Correlations of Factor ETFs
August 28, 2013
Fairholme Fund Reopens
BlackRock has recently introduced a set of four iShares ETFs that follow factor indices. They are:
The first three of these ETFs debuted on April 16, 2013, while the fourth one three months later. Therefore, as of this writing, there are only 91 and 28 trading day data available for these ETFs, respectively. Traditionally, at least three years worth of data (a minimum of 36 monthly data points) are required to calculate a return correlation between two investments. However, it may be helpful to take an early look on how the return correlations among these ETFs and the iShares Core S&P 500 ETF (IVV) are shaping up so far:
Since daily returns are assumed to contain a substantial amount of “noise,” and the observation period is very limited, the above figures certainly cannot be considered very reliable. However, there is an early indication that the majority of correlations are lower than 0.6, which should aid in portfolio diversification. A research paper from BlackRock shows that idealized zero-net-investment factor portfolios constructed using Fama-French approach* can have much lower long-term correlations:
*MktRf = market, SMB = size, HML = value, CME = quality.
Only time will tell whether these new factor ETFs provide low inter-correlations and sufficient returns to truly benefit an investment portfolio. However, early signs are encouraging.
August 28, 2013
On Factor Investing
A number of articles from The Wall Street Journal, Barron’s, and InvestmentNews covered a recent reopening of the Fairholme Fund, closed to new investors since the end of February 2013. Should potential investors care?
In a continuation of a prior Alpholio™ post, here is an updated cumulative RealAlpha™ chart for the fund:
The trend of a relatively flat cumulative RealAlpha™ since early 2012 has persisted through the end of July 2013. Here are the updated reference weights for the fund:
Although the equivalent position in the Vanguard Financials ETF (VFH) decreased from the peak of almost 97% in April to about 57% in July, it still dominates the fund’s portfolio. At 22%, the PowerShares Dynamic Market Portfolio ETF (PWC) has the highest second weight, followed by the Vanguard Energy ETF (VDE; 14%).
According to the most recent filing, as of May 31, 2013 the fund held almost 90% of its assets in just ten positions:
This highly concentrated portfolio led to a much higher volatility of the fund compared to that of the reference portfolio. As the above analysis demonstrates, existing investors were hardly compensated for the elevated risk carried by the fund. To most prospective ones, the fund’s reopening should be a non-event.
August 22, 2013
No Turnaround in Sight
Quite a few of recent industry articles focus on factor investing.
Rick Ferri has a two-part article on the topic, with the first part covering the history of multi-factor models, and the second part delving into more practical considerations. According to the author, factor investing has the following benefits:
- Outsized performance (returns) compared to a single-factor (market) portfolio, e.g. as historically observed for small-cap stocks
- Combination of uncorrelated factors leads to a higher risk-adjusted performance of the portfolio
- Intellectual enrichment and academic stimulation that stems from studying of multi-factor models.
The author also points out the disadvantages of factor investing:
- Cost of factor vehicles [although the actual expense ratio of VTI is 0.05% and not the cited 0.15%]
- Historical lack of risk premium persistence of factors such as size.
The author comes to the same conclusions about factor-based products from DFA as Alpholio™ already did in one of the previous posts. However, in his zeal to defend pure market-based factor investing, the author confuses terms:
Finally, tracking error is the name give [sic] to a strategy that falls short of a market benchmark. It could mean the downfall for many multifactor investors.
A tracking error of a factor investment vehicle is in reference to the theoretical index of this vehicle and not to the market benchmark. For example, a momentum factor does not purport to track the S&P 500® index.
The author goes on to combat the term “smart beta” in his post on InvestmentNews. Despite a more pragmatic approach from Arnott, the author quotes noted academics (Sharpe, Fama, and French) to instill the message of beta purity, :
I believe the original definitions are best left unchanged. Beta is non-diversifiable market risk, other return dimensions are defined as additional risk factors, and putting these risks together in a portfolio is multifactor investing.
In the end, does it really matter if factor coefficients in a multiple regression are labeled beta1, beta2, …, or beta, gamma, delta, …? Sure, “smart beta” may sound like a marketing gimmick from fund providers to peddle their latest products, but it is a simple way of conveying the difference of these factors from plain market-cap based ones.
Finally, an article in Morningstar focuses on the scientific background of multifactor investing. It also presents two points of view on factors: from the perspectives of efficient and not perfectly rational market. The author leans toward the second interpretation, which is supposedly supported by the following “evidence:”
It’s also hard to reconcile them all [factors] as representing risk because if you lump them all together, you get an eerily smooth return stream.
