Performance measurement of hedge funds using data envelopment analysis

Abstract

Data envelopment analysis (DEA) is a nonparametric method from the area of operations research that measures the relationship of produced outputs to assigned inputs and determines an efficiency score. This efficiency score can be interpreted as a performance measure in investment analysis. Recent literature contains intensive discussion of using DEA to measure the performance of hedge funds, as this approach yields some advantages compared to classic performance measures. This paper extends the current discussion in three aspects. First, we present different DEA models and analyze their suitability for hedge fund performance measurement. Second, we systematize possible inputs and outputs for DEA and again examine their suitability for hedge fund performance measurement. Third, two rules are developed to select inputs and outputs in DEA of hedge funds. Using this framework, we find a completely new ranking of hedge funds compared to classic performance measures and compared to previously proposed DEA applications. Thus, we propose that classic performance measures should be supplemented with DEA based on the suggested rules to fully capture hedge fund risk and return characteristics.

Data envelopment analysis models of investment funds

“…dea can model different investor preferences and distributional shapes by using several measures of risk and return such as mean, median, standard deviation (Gregoriou et al, 2005a), lower and upper semivariance and semiskewness (Gregoriou et al, 2005b), skewness (Wilkens and Zhu, 2001), excess kurtosis (Nguyen-Thi-Thanh, 2006), time horizons (Galagadera and Silvapulle, 2002), percentage of periods with negative returns, skewness (Wilkens and Zhu, 2001), value at risk, conditional value at risk (Chen and Lin, 2006), downside absolute standard deviation, weighted absolute deviation from quantile, and tail value at risk (Lozano and Gutiérrez, 2008a). Eling (2006) reviews the measures used and concludes there is no single standard choice. Lozano and Gutiérrez (2008a,b) try to account for rational investor behavior using stochastic dominance (Levy, 1992).…”

Resampling DEA estimates of investment fund performance

“…Gregoriou and Zhu (2005); Lozano and Gutiérrez (2008a,b); Briec and Kerstens (2010)) and choice of measure (e.g. Eling (2006)). Our findings are valid for a wide range of models and measures and so we do not discuss these in detail.…”

Provincial preferences in private equity

“…A list of investors that are members of the Canadian Venture Capital Association is available at http://www.cvca.ca/full_members/index.html. 30 See, e.g., Kassberger and Kiesel (2006), Eling (2006), Le Moigne and Savaria (2006) for related work on hedge funds. transactions), 84.42% of investments involved an investor and entrepreneur that resided in the same province.…”