The Economy, Accountability, and Public Support for the President of the European Commission

European Union Politics [1465-1165(200006)1:2] Volume 1 (2): 173-200: 012600 Copyright 2000 SAGE Publications London, Thousand Oaks CA, New Delhi The Economy, Accountability, and Public Support for the President of the European Commission Ronald D. Gelleny Binghamton University (SUNY), USA Christopher J. Anderson Binghamton University (SUNY), USA ABSTRACT Using public approval for the President of the European Commission as an example, we investigate whether there are economic effects on public support for political authority: (1) when democratic accountability is absent; and (2) when public information about authorities is low. We find that economic effects as traditionally defined are weak, but that people make utilitarian assessments even when authorities cannot be held directly accountable. Moreover, the analyses suggest that voters with low levels of information, interest or awareness behave in ways that cannot be empirically distinguished from the behavior of well-informed voters. The findings imply that citizens employ utilitarian considerations when assessing political authorities even when they cannot hold them accountable or when they have little information about them. KEY WORDS accountability economic voting European Commission presidential approval public opinion 173

174 European Union Politics 1(2) 1 Introduction Recent years have seen a resurgent interest in the determinants of Europeans perceptions about the operation of and benefits derived from the European Union (EU) (Eichenberg and Dalton, 1998; Gabel, 1998a). In all likelihood, this is at least in part due to the greater role that European mass publics have started to play in the integration process. Thus, public opinion has captured the attention both of scholars investigating European institutions and policy processes as well as those interested in political behavior. While the former view mass publics as an important part of the development and functioning of EU integration and institutional evolution (e.g. Carrubba, 1997; Hooghe and Marks, 1999; Schneider and Weitsman, 1996), the latter see the EU and mass political behavior in the EU as a fascinating laboratory for testing theories that have been developed in the context of stable nation states (see e.g. Van der Eijk and Franklin, 1996). Most of the literature on public opinion about the EU that has developed as part of this research agenda has focused on support for the integration process (see e.g. Gabel, 1998a). Centering around utilitarian models of attitudes about integration, a number of studies have suggested that Europeans are more supportive of the EU and the integration process if there are positive payoffs to be derived either for their nation state or for them personally (Anderson, 1995a; Eichenberg and Dalton, 1993; Gabel, 1998a). This article seeks to contribute to this literature by examining the effects of utilitarian considerations on support for the President of the European Commission, an area that previously has not been examined by researchers. Specifically, we question whether citizens should have such utilitarian considerations in the first place, especially when they are ill informed about the Commission and its President, and when the President and the Commission cannot be held directly accountable. As such, the article is also intended to contribute to the more general debate in comparative politics on utilitarian or economic explanations of public support for political authorities (cf. Anderson, forthcoming). This article reports on a case study of public approval for European Commission President Jacques Delors. Using data on the popularity of Delors during his last year in office (1994) collected in the (then) 12 member states of the EU, we find that Europeans are simultaneously utilitarian and illinformed when it comes to judging the performance of the President of the Commission. While we find that roughly half of all Europeans do not express an opinion on the performance of the Commission President, our analysis also shows that utilitarian considerations powerfully affect whether Europeans approve of the job the President of the Commission has done. The next section reviews the literature on economic and utilitarian effects

Gelleny and Anderson Support for the President of the EU 175 on public support for European integration and institutions. We argue that there are two ways to view people s attitudes toward the President of the Commission utilitarian and uninformed and that these may affect the formation of attitudes toward the President of the Commission in different ways. We then describe the data and discuss people s perceptions of the President of the Commission. Next, we develop a model of public support for the President that we estimate with the help of a series of multinomial logistic regressions. We conclude by discussing the results and suggesting avenues for future research. 2 Utilitarian Models of Public Opinion toward Europe Much of the research on public support for European integration has sought to show that mass opinion is driven by macro-economic conditions and people s cost-benefit calculi (Anderson, 1995a; Eichenberg and Dalton, 1993; Gabel, 1998a). This line of inquiry, at least in part, grew out of the economic voting research, which has shown that macro-economic conditions are systematically related to government popularity and electoral support (cf. Anderson, 1995b; Lewis-Beck, 1988). It is also grounded in theories about gains to be captured from free trade among the European states (Gabel, 1998a). Given that European politicians have extolled the economic benefits of integration for the member states, one stream of this research contends that citizens will weigh the costs and benefits of integration for their domestic economy. Consistent with this argument, a number of empirical studies have found that macro-economic indicators and people s economic perceptions are systematically associated with cross-national differences in support for an integrated Europe (see e.g. Anderson and Kaltenthaler, 1996; Eichenberg and Dalton, 1993; Gabel and Whitten, 1997; Smith and Wanke, 1993). While such quasi-economic voting models of support are difficult to attack on the basis of their empirical validity, their underlying micro-logic has seldom been explicitly argued or examined. We suggest that traditional economic voting models (or what we prefer to classify as utilitarian models more generally) may not apply ipso facto to the European Union, or may apply only with some modification. Specifically, two frequently implicit assumptions underlying the economic voting literature may not hold in the context of the European Union. First, researchers have long assumed that economic voting takes place because citizens seek to hold their governments responsible for the management of the economy (Key, 1961). However, citizens may not find it useful to judge political actors such as the European Commission on the basis of economic or any other policy outcomes when these actors cannot be

