Two essays in time nonseparable preferences

My dissertation analyzes asset pricing in a general equilibrium representative agent model in which preferences exhibit endogenous shifts. In addition to current consumption, current wealth, as a proxy for a consumer's standard of living, is an argument in the within-period direct utility function. As wealth changes over time, marginal utility expressed as a function of consumption shifts over time. One may view this utility specification as displaying habit persistence. When optimizing intertemporally, the agent takes the wealth effect into account. Moreover, the wealth-dependent preference function includes CRRA utility as a special case and permits easy comparisons.;In my theoretical paper, I combine the assumptions of isoelastic utility and lognormal endowment to derive closed-form solutions for equilibrium asset prices, which are functions of preference and technology parameters. The property defined risk premium increases with risk aversion and the variance of the log shock. Properties of equilibrium are analyzed in terms of the effects on asset returns of changes in the first two moments of endowment growth. I decompose these total effects into their essential components and relate them to the basic properties of the value function. I find that risk aversion, precautionary motives and the magnitude of preference shifts jointly determine asset returns.;In my empirical paper, I investigate the testable restrictions implied by the wealth-dependent preference function and the role played by this form of utility in consumption-based asset pricing models. Here, some of the assumptions made in my theoretical paper are relaxed. Using the generalized method of moments, I estimate model parameters and test the Euler equation in terms of conditional and unconditional moments. In addition, I fit the marginal rates of substitution implied by the different utility models into the Hansen-Jagannathan bound. I find that the preference shift parameter is insensitive to the choice of instruments and the consumption measure. Switching from the CRRA to the wealth-dependent utility model, although the variance of the marginal rate of substitution increases, the mean tends to decrease. In general, the wealth-dependent utility model matches the data well.