Anthony Downs' landmark Economic Theory of Democracy, published in 1957 but circulated earlier, deals exactly with what the title says. Not only does Downs draw from an 'economic' theoretical perspective, but his style – and one might also say his method – is deductive, though not in the rigorous and formalised fashion of current deductive theorising. Since Downs assumes that citizens in a democracy primarily act to maximise their self-interest and their utility income, that [parties formulate policies in order to win elections, rather than win elections in order to formulate policies,] and that the primary objective of politicians [is to be elected,]1 it is consequential to these premises that a theory of elections is very central to his theory of democracy. Thus Downs can be read and developed along three perspectives: (i) within the general context of theories of democracy, (ii) in terms of a better formal and deductive fit between economic premises and the theory of elections, and/or (iii) by isolating his spatial model of party competition and testing it against empirical findings.
As a theorist of democracy it suffices to note that Downs does not even begin to explain how a democracy comes into being, but that his interpretation appears more convincing the more we read it as an explanation of how democracies inevitably deteriorate and end up performing as meanly as they do. This is, however, the least pursued way of developing Downs.2 The second reading focuses on his spatial theory of elections, but in the light of more rigorous and mathematically formalised premises. Much of this development hinges, in fact, on rational choice theory of the game-theoretic variety. As has been concisely stated: [Spatial theory is but a particular formulation of elections as a game in the Von NeumannMorgenstern sense….]3 The third reading of Downs neglects the premises (the rational-action assumption), does not seek a more formalised model, and tests the spatial model of party competition against the evidence on voting behaviour. Donald Stokes was the first to take issue with the Downsian model with reference to its empirical applicability, while Philip Converse has been especially sensitive to its bearing on the interpretation of the data.4
My own interest in Downs is very close to the one of Stokes and Converse,