176 European Union Politics 1(2) held directly accountable through democratic elections. Second, economic voting analyses assume that citizens can gauge the effects of government actions on policy outcomes and therefore are able and willing to draw a connection between what the government does and economic conditions that exist in a country. Because Europeans frequently are ill informed about the European Union (cf. Anderson, 1998), and because the Commission and its President may have only a small impact on the European economy, citizens may not know or believe that economic outcomes are affected by European Union policies. If they do not link economic outcomes and evaluations of the Commission President, then economic voting should not exist. 3 Utilitarian Considerations and Public Support for the President of the European Commission To examine the validity of utilitarian approval models, we examine support for the President of the European Commission. In contrast to the extensive research on approval of the US President and other heads of government, no research to date has examined the popularity of the most visible representative of the European Union. This is surprising, given that public approval or the absence of disapproval are important political resources for Commission Presidents to conduct their job effectively. The recent debacle surrounding the resignation of the Commission under the unpopular President Santer is a case in point. With the increased transparency expected to accompany recent reforms of the Commission contained in the Amsterdam Treaty of June 1997, the Commission President s behavior will increasingly attract media and political attention, and Presidents will need public support in order to weather greater scrutiny and maintain authority vis-à-vis national delegates in the Council of Ministers. Two competing visions of the relationship between the Commission and the European public seem particularly plausible to us. The first focuses on the importance of the European Commission in the political economy of the EU. This perspective realizes that the Commission s actions can significantly affect economic conditions in the member states in a variety of ways. According to this view, people should evaluate the President of the Commission on the basis of economic outcomes. Because the President of the Commission is not elected by the people of Europe and thus cannot be held directly accountable by them, we refer to this model as a quasi-economic voting model. 1 The second perspective argues that Europeans are notoriously uninformed about the EU in general and the Commission in particular (Anderson, 1998; Janssen, 1991). According to this view, it may be rational for

Gelleny and Anderson Support for the President of the EU 177 Europeans not to pay too much attention to the Commission, especially because the Commission is not directly accountable to the people of Europe. Under conditions of low accountability and low information, it would not be particularly sensible for citizens to evaluate the Commission and its President on the basis of economic outcomes. 3.1 The Case for Quasi-Economic Voting: The Commission and the European Economy In the ongoing study of the European Union, the importance of the European Commission has often been overlooked. Part of the lack of attention paid to the Commission has been the theoretical preoccupation with the view that member states are the principal engines driving and controlling EU integration. This intergovernmentalist perspective on the integration process views the subordinate Commission as simply carrying out the wishes of the member states (Hoffmann, 1994; Keohane and Hoffmann, 1991; Moravcsik, 1991). However, a number of scholars have recently started to argue that the Commission is an important political actor that advances its own political goals through influencing the policy preferences of the membership (Christiansen, 1996; Cini, 1996; Marks et al., 1996; Sandholtz, 1996; Sandholtz and Zysman, 1989; Smyrl, 1998). As the executive branch of the European Union, the Commission has the responsibility of proposing initiatives, planning the budget as well as administering policies and regulating their compliance. In addition, the Commission acts as a mediator between member states, speaks for the organization in international trade negotiations and represents the conscience and voice of the EU at large (Cini, 1996). Thus, the formal functions of the Commission are fairly similar to those of national executives, especially those of federal states. If the European Commission and its President have established a significant and public role in the economic domain of the EU, then questions regarding economic voting and public opinion about the President become all the more relevant. Although it is difficult to assess the direct influence of the Commission on the economies of the member states in the EU, the power to initiate policies and monitor the workings of the internal market afford the Commission a central role in the EU policy process (it issues between six and seven thousand legislative instruments a year). In fact, former Commission President Jacques Delors boldly predicted that about 80 percent of social and economic legislation would be eventually generated from Brussels (Economist, 1997; see also Hitiris, 1994). The Commission s influence on the national economies of member states

