Book
Two Essays On Portfolio Performance Evaluation
Arizona State University
2001
Abstract
Regression models that estimate a portfolio manager's selectivity (i.e., stock-picking) and timing ability make implicit assumptions about the manager's timing strategy and the timing benchmark. Given that these assumptions are rarely verifiable, specification error is almost inevitable in empirical studies. In the first essay of this dissertation, the researcher makes the assumption that benchmark returns are normally distributed and then performs Monte Carlo simulations to examine how inferences on portfolio performance are affected by the specification errors. Simulation results indicate that: (1) timing strategy misspecification results in severely biased measurement of selectivity and timing ability, while bias in measures of overall ability are economically insignificant and (2) benchmark misspecification results in severely biased measures of selectivity and overall ability, while measures of timing ability are relatively less biased. Detailed investigation of the sources of bias suggests that the current practice of adapting the latest advances in asset pricing to performance evaluation does not guarantee an unbiased estimate of ability.
The second essay incorporates the fact that most commonly employed benchmarks exhibit negative skewness and fat tails. Simulation results indicate that: (1) deviations from normality do not, in general, alter the bias due to timing strategy misspecification; (2) kurtosis does not, in general, alter bias due to benchmark misspecifcation, while skewness does; and (3) benchmark bias obtained under bootstrapped distribution is significantly different from that obtained under a noncentral t distribution, which suggests that factors other than the first four moments affect benchmark bias.
Metrics
1 Record Views
Details
- Title
- Two Essays On Portfolio Performance Evaluation
- Creators
- Naveen D DanielArizona State University
- Publisher
- Arizona State University
- Resource Type
- Book
- Language
- English
- Academic Unit
- Finance
- Identifiers
- 0493268944; 9780493268941; 991020537627304721