Presidential Popularity and Macroeconomic Performance: Are Voters Really So Naive?

F OLLOWING the pioneering work of Kramer (1971), numerous studies have analyzed the relationships between the macroeconomic performance of political administrations and their popularity or vote-getting ability. Kramer's work showed that the aggregate congressional vote for an incumbent party can be explained rather well by variables indicating recent economic performance. Subsequent works have extended Kramer's research by explaining presidential vote or Gallup Poll popularity data with similar models.' On one point almost all of these studies agree:2 votes and popularity can be explained well by models which suppose that voters judge policymakers on the basis of retrospective evaluation of past macroeconomic outcomes. Most authors interpret their results as being supportive of the hypothesis of "rational" voters, since the evidence indicates that voting decisions are not based purely upon personalities, party affiliation, or chance. Kramer argued that if voters regard past macroeconomic performance of an administration to be indicative of future performance, then a concern with past performance indicates a concern for future performance. Voters who simply extrapolate from the past to predict the future are rather naive, however. A "sophisticated" voter, who has an understanding of the important intertemporal constraints embodied in the structure of the economy, would be able to forecast future consequences of current policy choices more accurately than one who simply extrapolates. If votes and popularity are simply functions of past outcomes, then voters must either be ignorant of the basic structure of the economy, or unconcerned with the future. The supposition that voters are so naive might run counter to the intuition of most economists, who generally assume individual rationality as a central behavioral postulate. Nevertheless, the empirical results noted above have been accepted by many as evidence that voters do misperceive their own long-run interest. The implications of the naive voter hypothesis for macroeconomic policymaking are disturbing. Nordhaus (1975) has developed a model which shows that voter naivete can encourage vote maximizing politicians to induce political business cycles and an inflationary bias. There is some, but not unanimous, empirical support for the existence of politically motivated business cycles, and casual empiricism also suggests that politicians themselves believe in the naive voter.3 While studies show that empirical evidence is consistent with the notion that voters are concerned with the recent past, the results do not reject a "sophisticated" voter hypothesis. None of the aforementioned studies has tested a model consistent with a sophisticated voter hypothesis, i.e., a model in which voters understand what outcomes are feasible in the long and short runs, and are concerned with future as well as current and past economic performance. In this paper we develop such a model and estimate it using Gallup Poll data on presidential popularity. Results of this model are compared to results from a "naive" voter model similar to those employed in previous research. Received for publication February 4, 1982. Revision accepted for publication July 30, 1982. University of South Carolina. I am grateful for helpful comments provided by Ray Fair, Douglas Hibbs, William Keech, Daniel Richards, Abel L. Da Costa Fernandes, and an anonymous referee. I also wish to thank William Keech for providing me with the Gallup Poll data on presidential approval rates. Aurel Schubert provided excellent research assistance. ' Fair (1978a), Bloom and Price (1975), Arcelus and Meltzer (1975), and Pollard (1981 a, 1981 b) have all estimated models to explain voting, while Frey and Schneider (1978) and Hibbs (I 982a, 1982b) have developed models to explain presidential popularity. 2 An exception is provided by Arcelus and Meltzer (1975). 3Nordhaus (1975) and Tufte (1978) have provided evidence of political business cycles for a number of Western democracies, and Frey and Schneider (1978) and Maloney and Smirlock (1981) have estimated policy "reaction" functions suggesting that the timing of elections affects policymaking. See Tufte for a discussion of casual empirical evidence indicating that politicians believe that current economic conditions affect electoral outcomes.