178 European Union Politics 1(2) may be felt most directly in the agricultural and steel sectors, where the treaties give the Commission a direct role to set prices and manage production output. Furthermore, the role of monitoring the compliance of EU legislation and regulating competition policy can receive significant public and political attention in particular when the Commission requires state aid to be returned and/or fines are levied against firms. Moreover, the Commission can influence the economy through the budget in the form of structural funds. Lastly, the Commission may be given the highly visible task of drafting initiatives to overcome economic problems by the member states. In a variety of ways, the European Commission thus has established itself as an important actor within the EU economic arena. Although this is not to say that the Commission has preempted the role of the member states to influence European economic policy and the performance of the European economy, it also must be recognized that the Commission has acquired responsibilities that had previously rested chiefly in the domain of the member states (such as competition policy). This has limited member states abilities to do all the things they could in the past. Citizens may respond to this role of the Commission in the European economy in two ways: They may evaluate a country s macro-economic conditions and relate these to how they view the Commission and its President; or they may look more specifically at particular policy outcomes (such as the Single Market) and evaluate the Commission President based on whether they view them as a success or failure. 3.2 The Case Against Quasi-Economic Voting: Low Accountability and Low Information Given the lack of direct influence of Europe s citizens over the Commission and the President of the Commission, it is clear that they are not subject to accountability in the same sense as are national governments. In the absence of such direct electoral accountability, it is open to debate why European mass publics would have an incentive to pay much attention to the Commission and its policies effects on the European economy (see also Duch and Taylor, 1997). Moreover, Europeans are notoriously uninformed about the EU and the Commission. Anderson (1998: 572 3) reports, for example, that a significant majority roughly 65 to 70 percent typically feels uninformed about the EU and that only about a third qualified as knowledgeable about the EU. Numerous polls have also shown that citizens in advanced democracies believe that national governments have responsibility for economic outcomes. In fact, this belief has been increasing since the Second World War (Alt and

Gelleny and Anderson Support for the President of the EU 179 Crystal, 1983). Thus, in the absence of effective mechanisms of accountability and reliable information on the part of European citizens, it would be reasonable to conjecture that there should not be any significant economic effects on support for the President of the Commission. In fact, we would argue that the absence of accountability engenders lower levels of awareness. Support for the President thus should be driven by variables that are unrelated to people s information or their perceptions of economic conditions. 4 A Case Study of Jacques Delors Below, we investigate the determinants of public support for Jacques Delors, President of the European Commission from 1985 to 1995. Delors long and successful presidency has been applauded as an example of what a Commission President can accomplish. Several factors contributed to his success. First, Delors assumed the Presidency with an extensive political background in European and French domestic politics that helped him successfully operate in the EU s political environment. Moreover, Delors was able to provide strong leadership within the European Union because he made skillful use of the institutional advantages that accompany the office. These advantages were clearly exploited by Delors with the able assistance of Pascal Lamy, who earned a reputation of forcefully promoting Delors agenda. Third, the Secretariat General and the Legal Service each report to the President of the Commission, providing an administrative, political and public relations power base of enormous skill and intellectual agility (Spence, 1994: 101). Hence, the President is the person best placed to know what is going on inside the Commission. Lastly, the President s power to determine portfolio distribution within the Commission provides an opportunity to control crucial policy areas. Although each Commissioner has one vote, those with greater expertise are given added consideration. Delors used this tool very effectively, for example by claiming monetary matters for himself and dividing Commissioner Cockfield s internal market portfolio among three new Commissioners. In part as a result of Delors successful presidency and to give the president greater powers vis-à-vis the national governments, the EU in the Amsterdam Treaty sought to deliberately create a Delors-style Commission President through reforms of institutional mechanisms underlying presidential powers. Specifically, the EU has turned the Commission into more of a classic cabinet government model led by a first among equals (the President). As a result of these changes the President now has the power to veto national government nominees for individual Commissioners and is responsible for allocating and

180 European Union Politics 1(2) reshuffling the portfolios of the other Commissioners. These reforms are likely to have the implication that the President will be exposed to greater scrutiny and thus also will be in greater need of public support and legitimacy than ever before. 5 Measuring Approval of the President of the Commission Our data on public support for the Commission President come from Eurobarometer 42 (EB 42), which was conducted in the member states in November and December of 1994. The countries included in our study were Belgium, Denmark, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain and the United Kingdom. We define support for the President of the European Commission by measuring how respondents evaluate Delors performance. The question is as follows: What is your view about the job that Jacques Delors has done as President of the European Commission? Do you think he has done a very good job, a fairly good job, a fairly bad job, or a very bad job? The answer categories ranged from very good to very bad and also included a category for don t know. Taking the positive ( very good and fairly good ) and negative ( fairly bad and very bad ) responses together, Figure 1 shows the distribution of opinions regarding public approval of Jacques Delors presidency. The graph Figure 1 Public Support for the European Commission President.

