Morningstar Advisor - Summer 2007 - (Page 17)

see if the funds behaved as they were labeled. The definitions of the categories were: Equity net long funds invest more assets in long positions than short positions and exhibit a discernable net-long bias over time. Equity net short funds behave in just the opposite way. They exhibit a discernable netshort bias over an extended period. Equity neutral funds divide their exposures evenly between long and short positions, aiming to produce returns that are uncorrelated with the stock market. Because these funds don’t lean long or short, their returns should be less volatile than those of funds in the other equity categories. Equity variable funds vary their long or short biases. We expect the returns of funds in this group to have little correlation with the stock market. However, because they are rarely neutral, their returns should be more volatile than those of equity neutral funds. The tests consisted of running five-year monthly regressions (from March 2001 to February 2006) against each fund’s returns. The S&P 500 was used as the benchmark for the U.S.-based funds; for overseas funds, we used the appropriate international or regional benchmarks. The results were: 1 Equity net long: The fund must have a statistically significant positive beta of at least 0.2. Equity net short: The fund must have a statistically significant negative beta of at least –0.2. Equity neutral: The fund has a low or statistically insignificant beta, with a relatively low standard deviation that reflects its market neutral status. Equity variable: The fund has a low or statistically insignificant beta, with a relatively high standard deviation that reflects its proclivity for taking directional bets. We found a major disparity between how many funds self-categorize and how they actually behave. Overall, our returns-based categorization differed from that obtained by self-categorization for greater than 40% of the funds. The highest number of discrepancies occurred for funds that selected the variable category, most of whom actually exhibited a significant net-long bias. We found even stronger results when taking into account the tendency of hedge funds to “smooth” returns. This time, we regressed hedge funds’ returns against not just the corresponding months’ S&P 500’s returns, but also against the one-month, two-month, and three-month lags of the S&P’s returns. Using this autoregressive model, 48% of funds selfselected into the wrong category. Again, the culprits were net long funds touted as variable. Follow-up interviews with managers suggested a variety of reasons for these findings, most of which revolved around differing perceptions of the definitions. One manager selected equity variable as his fund’s category. Judging from the fund’s description in its literature, the strategy could easily have been equity neutral instead: “The fund does not place bets on the direction of the market… [T]he fund’s goal is to make superior returns for its investors regardless of market conditions.” However, the fund’s behavior resembled neither type of fund. Instead, the fund notched a significant beta of 1.58 with an accompanying R-squared of 48%—a net long fund if there ever was one. A conversation with the fund’s portfolio manager explained those statistics. Since its 1997 inception, the fund has been long for all but one month. While it’s conceivable that the fund’s bias will not be so consistently long in the future—the reason why the manager called the fund variable—the fact remains that the fund has been long for almost 10 straight years, including through a major bear market. To date, it has been variable in theory, but net long in practice. Similarly, a self-described Faulty Peer Groups The chart shows the percentage of hedge funds that were wrongly self-categorized in four peer groups. 78% 56% 50% 43% 17% All Funds Net Long Net Short Neutral Variable Self-Selected Categories neutral fund turned up as net long with our test. Indeed, the fund has carefully balanced its long and short positions. Nevertheless, the fund has not behaved in a market-neutral fashion, because it has made little attempt to hedge its factor exposures and has consistently been long higher-volatility stocks and short lower-volatility securities. As a result, the fund has posted a consistently positive beta and a standard deviation that is within spitting distance of the S&P 500. This study strongly suggests that a categorization system built on hedge fund self-definitions fails at the basic task of classification: to group funds according to broad similarities. As a result, peer-group evaluations that are based on such categories are necessarily compromised. The ideal approach would be a direct examination of the portfolio’s holdings. In the absence of such information, a returns-based approach and conversations with managers should prove an acceptable alternative. K 2 3 4 John Rekenthaler, CFA, is Morningstar’s vice president, research and new product development. MorningstarAdvisor.com 17 http://www.morningstaradvisor.com

Table of Contents Feed for the Digital Edition of Morningstar Advisor - Summer 2007

Morningstar Advisor - Summer 2007 Contents Letter from the Publisher Get to Know the Bond “All-Stars” Research Briefs Our Stewardship Test Gets Tougher Save It for Later Too Many Oranges Best of Both Worlds No Two Clients Are Alike When It Comes to VAs, It’s Seller Beware Hands On: Finding the Suitable Fund Share Class for Your Client Fund Style Can Drift Off Course Sticking to Their Style, Through Thick and Thin Marathon Man Growth, With a Chaser A Menu of Ideas to Fill Five Market Baskets Funds: Finding Funds with Conviction Stocks: So You Say You Want Income? All Eyes on China Mutual Fund Analyst Picks Undervalued Stocks Most Popular Variable Annuities What’s New at Morningstar and on the Web The Advisor’s Economic Moat

Morningstar Advisor - Summer 2007

For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.