Clive W. Granger University of California - San Diego SBR-9600674 Forecasting is a widely used tool in policy analysis. The P.I. will apply recent developments in nonlinear time-series models to improve forecasting models. Among the areas to be investigated include error-correction models and duration models. Error- correction models will be generalized to include nonlinear forms. Duration models, which relate to the time period "news" has effect on a market, will be investigated to determine their relevance for forecasting.
