Referendums and Ethnic Conflict

7. EU Referendums: Nationalism and the Politics of Supranational Integration

C h a p t e r 7 EU Referendums: Nationalism and the Politics of Supranational Integration Jean Monnet, the founding father of the European Union, was as honest as he was unequivocal about the involvement of citizens in the process of making an ‘‘ever closer union.’’ He confessed that he ‘‘thought it wrong to consult the people of Europe about the structure of a Community of which they had not practical experience’’ (Monnet 1978: 367). So why have political elites not heeded the call of the founding father of the European ‘‘project ’’? And why have they been willing to ‘‘introduce an additional veto player’’ (Tsebelis 2002 116; see also Hug 2002)? Given that governments, presumably, ‘‘only hold referendums when they expect to win’’ (Lijphart 1984: 148), it is, as one political scientist put it, a bit of a ‘‘puzzle that governments would voluntarily introduce another veto player in the decision-making process’’ (Hobolt 2006: 157). This chapter seeks to answer this puzzle using a formal model, which subsequently will be tested. In addition, the chapter analyzes why referendums are won. When Are Referendums on European Integration Held? Perhaps surprisingly, there have—with certain notable exceptions in recent years (e.g., Dür and Mateo 2011; Nielsen 2009)—been relatively few studies of why EU referendums have been held. One of the few exceptions in the standard literature on referendums is Tor Bjørklund’s much-cited article ‘‘The Demand for Referendum: When EU Referendums 115 Does It Arise and When Does It Succeed?’’ (Bjørklund 1982); another example is Laurence Morel’s taxonomy of referendums (Morel 2005). These models are presented and subsequently reformulated into a formal model. According to Bjørklund, referendums are held in order to paper over differences within a multiparty coalition. Thus, the Norwegian coalition government decided to hold a referendum on European Economic Community (EEC) entry in 1972 in order to avoid a split (Bjørklund 1982). This model applies to other cases too. Indeed, the Swedish bourgeois parties followed the same logic when they agreed to disagree over EU membership in 1994 (Ruin 1996). Bjørklund’s model, then, is clear: when a coalition is ‘‘divided over an important issue . . . [it] may embrace the referendum as a mediating device’’ (Bjørklund 1982: 248). The other explanation is provided by Laurence Morel, who—in empirical work—has analyzed what she calls a ‘‘politically obligatory referendum’’ (Morel 2005). The logic of this model is that a political party facing considerable competition feels forced to hold a referendum to show that it acknowledges the views of the voters. For example, the Conservative government of John Major faced considerable competition from the Labour Party in 1996, and opinion polls suggested that the Conservatives—on this issue—were closer to the median voter than was the Labour Party (McAllister and Studlar 2000; Qvortrup 2006). This model can be formalized. Taken as a whole, an actor, i, is likely to benefit from a referendum if there is considerable competition, C, and if the squared distance between the actor’s preference point and the preference point of the median voter is small. In other words, as noted several times before, Pref 冘 n i1 C (ImPi)2 Empirical Testing of the Formal Model Morel’s and Bjørklund’s explanations both have intuitive merits. However, they are based on relatively few—and perhaps selective—cases. The question is if we can find an overall pattern. To test the respective hypotheses, we need to operationalize the variables. Bjørklund’s model basically states that there is a positive correlation between the number of parties in a 116 Chapter 7 coalition and the propensity to submit issues regarding European integration to the voters. This can be measured by the number of parties in the coalition. Consequently, if there is a positive correlation between the effective number of parties in the government over the period in question and the number of referendums, then we can regard the Bjørklund model as partially corroborated. It is slightly more difficult in the case of Morel’s model. But two factors may lend themselves to an explanation: the electoral volatility and support for integration using Eurobarometer data for the years when a referendum was held. Political competition is often analyzed using the Pedersen Index of Volatility, which measures the net changes in voters’ party preferences— how many actually change their support...