This award is for continued research on the nonparametric analysis of regression models. The results under the previous award are important because the investigator has developed estimating methods that allow the relaxation of certain extreme assumptions. The purpose of this study is to extend ythe estimation techniques and apply them in different contexts. The first extensions employs the use of a duration model to analyze unemployment data available from the Panel Study on Income Dynamics. The second extension will use competing risk models to conduct an empirical study of unemployment spells. In these models, the risks of finding a new job verses recall in an old job are considered. The third extension will involve the use of a multinomial discrete choice model. Housing mobility of the elderly will be analyzed with this model.