The purpose of this research is to develop several applications of the Likelihood Principle for investigating macroeconometric problems that have been difficult to address using classical methods. Further, methods will be devised for evaluating the performance of these procedures in repeated samples, and for illustrating the impact of a broad range of different assumptions on the results. The first consideration is whether macroeconomic time-series-data are best described as integrated or trend- stationary. The performance of Classical integration and co- integration tests will be compared against trend-stationary alternatives. In addition, a method will be devised for examining the performance of macroeconomic models from a Bayesian perspective, and to seek the appropriate interpretation of classically obtained results. Finally, the theory of dynamic games will be extended to models in which the structure of the relevant model is not known with certainty. This research is important because it will provide better methods for analyzing time-series-data, which is the type of data used in much of the empirical research in economics and the other social sciences.