Gelleny and Anderson Support for the President of the EU 181 reveals that there are significant differences in approval rates across countries, ranging from a high of almost 75 percent in Ireland and Luxembourg to a low of less than 30 percent in Italy and Spain. The average approval score across the member states is about 50 percent (47.6%), indicating that about half of all Europeans approve of the way Jacques Delors was handling his job as President of the Commission. It is noteworthy that very few Europeans disapprove of the President of the Commission on average fewer than 10 percent (7.1%), with 15.9 percent of the Danes disapproving at the high end of the distribution and 0.8 percent of the Portuguese at the low end. Thus, instead of saying they disapprove of the President of the Commission, Europeans may be saying that they do not know. In fact, Europeans on average are as likely to say they don t know how to evaluate the President (45.3%) as they are to approve (47.6%). The highest numbers of Europeans saying they don t know whether they approve or disapprove can be found in Italy, Spain and Portugal at over 60 percent, whereas the lowest numbers are found in Denmark, Ireland and Luxembourg at around 20 to 25 percent. 6 Modeling Support for the Commission President To explain approval of the Commission President, we developed and estimated a series of multinomial logit models that included three main elements: utilitarian considerations, variables accounting for differences across respondents with regard to information about the European Union, and considerations specific to Jacques Delors. First, we examined the impact of these variables on approval for the President. Second, we investigated whether there are differences in the effects of the independent variables across well and badly informed respondents given that economic voting analyses usually assume that voters are fairly sophisticated when making judgments about political authorities based on economic considerations. Whether more informed voters make more use of economic information when judging the President is therefore examined as well. We sought to estimate utilitarian considerations in two ways. First, by measuring people s perceptions of general personal and macroeconomic conditions; second, by using people s evaluations of whether the country has benefited from being part of the EU and their attitudes about the Single Market. Europeans who have positive economic assessments, believe that their country has benefited from EU membership, and who are favorably disposed toward the Single Market were expected to approve of the way Jacques Delors was handling his job. In addition, individuals of high levels of human

182 European Union Politics 1(2) capital were hypothesized to have positive attitudes toward the President (Gabel, 1998a). Second, we needed to account for the significant number of respondents who stated that they did not know when asked about whether Delors was doing a good or bad job as President. Were these responses reflective of people s feelings that they lacked sufficient information to express an opinion about Delors, or were they in fact simply negative responses? If Europeans used the don t know response category as a way to express disapproval, negative utilitarian assessments should be significantly associated with presidential disapproval. If, however, the Don t Knows simply did not know and said so, we expected significant relationships between indicators of awareness of European Union politics on the one hand and the willingness to express both a positive or negative opinion about the President on the other. Finally, we hypothesized that people s views of Delors may be affected by factors regarding his personal and political traits as well as demographic differences across respondents. We expected that the fact that he was French and a prominent Socialist Party politician before becoming President would affect people s views about him. Specifically, we hypothesized that respondents who felt close to France and who supported socialist parties were more supportive of Jacques Delors as President of the Commission. In addition, we controlled for the usual demographic variables of age and gender. 6.1 The Dependent Variable If respondents thought that Delors had done a very good or fairly good job, they were coded 2; a fairly bad or very bad job, they were coded 1, and if they said they did not know, they received a 0. Note that our dependent variable thus did not rank-order respondents answers about the President. 6.2 Utilitarian Independent Variables 6.2.1 Economic Perceptions Economic evaluations were measured with the help of both egocentric (pocketbook) and sociotropic (national) economic perceptions. Because questions about the President of the Commission asked respondents to evaluate the job Jacques Delors had done, we employed retrospective economic assessments in our multivariate models. The question wording was: Compared to 12 months ago, do you think that the general economic situation in this country now is...? and Compared to 12 months ago, the financial situation of your household now is...? The variables were coded from 1 ( much

