Most economists agree that the decline in the savings rate in the United States over the past two decades is a cause for concern. But there is conflicting evidence as to the reasons U.S. saving are so low. Much of this evidence is based on models of savings under perfect certainty because realistic intertemporal models of saving under uncertainty are intractable. The contribution of this project comes from developing software for vector calculations on a supercomputer that meet the substantial computing and programming requirements of an intertemporal model of consumption and labor supply with overlapping sources of uncertainty and a realistically large number of periods in the model. Numerical methods will be used to estimate the magnitude of precautionary saving, that is, saving caused by uncertainty about future events. The magnitude of precautionary saving depends in an important way on the extent that existing private and public insurance programs reduce individual uncertainty. Existing and potential future government insurance programs, such as Social Security, unemployment insurance, Medicaid, Medicare, and catastrophic health insurance, create highly nonlinear or kinked budget constraints which are impossible to model in the standard analytic framework. The numerical solution techniques developed by this project will permit the effects of these nonlinear insurance and tax programs to be measured accurately.
