We spend a lot of time "seeking alpha" on this website. But many don't believe that it is possible for active managers to generate alpha - or returns in excess of the market risk premium - with any consistency. They reference the efficient market hypothesis, which posits that security prices are not predictable and instantaneously adjust to all available information, and cite research showing that the vast majority of mutual fund managers do not outperform their benchmarks.
So, is it possible for managers to generate positive alpha more often than not? And if so, is it possible to identify those managers beforehand with any confidence?
Based on an excellent recent article in the Financial Analysts Journal by Robert Jones, CFA, and Russ Wermers - Active Management in Mostly Efficient Markets - the short answer to both questions is yes. The authors would argue, though, that choosing superior active managers is difficult and that investors should carefully assess the expected returns and risk of active strategies and adopt a disciplined budget for active vs. passive risk in their portfolios.
Jones and Wermers review over 30 academic papers written between 1999 and 2011 that shed light on the potential identification of superior active managers. They boil down the findings into four main categories of analysis that can help to identify superior active managers:
1. Past Performance (suitably adjusted)
2. Macroeconomic Forecasting
3. Fund/Manager Characteristics
4. Portfolio Holdings
In this article, I apply some of Jones' and Wermer's recommendations to a sample of 15 equity closed-end funds (CEFs) that have recently been rated using Morningstar's new, forward looking CEF ratings methodology. My goal is to try to identify potential superior active managers for investment purposes and also examine the relative performance of funds in the various Morningstar CEF rating categories.
Closed-End Fund Sample
I started with all CEFs rated by Morningstar's CEF analyst team and sorted by Morningstar category, selecting CEFs that primarily focus on U.S. listed equities. I did not include funds in the "World Stock" category or that focused on Master Limited Partnerships, REITs, precious metals or any fixed income category. To review, a Closed End Fund is a mutual fund with a fixed amount of shares that trade like stocks. Because CEF managers do not have to worry about providing daily liquidity to investors, they can use leverage and pursue more esoteric strategies involving options and less liquid asset classes. Also, because the number of shares is fixed and the value is market determined, CEF shares can trade at a significant premium or discount to the value of the underlying assets held by the fund.
I sorted the remaining 15 funds by Morningstar rating (Gold / Silver / Bronze / Neutral / Negative) and gathered monthly price data - adjusted for splits and dividends - on all of the funds from Yahoo Finance. While some funds had much longer return series, the majority of funds started between September 2003 and September 2005. So, I used September 2005 as my starting point of analysis, providing over seven years of monthly returns data that covers both cyclical bull and bear market periods. The only exception was the Eaton Vance Risk-Managed Diversified Equity Income (NYSE:ETJ) fund, the lone Negative rated fund, which started in August 2007 and provided a roughly 5-year return series.
The following tables provide basic descriptive statistics on the 15 CEFs in the sample, as well as the Vanguard 500 Index and Vanguard Small Cap Index funds, which I use for comparison. The tables include information from Morningstar's excellent CEF section on style (eg. large vs small cap, value vs growth, etc), strategy, expenses, premium or discount of the fund's price to its Net Asset Value, leverage and performance. One note on strategy - within their styles, six funds pursued a plain-vanilla, long-equity strategy, one a long-equity /dividend capture strategy (eg. selling stocks shortly after receiving dividends) and eight pursued a Buy-Write strategy, selling out-of-the-money call options on stocks they held in their portfolios or on the indexes in which those stocks reside. The table also includes two snapshots of the Morningstar ratings for each fund, taken on 9/1/2012 and on 11/30/2012; as of 9/1/2012, the group included two Gold rated funds, four Silver, four Bronze, four Neutral and one Negative.
Note - The Z-score of the premium (discount) is an indicator, calculated by Morningstar, that takes into account the current market price of the CEF versus its Net Asset Value, as well as the historical average premium(discount) over the previous year and the volatility of that premium(discount).