Gelleny and Anderson Support for the President of the EU 183 worse ) to 5 ( much better ), and we expected positive relations between perceptions and presidential support. 6.2.2 Benefit of EU Membership: Collective Benefits To gauge the effects of people s specific EU-related utilitarian perceptions of whether their country has benefited from EU membership on support for the President of the Commission, we made use of the following question: Taking everything into consideration, would you say that our country has on balance benefited or not from being a member of the European Union? Respondents who indicated that their country had benefited were coded 1, and 0 otherwise. 6.2.3 The Single Market Probably the single biggest achievement of Jacques Delors tenure that significantly affected the European economy was the Single European Act. To measure people s feelings about the Single Market, the Eurobarometer asked: Personally, would you say that the Single European Market, which came about at the beginning of 1993, makes you feel very hopeful, rather hopeful, rather fearful, or very fearful? Responses range from 1 ( very fearful ) to 4 ( very hopeful ). Thus, if people associated their feelings about the Single Market with Jacques Delors performance in office, we expected to see a positive relationship between favorable attitudes about it and approval of the Commission President. 6.2.4 Human Capital (Education): Personal Benefits Using people s levels of education as an indicator of human capital, these studies examined attitudes toward integration from the perspective of which individuals are likely to have the skills and resources to take advantage of an integrated European economy (see e.g. Gabel, 1998a, b, c; Gabel and Palmer, 1995; Gabel and Whitten, 1997; see also Anderson and Reichert, 1996). Those who are likely to benefit from further integration individuals with high levels of education also should be more supportive of a President of the Commission whose single most important legacy was the completion of the internal market. Education was measured on a scale from 0 (low) to 10 (high), and we expected a positive relationship with support for Delors. 6.3 Information-Related Independent Variables To examine whether Europeans used the don t know response category as a way to express disapproval or simply did not know and said so, we also investigated the relationships between indicators of awareness and

184 European Union Politics 1(2) information as well as interest in European Union politics on the one hand and the willingness to express a positive or negative opinion about the President on the other. We hypothesized that those who were interested in EU politics and indicated higher levels of subjective knowledge as well as awareness about the EU and the Commission would be significantly less likely to answer that they did not know when asked about the Commission President. 2 This means also, however, that those who said they were interested and informed also should be more likely to have both a negative or positive opinion that is, to have an opinion in the first place. 6.3.1 Interest in EU Politics We hypothesized that people who are motivated to seek out information about the EU would also be more likely to have an opinion (either negative or positive) when asked about the Commission President. We measured interest in EU politics with the help of the following question: To what extent would you say you are interested in European politics, that is to say matters related to the European Union (European Community): a great deal, to some extent, not much or not at all? The scale ranged from 1 ( not at all interested ) to 4 ( very interested ) and we expected a negative relationship with the don t know category and a positive relationship with the two presidential approval categories ( approve or disapprove ). 6.3.2 Subjective Levels of Information about the EU Respondents were also asked to rate their own levels of information about the EU: All things considered, how well informed do you feel you are about the European Union, its policies, its institutions? The answers were coded from 1 ( not at all well ) to 4 ( very well ), and we expected a negative relationship between this variable and the don t know category, as well as a positive relationship with the two other answer categories. 6.3.3 Heard about the Commission To gauge people s specific awareness of the European Commission, we utilized a question that asked respondents: Have you recently heard or read about the European Commission in Brussels, that is the Commission of the European Union? If respondents indicated they had heard of the Commission, they were coded 1, 0 otherwise. As in the other cases, we expected a positive relationship between this variable and a willingness to give an opinion about the Commission President. 3

Gelleny and Anderson Support for the President of the EU 185 6.4 Delors-Specific Control Variables: Feelings toward France and Socialist Party Support Finally we sought to control for factors associated with Jacques Delors personal and political characteristics as well as demographic differences across the respondents that may affect people s approval of how well Jacques Delors was handling his job as Commission President. First, we hypothesized that people who felt close to France also would have more positive feelings about Delors. In addition, we expected that socialist party supporters would be more positive toward Delors. We measured feelings toward France with the help of the following question: Please tell me for each of the 12 countries of the European Union, whether you feel close to it or not. If respondents indicated that they felt close to France, they were coded 1, and 0 otherwise. Socialist party support was constructed from the following question: If there were a General Election tomorrow, which party would you vote for? If respondents said that they would vote for a socialist party (or one of the socialist parties) they received a score of 1, and 0 otherwise. 6.5 Demographic Control Variables 6.5.1 Age Following Inglehart s (1970; Inglehart and Rabier, 1978) argument that younger generations would be more cosmopolitan in their outlook, would abandon chauvinistic attitudes toward the world, and would embrace efforts to achieve international understanding and European integration, we hypothesized that age would be negatively associated with support for the Commission President, provided that the President is seen as a symbol of the EU. Age was measured as the actual age of the respondent. 6.5.2 Gender Based on research which shows that women are among the losers of the integration process because of their position in the labor market, and because the fiscal tightening that went along with the run-up to the euro has disproportionately affected women through cuts in the welfare system, we also expected a gender gap in attitudes toward the President of the Commission (Liebert, 1998). We thus hypothesized a negative relationship between gender (coded 0,1; 1 = female) and attitudes toward Jacques Delors handling of the presidency.

