Test for Zero Mean of Errors In An ARMA-GGARCH Model After Using A Median Inference

Test for Zero Mean of Errors In An ARMA-GGARCH Model After Using A Median Inference Abstract: The stylized fact of heavy tails makes median inferences appealing in fitting an ARMA model with heteroscedastic errors to financial returns. To ensure that the model still concerns the conditional mean, we test for a zero mean of the errors using a random weighted bootstrap method for quantifying estimation uncertainty. The proposed test is robust against heteroscedasticity and heavy tails as we do not infer the heteroscedasticity and need fewer finite moments. Simulations confirm the good finite sample performance in terms of size and power. Empirical applications caution the model interpretation after using a median inference.