Event studies with conditionally heteroscedastic stock return

Empirically, we show that the proportion of stocks exhibiting conditional heteroscedastic residuals, is high. We suggested to use the market model with GARCH(1,1) residuals in order to describe daily stock returns and derived a test statistics for the null hypothesis of no abnormal returns, which is an extension of Boehmer et al. (1991) test statistics. Monte Carlo simulations and simulations on real data show that the test statistics accounting for conditional heteroscedasticity, Boehmer et al. (1991) test and the Generalized Sign test proposed by Cowan (1992) are well specified. The test statistics accounting for conditional heteroscedasticity dominates the previous tests in terms of power. Interestingly, Monte Carlo simulations and simulations on real data lead to very close results concerning the specifiation and the power of the test statistics so that the market model with GARCH (1,1) residuals can be seen as a reasonable approximation of the true data generating process.