A tale of two investors : estimating risk aversion, optimism, and overconfidence

We combine two approaches to the pricing kernel, one empirical and one theoretical, which relax the restriction that the objective return distribution and risk neutral distribution share the same volatility and higher order moments. The empirical approach provides estimates for the evolution of the pricing kernel projection onto S&P 500 returns for the period 2002 through 2009. The theoretical approach provides a framework for extracting estimates of sentiment from the results of the empirical analysis, along with estimates of risk aversion and time preference. These estimates of sentiment turn out to be highly correlated with external measures such as the BakerWurgler sentiment index, the Yale/Shiller crash confidence index, and the Duke/CFO survey responses. We analyze the manner in which the three external measures reflect biases such as excessive optimism and overconfidence. Our analysis points out three significant issues related to overconfidence. The first issue is that the BakerWurgler sentiment index robustly reflects excessive optimism, but not the component of overconfidence that is uncorrelated with excessive optimism. The second issue is that overconfidence is strongly related to the presence of an upward sloping portion in the graph of the pricing kernel, a key feature of the pricing kernel puzzle.ʺ The third issue is that the time series properties of excessive optimism and overconfidence appear to generate a negative relationship between perceived risk and return. Sentiment ; Risk Aversion ; Pricing Kernel ; Optimism ; Overconfidence