Performance Evaluation with Stochastic Discount Factors

We study the use of stochastic discount factor (SDF) models in evaluating the investment performance of portfolio managers. By constructing artificial mutual funds with known levels of investment ability, we evaluate a large set of SDF models. We find that the measures of performance are not highly sensitive to the SDF model, and that most of the models have a mild negative bias when performance is neutral. We use the models to evaluate a sample of U.S. equity mutual funds. Adjusting for the observed bias, we find that the average mutual fund has enough ability to cover its transactions costs. Extreme funds are more likely to have good rather than poor risk adjusted performance. Our analysis also reveals a number of implementation issues relevant to other applications of SDF models.