186 European Union Politics 1(2) 7 Results We estimated multivariate regression models of support for the President of the Commission (coded 0 = don t know ; 1 = bad job ; 2 = good job ). Because the dependent variable had three unordered, multiple choice response categories, we relied on multinomial logit estimations (cf. Whitten and Palmer, 1996). The independent variables were the utilitarian and information measures discussed above, along with the controls for direct Delors effects and demographic variables. 4 With k answer categories, multinomial logit estimates k-1 equations, with the remaining category serving as the baseline. In our case, the baseline category is 0 that is, answering don t know to the approval question. 7.1 Multinomial Regression Results In our case, the multinomial logit coefficients indicated the effects of an independent variable on increasing the likelihood that the respondent chose good job (2) or bad job (1) over don t know (0). Thus, a positive coefficient for any of the independent variables indicated that a respondent scoring high on this variable was more likely to say that the President had done a good job (Column 1, Table 1) relative to expressing no opinion, or that the respondent scoring high on this variable was more likely to disapprove of the President relative to expressing no opinion (Column 2, Table 1). The probability of choosing the don t know category is 1 minus the probability of choosing the two other answer categories that is, good job and bad job. 7.1.1 Utilitarian Variables The base category was no opinion. The results show that utilitarian considerations played a significant role in determining citizens attitudes toward Jacques Delors performance in office. Specifically, we found that those who believed that their country had benefited from EU membership were also significantly more likely to say that they approved of the Commission President; conversely, the results for the disapprove/don t know analysis showed that those who said their country had benefited were also much less likely to say they disapprove relative to those in the no opinion category when asked about the President. This means that those who had a positive attitude about the benefits of EU membership were more likely to say that they did not know than to say they disapproved. Overall, the results revealed that those who considered EU membership to have benefited their country were significantly more likely to approve and less likely to disapprove of the President. We also found that feelings about the Single Market had powerful effects

Gelleny and Anderson Support for the President of the EU 187 Table 1 Multinomial Logit Models of Approval of the President of the European Commission Independent variable ln [Approve/Don t know] ln [Disapprove/Don t know] EU country benefit.383***.607*** (high = has benefited) (.063) (.104) Single European Market.294***.503*** (high = hopeful) (.038) (.062) Sociotropic economic.098**.048 assessments (retrospective) (.030) (.050) (high = positive) Egocentric economic.035.074 assessments (retrospective) (.034) (.056) (high = positive) Information about the Commission.735***.610*** (high = heard) (.055) (.097) Interested in EU politics.377***.386*** (high = interested) (.034) (.056) Subjective information about EU.429***.182** (high = informed) (.039) (.065) Close to France.523***.435* (high = feel close) (.105) (.206) Socialist party support.203**.266* (high = yes) (.063) (.111) Human capital (education).029**.021 (high = educated) (.010) (.017) Age.012***.010*** (in years) (.002) (.003) Gender.355***.584*** (1 = female) (.052) (.090) 2LL 6522.631 % correct 67.0 Reduction in error 24.3 Chi-squared (d.f.) 3027.346 (46) N 8960 Notes: Don t know serves as the baseline category. ***p <.001; **p <.01; *p <.05; significance tests are two-tailed. France is the reference category. on approval of the Commission President. In line with the quasi-economic voting expectations, we found that positive attitudes toward the Single Market engendered approval of the EU President and significantly reduced the likelihood of disapproval. Thus, both utilitarian questions about the

188 European Union Politics 1(2) benefits of European Union membership and prospective judgments about the Single Market had significant effects on presidential approval effects that were consistent with the quasi-economic voting thesis. These results are particularly noteworthy because the questions measuring people s perceptions of national and personal economic conditions had much less significant effects. In fact, the analysis revealed no significant relationships between personal economic conditions and approval of the EU President, and the sociotropic question had only weak substantive effects on the probability of approval, and no effect on the probability of disapproval relative to having no opinion. Thus, we found no evidence of quasi-pocketbook economic voting with regard to approval of the President of the Commission, and the sociotropic effects were not particularly strong substantively. Taken together, our analyses discovered systematic evidence for the effects of utilitarian variables on presidential approval in the EU. However, those variables measuring utilitarian considerations related specifically to the EU were more successful at explaining support for the Commission President than general perceptions of economic performance. 5 7.1.2 Information and Awareness Variables The variables measuring informational and awareness differences across the European public displayed extremely consistent results. Those who said they were interested in EU politics, said they had heard of the Commission lately, and those who felt that they were aware of EU politics were significantly more likely to say they either approved or disapproved of Delors performance in office. Conversely, this indicated that those with high subjective levels of information were much less likely to say they did not know when asked about the performance of the Commission President. These results strongly suggest that higher levels of information, awareness and interest significantly increased the probability that respondents were willing to voice an opinion about Delors in the first place. That is, those who were motivated to seek out information about the EU were more willing to say what they thought about the Commission President. 7.1.3 Control Variables When we examined the results for the Delors-specific and demographic control variables, a few stood out. Those who felt close to France also were significantly more likely to approve of Delors performance; however, they were also slightly more willing to say that they disapproved. This indicated that feelings of closeness to France both increased the chances that respondents were satisfied with Delors and their willingness to voice an opinion about him. Thus, closeness bred both content and a willingness to express discontent.