The sample achieved, on average, a 2.0% compound annualized return for the period studied with a 22.6% average standard deviation. Return and standard deviation were both between the results for the Large Cap (0.9%, 19.2% st dev) and Small Cap (2.7%, 25.1% st dev) index funds. The average CEF also has a 1.2% expense ratio, has a distribution yield of 8.5% of share price annually, and currently trades at a -8.1% discount to its NAV, which is about 0.05 standard deviations above the regular discount. On average, over the last five years, 52.1% of distributions were from Return of Capital, 33.2% from Income, 6.4% from short-term capital gains and 8.4% from long-term capital gains (see table below).
For good measure, I also produced charts that show the cumulative performance of each of the CEFs against their benchmarks, proxied by the Vanguard S&P 500 Index Fund (VFIAX), the Vanguard Small Cap Index Fund (VSMAX) or the CBOE S&P 500 Buy-Write index (BXM), since September 2005 (see below). Starting in September 2005, each of the asset classes (Long Equity and Covered Call) has roughly an equal number of funds outperform and underperform its respective benchmark on a cumulative basis.
Now that we are familiar with the sample, let's apply Jones and Wermer's findings to try to find some superior active managers.
Past Performance - Using Multiple Factors to Calculate More Robust Alphas
Investors often evaluate the performance of active managers based on their prior 3-5 year performance relative to their peers. Indeed, Morningstar's "Star" ratings methodology was based entirely on risk-adjusted past performance relative to a peer group. However, past relative performance itself is a poor indicator of future performance, with results of both outperforming and underperforming funds showing a strong tendency to revert to the mean. For an excellent study on the pitfalls of focusing solely on past performance when making investment decisions, see Charles Ellis, Murder on the Orient Express: The Mystery of Underperformance.
Jones and Wermers did, though, find research showing that past relative performance (eg. alpha) can shed light on future performance if it is suitably adjusted. But, how should it be adjusted?
What is Alpha? As a review, the concept of "alpha" comes from the "Security Characteristics Line" of the Capital Asset Pricing Model, which is written:
Eg. (Portfolio Return - Risk Free Rate) = Alpha + Beta x (Market Return - Risk Free Rate) + Error
In this equation, beta represents the role of overall market returns in the investment portfolio's return and alpha represents the portion of the portfolio's return that is not explained by the market beta (eg. the portion attributable to the manager's skill during that period).
Improvements on CAPM Single-Factor Alpha: While CAPM was a giant leap forward in investment theory, research in the ensuing half-century has identified additional risk factors other than the market return that are significant in explaining equity returns. The most famous additional factors are the "Fama-French" factors, which include Size (small-cap shares outperform large-cap) and Value (high "book value-to-price" shares outperform low "book value-to-price" shares). Other notable factors include Carhart's Momentum factor (outperformance or underperformance exhibits "persistence" over time) and the Liquidity factor of Pastor and Stambaugh.
Alpha Snapshot - Sep 2005 to Dec 2011: The research cited by Jones and Wermers found that adding additional factors into the SCL equation results in a more reliable estimate of a manager's skill, as the factors strip out - in the words of Richard Levich and Momtchil Pojarliev - the results of "beta grazing" in order to identify the manager's talent for "alpha hunting."
Put simply, managers with positive, statistically-significant multi-factor alphas are more likely to continue to be superior active managers in the future, a finding known as "persistence." Also, Jones and Wermers cited research that taking into account the "skewness" of a manager's alpha - or the likelihood that a fund would have periods of large positive or negative alpha - was helpful in identifying future outperformance. In short, a fund that achieves strong positive alpha and has positive skewness in its alpha distribution is a good bet as a potential SAM, especially if both readings are statistically significant.
To analyze the fifteen funds in the sample, I first downloaded monthly factor returns for the Market, Size, Value and Momentum factors from Kenneth French's website and for the Liquidity factor from Lubos Pastor's website. The last month available in the Market, Size, Value and Momentum factors at the time of analysis was October 2012 and the last month available in the Liquidity factor was December 2011. I then conducted five-factor multiple-regressions in MS Excel on the 15 CEFs for the October 2005 - December 2011 period and, as a check, the two index funds using the factors and the monthly total-return series discussed above. The readings for alpha and regression strength are in the table below, with statistically significant factor readings highlighted in green.
