Over the past ten years, my research has focused on two problem areas --- one in statistical decision theory and the other in predictive inference. The first of these concerns the relationship between the admissibility of formal Bayes decision rules and the the recurrence of an associated symmetric Markov chain. The transition function of this Markov chain is essentially the expected value under the model of the formal posterior distribution. This approach to studying improper prior distributions has yielded useful insights into formal Bayes arguments and the corresponding induced inferences.
The second area of research regards foundational issues in prediction theory. Recent results show that some classical predictive procedures suffer from what de Finetti might have called incoherence (also called strong inconsistency and directly related to the notion of a Dutch book). These results cover multidimensioal prediction problems based on normal models. Alternatives to the classical procedures have been proposed and and evaluated under differrent optimality criteria.