Implications of temporally dependent preferences for asset pricing puzzles and the welfare cost of business cycles: A spectral investigation

The dissertation is an investigation of the economic implications of time-non-separable preferences (TNS). Chapter I develops a set of tools, spectral utility and spectral welfare cost functions, to illustrate the quantitative implications of TNS preferences. Chapter II addresses a troubling feature of welfare cost calculations with TNS preferences: that there is very little discipline in the literature on the choice of preference parameters governing the non-separabilities. Here, discipline is placed on the preference parameters by requiring that the form for the non-separability be consistent with the aggregate fluctuations that we observe in the US. Chapter III adopts the spectral view of TNS preferences to show how these preferences resolve the 'Equity Premium Puzzle'.;Chapter I illustrates the interaction between consumption volatility and TNS preferences with spectral utility functions, which measure expected utility frequency by frequency. The functions decompose agents' preferences for consumption smoothness into preferences for smoothness by frequency. The spectral utility and welfare cost functions are applied to Lucas's (1987) measure of the welfare cost of business cycles.;Chapter II estimates the welfare cost of business cycles using TNS preferences. The representative agent is placed in a theoretical world, and he or she responds optimally to exogenous shocks, given the frictions in the economy. The agent's preference parameters are estimated using Markov Chain Monte Carlo methods. Two main results emerge from the chapter. First, the form for the time-non-separability estimated in this paper is very different than the forms suggested and used elsewhere. Second, the welfare cost of business cycles is small.;Chapter III explores how the introduction of habit preferences into the simple intertemporal. consumption-based capital asset pricing model "solves" the equity premium and risk-free rate puzzles. Using the spectral utility functions from Chapter I, we show that the size of the equity premium in the habit model is determined by a relatively insignificant amount of high-frequency volatility in U.S. consumption. Further, the model's premia and returns are very sensitive to changes in characteristics of the stochastic process for consumption.