The readings from the Large-Cap and Small-Cap index funds suggest that the regressions were working. Both show extremely low annualized alphas, betas to the Market factor of almost 1.0, and very high regression significance (R-squared readings of 0.99), all of which would be expected with index funds. Also, the Small Cap fund has a significant positive beta to the Size factor and the Large Cap fund has a negative beta to the Size factor (see Factor table below), both of which make sense since the factor measures the outperformance of small-cap stocks versus large-cap stocks.
Regression strength, as measured by Adjusted R-squared, saw a wide range, though. The regressions were statistically significant for all funds except for the Eaton Vance Risk-Managed Diversified Equity Income . However, in contrast to the 0.99 R-squared readings on the index funds, regression fit ranged from a low of 0.27 for DNP to a high of 0.93 for Adams Express (NYSE:ADX), with only four other funds above 0.7. This suggests that, as a group, there are a lot of idiosyncratic elements playing into the CEF returns, potentially related to their leverage, option strategies, or active management styles.
The key takeaway here is that only five (5) of the 15 funds - GAB, CII, EVT, HTD, and DNP - showed a positive alpha as a whole over the October 2005 - December 2011 period. However, none of the alpha readings were statistically significant, providing some comfort to the negative alpha funds (and taking some away from the positive alpha funds).
Moving on from alpha, the regression readings for each of the factors are in the table below.
The average Market beta among the CEFs studied was 0.85, suggesting that the CEFs as a group provide some diversification and risk-reduction benefit relative to market returns. However, the CEFs also exhibited a wide range of statistically significant betas to the market factor, ranging from 0.35 for DNP Select Income (NYSE:DNP) to 1.37 for Eaton Vance Tax Advantaged Dividend (NYSE:EVT). On the whole, funds using large amounts of leverage (GAB , RVT, EVT) had higher betas to the market factors, though HTD and DNP also used significant amounts of leverage and had relatively low betas.
With respect to other factors, Royce Value Trust (NYSE:RVT) showed a significant positive beta to the Size factor, in line with expectations since RVT is a Small Blend fund. However, despite five funds showing up in the Large Value category (and Royce Value Trust having the word in its name), no fund had a significant positive beta to the Value factor, suggesting possible "style creep" among the funds. In fact, ETJ, which Morningstar categorizes as a Large Blend fund, also had a significant positive beta to the small cap (Size) factor.
Three funds - RVT, Gabelli Equity Trust (NYSE:GAB) and DNP - had significant negative betas to the Momentum factor, suggesting that they do better in environments where momentum strategies do poorly (eg. periods of sharp trend reversals around market tops or bottoms). Interestingly, one of those funds also had a significant positive beta to the Liquidity factor, suggesting that it does well when liquidity is plentiful, even though environments of plentiful liquidity are typically positive for momentum strategies.
Alpha Skew: To get a sense of the "skewness" of each fund's multi-factor alpha, or the likelihood that manager's would achieve significant positive or negative alpha, I took alpha readings across rolling one-year periods. Because the Liquidity factor had data only through December 2011, I calculated four-factor alphas (Market, Size, Value, and Momentum) rather than five-factor alphas for this part of the analysis. I then used the "Skew" function in Excel to calculate the skewness of rolling alpha readings over time. To review, a positively skewed distribution suggests a large number of small positive or negative returns and a few large positive returns, whereas a negatively skewed distribution suggests the opposite. The results are repeated in the table below.
Six of the CEFs had statistically significant negative skew in their rolling alpha readings and two - Blackrock Enhanced Dividend Achievers (NYSE:BDJ) and Adams Express - had significant positive skew. None of the funds with positive alphas over the full period also had statistically significant positive skew in those alphas over time. Indeed, only Gabelli Equity Trust had both a positive alpha reading and a positive alpha skew, though neither was statistically significant.
Robustness Test - Longer, Out of Sample Window: Five of the funds - GAB, RVT, DNP, ADX and ZR - had total return series going back to at least January 1995. In order to see whether a longer historical series would lead to a different conclusion, I performed the five-factor regressions on these funds for the period January 1995 to December 2011 (204 months). Despite the longer return series (204 months vs 75 months), none of the alpha readings were statistically significant. Four of the five funds had alphas for the same sign (eg. Positive or negative) for both periods, while ADX showed a mild positive alpha in the longer period and a negative alpha in the shorter period (see table below).
