Testing for Nonlinearity in Mean in the Presence of Heteroskedasticity

This paper considers an important practical problem in testing time-series data for nonlinearity in mean, namely, the distortion in the size of the test encountered if the the data are heteroskedastic. It is shown that using a heteroskedastic consistent auxiliary regression together with the wild bootstrap is an effective way of dealing with the problem. Simulation results indicate that significant improvements in empirical size are obtained, particularly in small samples.