Previous |Next |Title |Contents
In this section, we analyze our data on preliminary references, and evaluate the approaches reviewed in the previous section. We begin by examining the aggregate picture, focusing on time-series trends and crossnational variance. We then break the data down by subject matter (domain of EC law) and court of referral.
Figure 1 plots annual levels of art. 177 activity for the 1961-94 period, with no crossnational element. The increase in preliminary references after the ECJ's announcements of supremacy and direct effect (consolidated in judgements rendered between 1963 and 1974), provide solid support for the doctrinal emphasis on the importance of those doctrines, and at least indirectly support the dominant "judicial empowerment" thesis elaborated best by Weiler and Burley-Mattli. It would be all but impossible to explain the steady and dramatic expansion of art. 177 activity without reference to the Court's innovative case law. It would be as difficult to argue that these doctrines did not fundamentally alter, in a pro-integrative direction: (1) the strategic context in which national judges operate; and, (2) how judges are expected to perform their duties.
Figure 2 reports the average number of references per year for the EC-12 since 1961 (or a subsequent date of entry into the Community for new members). The figure graphically depicts what has been the most important puzzle confronting scholars working in this area, namely: what accounts for the wide variance in preliminary references? At this level of analysis at least, it is obvious that distinctions between monist and dualist systems, and between those systems which permit or forbid judicial review have no systematic effect that is measurable by our data.
The results of tests of our own theory have gone a long way towards solving the puzzle (for an extended analysis, see Stone Sweet and Brunell, forthcoming 1998). We gathered data on a range of variables plausibly associated with levels of references, including population, GNP, intra-EC trade, and the "diffuse support" for the ECJ and EC law (see Caldeira and Gibson, 1995), and regressed this data (the independent variable) on art. 177 activity (the dependent variable). Given our theoretical priorities, we expected to find a high correlation between levels of transnational exchange and references. Unfortunately, systematic data on a number of factors that are associated with transnational activity, such as the growth of transnational interest groups, investment, and so on, do not exist or are incomplete. We therefore used intra-EC trade as a proxy for transnational exchange. The choice is defensible given the primary importance of trade and the common market to the European integration as a whole; further, the EEC treaty required the abolition of national measures restricting trade as of December 31, 1969, and traders, by virtue of the doctrine of the direct effect of Treaty provisions (initiated in ECJ 1963), could then ask national judges to set aside such measures.
Figure 3 depicts the linear relationship between average annual levels of intra-EC trade and average annual levels of art. 177 activity across the member states. This relationship is nearly perfect (the adjusted R2 is .92), indicating that almost all of the crossnational variance in preliminary references is accounted for by our independent variable. As the notes attached to the figure show, this model widely outperforms rivals.
We then analyzed the relationship between annual levels of intra-EC trade and preliminary references for the EC as a whole, with no crossnational element. A dummy variable is included in this regression model to account for the linked effects of the constitutionalization of the treaty system and the entry into force of common market provisions of the EEC Treaty. We coded the dummy variable as "0" for the years 1961-69, and "1" thereafter. Figure 4 presents our results. The Predicted line - generated by a regression model in which the dependent variable is the number of preliminary references per year, and the independent variables are intra-EC trade and the post-1969 dummy - plots the annual number of references predicted by our two independent variables; the Actual line plots the actual number of references registered per year. The connections between our independent and dependent variable remain enormously powerful; the adjusted R2 is .91, indicating that the model explains more than 90% of the variance in annual art. 177 activity.
We cross-checked these results in a series of "pooled," regression models. In this type of analysis, crossnational data is combined with time-series data to increase the number of observations being examined, providing a more rigorous test of hypothesized relationships among variables (see Beck and Katz, 1995). What is being examined in our pooled model is the causal relationship between intra-EC trade and references, for each member state, for each year. Table 1 summarizes the results. The correlation between our variables remains remarkably robust, and both the trade variable and the post-1969 dummy remain highly statistically significant.
