High net worth individuals, university endowments, and public pension funds have heavily invested in long/short equity hedge funds. But is the long/short equity asset class benefiting investors? The pitfalls of expensive financial products are well documented in the academic literature; however, the research on the subject of the “high cost of active management” is primarily focused on long-only stock-picking equity allocations. Less research has been done on alternative asset classes, which has largely avoided excess academic criticism over fees and performance.(1)
One component of alternatives that I study in this short piece is long/short equity. This research project attempts to assess the investment merit of long/short equity allocations. My study is far from perfect, as the sample period I examine is relatively short, the data I examine are not perfect, and my methods are not flawless. However, my core results, albeit imperfect, are worth consideration: a simple combination of passive indices have outperformed the median long/short equity fund from 2006-2016. Moreover, the analysis in this paper suggests that there is a lack of persistence among top quartile long/short equity managers, which would suggest endowments, pensions, and other investors may benefit from avoiding active long/short management altogether. (2)
Long/short equity funds advertise limited drawdowns, an ability to time and hedge against market volatility, and, of course, manager skill. On the long side, expert managers promise to curate portfolios of the most attractive equities. The short side promises insurance in times of market downturn – as well as the ability to bet against frauds and bankruptcies. This two-pronged approach attracted significant capital inflows into these strategies in the 1990’s; at the beginning of the decade, long/short capital comprised less than 10% of hedge fund assets, but by 2000, that number had reached 45% (Lamm, 2004).
The central question of this paper is whether long/short equity funds have met this promise – or whether these promises sound better than they work. In addition, I ask a basic question: Are there superior lower-cost alternatives?
Research has shown that passive investing is generally superior to active management for long-only mutual funds. The findings include that 70% of active money managers fail to beat their benchmark on average and just 2.3% deliver excess returns of more than 2.5% (Fama and French, 2010). Mutual fund managers are not the only ones making bad investment decisions, however. Evidence suggests that other financial experts also destroy value.
Brokerage accounts advised by financial advisors achieve lower net returns and inferior risk-return tradeoffs than self-directed accounts (Hackethal, Haliassos, and Jappelli, 2012); Expensive investment consultants who advise the world’s biggest pools of money have shown limited ability to pick funds (Jenkinson, Jones, and Martinez, 2013); the best metric for picking mutual funds is the expense ratio (Kinnel, 2010). The general lesson of academic research into active management is that the less you pay experts to manage your money, the more you keep.
But little research has tested this premise when it comes to long-short equity strategies (here is an example I know of). I compare the returns of long/short equity strategies to simple stock/bond balanced portfolios to test whether these active strategies outperform a passive alternative that replaces the long side with a passive stock index and the short side with a passive bond index. I also test performance persistence to see if the long/short managers who exceed the benchmark have proven skill in doing so.
My analysis used data from Preqin’s hedge fund database. I used monthly returns on 538 hedge funds which identity as long/short funds over the period 12/31/2005 to 12/31/2016. I obtained data on the Vanguard Total Stock Market ETF (VTI), Vanguard Total Bond Market ETF (BND), Vanguard Extended Duration Treasury ETF (EDV), and iShares TIPS Bond ETF (TIP) from CapitalIQ. For the passive portfolios, all statistics in this paper were calculated using annual rebalancing.
The hedge fund data is inevitably not comprehensive. The first issue involves the nebulous nature of the term long/short. The screening process for identifying long/short funds out of the equity strategy universe can never be perfect. Secondly, because of the legal structure of hedge funds, all data provided was received voluntarily from hedge funds. This results in a “self-selection” bias, where new hedge funds with strong performance are incentivized to provide data in order to attract more capital. Thirdly, funds with poor performance tend not to provide data, which results in a “survivor bias” in most datasets. Amin and Kat (2003) found that over the period 1994 to 2001, survivor bias led to an overstating of actual performance by an average of 1.9% annually.
