The last decade has experienced a rebirth of interest in the role of social class as an explanatory factor of voting preferences. The reappraisal of this classic notion has been mainly built on the redefinition of class (Erikson and Goldthorpe, 1992), method refinement (Nieuwbeerta and De Graaf, 1999), and better specification of party preference (Evans, 1999b). Despite these profound empirical rearrangements, the basic theoretical mechanism linking social class and vote has remained basically intact. However, a growing volume of political economy research on the redistributive and political effects of welfare states has questioned the unidirectional causal mechanism of classic class voting supporting the view that voter preferences are actually endogenous to party competition. According to this, the ability of political elites to gain and retain office is a function of their capacity to forge wide electoral coalitions that often overcome class boundaries (Przeworski and Sprague, 1986; Esping-Andersen, 1985; Iversen, 2006). 

At the same time, another body of literature has found empirical support to the notion that the economic and political geography of political unions in
uence political actors' preferences on inter-personal and inter-regional redistribution and its institutional arrangement. In particular, Beramendi (2012) set up a political economic model in which individual income and the geography of wealth and inequality shape preferences for redistribution and fiscal structures under decentralization. Although Beramendi's work built on previous solid political economic research on the determinants of preferences for inter-personal redistribution (Alesina and Giuliano, 2009), the need to account for the multidimensionality of the political space to understand actors' strategies to forge voter coalitions over redistribution (Roemer, 2001, 1999; Iversen, 2006), and the effect of institutions on redistribution (Iversen and Soskice, 2006), his account of the effects of economic geography in shaping preferences for redistribution has proved effective and fruitful to widen our knowledge of the mechanisms at work within processes that lead to specific institutional designs of decentralized political unions. For one, it has set the pace for others to improve our understanding of the determinants of individual preferences for inter-regional redistribution (Amat, 2012; Balcells et al., 2015). 

However, the mechanism linking class (or income) and party preference has not yet been reappraised under this new political geographic and economic framework, and as a result, the specific connection between class-based voting and political decentralization is still in need of further development. This paper deals with the fragmentation of social class mobilization in political unions due to the e ect of economic geography.