Models of the `political business cycle' assume that politicians are solely `office motivated' and enhance their chances of retaining office by generating pre-election booms. In recent versions including private agents' rational expectations, policy- makers' informational advantages enable them to create a temporary illusion of competence in maintaining low unemployment and inflation. In `partisan' models, in contrast, political parties' ideological differences also matter: `conservative' parties choose lower inflation than `socialist' parties and therefore tolerate higher unemployment. In rational expectations versions, wage setters anticipate policy-makers' incentives, so inflation is higher under `socialist' administrations, while unemployment remains at its `natural' rate under either government. If nominal wages are set before the announcement of an election result, however, they are based on a weighted average of inflationary expectations for the two regimes. If the `socialists' win, these expectations prove to have been too low, so real wages and unemployment fall; if the `conservatives' win, real wages and unemployment correspondingly rise. In either case, unemployment returns to its `natural' rate during the rest of the administration.

In Discussion Paper No. 547, Research Fellow George Alogoskoufis and Apostolis Philippopoulos extend the `partisan business cycle' approach to allow for unemployment dynamics and test their model on the Greek economy for 1958-89. Nominal wages are set one period in advance to achieve an employment target, which depends asymmetrically on the employed and the unemployed so unemployment returns only gradually to its natural rate. Their results which are mainly based on structural wage equations offer qualified support for the `partisan' model. The identity of the current government exerts a stronger influence on expected price inflation and hence nominal wage growth than the date of the most recent election, which suggests that there has been little or no electoral uncertainty.

Alogoskoufis and Philippopoulos suggest two explanations for these results: most elections have been held at the end of the year, shortly before the traditional January start of wage negotiations; and the the eventual winner's identity has almost always been easily predictable well in advance. The identity of the administration has therefore been more important than the occurrence of elections: on average `socialist' governments have generated inflationary expectations five percentage points higher than `conservative' governments. The authors find no evidence of temporary partisan correlations between elections and unemployment, which appears to depend on other factors.

Political Parties, Elections and Inflation in Greece
George S Alogoskoufis and Apostolis Philippopoulos

Discussion Paper No. 547, June 1991 (IM)