This paper extends empirical literature on political economy of intergovernmental transfers to multiparty systems that are typical for most European countries. It proposes and uses new methods of estimating the number of swing voters from survey data. The first method estimates densities at the cutpoints, where a voter is equidistant to competing parties. To take into account bi-dimensionality of Spanish politics for three party regions, we estimate bivariate densities at the cutpoints on the left-right and nationalist dimensions. The second method counts voters with similar predicted likelihoods of voting for parties in the regions. The likelihoods of voting are estimated with the multinomial probit technique and include additional controls for the nationalist sentiment. We find that political variables enter significantly into allocation of state subventions in Spain, and the magnitude of the effect is comparable to that of economic variables. In particular, we find strong evidence for the loyal hypothesis and no evidence for the swing hypothesis. In line with the explanation suggested by Cox and McCubbins (1986), the risk-averse incumbent prefers investing in loyal regions, where he knows better preferences and numbers of their supporters.