Significant Alphas in Real Estate Funds: An Empirical Comparison of Alternative Estimators

Abstract

Executive Summary Real estate returns recurrently show heteroscedasticity. Using Jensen's alpha as a measure of risk-adjusted returns, we compare test statistic sensitivity to alternative estimates of the standard errors. Utilizing several robust estimators, we find the wild bootstrap consistently provides the most conservative result in real estate mutual funds and REITs. Surprisingly, the Newey-West standard error increases the percentage of REITs exhibiting significant alphas. Sensitivity to specification error in the model is examined. Explanatory variables failed to systematically attenuate significant alphas. When using the wild boot-strapped HC3 standard errors, significant alphas in REITs are no greater than random chance. Our results suggest appropriate adjustment for heteroscedasticity in real estate returns would minimize the potential for erroneous interpretation.