Redistribution, Polarization, and Medians: Bringing Data to Downsian Puzzles

Abstract

In An Economic Theory of Democracy, Anthony Downs made two widely discussed predictions about political outcomes in a democracy: first, that where the principle of one person, one vote obtains, redistributive economic policies are likely to ensue; and second, that rational, vote-maximizing parties -- and, presumably, their candidates -- in two-party systems have incentives to converge at the preferences of the median voter. Without modification, however, these predictions have not fared well empirically as explanations of political outcomes in America, where policies are not especially redistributive, and parties seem to be increasingly polarized at both the elite and mass level.

In this paper we use survey data to investigate the consequences for the Downsian model of the distribution of electoral participation by voting, working in campaigns, or making campaign contributions. We consider two characteristics of citizens that affect both their issue attitudes and their level of participation – their incomes and their level of religious attendance. We show that the medians of the distribution of income and of religious attendance for those who participate are always higher than for those who do not participate in politics. We do this by using a strong property of distributions, first-order stochastic dominance, which implies the median result and which turns out to offer some technical advantages for empirical work.

Because the medians for income and religious attendance are substantially higher for participators than those for the general population and because there are reasons to believe that participators have a greater impact on policy outcomes than non-participators, it is not surprising that there is not much income redistribution in America. Those citizens who work in or donate to campaigns -- whose economic position, economic needs, and policy preferences do not reflect those of the median voter -- are in a position both to have additional influence on the outcome of an election and to send direct messages to candidates about their preferences. Moreover, whether Republican or Democratic, the contributors of the median partisan dollar are, in spite of their policy differences, much more affluent than the identifiers, voters, or workers in either party. Vote-seeking candidates will, therefore, not converge on the median voter. Rather, the requirements of running and funding a campaign force parties and candidates to be responsive to political activists, whose circumstances and perspectives have been communicated to them and who do not typically favor redistribution.

We also show that the parties differ in their internal medians on both dimensions, and they have become more polarized over time, especially along the dimension of religious attendance and especially with respect to campaign workers. Whereas campaign workers within the Republican Party have high levels of church attendance, those within the Democratic Party have very low levels. Yet despite the emergence of the religious attendance dimension in the past forty years and the increasing polarization along it, the parties remain somewhat more polarized along the income dimension. But it seems likely that the focus on social, cultural, and moral issues fostered by the sorting of party identifiers along the dimension of religious attendance has distracted policy-makers from the economic dimension just as civil rights did in an earlier period. This suggests another way that redistribution has been thwarted in America.

Keywords: participation, polarization, redistribution, medians, Downsian model