This proposal is about developing nonparametric methodology for applications in economics and finance. It is broadly in two parts. The first part is related to my work on separable nonparametric models, which is at the frontier of econometric methodology. I am working on a number of projects that extend the applicability of these methods to problems of current interest. The second part is related to my work on testing stochastic dominance, which is a topic of considerable current interest and considerable scope of application. This work involves several extensions of existing methodology and its application to new problems. Both sets of projects involve theoretical work in terms of defining estimators and test statistics and analyzing their properties. They also involve application of the methodology to simulated and real data. This involves developing efficient computer programmes and running them on state of the art machines. Because the procedures we develop are complicated functions of the data they are very time consuming to implement, especially time consuming to implement well.