The reason for this smoothing is that excess returns of factors are generally un- or low-correlated, and thus tend to cancel “bumps” in portfolio returns. This does not mean that each factor does not represent risk.
The author then proceeds to cover the problem of redefinition of alpha, which results from the introduction of multiple factors, and concludes that from his perspective such an adjustment is inappropriate because it “redefines outperformance:”
From my perspective, the mountains of studies purporting to show that active equity managers can’t beat the market are really showing that much of their excess returns can be replicated by a handful of factor strategies.
Regardless of semantics and opinions, if a manager’s excess returns can be replicated by cheaper and readily accessible instruments (such as factor ETFs), then there is no need to pay the excess management fee.
From Alpholio™’s perspective, all of the above discussions are academic. Whether or not factors have persistent risk premia, market is efficient or not perfectly rational, what truly matters is whether or not active management adds value over a reference portfolio after all fees have been taken into account. Factor ETFs provide yet another set of potential explanatory variables that squeeze the alpha to its essence, the RealAlpha™.
August 19, 2013
Are Stocks Overvalued?
An article in Barron’s describes a “turnaround” of the $537M Loomis Sayles Growth fund under the new manager in the last three years. Apparently, the manager is searching for deep value within growth sectors:
Despite his obvious mandate for growth, Hamzaogullari takes a surprisingly value-oriented perspective when scrutinizing potential investments. Sure, he looks for companies with a sustainable competitive advantage, good profitability, and quality management. But he only buys stocks that are trading at a 40% discount to his estimate of the company’s intrinsic value.
Let’s analyze the fund’s performance from the Alpholio™ perspective. Here is the cumulative RealAlpha™ chart for the fund since early 2005:
The chart clearly shows that the downward trend in cumulative RealAlpha™ was unbroken since the current manager took over the fund in May 2010. An a truly risk-adjusted basis, the fund did not generate any value for its shareholders. This is compounded by the maximum 5.75% front load of the fund, which is not taken into account in the above analysis.
The reference weights chart depicts the major investment themes of the fund in the same analysis period:
The fund’s current top three equivalent positions are in the Vanguard Health Care ETF (VHT; weight of 22.8%), iShares North American Tech ETF (IGM; 20.2%), and iShares Morningstar Mid-Cap Growth ETF (JKH; 13.6%). As of June 30, 2013, the fund had 35 equity holdings. While the manager personally visited each company, this evidently did not add much value.
Here is how the fund fared in the three years to July 31, 2013 against its stated primary benchmark, the Russell 1000® Growth Index, for which a proxy is the iShares Russell 1000 Growth Index ETF (IWF); data from Morningstar:
|Loomis Sayles Growth
|iShares Russell 1000 Growth ETF
Despite a slightly higher return, the fund had a higher volatility and, consequently, a lower Sharpe Ratio than a practical realization of its benchmark. Again, the return figure does not take into account the front load of the fund; if it did, the annualized return would be 15.55%, further depressing the Sharpe Ratio.
In sum, while it may be true that the current manager did better than his predecessor, the fund’s risk-adjusted performance continued to be unimpressive. Since early 2006, investors would have been better off with a reference portfolio of ETFs calculated by Alpholio™. So far, there is no indication of a change in this long-term trend.
August 18, 2013
According to an article in The Wall Street Journal, stocks are currently overvalued. First, the author compares the price-to-earnings (P/E) ratio of the S&P 500® index based on reported (i.e. net income) trailing twelve month (TTM) earnings to a 140-year median value. Then, the author admits that forward-looking, i.e. next twelve month (NTM), earnings estimates predict operating income that is higher than the net income, which suppresses the P/E ratio. To overcome this discrepancy, the author extends the average relation of the NTM P/E being lower than the TTM P/E by 24%, as observed from 1976 to 2003, to the entire 140-year historical period. (The 24% discount encompasses three factors: the predicted growth of earnings from TTM to NTM, difference between operating and reported earnings, and over-optimism in earnings forecasts.) This sleight of hand enables the author to conclude that in either case, stocks are currently overvalued by about 25% compared to a historical P/E median.
This mixing of reported and operating earnings, coupled with an arbitrary extension of a medium-term observation to a very-long-term historical period, leads to dubious conclusions.
Here is an alternative point of view from the July 22, 2013 edition of the S&P The Outlook:
From a fundamental perspective, S&P 500 valuations continue to look attractive. As of July 12, the S&P 500 is trading at 15.9 times trailing 12-month operating results, including the June 2013 EPS results projected by Capital IQ consensus estimates. This multiple represents an 11% discount to the median P/E of 17.8 times since Wall Street started looking at operating results in 1988. What’s more, the market is trading at a multiple of 15.2 times and 13.7 times trailing 12-month operating EPS for year-end 2013 and 2014 results, respectively.
