9707771 Hamilton This project develops new tests of whether an economic relation his nonlinear and a new method for choosing a particular nonlinear specification to use to describe the data. The idea is to view the underlying relation as the outcome of a latent Brownian field. The population parameters of this latent field are estimated by maximum likelihood. Once its probability law has been specified, optimal inferences about the underlying relation can be obtained. Potential economic applications include arriving at a better statistical description and understanding of business cycles, modeling the effects of oil shocks, analyzing how monetary policy affects the economy, and studying the convergence of income levels across different countries. ??