How Much Does Money Matter? 'Buying' Votes in Japan, 1967-1990

Abstract

Japanese elections are notorious for the money that flows between contributors, politicians, and voters. To date, however, nobody has estimated statistically the impact of this money on electoral outcomes. Students of American politics have discovered that this question is difficult to answer because, while performance may depend on spending, spending may also depend on expected performance - so that there is simultaneous causation to deal with. In this paper, we specify a two-stage least squares model that explains the vote shares of LDP candidates as a function of their own spending, spending by other candidates, and a battery of control variables. Interestingly, the multiple candidate nature of Japanese elections means that district-level demographic variables are largely unrelated to any particular LDP candidate's vote share, allowing us to use these variables to create instruments for campaign spending. Finally, in a necessarily tentative comparison, we find that the marginal dollar of campaign spending buys the spender a great deal more in Japan than is true in the U.S.