Economic performance and U.S. Senate elections, 1958?1986

Abstract

Despite the use of several alternative estimation model specifications, no relationship was found between state economic performance and U.S. senatorial votes. On the other hand, a significantly positive relation was found between election year national real per capita income growth and the vote share of candidates of the incumbent president's party. Both of these results are broadly consistent with earlier findings for other state level elections. Upon further examination we find this effect to be operative only in the negative sense that the presidential party's senatorial vote share declines when economic performance is poor, but is not positively associated with above average economic performance. Further, the vote share of incumbent candidates, independent of party affiliation, is found to be lower in slow growth pre-election years than in years of above average growth. The estimate of the difference is such as to suggest a large and direct impact of national economic performance on votes for incumbent candidates themselves, independent of effects acting through party affiliation. Such an effect, despite its intuitive plausibility, has been sought (and found) in one previous study — Adams and Kenny. Since the high growth and the low growth sample means differ by about three percent, the 4 to 6 percent difference in vote plurality associated with incumbent candidacy in good versus poor performance periods translates into about the same impact as that operating through presidential party affiliation. Taking these effects together one may surmise than an incumbent senator of the president's party may rightly fear a pre-election recession.