A Two-Stage Model of the Effect of Economic Conditions on Election Outcomes
Levernier, William, Atlantic Economic Journal
A commonly held political-economic belief is that macroeconomic conditions are an important determinant of election outcomes. A growing economy is typically thought to benefit the incumbent party, while a declining economy is typically thought to benefit the opposition party.
The above hypothesis has been widely tested in the economic and political science literature during the last 30 years.(1) Most of the literature examines the effect of economic conditions on the outcome of U. S. congressional elections [Arcelus and Meltzer, 1975; Erikson, 1990; Jacobson, 1990; Kramer, 1971; Rees et. al, 1962; Stigler, 1973; Tufte, 19751 or the effect of economic conditions on presidential elections [Fair, 1978; Galeotti and Forcina, 1989!. Other papers relate economic conditions to election outcomes in Canada [Archer, 1987; Archer and Johnson, 1988; Happy, 1986] and Western Europe [Lewis-Beck, 1986]. Only fairly recently, however, have economists and political scientists begun to study the impact of economic conditions on U.S. gubernatorial elections [Adams and Kenny, 1989; Chubb, 1988; Peltzman, 1987; Stein, 1990!.
Despite the existence of a large number of studies on the subject, there is no consensus as to whether or not economic conditions actually affect election outcomes. Some researchers have found economic conditions to be important determinants of election outcomes [Chubb, 1988; Fair, 1978; Happy, 1986; Kramer, 1971; Rees et. al, 1962; Tufte, 1975!, while others have not [Adams and Kenny, 1989; Arcelus and Meltzer, 1975; Erikson, 1990; Peltzman, 1987; Stigler, 19731.
Heretofore, a weakness of most of the literature is that voting is implicitly treated as a one-stage process, whereby an individual simply decides whether to support the incumbent party or the opposition party. In reality, though, voting is a two-stage process, whereby an individual must first decide whether or not to vote. Only after the decision is made to participate in the voting process can the decision of which candidate to support be made.
In this paper, the influence of economic conditions on gubernatorial elections is examined. Elections are treated as a two-stage process in the framework of Arcelus and Meltzer . The specific purpose of this paper is to determine the influence of economic conditions on voter participation rates and on election outcomes in gubernatorial elections. The results of this study indicate that economic conditions, as measured by per capita income growth, significantly affect the voter participation rate, but that economic conditions have only a minor effect on the share of the vote received by the candidate representing the incumbent party.
II. A Model of Voter Participation Rates and Election Outcomes
The classic model of individual voting behavior was developed by Riker and Ordershook . In the Riker-Ordershook model, the individual is viewed as a rational decision maker who will participate in voting if the expected benefits of voting exceed the cost of voting. The expected benefits of voting are determined by the individual's perceived probability of affecting the election outcome and by the increase in his utility that will result if the candidate he votes for wins the election. The cost of voting is the time and money cost involved in obtaining information on the candidates and in going to the polling location.
(A) A Model of Voter Participation Rate
The Riker-Ordershook model implies that the likelihood of an individual voting increases as the cost of voting decreases and as the expected benefits increase. The cost of voting is likely to be lower when an incumbent or former candidate is running for office. If an incumbent or former candidate is a contestant in an election, voters will already have some knowledge about the contestant from having observed him in office and in prior campaigns. In this situation, the cost of acquiring information about the contestant is very low and may, in fact, approach zero. …