Political business cycles and endogenous elections.

• Author: Heckelman, Jac C.

1. Introduction

   The political business cycle (PBC) model as developed by Nordhaus (1975) suggests incumbent politicians have an incentive to stimulate the economy to bolster their reelection chances. This coincides with empirical evidence that favorable economic conditions impact upon incumbent popularity (Frey and Schneider 1978; Kirchgassner 1985; Chappell 1990) and voting patterns (Kramer 1971; Fair 1978; Chappell 1990) in an advantageous manner. If voters do behave in this fashion, politicians should be aware of this and consequently help themselves by spurring growth prior to an election. However, any stimulus the politicians can provide is expected to be short-lived; economic booms during the election period are followed by busts in the next period as the reelected government attempts to eliminate the newly created inflationary pressures on the economy.

   PBC models have come under close scrutiny with mixed empirical results. Evidence of PBCs have been found in the U.S. for transfer payments (Tufte 1978) and monetary policy (Grier 1987) and in Britain for government expenditures and unemployment (Keil 1988). PBC models have been rejected in the U.S. for unemployment (McCallum 1978; Hibbs 1987a) and GNP (Hibbs 1987a; Heckelman and Whaples 1996) and in Japan for fiscal and monetary manipulation (Ito and Park 1988). In a comprehensive study of several OECD countries, Alesina, Cohen, and Roubini (1992, 1993) find limited supporting evidence for a variety of macroeconomic cycles and conclude by rejecting the implications of PBC. Nordhaus's (1975) original tests on unemployment cycles also yield mixed results for a few industrialized democracies.

   However, another explicit assumption in Nordhaus's PBC model is that the election occurrence is fixed. As noted above, studies of the Nordhaus PBC have often tested his model on OECD countries where this assumption is violated. In most parliamentary-style nations, the [TABULAR DATA FOR TABLE 1 OMITTED] current government has a wide degree of latitude in determining when the next election will take place (subject to a maximum term length), that is, the election calendar is flexible and the actual government term lengths are variable. Using the election calendars in Mackie and Rose (1991), Table 1 shows the prevalence for member OECD nations to hold an election prior to the legal limit. Only in Norway and Switzerland is the government unable to call for early elections, and in the remaining countries, only Luxembourg and New Zealand have failed to hold at least one election more than a year before the current government's term was due to expire. In fact, 61 out of the 146 (42%) elections held in the flexible-term nations between 1961 and 1988 were held at least one year early, and an additional 14 elections (51% total) were held at least six months early.

   Chappell and Peel (1979) and Lachler (1982) have developed theoretical models to suggest incumbent politicians with the authority to determine when an election is to be held can also use this control to their advantage. Incumbents may time the election to take place when the economy is doing well to bolster reelection. This hypothesis is referred to as opportunistic election timing (OET). Therefore, incumbents with variable electoral terms can behave strategically in either of two manners: manipulation of the economy prior to an election (PBC) or calling for an election under favorable economic conditions (OET). OET suggests a strong endogeneity problem in traditional tests of PBC models which do not allow for the possibility of OET.

   Ito and Park (1988) present evidence that elections in Japan are more likely to be called when economic growth is high. In a series of regressions on various OECD democracies, Alesina, Cohen, and Roubini (1993) also suggest Japanese elections are endogenous to GDP growth, although they reject OET for all other countries in their sample.

   There are two serious problems with interpretation of the standard single-equation methodology. First, if elections are indeed endogenous to PBC, much of the empirical literature will be untrustworthy because regression results that test PBC will be biased. Second, since Ito and Park (1988) and Alesina, Cohen, and Roubini (1993) also used single-equation regressions to test OET, these results are as likely to be biased as the PBC tests. Thus, it is still not clear which countries actually engage in OET. Although in this paper we do not test directly for the determinants of election timing, our focus remains on the importance of the OET implication for the testing of political cycles.

   In this paper, we treat the endogeneity issue formally. We present standard tests for PBC and then account for endogenous elections using an instrumental variable (IV) technique. Our empirical results do not yield any support for PBC in single-equation regressions, but we uncover hidden evidence of a political monetary cycle in Japan and an inflation cycle for Britain in IV estimation. Also, both countries exhibit strong evidence of OET.

   The rest of the paper is structured in the following format. In section 2, we outline the PBC models and detail the endogeneity problem. Section 3 presents the empirical results. Conclusions appear in section 4.

2. Political Business Cycle Framework

   Institutional Considerations

   The PBC suggests an incumbent will exploit the short-run Phillips Curve to stimulate growth and reduce unemployment in the election period even though it may result in higher inflation. In testing for political cycles, it is important to recognize institutional factors as well. For example, as mentioned above, the notion of fixed versus flexible election dates can create econometric difficulties which we address further below. In addition, election results in Japan are unlike those in most other industrialized nations. The Liberal Democratic Party (LDP) has been the majority party in parliament and has thus controlled the prime ministership and set the next election date since its inception in 1955 until 1993. The only exception was following the July 1977 election. Although the LDP held a large plurality in both the House of Consillors (upper house) and House of Representatives (lower house), they did not actually hold a pure majority of seats in either. This marks the only time prior to 1993 that the LDP was not the pure majority party in parliament,(2) but support from several independents allowed the LDP to retain control of the Diet and they successfully regained a pure majority of seats in both houses following the June 1980 election (Banks 1993). Only once during this time period did they decline to call an early election - when the December 1976 election was held the maximum four years after the previous election.

   Under the hypothesis set forward here, the LDP may have been very successful at retaining power due to effective manipulation of the economy. The government of Japan is in an especially good position to alter aggregate variables due to their very dependent central hank (Cukierman 1992). Alternatively, the prosperity of the LDP may be due to successful manipulation of the...