Voting and turning out for monetary integration: the case of the French referendum on the Maastricht treaty

Abstract

This article analyses the voting and abstention patterns in French departments in the 1992 referendum on the Maastricht treaty, in light of the potential impact of monetary union. We observe that departmental characteristics implying either greater benefits or lower costs from monetary union are significantly correlated with the approval rate. This supports the view that the voting behaviour of individual agents depended on their self-interest. The impact of economic characteristics on the abstention rate is less clear. Indeed, the variable that is most significantly correlated with abstention in the referendum is average abstention in other elections.

Notes

¹In France, départements are the most important administrative units between municipalities and regions.

²This was however unexpected at the time, as De Boissieu and Pisani-Ferry (1998) or Franklin et al. (1994) recall.

³In line with Frieden's (1991) argument, stands a survey commissioned by the ‘Mouvement des citoyens’, a private pro-European association (Institut Louis Harris, 1993). According to that survey, in 1993, 81% of French farmers declared that they viewed the fluctuations of the French franc vis-à-vis other European currencies as detrimental to their activity. More to the point, 72% of French farmers stated that they were favourable to the adoption of a common currency.

⁴Strikingly, the result on the unemployment rate may seem at odds with the finding of Doyle and Fidrmuc (2004) that higher unemployment was associated with greater approval in referenda on EU accession held in candidate countries. One must recall however that the scope of those referenda was larger than the scope of the Maastricht referendum. In addition, participation in the EU does not imply, at least in the short run, adoption of the euro.

⁵Another interpretation can be conceived if one assumes that knowledge about the European Union increases with education. The negative correlation between lack of education and the share of yes votes may then be interpreted as a consequence of Hayo's (1999) finding that support for European integration is correlated with EU knowledge.

⁶Opp (2001) also observed a positive impact of education on participation.

⁷In addition to being estimated in a SURE model, abstention Equations 7 and 8 were estimated on their own. It results that the adjusted R ²dropped from 66.05% to 11.88% between the two regressions. In other words, dropping average abstention from the set of explanatory variables results in a 54.17 percentage point drop in the explained variance of the abstention rate. Moreover, average abstention explains up to 48% of the variance of abstention observed in the referendum in a simple bivariate regression.

⁸They however contended that using aggregate abstention figures, as we do here, instead of individual observations may raise the share of variance explained by past abstention.

⁹It should also be noted that the coefficient of correlation between any two of those variables, although always significant, never exceeds 0.45 (0.27 between unemployment and long-term unemployment, 0.21 between unemployment and the share of population that has no qualification, and 0.44 between long-term unemployment and the share of the population that has no qualification).

¹⁰In France, departments are officially referred to either by their name or by their number, which follows the alphabetical order. Number one is for instance Ain and Yonne is 89. The only exceptions are the metropolitan departments whose numbers range from 90 to 95, which were created after the others, and are mainly located in the Paris area. Focusing on uneven-numbered departments is therefore equivalent to a random draw.