A Bayesian estimate of the pricing kernel
Our research plan aims initially to improve the estimates of the pricing kernel in the literature, using Bayesian methods to combine the prior information available from different security markets. Current studies produce conflicting and often puzzling results. We adopt a non-parametric approach motivated by the flexibility of the resulting model that should lead to better performance compared to the corresponding parametric counterparts.
The potential of our estimates to solve the puzzles in the asset pricing literature will be tested on American index and index option data. Our choice is motivated by the large number of liquid index options available in the U.S. market, that make it an ideal environment to test our model.
Later we plan to extend our Bayesian non-parametric pricing kernel to the existing behavioral models. In particular we will re-examine the estimates of excessive optimism and overconfidence in Barone-Adesi, Mancini and Shefrin (2013), compare them with other indicators in the literature and assess their potential as early warning indicators of financial bubbles.
The implications of our research are likely to be relevant for risk management, regulation and financial policy. Asset management will also benefit from improved understanding of the pricing kernel.