Abstract
In this paper we show the effects that outliers have on estimation and inference for ARCH models. We propose an empirically tractable solution to this problem by replacing outliers with their conditional expectations (optimal forecasts) in the likelihood function. This solution works well in both simulations and applications for a wide class of ARCH models. We demonstrate the accuracy of the procedure for parameter estimation, forecasting, and asset pricing. The empirical examples include U.S. interest rate, foreign exchange rate and stock index data. In addition, we offer a robust bootstrap for outliers.