2014 Theses Doctoral

Essays in Asset Pricing and Mutual Fund Behavior

Argyle, Bronson

This dissertation consists of three essays in asset pricing. The first essay demonstrates the application of a Bayesian methodology of regressor selection to factor pricing models. Bayesian Variable Selection (BVS) algorithms offer robust, intuitive methods for determining the inclusion of specific regressors within a regression framework. These algorithms can be vastly more efficient than traditional, frequentist approaches. After constructing a BVS framework, I apply this methodology to test a factor asset-pricing model, using 270 different portfolios, spanning 8 different sorting characteristics, onto 5 popular factors - the Fama-French factors (SMB, HML, MKT), a measure of aggregate "liquidity" (the Pastor-Stambaugh liquidity measure from Pastor and Stambaugh (2003)), and the Carhart measure of momentum (UMD). Results show that the Fama-French three factors have average inclusion probabilities of .93, .69, and .66, respectively. There is marginal evidence that the momentum risk factor is priced (.22 probability of inclusion). There is little evidence that the measure of liquidity is priced (.11 probability of inclusion). The apparent nonpricing of liquidity may be due to even liquidity-spreading across portfolios. Applying BVS to the CRSP universe of individual securities from 1962 to 2008, I find a notable reduction in the MKT factor (an average inclusion of .19), an increase in the intercept (the stock ɑ), a reduction in the inclusion of the momentum factor, and an inclusion of the liquidity factor of .21. This suggests that the liquidity measure is more relevant when pricing individual securities, and the momentum factor is more relevant when pricing portfolios.
In the second essay, I explore one potential channel in which firms are exposed to the idiosyncratic shocks to the returns of other, seemingly unrelated, firms. This essay expands our understanding of flow-related price pressure by demonstrating that induced flow is one channel in which idiosyncratic shocks can affect seemingly unrelated firms (controlling for common factor and industry shocks). Looking at mutual fund portfolios and instrumenting to address potential flow/return endogeneity, I find that the shocks to other firms in common mutual fund portfolios induce future portfolio flows, which induce portfolio rebalancing and result in temporary price pressure on a given firm. A one standard deviation increase in the flow-induced price pressure corresponds to a .15-.6% increase in daily abnormal firm returns. This pressure fully reverses in 5-6 days, and the magnitude is larger if funds experience a net outflow than if they experience a net inflow. Liquid firms are more sensitive than illiquid firms to this price pressure. These findings are consistent with the hypothesis that managers experiencing a portfolio return shock adjust the most liquid assets in expectation of fund flows. If investors are unable to properly estimate the correlations induced by being in common portfolios, they are unable to fully diversify away idiosyncratic risk.
Finally, the third essay, co-authored with Li An, further expands our understanding of mutual fund managers and the pricing effects that result from their trading behavior. This study investigates a V-shaped disposition effect - the tendency to sell relatively big winners and big losers - in the trading behavior of mutual fund managers. We find that a 1% increase in the magnitude of unrealized gains (losses) is associated with a 4.2% (1.6%) higher probability of selling. We link this trading behavior to equilibrium prices and find, consistent with the relative magnitude found in the selling behavior regressions, that a 1% increase in the magnitude of gain (loss) overhang predicts a 1.4 (.9) bp increase in future returns. An overhang variable capturing the V-shaped disposition effect strongly dominates the monotonic capital gains overhang measure of previous literature in predictive return regressions. Alternative V-shaped overhang measures produce similarly consistent results. One of the major contributions of this essay is to step beyond simply documenting the existence of this behavior among mutual fund managers and to shed light on the other predominant characteristics of the managers that are most likely to manifest this effect. Funds with higher turnover, shorter holding period, higher expense ratios, and higher management fees are significantly more likely to manifest a V-shaped disposition effect.