Conclusion: None of the funds had statistically significant positive alphas over the periods studied. And, only two showed positive, statistically significant skewness in their rolling four-factor alphas. Based on these results, we cannot conclude that any of the managers studied is a superior active manager (SAM).
Macroeconomic Forecasting - Can an IAM Become a SAM Under the Right Conditions?
Jones and Wermers cite a number of studies showing that the average active manager outperforms passive benchmarks during periods of economic stress and indicators of stress, including the credit default spread, term structure of interest rates and CBOE Volatility Index (VIX) can be used to select funds. The implication here is that inferior active managers (IAMs) can become SAMs during periods of economic uncertainty or, more generally, periods that are favorable to their particular style.
To try to evaluate how the CEFs performed during different market environments, I graphed the rolling four-factor alphas over time. Because the chart of 15 rolling alphas looked a bit like spaghetti, I grouped them by their 9/1/2012 Morningstar rating into "High" rated (Gold/Silver), "Medium" (Bronze) and "Low" rated (Neutral/Negative) buckets, as well as buckets by strategy (eg. Long Equity or Covered Call).
The data seems to support the idea that IAMs can become SAMs during different market environments. Indeed, the following chart shows significant trends in the annualized four-factor alphas for the long-equity and covered call strategies.
Long Equity CEFs seem to achieve the best relative performance during periods of positive market outcomes and plentiful liquidity, while experiencing significant drawdowns and negative alpha during periods of market stress. This tendency to perform worse during bad times may be due to their use of leverage; on average, borrowed funds account for about 18.5% of their portfolios. Covered call CEFs, by contrast, use no leverage and tend to perform better during bad times. Their call selling strategy provides option premium that helps cushion returns during bear markets, though also limits returns in bull markets as shares get called away below current market value.
The tables below show performance statistics for the Long Equity and Covered Call composites during the 20 worst and best months for the S&P 500 since February 1996.
Long Equity - Best in Good Times: The Long Equity composite starts in February 1996 and the Covered Call composite starts in July 2004, with both sub-periods covering a little more than eight years. During the most recent sub-period, the Long Equity composite outperformed the broad market during the best months and underperformed during the worst months, while the Covered Call composite did the opposite. During the earlier period, the Long Equity CEFs actually underperformed during good months and outperformed during bad months. However, the significant positive alpha generated by the composite - after adjusting for value, size and momentum factors - during the 1996-1998 and 1999-2001 periods suggests that the underperformance during good months in the earlier period was due to their value tilt amid a growth stock bubble.
Covered Call - Best During Bad Times: The tables also affirm the finding that covered call CEFs perform better during turbulent markets and worse during smooth markets. The chart below finds that, on average, covered call CEFs were more stable than Long Equity CEFs during the period studied, with lower standard deviation and market beta.
During the September 2005-December 2011 period as a whole, they saw lower annualized return and alpha as well. However, the lower relative return and alpha may be due to the fact that the relative performance of the buy-write (covered call) strategy moves in cycles, outperforming during periods of turbulence - denoted in the chart below with spikes in the VIX index - and underperforming during periods of calm.
Morningstar Fund Ratings: Interestingly, while Morningstar has been attempting to make its CEF ratings (and broader mutual fund ratings) more forward looking, the relative performance of Covered Call CEFs versus Long Equity CEFs seems to have been influencing the ratings. Indeed, while Covered Call CEFs made up 53% of the overall sample, they accounted for 78% of the Bronze/ Neutral/ Negative rated funds as of 9/1/2012.
Then, as the relative performance of Covered Call CEFs shot-up during the fall, driving the Bronze and Neutral/Negative composites sharply higher in the chart below, three of the Covered Call CEFs were upgraded (JSN to Silver and EOI and EOS from Neutral to Bronze). The chart below shows the historical relative performance of funds rated High (Gold/Silver), Medium (Bronze) and Low (Neutral/Negative) as of 9/1/2012.
