The econometrics of rational partisan theory
This study develops an econometric intervention model representing the standard empirical approach to testing Alesina's (1987) Rational Partisan Theory implication that elections lead to short-term changes in output growth and unemployment. This intervention approach may be subject to two econometric difficulties. First, the cyclical nature of the autoregressive variables suggest the regression residuals may be serially correlated. Second, the election intervention variable may be endogenous to the cyclical variables. Empirical support for the model is mixed. Ordinary Least Squares estimates for both series produce a coefficient for the intervention variable which is of the predicted sign but not significant. The output growth regression results are robust to serial correlation and endogeneity concerns. For unemployment, controlling for serial correlation generates a significant coefficient, but adjusting for endogeneity does not.