Loss aversion and the equity premium puzzle : evidence from quantitative experiments /

Abstract

This dissertation undertakes both quantitative and empirical analyses of a Dynamic Stochatic General Equilibrium (DSGE) asset pricing model addressing an issue that has contributed to the empirical failure of these models: household inter-temporal preferences. The preferences studied have two parts: loss aversion and narrow framing. Loss aversion implies an agent's utility falls by more from a loss than it rises from an equal sized gain. Narrow framing is modelled by preferences that are defined over the differences in equity and risk-free bond returns. Towards these goals, this dissertation makes two advancements. First, a Hybrid Perturbation-Projection Method is introduced and evaluated to illustrate problems with current solution methods in estimation. Second, a prior predictive analysis is undertaken to find the distribution that describe the parameters of loss aversion and narrow framing preferences.

The Hybrid Perturbation-Projection Method, that combines in a less precise but fast perturbation method with a change of variables and projection algorithm, is shown to be an accurate and speedy mechanism well suited for structural estimation. The structural estimation conducted in a prior predictive analysis indicates that loss aversion and narrow framing are not a global solution to the Equity Premium Puzzle. That is, other theories must be incorporated with loss aversion and narrow framing for the equity premium to be reconciled.