Analysis of the data provides overwhelming support for our proposition that transnational activity has driven art. 177 activity and, therefore, the construction of the EC legal system. We also theorized that tight causal linkages between three variables - transnational exchange, European rules, and transnational dispute resolution - would develop as European integration proceeds, and would sustain integration over time. A series of propositions derived from this claim have been tested, quantitatively and in case studies of litigation in specific policy sectors, elsewhere (Stone Sweet and Brunell, forthcoming 1998; Stone Sweet and Caporaso, forthcoming). Here we present only our analysis of the statistical relationship between annual levels of intra-EC trade (the proxy for transnational activity), European secondary legislation adopted (the proxy for European rules), and art. 177 activity (the proxy for transnational dispute resolution). Figure 5 plots the actual and predicted levels of art. 177 references. The Predicted line - generated by a regression model in which the dependent variable is the number of preliminary references per year, and the independent variables are intra-EC trade, EC secondary legislation, and the post-1969 dummy - plots the annual number of references predicted by our two independent variables; the Actual line plots the actual number of references registered throughout the EC per year. The adjusted R2 is .92, and the coefficients for all three variables are positive and significant.
The data can also help us to evaluate Golub's critique of the Weiler synthesis. Recall that Golub, focusing on the experience of the UK, derived three propositions from the "traditional model," and found an absence of empirical basis for all three. Figure 6 plots the number of references issuing from British courts since 1973, and table 2 reports the average number of references per year for the EC-12 over four time periods: 1961-73; 1974-80; 1981-87; and 1988-94. Although as formulated, Golub's first statement is not testable, the data appear to support Golub's (p. 365) second statement, that "the frequency of references from Britain followed a learning curve, increasing steadily over time." The average number of references generated by UK courts has doubled in each period, while the number of Italian references fell during the 1981-87 period, and French, Greek, Irish, and Luxemburg references declined in the most recent period.
Given the numbers, Golub's central claim - that British judges "are loathe to make referrals and unwilling to co-operate with the ECJ in promoting European integration" (p. 368) - is unconvincing. In the next section, we show that it cannot be sustained.
To this point, we have examined patterns of art. 177 activity revealed when we look at the data organized temporally and crossnationally. A different set of patterns appear when we attend to two other pieces of information: the subject matter of the referral, and the court (or court system) generating the reference.
National judges interact with the European Court within specific domains of EC law. The density of any sector of EC law, and hence that sector's importance relative to other domains, is partly indicated by the quantity of legal rules that comprises it. At least as important is the relative intensity with which actors, both public and private, cooperate and dispute under these rules. Disaggregating art. 177 activity by legal domain provides a good measure of this intensity.
Our data set contains nearly 3,000 preliminary references, and these references invoke EC rules slightly less than 4,000 times (some references ask for interpretations of legal provisions that fall within two or more areas of EC law). Table 3 charts the evolution of the substance of art. 177 activity over six time periods. Most striking is the extraordinary expansion of EC law into new areas. In the 1971-75 period, more than half of all references concerned just two sectors: the free movement of goods and agriculture. In the 1991-95 period, these two domains accounted for only 27% of total references, while new areas, such as environmental protection, commercial policy, and competition, are becoming increasingly important.
Table 4 presents the data on the subject matter of preliminary references organized by member state. The table depicts crossnational patterns of litigation in a manner that is particularly useful for purposes of comparison. To compile the table, we separated art. 177 references into categories corresponding to subject matter, assigning to each member state the number of instances within each category that would have resulted had there been no difference between percentages of rates of litigation overall and rates of litigation in specific legal areas. Thus, if a member state accounted for 12% of total references, we assigned, as its proportional share, 12% of the references in each legal domain to that member state. We then subtracted, for each category and for each member-state, the proportional number of subjects from the actual number of subjects referenced. The percentage difference highlighted in bold is the difference between the actual and predicted level of references for each country standardized, by percentage, in terms of the total number of cases in each column. The process may seem complicated, but these calculations make comparing art. 177 across member-states and across legal domain simple and instantaneous. Cell entries in bold consist of positive and negative percentages: a high positive value indicates that litigants are attacking the rules of a particular country in a particular legal domain more frequently relative to other countries and other areas; a negative value indicates that a country is not being dragged to the ECJ as often as we might expect based on overall litigation rates relative to other member-states and policy areas.
