On the robustness of week-day effect to error distributional assumption: International evidence

We examine the robustness of the week-day effect both in the mean and conditional volatility using 51 stock market indices while controlling for volatility clustering and using three specifications of the error distributions. We show that the evidence of week-day effect in mean and conditional volatility is sensitive to the choice of the underlying distribution. Our results are not limited to the classic setting of examining week-day effect but also extend to settings that account for conditional and unconditional market risk. We extend our analysis to “wandering week-day” effect, and find that it also varies with the error distributional assumptions. Moreover, we document that the 2008 financial crisis lessened the week-day effect in the mean but at the same time magnified such effect in the conditional volatility. Our findings are robust to the presence of outliers, to different model specifications, and to the presence of structural breaks. Our study suggests that empirical regularity in financial series may be econometrically fragile in the sense of Leamer (1985). It also corroborates the calls that followed the recent financial crisis about the hazard of using models that are based on unrealistic assumptions.