David Swensen, in Pioneering Portfolio Management, describes an empirical study of long/short hedge funds; his conclusion is that, even for median performing funds, is as follows:
The fee burden shaves the net return to 3.9 percent, a result disturbingly close to the 3.7 percent return of simply holding cash.
Thus, I proceeded with an analysis of passive portfolios containing a mixture of VTI and cash. The top row in the tables below indicates a one-year holding period ending on the written date.
Figure 1: VTI-CASH Portfolio Sensitivity Analysis
The results of the VTI-Cash portfolio were not promising, as each incremental shift toward VTI increased the Sharpe Ratio. Logically, the next step examined the traditional equities-bonds portfolio.
Figure 2: VTI-BND Portfolio Sensitivity Analysis
Despite the apparent benefit from allocating more capital to BND, especially in times of poor market performance, the Sharpe Ratios and annual returns obtained were not encouraging. Next, I analyzed inflation-protected bonds through the iShares TIPS Bond ETF (TIP).
Figure 3: VTI-TIP Portfolio Sensitivity Analysis
The results of the VTI-TIP portfolio were nearly identical to the VTI-BND portfolio in Sharpe Ratio, but with higher returns and higher volatility. Again, the results were not encouraging.
Finally, I considered EDV, which is an extended duration zero-coupon treasury strip ETF. Firstly, the extended duration lends more volatility to EDV due to the time value of money. Secondly, the strips component means that the interest payments have been removed so the return is purely based upon the principle. As a result, the duration is further increased and these financial instruments tend to act inversely to the market in times of crisis. EDV provides the best alternative to shorting as a result of these characteristics.
See, for instance, the performance of EDV during the financial crisis:
Figure 4: EDV vs. SPY Share Pricing (1/1/2008 – 10/29/2010)
I proceeded with the same analysis of a portfolio constructed from VTI and EDV, with encouraging results.
Figure 5: VTI-EDV Portfolio Sensitivity Analysis
Due to the Sharpe Ratios and worst year of the VTI-EDV, particularly in the 60-40% portfolio, I proceeded in my comparative analysis with that portfolio in mind.(3)
An annual analysis of the performance of the universe of long/short hedge funds yields interesting results. First, I looked at the annual performance to see the percentage of long/short funds that beat the VTI-EDV portfolio in a given year.
Figure 6: Long/Short Fund Performance (1 Year)
The results look promising for active management; however, consider the original purpose of long/short strategies: insurance against market downturns. In the year when the market was down most, 2008, only 31% of actively managed funds beat the passive strategy.
Considering similar analysis done on rolling three and five-year periods paints an even bleaker picture of active management.
Figure 7: Long/Short Fund Performance (3 Year)
Figure 8: Long/Short Fund Performance (5 year)
While long/short funds performed particularly well in the year priors to the financial crisis and the year after the financial crisis, their poor performance during the crisis is evident in the five-year period which began in 2008, where only 25% of long/short funds beat the passive portfolio. The insurance supposedly provided by long/short strategies appears to be better implemented through the passive strategy.
These analyses do not directly address a key aspect of long/short active management: dispersion between the top and bottom quartile managers. My analysis confirms previous findings of such dispersion.
Below is a summary of the VTI-EDV portfolio versus the long/short universe by quartile.
Figure 9: Long/Short Quartile Analysis (10 year)
The top quartile managers outperformed the passive strategy in annualized return by 3.38%, but had a lower Sharpe Ratio and worst year by 0.27 and -2.1%, respectively. However, the passive strategy drastically outperformed the bottom quartile of long/short funds in Sharpe Ratio and worst year. To address analysis left out of this paper, I performed statistical significance tests on the 10-year performance assuming normal distributions and using averages instead of medians. The VTI-EDV portfolio outperformed the universe of long/short funds by 0.43% with respect to annualized return. This result was not statistically significant with a t-stat of 0.88. More interestingly given the insurance against downturns that long/short funds promise, the worst annual performance of the VTI-EDV portfolio was 13.81% higher than it was for the universe of long/short funds and the Sharpe Ratio was 0.6 points higher. This was highly statistically significant with t-stats of 9.12 and 6.41, respectively. Quartile analyses were favored in this paper due to the practical importance, from an institutional investor’s perspective, of attempting to select a high-performance manager.
