NBER Working Papers and Publications
|April 1999||Bayesian Performance Evaluation|
with Andrew Metrick, Jessica Wachter: w7069
This paper proposes a Bayesian method of performance evaluation for investment managers. We begin with a flexible set of prior beliefs that can be elicited without any reference to probability distributions or their parameters. We then combine these prior beliefs with a general multi-factor model and derive an analytical solution for the posterior expectation of alpha', the intercept term from the model. This solution can be computed using only a few extra steps beyond maximum likelihood estimation and does not require a comprehensive or bias-free database. We then apply our methodology to a sample of domestic diversified equity mutual funds and ask what prior beliefs would imply zero investment in active managers?' To justify such a zero-investment strategy, we find that a mean-variance...
Published: Newly titled "Should Investors Avoid All Actively Managed Mutual Funds? A Study in Bayesian Performance Evaluation", Journal of Finance (February 2001).