in relation to globalization as much within than outside the EU. EU voters have so far been immune, or less vulnerable, to the global call for “exit” from partisan politics, exhibiting instead greater partisan loyalty. Last, we found evidence for political voluntarism in the centrifugal effects of the EU structural funds policy. This effect runs contrary to the consolidating effect at the national level that greater economic openness seems to generally have on state budgets in response to the rising disparity between local economies. In sum, the overall effect of globalization on interest representation is negative in the labor and capital markets; it demobilizes voters and breaks the subnational governments’ common front in relation to the political center. The effect of Europeanization on interest representation is equally demobilizing in factor markets (actually worse in the labor market) where prospects for corporatism seem remote. Absent any centralized system of interest representation and bargaining, it is unlikely that EU policy in factor markets can take a path different from that pursued by non-EU governments in response to the challenge of global market integration. In contrast, European countries are maintaining a comparatively vibrant system of voter and local government representation. Voter representation is national and intergovernmental, whereas local government representation is supranational. The existence or endurance of these two systems of interest representation is the product of EU institutions and policies. Yet, they may feed back on European integration, orienting it in an institutional and policy direction different from that pursued by non-EU countries. Indeed, by nursing or preserving a market-correcting capacity, EU countries are likely to be tempted to use it accordingly.

260

COMPARATIVE POLITICAL STUDIES / April 2001

REFERENCES Alesina, A., & Spolaore, E. (1997). On the number and size of nations. Quarterly Journal of Economics, 112, 1027-1056. Alesina, A., & Wacziarg, R. (1999). Is Europe going too far? (NBER Working Paper No. 6883). Cambridge, MA: National Bureau of Economic Research. Amin, A., & Thrift, N. (1992). Neo-Marshallian nodes in global networks. International Journal of Urban and Regional Research, 16, 571-587. Bartolini, S. (1998, October 17-18). Old and new peripheries in the process of territorial retrenchment/expansion: Europe from state formation to regional integration. Paper presented at the conference Beyond Center-Periphery or the Unbundling of Territoriality, Berkeley, CA. Bates, R., & Lien Da-Hsiang, D. (1985). A note on taxation, development and representative government. Politics & Society, 14, 53-70. Beck, N., & Katz, J. N. (1996). Nuisance vs. substance: Specifying and estimating time-series-cross-sectional models. Political Analysis, 6, 1-36. Bolton, P., Roland, G., & Spolaore, E. (1996). Economic theories of the break-up and integration of nations. European Economic Review, 40(2), 697-705. Castells, M. (1996). The information age: economy, society, and culture: Vol. 1. The rise of the network society. Oxford, UK: Basil Blackwell. Cerasi, V. (1996, July). An empirical analysis of banking concentration (Working Paper No. 96-04). Milan: Universita degli Studi di Milano, Dipartimento di Economia Politice e Aziendale. Cerasi, V., Chizzolini, B., & Ivaldi, M. (1998). Sunk costs and competitiveness of European banks after deregulation (Working paper). London: London School of Economics Financial Markets Group. Chase, K. A. (1998, September 3-6). Capital flows, domestic coalitions, and the politics of trade policy. Paper presented at the annual meeting of the American Political Science Association, Boston. Commission of the European Communities. (1994). Die Durchführung der Strukturfondsreform 1992. Vierter Jahresbericht. [The implementation of the reform of the structural funds. Fourth annual report]. Luxembourg: Office for Official Publications of the European Communities. Crouch, C. (1993). Ambiguities of decentralization. In International Industrial Relations Association (Ed.), Economic and political change in Europe: Implications on industrial relations (pp. 881-894). Bari: Cacucci Editore. Deeg, R. (1998). Finance capitalism unveiled: Banks and the German political economy. Ann Arbor: University of Michigan Press. Eijk, C. van der, Franklin, M., & Marsh, M. (1996). What voters teach us about Europe-wide elections: what Europe-wide elections teach us about voters. Electoral Studies, 15, 149-166. Ferner, A., & Hyman, R. (1992). Introduction. In A. Ferner & R. Hyman (Eds.), Industrial relations in the New Europe (pp. xvi-xlix). Oxford, UK: Basil Blackwell. Franklin, M., Marsh, M., & McLaren, L. (1994). Uncorking the bottle: Popular opposition to European unification in the wake of Maastricht. Journal of Common Market Studies, 32, 455-472. Friedrich, R. J. (1982). In defense of multiplicative terms in multiple regression equations. American Journal of Political Science, 26(4), 797-833.

