Collaborative Research: Temporal Dependence and Economic Fluctuations -- A Spectral Investigation

The proposed research explores the economic implications of non-time-separable preferences and technology. The research is motivated by experience with spectral utility for time-separable constant-relative-risk-aversion utility functions, as well as non-time-separable functions. Spectral utility functions decompose agents' preferences for consumption smoothness into preferences for particular types of smoothness by frequency. Smoothness at low frequencies (the long run) may be more or less important than smoothness at high frequencies. For time-separable preferences, utility does not vary by frequency, but for non-time-separable cases it does. This carries implications for a variety of economic phenomena that we propose to study. We will consider how models with habit formation in preferences produce observed equity premia and risk-free rates of return and whether such temporal nonseparabilities could play a role in explaining the 'excess volatility' puzzle of stock prices. We will also analyze the implications for dynamic diversification; optimal portfolio allocation, fund separation, and other results require modification when returns are serially correlated and/or preferences are temporally nonseparable. We expect that the spectral properties of evaluation devices like the Hansen-Jagannathan bounds will provide additional insight into behavior of economic models even when preferences are time separable. Finally, we will explore whether or not these devices are useful for assessing the successes and failures of business cycle models that use time-to-build and other technologies.