Gelleny and Anderson Support for the President of the EU 189 In addition, the variable measuring socialist party support behaved as expected: socialist party supporters approved of Delors and also were less likely to disapprove of his performance. Finally, gender had a significant impact on evaluations of Delors performance. The coefficient for women was negative in both the approval and the disapproval categories, indicating that women were significantly more likely to say that they did not have an opinion when asked about the performance of the Commission President. 7.2 Marginal Effects In nonlinear models such as multinomial logit models, the reported coefficient is not equal to a marginal effect, or derivative of an expected value with respect to a variable. Thus, to get a sense of the overall independent effect of a variable, it is necessary to mix the results of two sets of estimates to obtain the marginal effects. Marginal effect estimates are computed at the overall means of the data set. The marginal effects calculated from the model estimations confirmed the story told by the coefficients. Positive EU-related utilitarian considerations EU membership benefit and attitudes toward the Single Market significantly increased the probability that Europeans approved of the President and decreased the probabilities that they either disapproved or expressed no opinion. Moreover, we found that positive national economic assessments increased support for the President and decreased the probability of no opinion to almost equal degrees. The information variables worked in two particular directions: first, more information increased the probability that citizens approved of the President; second, high levels of information and awareness decreased the probability that they did not express an opinion. Interestingly, the effects of informationrelated variables on the probability of expressing a negative opinion were negligible. 7.3 Do Awareness and Quasi-Economic Voting Interact? When asked about the President of the Commission, Europeans fall into one of two camps: either they approve of the job the President is doing or they say that they do not know. Given this bifurcation and given the strong utilitarian and informational effects on approval of the President found in the multivariate estimation models, we sought to submit the utilitarian quasieconomic voting model to an additional test. Specifically, we examined whether quasi-economic voting was stronger among citizens with different levels of information, interest and awareness.

190 European Union Politics 1(2) Table 2 Marginal Effects of Utilitarian and Information-based Variables on Approval of the President of the European Commission Dependent variable category Independent variable Approve Disapprove Don t know EU country benefit.112***.046***.066*** (high = has benefited) (.014) (.006) (.014) Single European Market.066***.035***.032*** (high = hopeful) (.009) (.004) (.008) Sociotropic economic assessments.022**.001.021** (retrospective) (high = positive) (.007) (.002) (.006) Egocentric economic.006.003.009 assessments (retrospective) (.008) (.003) (.007) (high = positive) Information about the Commission.153***.008.160*** (high = heard) (.013) (.005) (.013) Interested in EU politics.076***.008**.083*** (high = interested) (.008) (.003) (.008) Subjective information about the EU.095***.005.090*** (high = informed) (.009) (.003) (.009) Close to France.109***.005.114*** (high = feel close) (.023) (.011) (.023) Socialist party support.058***.022***.036** (high = yes) (.014) (.006) (.014) Human capital (education).006**.000.006** (high = educated) (.002) (.001) (.002) Age.002***.000.003*** (in years) (.000) (.000) (.000) Gender.064***.019***.083*** (1 = female) (.012) (.005) (.012) Notes: Marginal effects for the independent variables were calculated by varying one independent variable s value while holding all other independent variables constant at their means. ***p <.001; **p <.01; *p <.05; significance tests are two-tailed. Because utilitarian models make certain informational assumptions that citizens can identify benefits and relate these to their attitudes about the EU we hypothesized that quasi-economic voting, that is, the use of utilitarian considerations to express approval or disapproval of the President, would be stronger among those with higher levels of information. The more aware citizens were, the more likely it should be that economic considerations affected

