| This dissertation explores the effects of saving motivated by uninsurable income risk and tight borrowing constraints in lifecycle models with either quadratic or constant relative risk aversion (CRRA) preferences.; The traditional picture of precautionary saving at the individual level is dependent on the third derivative of the period utility function being positive. Nevertheless, aggregate precautionary saving, demonstrated by a lower rate of return on capital, has been found with idiosyncratic risk and borrowing constraints independent of the third derivative. Chapter One employs a model without capital to examine the source of interest rate reductions under quadratic utility. The interest rate is shown to fall primarily as a direct consequence of agents borrowing less. However, agents also save more in response to the possibility of a potentially binding future borrowing constraint. Only some of this extra saving is associated with uncertainty. The rest can be termed anticipatory saving.; Informed by the results of Chapter One, I then employ a more realistic model, calibrated to match properties of the U.S. economy, to examine the contribution of uncertainty and tight borrowing constraints on several macroeconomic observables. A model with a borrowing constraint and heterogeneity but no uncertainty can account for most aspects of the complete buffer-stock model. At high degrees of risk aversion, uncertainty and precautionary saving are needed to explain aggregate saving and the consumption profile, though not the marginal propensity to consume.; With CRRA preferences, there is no analytic solution for the decision problem of an agent in the kind of model studied in this dissertation. In Chapter Three, I focus on the use of Taylor expansions to obtain approximate analytic expressions for consumption functions. I find there is a threshold interest rate associated with each order in the Taylor expansion such that this order term diverges in the infinite horizon if the interest rate is below the corresponding threshold. Furthermore, the equilibrium interest rate must lie below the threshold for some order of the expansion; so there is a bound on the degree of accuracy that can be achieved with this approximation method. |