Among other things, the data reveal the flaws in Golub's thesis. UK judges are not "loathe" to cooperate with the ECJ compared with judges in other member states. UK judges refer more references in some areas than some other national judges, and fewer references in some areas than some other national judges. Golub supports his arguments with reference to selected cases of litigation in environmental protection. But, in fact, UK judges have referred more cases in this sector than have judges in all but France, Italy, and the Netherlands.
Because Golub has wrongly identified the puzzle, proposed solutions miss the mark. UK litigation in the area of gender equality is a dramatic example. Golub suggests that judges take their cues from national governments, noting that Conservative governments have blocked moves to deepen integration. During the 1980s and beyond, Tory governments constituted the crucial veto point in the deliberations of the Council of Ministers on proposed directives on gender equality. In that area, which is coded by the ECJ as social provisions (art. 119 EEC, and directives on sex equality and non-discrimination), UK judges refer far more cases per year to the ECJ than do judges in any other member state. Indeed, the percentage difference figure in this cell is the highest registered in any legal domain for any member-state; in this area UK judges use their discretion to refer to the ECJ not to support, but to undermine, the government's position. What happened is well known. The UK withheld its approval from any provision that would have enshrined, as EC law, rights and remedies not already available under UK law. But judicial cooperation between the ECJ and national judges served to upgrade these rules, forcing the UK parliament to revise its domestic law on a number of important occasions (see Ellis, 1991; Kenney, 1992, 1996; and Stone Sweet and Caporaso, forthcoming).
Are UK judges more willing to make referrals in some areas of EC law than in others and, if so, why? What accounts for the wide crossnational variation in references in any given area of EC law? Once formed as a set hypotheses, proposed answers to these questions would naturally generate a research agenda of great importance. For example, taking a cue from political economy perspectives, one might propose that variance in references is, to some significant extent, explained by variance in how potential litigants in that area are organized, and the level of resources that they command. EC rules on equal pay for equal work and non-discrimination, for example, lay down certain minimal standards for treatment in the workplace; but these rules are more or less advantageous to any given female worker, relative to national rules and practices, depending on where she lives and works. Does the fact that UK litigation is generating the lion's share of art. 177 activity in this area mean that women are subject to greater, and worse forms of, job-related discrimination in the UK than are women in every other member state? We doubt it. It must make a difference that women workers in the UK are better organized and better supported institutionally (e.g., by trade unions and the Equal Opportunity Commission) than are women elsewhere. If this and related propositions are generally borne out by comparative studies of litigation, theories of integration that focus on the crucial role that litigants have played would be much improved. At the very least, such studies would show us if and how we ought to attend to variation in the configuration of state structures, and the organization of private interests. Other propositions concerning, for example, crossnational variance in access to justice, in how specific lines of case law have developed in specific areas of the law, in legal cultures, and so on, also immediately come to mind.
We end by examining which courts have made referrals, and in which areas of EC law. As we noted above, in our coding system we distinguished between three levels in national judicial hierarchies: lower (L) , intermediate (I), and high (H) courts. Figure 7 depicts the number of references generated by each level, by decade, with no crossnational component. Lower courts have produced a minority of references, and in recent decades have collectively made fewer references than have intermediate courts. Figure 8 organizes this data crossnationally, with no time element. In one member state - Luxembourg - a single high court (the Conseil d'état) dominates art. 177 activity; in seven member states - Denmark, France, Ireland, Italy, the Netherlands, Spain, and the UK - intermediate courts are the most active; and in Belgium, Germany, Greece, and Portugal, lower courts refer the highest proportion of references.
This data bears directly on Alter's line of research, but only indirectly on her thesis. The claim that the lower courts have been "the motors of EC legal integration" does not necessarily imply that lower courts have made more references. Alter argues that the lower courts went to the ECJ to subvert the control of appellate courts above them, facilitating the penetration of EC law into national legal orders in the process; and the higher courts began associating with the ECJ only when EC law had become so great a part of national law that they had little choice (Alter, 1996, pp. 466-71).