Gelleny and Anderson Support for the President of the EU 191 their levels of presidential support because they know how the Commission affects economic conditions in the European economy. 6 A plausible alternative hypothesis to the claim that better informed Europeans would make more use of utilitarian considerations when judging the Commission President can be developed from political psychological work on the rationality of people s attitudes and the informational requirements for acting in a sophisticated fashion. Specifically, researchers have found that ill-informed citizens can mimic the behavior of more informed ones through the use of heuristics and cues in a way that is consistent with their self-interest (Lupia, 1994), such that it is virtually impossible to distinguish between sophisticated citizens and those who behave as if they were well informed (Zaller, 1992). To test whether well-informed, highly aware and interested Europeans differ from those who are less well informed about the EU, we examined the effects of the utilitarian variables on presidential approval among these two subgroups. Specifically, we categorized those individuals who said they had not heard about the Commission, that they were not interested in EU politics, and that they did not feel well informed about the EU as low information respondents and all others as high information respondents. Thus, if an individual fell into all three low information categories, he/she qualified as less informed. Incidentally, this split the sample approximately in half. We then estimated multivariate multinomial logit models of the effects of all other independent variables on presidential support separately for these two populations. To check for the robustness of the results, we then also estimated models using the full sample that included the information variable and interaction variables of information as well as the utilitarian measures. If respondents in the two informational categories responded systematically differently in terms of their use of utilitarian considerations to judge the President of the Commission, the interactive variables were expected to be statistically significant in their impact on approval. For simplicity, we only show the results of the analysis for the approval category. 7 The results, shown in Table 3, revealed that there were no differences across individuals of high and low levels of awareness, information about, and interest in EU politics. That is, the effects of the utilitarian variables on the probability of approving of the Commission President were virtually identical across the more and less informed subgroups of the population. Moreover, there were no differences in the effects of the other independent variables on the probability of approval. Unsurprisingly, the marginal effects (not shown here) bore out these conclusions as well. Finally, when we examined the effects of the interaction models based on the full sample, we reached identical conclusions. Regardless of levels of information,

192 European Union Politics 1(2) Table 3 Multinomial Logit Models of Approval of the European Commission President by Levels of Awareness Low Information High Information Full Sample ln [Approve/ ln [Approve/ ln [Approve/ Independent variable Don t know] Don t know] Don t know] EU country benefit.508***.522***.330** (high = has benefited) (.090) (.080) (.101) Single European market.497***.271***.185** (high = hopeful) (.053) (.047) (.063) Sociotropic economic.127**.151**.058 assessments (retrospective) (.041) (.038) (.048) (high = positive) Egocentric economic.039.055.081 assessments (retrospective) (.047) (.043) (.055) (high = positive) Information index (high =.557*** aware/informed/interest) (.136) EU country benefit.067 Information (.065) Single European Market.024 Information (.039) Sociotropic economic.035 assessments Information (.030) Egocentric economic.035 assessments Information (.035) Close to France.426**.699***.553*** (high = feel close) (.142) (.133) (.104) Socialist party support.196*.208**.217*** (high = yes) (.087) (.078) (.063) Human capital (education).106***.081***.036*** (high = educated) (.013) (.012) (.010) Age.023***.017***.012*** (in years) (.002) (.002) (.002) Gender.637***.502***.369*** (1 = female) (.073) (.065) (.052) (.221) (.200) (.153) 2LL 3181.116 3990.233 6554.585 % correct 65.5 65.1 67.6 Reduction in error 22.6 16.9 3.05 Chi-squared (d.f.) 1286.595 (40) 1365.395 (40) 2963.438 (50) N 4278 5336 8960 Notes: Don t know serves as the baseline category. ***p <.001; **p <.01; *p <.05; significance tests are two-tailed. France is the reference category.

Gelleny and Anderson Support for the President of the EU 193 awareness and political interest, utilitarian considerations played a significant role in helping Europeans to form assessments of the Commission President; and the most important of these considerations are attitudes about the Single Market and whether the country has benefited from EU membership. Moreover, our results strongly suggest that less informed, interested and aware citizens at least in the context of attitudes about the President of the Commission behave like those with more information. 8 Discussion The results reported in this study have implications for our understanding of European integration and how Europeans make judgments about European institutions. Moreover, the analysis points to some general conclusions about the nature of public opinion toward international institutions and policies, and for economic voting theories and issues of information and accountability. In contrast to most studies on public support for European integration, which have tended to focus on people s support for the integration process, their country s membership in the EU, or their feelings if the EU were dissolved, this study examined people s evaluations of the performance of the most visible policymaker the President of the European Commission. Consistent with the general argument in the literature on support for European integration, we found that Europeans systematically take utilitarian considerations into account when assessing the performance of the Commission President. Moreover, the results showed that having information about the EU is a crucial ingredient in models of public opinion toward European institutions. The distribution of responses showed that many Europeans do not give an opinion when asked about the President of the Commission. Those who had information (or felt they did) were much more likely to give an assessment of the performance of the President. This appears to indicate that a significant segment of the European public is not actively involved in Europeanlevel politics. And when these citizens lack the relevant information to make a judgment, they say they have no opinion because that is the sensible answer to give when one has less information. Although this phenomenon may be associated with the general problem of a democracy deficit in the EU, it also means that a significant share of the European public is involved and makes fairly sophisticated choices when asked about EU politics. Moreover, and maybe more importantly, when we analyzed well- and ill-informed Europeans separately (not shown here), we found no significant differences with regard to how they make use of