Weekday dependence of German stock market returns (Q2756667)

In this paper the so-called wild bootstrap is used to infer against weekday effects for German stock market data. The wild bootstrap procedure copes conveniently with heteroskedasticity. Nonparametric techniques are applied in order to give insight into the conditional and unconditional stochastic behaviour of German stock market returns. Compared to parametric models the approach provides a data driven framework to uncover important dynamic features which may remain undetected by parametric modeling. The author concludes that depending on the period of interest German stock market returns show significant correlation.NEWLINENEWLINENEWLINEAllowing for time-varying autoregressive dynamics periodic time-series models provide a general framework to account for time-dependent expectations. The wild bootstrap is adopted for inference in parametric models in the case of heteroskedastic errors which is relevant for analysis to empirical return data. Estimation of the nonparametric autoregressive model is essentially data driven and provides evidence in favour of significant autocorrelation of German stock market returns.