DESCRIPTION (provided by applicant): This project explores the roles of heterogeneous time preference, hyperbolic discounting and imperfect information in dynamic-stochastic models of retirement and saving. In contrast to the assumptions of simple life cycle models of saving and retirement, recent evidence suggests that many people may be poorly informed about their pensions and Social Security. To further explore the relation of imperfect information to retirement and saving, we will build the process of information gathering as a jointly determined endogenous outcome into a life cycle model of retirement and saving. We then will estimate models of retirement and saving in which there is heterogeneous time preference, and in which knowledge is endogenously determined. Hyperbolic discounting has been used to explain the failure to save, and recently to explain the joint determination of retirement and saving in the presence of sophisticated hyperbolic discounters (Diamond and Koszegi, 2003). We will extend to many periods the four period theoretical model of Diamond and Koszegi (2003). We then will estimate the model, producing for the first time estimates of a structural model of retirement and saving where there is hyperbolic discounting. By allowing a lengthy period of retirement, we can explore the implications of having a limited commitment device such as a DC plan, which cannot be used to constrain consumption beyond the time when a person becomes eligible to retrieve benefits at the beginning of the retirement period. As a baseline, we will specify a set of life cycle models without an explicit role for learning. Ultimately, we will integrate the strengths of each approach into a single model. The econometric analysis will be based on the first eight waves of the Health and Retirement Study. Health is closely related to retirement and wealth. It is a major determinant of each. Health measures reported by respondents are included in the model to be estimated.