Precautionary saving and asset pricing: implications of separating time and risk preferences

It is a topic of active research in consumer and capital theory to determine how to characterize preferences about uncertainty and intertemporal choice. Intertemporal extensions of expected utility theory produce serious theoretical and empirical difficulties in many economic modeling applications by artificially restricting the functional forms used to characterize risk attitude and intertemporal preference. An important consequence is that these two distinct preference concepts become indistinguishable within an expected utility index. Recent theoretical work has produced alternatives to expected utility which allow independent specification of functional forms and thus permit the separation of the two preference parameters. This study uses a simple two-period labor-leisure and consumption model which can distinguish between an absolute risk aversion parameter and the elasticity of intertemporal substitution to show the asymmetric way each parameter influences intertemporal factor allocation decisions, particularly precautionary saving behavior, under alternative stochastic income sources. Pure endowment, pure capital return, and pure wage income risk, as well as various combinations of them, are all addressed in the study, and for each case a set of closed-form comparative static decision rules is derived. It is shown that intertemporal preference plays a significant role in determining factor supply decisions when either the intertemporal price of consumption or leisure is affected by risk, while risk attitude dominates behavior when intertemporal preferences are assumed homothetic and both intertemporal prices remain undisturbed. It is also shown that possible income correlation can magnify, dampen, or even reverse the comparative static implications in the three univariate cases. Computer simulations are used to derive numerical solutions for the model which highlight these effects;These results are contrasted with those obtained from an isoelastic expected utility model which entangles the risk attitude and intertemporal preference parameters, and thus treats each symmetrically in motivating factor allocation decisions over time;This study also examines the usefulness of a popular class of recursive preferences developed by Epstein and Zin (1989, 1991) for asset pricing applications. This class of preferences also distinguishes between risk attitude and intertemporal preference, and has been shown in the received literature to improve upon the poor empirical performance of expected utility in consumption-based models of asset pricing. However, using the nonparametric test procedure of Hansen and Jagannathan (1991) as well as seasonally unadjusted US data, it is shown in this study that little improvement occurs when moving from expected utility to this recursive class of preferences, suggesting that the meshing of preferences in the former does not account for its inability to explain the high volatility of asset prices in the US. Thus, while the added analytical power derived by independently specifying preference functional forms can generate new theoretical insights, this freedom may have limited usefulness in other areas of economic research.