Twenty-five Years after Kramer: An Assessment of Economic Retrospective Voting based upon Improved Estimates of Income and Unemployment


D. R. Kiewiet Division of Humanities and Social Sciences California Institute of Technology, Pasadena, CA 91125, USA


Several studies in economics have derived estimates of economic activity in the years prior to World War II that differ significantly from official statistics. In this paper we examine the possibility that the impact of economic retrospective voting upon congressional election outcomes has been at least partially obscured by shortcomings in the official data series. To that end we re-estimate various specifications of the basic Kramer model with these revised measures of income and unemployment. Although the effect of changes in real per capita income does not depend much upon which particular income series we used, the observed impact of unemployment doubles in size when the new estimates of historical unemployment rates made by Romer (1986b) and by Darby (1976) are used. Although the evidence is less than definitive, the results of our analyses also lend credence to the proposition that economic conditions have a greater impact upon congressional voting in on-year elections than in off years, and that the incumbent party is bound to suffer a midterm penalty.