In this paper, we analyze the adequacy of using GARCH as the data-generating process to model conditional volatility of stock market index rates-of-return series. Using the Hinich portmanteau bicorrelation test, we find that a GARCH formulation or any of its variants fail to provide an adequate...