In theory, if endowments, pension funds, and wealthy individuals could consistently select top quartile fund managers, the VTI-EDV portfolio would be of less value.
This notion begs the ultimate question of long/short active management: Is performance persistent?
Testing for Persistence Among Long/Short Equity Funds
I find some evidence for persistence in my dataset; in fact, there are indications that prior performance may be indicative of future performance, particularly when longer performance histories are considered. However, the persistence numbers still suggest institutional investment managers have at best a 50% of choosing a winning manager. This lack of persistence could be explained by the capital inflows to well-performing funds in the prior period, which subsequently destroys value in the following period.
The one-year persistence analysis below was performed as follows: I identified the universe of managers in the top quartile in year t, and then calculated the proportion of that universe that was also in the top quartile in year t+1. The three-year and five-year persistence statistics follow the same process except the periods are non-overlapping three and five-year periods.
Figure 10: Long/Short Fund Persistence
Given one-year of performance history, an institutional investor has little better than random chance (25%) at choosing a top quartile manager in the following year. The statistics are more promising for three and five-year performance histories, as the chance a top quartile manager in one period remains in the top quartile in the following period, are nearly double random chance, but still less than 50%. It must be noted these statistics are likely inflated by survivor bias. For instance, consider a fund that was a top-quartile performer in one period and offered to provide performance data, but performed poorly in the subsequent period…this fund might not offer performance statistics on the period in which returns were poor.
University endowments, public pension funds, and wealthy individuals have entrusted long/short equity managers with their capital for the past three decades. These managers promise limited drawdowns and hedges against market volatility through expert stock selection management, but their promises may be better fulfilled via lower-cost alternatives. Moreover, the managers who do deliver impressive returns, show marginal ability to replicate their success. One insight from my preliminary research is that access to lower cost investment products has dramatically changed the marginal benefits of long/short equity hedge fund exposure. By eschewing long/short managers, investors may benefit from reduced fees and increased portfolio liquidity; additionally, they will save precious time that otherwise is devoted to the difficult, if not impossible, task of selecting top-quartile long/short managers. Most importantly, by choosing a well-formed lower-cost portfolio, universities, pensions, and families can arguably meet their short-term and long-term capital requirements with insurance against a market downturn.
- The views and opinions expressed herein are those of the author and do not necessarily reflect the views of Alpha Architect, its affiliates or its employees. Our full disclosures are available here. Definitions of common statistics used in our analysis are available here (towards the bottom).
- Join thousands of other readers and subscribe to our blog.
- This site provides NO information on our value ETFs or our momentum ETFs. Please refer to this site.
References [ + ]
|1.||↑||Here is an example.|
|2.||↑||In choosing passive portfolios, Vanguard VTI was the obvious choice for equity market exposure. For a synthetic passive short, cash was initially considered under the hypothesis that shorting loses money on average. The Vanguard BND was a natural second step as treasuries are a classic equity portfolio hedge. Finally, inspiration for EDV was drawn from Risk Parity literature – the analysis sought a more “levered” product that could match the risk/return magnitudes of equity markets. This portfolio would certainly have been difficult to construct ex ante. It’s performance is merely indicative of the poor performance of long/short funds in aggregate during the past ten years. The author is himself working for a long/short fund, and does not intend to categorically condemn long/short active management, but rather begin a dialogue on the performance of long/short funds when investors needed it most, and whether long/short managers deserve the fullness of their current capital allocation in equilibrium.|
|3.||↑||There is no doubt an element of hindsight bias here. Perhaps bonds won’t provide “crisis alpha” next go around? Plausible. But, this hindsight bias is assumed to roughly offset the similar upward biases in the L/S equity database returns.|