It is difficult for active managers to achieve sustainable alpha in a mostly efficient market. Indeed, in a recent interview, Nobel-prize winner, Eugene Fama said: "After costs, only the top 3% of managers produce a return that indicates they have sufficient skill to just cover their costs, which means that going forward, and despite extraordinary past returns, even the top performers are expected to be only about as good as a low-cost passive index fund. The other 97% percent can be expected to do worse."
This article applies research surveyed recently by Jones and Wermers (see citation at top) finding that analysis of adjusted past performance, macro forecasting, fund characteristics and portfolio holdings can increase the odds of choosing superior active managers.
After adjusting past performance for multiple beta factors, I found that none of the CEFs studied achieved statistically significant alpha over a 75 month period - or, for five of the CEFs with longer return series, a 204 month period - and none had the desirable combination of positive alpha and significant positive skewness. I did find, though, strong trends in the relative performance of Long Equity and Covered Call CEFs, both versus each other and versus the market more broadly, suggesting that dynamic allocation between CEFs based on macro forecasting may add value.
In addition to past performance and macro forecasting, the authors found as well that publicly available information about the fund and its managers can help identify SAMs. Examples of characteristics that may help predict outperformance include the average standardized test scores in the manager's college or graduate programs, investment by managers in their fund, low cash balances, lower expense ratios, and manager training. Also, in addition to characteristics of the manager or fund, analysis of a fund's holdings over time can also predict future outperformance. Specifically, the authors found that funds with contrarian holdings and which take larger active positions (eg. more variance from their benchmark) outperform and that funds that engage in "window dressing" or that change risk profiles underperform.
I did not focus on the fund characteristics and portfolio holdings because of the difficulty in finding longer time series of data to work with. However, I have found the Morningstar CEF section to be an excellent source of snapshots of this type of information. Indeed, the Morningstar CEF ratings are partially based on information like management characteristics, corporate governance, the culture of the fund's parent, and the fund's investment process. However, the ratings, performance and subsequent upgrades of several covered call CEFs suggest that the Morningstar ratings may still be too heavily influenced by recent relative performance, which could be a topic for future research and potential improvement.
What to Do?
The lack of statistically significant positive alpha over a 6-16 year horizon does not mean that the active managers in question do not have superior skills, just that it is not possible to conclusively prove one way or another based on the returns data available. Indeed, Eugene Fama goes on to say: "Even over a 20-year period, the past performance of an actively managed fund has a ton of random noise that makes it difficult…to distinguish luck from skill."
Strategically, this makes it more difficult to allocate funds to Long Equity CEFs, which charge 1.4% average annual expenses and are competing with index ETFs like Vanguard Value, which charges a 0.10% expense ratio, or iShares Russell 1000 Value (NYSEARCA:IWD) which charges a 0.20% expense ratio.
A stronger case can be made for allocating to Covered Call CEFs, which fill a gap in investable covered call products. Indeed, while there are many index ETFs and mutual funds in the Large Value and Small Value space that charge under 0.20%, the only available covered call index product is the iPath CBOE S&P 500 Buy-Write Index ETN (NYSEARCA:BWV) which, at 0.75%, is much closer to the average Covered Call CEF's 1.1% expense ratio. Covered Call CEFs, due to their option overlays, provide additional diversification benefits as well, with a 0.77 correlation of monthly returns with the S&P 500 versus a 0.89 correlation for the Long Equity composite.
Tactically, given the outperformance of Covered Call CEFs during turbulent times, one strategy might be to own leveraged, Long Equity CEFs after spikes in the VIX, then switch to Covered Call CEFs when the VIX dips below its rolling 5-year average. A momentum strategy, which would evaluate a basket of CEFs each month and own the one with the best six-month performance, or a trend strategy that holds a small basket of CEFs and sells a holding if it ends the month below its 200-day moving average, may allow investors to capture the momentum and carry aspects of the CEF structure while also protecting against deep draw-downs. Finally, a value strategy that buys funds which are strong on the qualitative pillars (eg. management team, process, fund parent) but are currently trading at a particularly wide discount (or low premium) relative to history, may also add value as well. Momentum, trend and value strategies have been shown to work across a number of asset classes (see Ilmanen, 2011, or Ilmanen and Asness, 2012), though back-testing their implementation in the CEF space is the subject of a different article.