Why does S&P look at operating results instead of reported earnings? Because of distortions caused by large, non-recurring, non-cash expenditures, and also by time misalignment of reported tax expenses with actual tax payments. Indeed, as the article’s title suggests, P/E ratios aren’t always what they seem.
August 13, 2013
Timing Hot Funds
An article from MarketWatch dismisses the latest fundamentally-indexed exchange-traded funds (ETFs) announced by Charles Schwab Investment Management as “different” as opposed to “better” mousetraps.
The question for investors is whether they are the rodent, getting snapped up by the contraptions that financial inventors are creating.
Catchy metaphors aside, according to a note from Reuters
San Francisco-based Schwab, long a leader in selling mutual funds to investors, already has five mutual funds based on fundamental indexing. Those five funds, which launched in 2007, had $4.5 billion in total assets under management as of June 30.
From Schwab’s website
Fundamental indexes use fundamental measures of company scale and success—such as adjusted sales, retained operating cash flow, and dividends + buybacks—to construct the indexes. These fundamental factors address the inherent bias of traditional indexes, which are based on market capitalization and can give too much weight to overpriced securities and too little weight to underpriced securities.
Hence the tilt towards value investing (i.e. higher weights of underpriced securities) inherent in these funds. While some research shows that fundamental investing can add a statistically significant value, apparently this is limited to a few specific indices and regions.
As the article states
Just as studies have shown that value investing holds a slight edge over growth investing over the long haul, fundamental investing has shown an ability to outperform its traditional cap-weighted peers, although that history is not long once you throw out back-testing (results “proven” by looking backward and seeing how the strategy might have performed had it existed years ago).
Since Schwab’s fundamentally-indexed mutual funds have been in existence for over six years now, and that period spanned a significant market downturn, it is worthwhile to take a look at their historical risk-adjusted performance, as measured by the trailing five-year Sharpe Ratio (all data from Morningstar):
In all cases, the Sharpe Ratio of the fundamentally-indexed fund was greater than or equal to that of the ETF that follows the best-fit index for the fund. While past performance does not guarantee future results, the above data certainly support the introduction of ETFs based on the same (and one more, the U.S. Broad Market) fundamental indices. They may be better mousetraps after all.
August 9, 2013
Waiting for Trauma to End
An article from MarketWatch claims that
The problem for most investors is that they have a tough time recognizing when a manager like Miller or Kenneth Heebner of the CGM Funds — another famous manager with a feast-or-famine record — is heating up, and only recognize it after the feed has started.
Heebner’s CGM Focus (CGMFX), for example, has the best record in the business among large-cap blend funds over the last decade, according to Morningstar Inc., with an annualized average gain of 10.75%. Over the last half of that decade, however, the fund is dead last in the category, with an annualized loss of 5.45%.
In short, it started the last decade hot, then fell off sharply.
To gain some insights about the performance of this fund, let’s take a look at CGM Focus (CGMFX) from the Alpholio™ perspective. First, a buy-sell signal derived from the smoothed cumulative RealAlpha™ for the fund since early 2005:
The fund significantly outperformed its reference portfolio from mid-2007 to mid-2008, but subsequently began to heavily underperform. (The latter period was punctuated by a pseudo-rebound in 2010, which generated a temporary buy signal. This illustrates that no performance prediction mechanism, even that of Alpholio™, is perfect.) To augment the performance analysis, the following chart shows components and weights of the reference exchange-traded product (ETP) portfolio for the fund:
The equivalent ETP positions with the highest weights in the mid-2007 to mid-2008 period were iShares North American Natural Resources ETF (IGE), iShares MSCI Brazil Capped ETF (EWZ), and Vanguard Materials ETF (VAW). The collective weight of these three positions averaged about 79% in that period, which clearly indicates that the fund was riding the boom in commodity prices. The peak of the cumulative RealAlpha™ coincided with the top of that boom in the summer of 2008. While the bet on commodities worked out, heavy trading saddled the fund’s investors with a large short-term capital gain distribution at the end of 2007.
The article further states that
Over the last year, however, Heebner has been back on top, gaining nearly 50% over the last 12 months.
This recovery was not reflected in the fund’s cumulative RealAlpha™, which was largely flat from mid-2012 till present. How is that possible? The fund had large equivalent positions in the Vanguard REIT ETF (VNQ) and the aforementioned VAW. In the last three months, the fund heavily tilted toward small-cap equities represented by iShares S&P Small-Cap Growth ETF (IJT). So much for being a “large-cap blend” fund and using the S&P 500® index as a relevant benchmark.
