Geoffrey C. Friesen

Heterogeneous Beliefs and Risk Neutral Skewness

This study tests whether belief differences affect the cross-sectional variation of risk-neutral skewness using data on firm-level stock options traded on the Chicago Board Options Exchange from 2003 to 2006. We find that stocks with greater belief differences have more negative skews, even after controlling for systematic risk and other firm-level variables known to affect skewness. Factor analysis identifies latent variables linked to risk and belief differences. The belief factor explains more variation in the risk-neutral skewness than the risk-based factor. Our results suggest that belief differences may be one of the unexplained firm-specific components affecting skewness.

An Empirical Examination of Jump Risk in U.S. Equity and Bond Markets

Abstract Actuaries manage risk, and asset price volatility is the most fundamental parameter in models of risk management. This study utilizes recent advances in econometric theory to decompose total asset price volatility into a smooth, continuous component and a discrete (jump) component. We analyze a data set that consists of high-frequency tick-by-tick data for all stocks in the S&P 100 Index, as well as similar futures contract data on three U.S. equity indexes and three U.S. Treasury securities during the period 1999-2005. We find that discrete jumps contribute between 15% and 25% of total asset risk for all equity index futures, and between 45% and 75% of total risk for Treasury bond futures. Jumps occur roughly once every five trading days for equity index futures, and slightly more frequently for Treasury bond futures. For the S&P 100 component stocks, on days when a jump occurs, the absolute jump is between 80% and 90% of the total absolute return for that day. We also demonstrate that, in the cross section of individual stocks, the average jump beta is significantly lower than the average continuous beta. Cross-correlations within the bond and stock markets are significantly higher on days when jumps occur, but stockbond correlations are relatively constant regardless of whether or not a jump occurs. We conclude with a discussion of the implications of our findings for risk management.

Sentiment and Stock Returns: Anticipating a Major Sporting Event

This study documents the effect of the Super Bowl on the stock returns of firms that are geographically associated with the competing teams. We find significant upward return drift in the 9 trading days leading up to the Super Bowl, a pattern consistent with investors trading in anticipation of the game itself. The “anticipatory behavior” among investors leads to widespread pregame returns, which is not documented in prior studies. These pre-event abnormal returns are positive and statistically and economically significant for all firms, and the size of pre-event returns varies according to each team’s favored status. In addition, firms associated with the winning team exhibit significant positive return drift over the 10-day period after their win. Firms associated with the losing team exhibit moderate downward drift. Our findings are strongest among the smallest quintile of firms and are robust to various risk adjustments and using a matched sample control group. The collective findings suggest that only by standing on the sideline will investors avoid winning around the Super Bowl.

Building a Better Mousetrap: Enhanced Dollar Cost Averaging

This article presents a simple, intuitive investment strategy that improves upon the popular dollar-cost-averaging (DCA) approach. The investment strategy, called enhanced dollarcost-averaging (EDCA), is a simple, rule-based strategy that retains most of the attributes of traditional DCA that are appealing to most investors yet adjusts to new information, which traditional DCA does not. Simulation results show that the EDCA strategy reliably outperforms the DCA strategy in terms of higher dollar-weighted returns about 90% of the time and nearly always delivers greater terminal wealth for reasonable values of the risk premium. EDCA is most effective when applied to high-volatility assets, when cash flows are highly sensitive to past returns, and during secular bear markets. Historical back-testing on equity indexes and mutual funds indicates that investor dollar-weighted returns can be enhanced by between 30 and 70 basis points a year simply by switching from DCA to EDCA.

A Note on the Dangers of Using Regressions to Analyze Earnings Forecasts

When data exhibit cross-sectional variation in scale and regression parameters, pooled regression parameters can exhibit severe biases. It is commonly assumed that normalizing per-share earnings data by a firms stock price eliminates cross-sectional variation in scale. This study shows that this is not always true. The hierarchical model explicitly allows for both types of variations in earnings data, and therefore provides more reliable parameter estimates. This study examines analyst exaggeration and herding behavior using empirical Bayes estimates from a hierarchical model. There are significant differences between hierarchical and conventional estimates, which sometimes substantially alter ones conclusions about analyst behavior.

Using Hierarchical Models to Shrink Alphas and Interpret Residual Covariance in Mutual Fund Returns

We present and estimate a Bayesian Hierarchical model of mutual fund returns. In our model, a funds alpha reflects not only that funds return history, but also information from other fund returns. Because parameters are estimated simultaneously for all funds, we can identify common residual covariation, and extract a factor to capture it. We show that the residual co-movement cannot be fully explained by common exposure to priced risk factors; rather, it is most likely attributable to common time-varying skill or fund manager herding behavior. In contrast to earlier studies, we find significant alpha-variation across funds, even after controlling for cross-sectional variation in expenses and turnover. This finding is robust even for prior beliefs that are strongly skeptical of alpha-variation, and has market efficiency and asset allocation implications. We demonstrate the strength of Bayesian framework by using posterior simulations to calculate 95% posterior confidence intervals for the fraction of mutual funds with alphas above various thresholds. Lastly, we conduct a mean-variance portfolio choice exercise, and demonstrate that Bayesian hierarchical alphas lead to optimal portfolios with much more reasonable allocations than their conventional counterparts.