Verdier, Breen / EUROPEANIZATION AND GLOBALIZATION

261

Fujita, M., Krugman, P., & Venables, A. J. (1999). The spatial economy: Cities, regions and international trade. Cambridge, MA: MIT Press. Garrett, G. (1995). Capital mobility, trade, and the domestic politics of economic policy. International Organization, 49(4), 657-687. Garrett, G. (1998). Partisan politics in the global economy. Cambridge, UK: Cambridge University Press. Golden, M., Lange, P., & Wallerstein, M. (1997). Union centralization among advanced industrial societies: An empirical study. Dataset available at http://shelley.sscnet.ucla.edu/data. Grahl, J., & Teague, P. (1989). The cost of neo-liberal Europe. New Left Review, 174, 33-50. Krugman, P. (1991). Increasing returns and economic geography. Journal of Political Economy, 99, 483-499. Kurzer, P. (1993). Business and banking: Political change and economic integration in Western Europe. Ithaca, NY: Cornell University Press. Lange, P. (1992). The politics of the social dimension. In A. N. Sbragia (Ed.), Euro-politics: institutions and policymaking in the “new” European community (pp. 225-256). Washington, DC: Brookings Institution. Leonard, D., & Natkiel, R. (1986). World atlas of elections. London: Economist Publications. Mackie, T., & Rose, R. (1991). The international almanac of electoral history. New York: Macmillan. Mackie, T., & Rose, R. (1997). A decade of election results: Updating the international almanac. Glasgow: University of Strathclyde, Centre for the Study of Public Policy. Majone, G. (1996). The European community as a regulatory state. Collected Courses of the Academy of European Law, 5, 321-419. Marks, G., Nielsen, F., Ray, L., & Salk, J. (1997). Competencies, cracks and conflicts: regional mobilization in the European Union. In G. Marks, F. W. Scharpf, P. C. Schmitter, & W. Streeck (Eds.), Governance in the European Union (pp. 40-63). London: Sage Ltd. Porter, M. E. (1990). The competitive advantage of nations. New York: Macmillan. Porter, T. (1993). States, markets and regimes in global finance. New York: St. Martin’s. Quinn, D. P. (1997). The correlates of change in international financial regulation. American Political Science Review, 91(3), 531-551. Quinn, D. P., & Toyoda, A. M. (1997). Measuring international financial regulation. Georgetown University. Unpublished manuscript, Georgetown University. Reif , K., & Schmitt, H. (1980). Nine second-order national elections: A conceptual framework for the analysis of European election results. European Journal of Political Research, 8, 3-44. Rodrik, D. (1997). Has globalization gone too far? Washington, DC: Institute for International Economics. Sabel, C. F. (1989). Flexible specialisation and the re-emergence of regional economies. In P. Hirst & J. Zeitlin (Eds.), Reversing industrial decline? Industrial structure and policy in Britain and her competitors (pp. 17-70). Oxford, UK: Berg. Saxenian, A. (1994). Regional advantage: Culture and competition in Silicon Valley and Route 128. Cambridge, MA: Harvard University Press. Scharpf, F. W. (1996). Negative and positive integration in the political economy of European welfare states. In G. Marks, F. W. Scharpf, P. C. Schmitter, & W. Streeck (Eds.), Governance in the European Union (pp. 15-39). London: Sage Ltd. Shepherd, W. F. (1994). International financial integration. Aldershot, UK: Ashgate. Smyrl, M. (1998). When (and how) do the commission’s preferences matter? Journal of Common Market Studies, 36, 79-99.

262

COMPARATIVE POLITICAL STUDIES / April 2001

Steinmo, S. (1993). Democracy and taxation. New Haven: Yale University Press. Storper, M. (1995). The resurgence of regional economies, ten years later: The region as nexus of untraded interdependencies. European Urban and Regional Studies, 2, 191-221. Strange, S. (1986). Casino capitalism. Oxford: Basil Blackwell. Streeck, W. (1987). The uncertainties of management in the management of uncertainties: employees, labor relations and industrial adjustment in the 1980s. Work, Employment and Society, 1, 281-308. Streeck, W., & Schmitter, P. C. (1991). From national corporatism to transnational pluralism: Organized interests in the Single European Market. Politics & Society, 19, 133-164. Swank, D. (1998). 18 nation pooled time-series database for the advanced capitalist democracies, 1955-1995. Milwaukee, WI: Marquette University. Thompson, J. K. (1995). Securitisation: An international perspective. Paris: Organization for Economic Co-operation and Development. Thurow, L. C. (1996). The future of capitalism: How today’s economic forces shape tomorrow’s world. New York: William Morrow. Tödtling, F. (1994). The uneven landscape of innovation poles: Local embeddedness and global networks. In A. Amin & N. Thrift (Eds.), Globalization, institutions, and regional development in Europe (pp. 68-90). Oxford, UK: Oxford University Press. Tsoukalis, L. (1993). The new European economy. The politics and economics of integration. Oxford, UK: Oxford University Press. Turner, L. (1996). The Europeanization of labor: Structure before action. European Journal of Industrial Relations, 2, 325-344. Underhill, Geoffrey R. D. (Ed.). (1997). The new world order in international finance. New York: St. Martin’s. United Nations. (1995). World investment report 1995. New York: Author. Verdier, D. (1995). The politics of public aid to private industry: The role of policy networks. Comparative Political Studies, 28, 3-42. Verdier, D. (2000a). State and finance in the OECD: Previous trends and current change. Politics & Society, 28, 35-65. Verdier, D. (2000b). The rise and fall of state banking in OECD countries. Comparative Political Studies, 33, 283-318. Woldendorp, J., Keman, H., & Gudge, I. (Eds.). (1993). Political data 1945-1990. Party government in 20 democracies. European Journal of Political Research [Special issue], 24, 1-119. Woldendorp, J., Keman, H., & Gudge, I. (Eds.). (1998). Party government in 20 democracies: an update (1990-1995). European Journal of Political Research, 33, 125-164.

Daniel Verdier is an associate professor in the Department of Social and Political Sciences at the European University Institute, Florence. He is working on a book on the politics of banking and finance in comparative and historical perspective. Richard Breen is an associate professor in the Department of Social and Political Sciences at the European University Institute, Florence, and professor of sociology at Nuffield College, Oxford University. His current research project is on national patterns of social mobility.