We are generally skeptical of Alter's arguments, for reasons that deserve brief elaboration. First, over the entire life of the Community, appellate courts have been more active than lower courts in referring questions to the European Court. If we consider the fact that there are many more lower than appellate courts, and that lower courts process the vast bulk of national litigation, this discrepancy is all the more striking. Because a core function of appellate judging is to resolve disputes involving legal interpretation and conflict of law, we would expect the appellate courts to be far more involved in the construction of the legal system than Alter imagines them to be.
Second, Alter's failure to discuss either the substance of the law or the outcomes of judicial rulemaking, in even a selective way, raises alarms. National courts may at times wish to protect their own solutions to legal problems, and they may communicate this wish to the ECJ in their referrals, or by refusing to make references; the ECJ, anxious to elicit the support of the national courts, and especially that of higher courts, may seek to accommodate these wishes. This is the essence of the intra-judicial dialogue that Weiler and Burley-Mattli argue has pushed and sustained legal integration. The crucial empirical questions are these: how often, where, in what areas of the law, for how long, and why did the intra-judicial dialogue fail to achieve an accommodation, blocking the effective enforcement of EC law before national courts? We cannot begin to answer them without intensive examination of the impact of supremacy (and of preliminary rulings) on specific lines of case law.
The evidence that Alter does provide often undermines her own thesis. For example, in what is virtually the only passage in the article that addresses the impact of the European legal system on specific policy outcomes or on specific areas of the law, Alter writes:
Because of the actions of lower courts, EC law expanded into new areas and came to influence national law. ... The influence of EC law spread to areas never envisaged by national politicians, such as the provision of educational grants to non-nationals, the provision of equal pay to men and women, industrial relations and the advertisement of British abortion services in Ireland (p. 468).
Did the leading decisions in these (and other) areas come pursuant to lower court references, as the passage implies? The Court's decision on abortion advertising responded to questions raised by an Irish court of appeal. The ECJ's decision announcing the direct effect of art. 119 EEC, Defrenne II (ECJ, 1976), was provoked by a reference from an intermediate, not a lower, court in Belgium. These references did not result from the fact that the relevant EC rules had been sufficiently aborbed into national law. The importance of Defrenne II was to create the conditions necessary for the steady penetration of art. 119 into the national legal order; and many of the Court's other landmark decisions in the area, including Dekker (ECJ, 1991a), Hertz (ECJ, 1991b), and Marshall II (ECJ, 1993) came pursuant to references from high courts, and the Barber controversy (ECJ, 1990) was referred by an intermediate court. Although more research is clearly warranted, it seems that Alter exaggerates the importance of lower courts, and makes too much of alleged differences between her "competition dynamic" and the judicial empowerment/intra-judicial dialogue dynamic of Weiler and Burley-Mattli.
We have proposed that variation in the intensity of the relationship between the ECJ and any given national court system is highly correlated, other things being equal, with the extent to which EC law is material to disputes that come before that court system. The more any given court deals with questions of EC law, the more that court is likely to interact with the ECJ in a constructive manner. Doctrinal support is easily found: as a number of contributors to the Slaughter, Stone Sweet, and Weiler volume (1997) document, Continental civil courts - including supreme courts (Cassation) - accepted supremacy early on, easily, and without complex, while administrative and constitutional judges exhibited more resistance. We would like to be able to test our proposition more fully by examining the relationship between art. 177 activity and the amount of litigation in which EC law is invoked by at least one of the parties, within any set of court systems over time. Unfortunately, even incomplete and sample data on the relative numbers of litigation involving EC law pleadings have not been compiled.
Finally, table 5 maps art. 177 activity with reference to national courts of referral, and subject matter. To save space and enhance relevance, we excluded those courts or court systems that had not generated at least three references and 5% of the total references in that member-state.
 Our time series data for intra-EC trade and art. 177 references are non-stationary, a common problem for data containing strong trends, requiring special treatment if regression methods are to be employed. We omit how we handled this problem here, but see Stone Sweet and Brunell (forthcoming 1998) for a full discussion.
 Lisa Conant, a Ph.D. candidate at the University of Washington (Seattle), is developing just such a framework.
 Alter cites two ECJ preliminary rulings here, one on the Irish abortion case, the other a German free movement of worker's case.
Previous |Next |Title |Contents
Top of the page