Active equity portfolio management by institutional investors

This dissertation comprises of two separate essays on active equity portfolio management by institutional investors.; The first essay is motivated by the following well known implication of the Sharpe-Lintner Capital Asset Pricing Model: the market portfolio is held by every aggregate investor. Using institutional holdings data compiled from 13F SEC filings and aggregated by institution type, we investigate whether and how stock characteristics explain the deviation of institutional portfolio weights from market weights. We find that aggregate institutional investors engage in characteristic-based portfolio tilting and exhibit considerable heterogeneity in their revealed preferences for stock characteristics. We perform characteristics based performance evaluation and attribution tests using both the Daniel-Grinblatt-Titman-Wermers holdings-based and the Fama-French-Carhart returns-based methodologies. We find that no type of aggregate institutional investor demonstrates an ability to select stocks or an ability to time stock characteristics. While our results are generally consistent with Fama's (1996) multidimensional market efficiency, they do reinforce Daniel and Titman's (1997) argument that the Fama-French factors do not subsume the role of the characteristics on which they are based.; In the second essay I answer the following question: Why would a rational investor choose to invest in an actively managed mutual fund when lower-cost index funds are widely available? In order to justify the fee differential between active and passive funds, a rational investor must demand that the active fund provides a return distribution that is, at the very least, sufficiently different from that provided by passive funds. We develop and make precise this simple intuition by analyzing a stylized model of Bayesian rational mutual fund investor behavior. Our analytical framework highlights the important role of ex ante tracking error when evaluating and investing in managed portfolios. To reduce the confounding effect of unobserved mutual fund heterogeneity, we derive testable implications that pertain to the determinants of the change in demand for actively managed funds. We use Robinson's (1988) double residual technique to estimate a semiparametric model of the change in demand for actively managed mutual funds. We find that investors in US domestic equity mutual funds appear to recognize the importance of ex ante tracking error in a manner that is consistent with our theoretical model.