Timing of any mutual fund characterized by large directional bets and rapid trading (the most recent turnover ratio for CGMFX was 360%) is certainly very challenging, but not entirely impossible, as the Alpholio™ analysis demonstrates. In addition to the automatically-generated buy-sell signals, investors should pay close attention to the overall trend of the cumulative RealAlpha™ curve, as well as the specific investment themes (market capitalization, sectors, industries, countries/regions, etc.) pursued by the fund’s manager. These rapidly changing bets, in turn, make it necessary to continually modify the benchmark to elicit the true risk-adjusted performance of the fund.
August 5, 2013
Trust Goes Down, Fees Go Up
An article in The Wall Street Journal describes the struggle of the $2.4B Fidelity® New Millennium fund to hold on to assets under management:
From the start of 2008 through the end of last year, the fund saw net outflows totaling $518 million, according to Morningstar. This year, through late June, the fund has taken in $9 million.
These asset losses are attributed to high correlations of stock returns:
In the long shadow of the financial crisis, global economic woes led many stocks to trade in lock step, making it hard for stock-fund managers to find stocks that would differentiate their returns from the swings in the broader market.
However, there may be light at the end of the tunnel:
Now Mr. Roth is hoping that both investors and the markets more broadly are in the final stages of working through the “trauma” of 2008. It’s encouraging that there is “less concern about systemic risk” in the U.S. economy and more focus on where we are in the economic cycle, says the manager. As long as correlations between stocks decline—meaning stocks move less in unison with the broader market—the current slow-growth environment for the U.S. economy can present a host of opportunities for stock pickers.
The problem is that the bull market has been going on for almost four and a half years now. The following chart show the cumulative RealAlpha™ for the fund in that and the prior period:
Despite this year’s breakthrough, from early 2005 through 2012 the cumulative RealAlpha™ for the fund oscillated in a roughly +/- 5% range. As a result, the overall statistics for the fund are rather unimpressive:
Although it is true that the S&P 500® component correlation hit a six-year low at the turn of the last year, this correlation was also low in prior periods during which the performance of the fund was not stellar:
So, while a low correlation supports active management efforts, it does not guarantee that the fund will outperform. In the end
Mr. Roth acknowledges, however, that investors need to see results. “The proof is in the numbers,” he says.
Alpholio™ could not agree more. That said, there is an issue of selecting a proper benchmark for the fund, i.e. one that would dynamically match its actual holdings instead of automatically default to a large-cap market proxy:
During his tenure, the fund has beaten the [S&P 500®] index 98% of the time on a rolling three-year basis, according to Morningstar Inc.
Since the beginning of 2013, the fund exhibited substantial equivalent positions in iShares Russell 2000 Growth ETF (IWO; average weight of about 26%) and iShares Morningstar Mid-Cap Growth ETF (JKH; 23%). This indicates the fund’s recent tilt towards small- and mid-cap growth stocks to boost its performance.
August 2, 2013
Eat Your Own Cooking
InvestmentNews reports that an annual survey of retail investors by State Street found that only 15% (down from about 33% last year) of respondents trust financial advisers as a group. The key issues are: performance, unbiased and high-quality advice, and transparency. Apparently, investors…
…don’t believe the fees they’re paying are commensurate with the return on their investments.
Granted, the lack of trust is evidently coupled with a lack of understanding of the finance industry or a sufficient interest in investments. However, the chief issue of performance remains.
So, how to fix this problem? By increasing fees, of course. That is what Bank of America just did by planning to raise fees on its managed-account (flat-fee) platforms at Merrill Lynch.
The current minimum fee schedule for equities on the most popular Merrill Lynch Personal Advisor (MLPA) platform with $152B under management is:
The new rate schedule will be:
Therefore, MLPA clients will face fee increases of 54-60%. Over 14,000 ML advisers have to implement that change by the end of 2015; the only way to reduce the fee hike for clients will be to cut their own compensation. Naturally, ML advisers are worried. So should be the clients. Luckily, Alpholio™ can easily show these investors whether advisers earn their fees by generating a sufficient RealAlpha™ on the managed accounts.
August 1, 2013
An article in The Financial Times shows the following statistics of mutual fund industry professionals who invest their own assets in low-cost passive vehicles, such as index funds and exchange-traded products (ETPs):
|Portion of Assets in Passive Products
||Percentage of Responders*
* Numbers do not add up to 100 due to rounding
The main reason why about two-thirds of these professionals have a sizable part of their assets in passive instruments is said to be the compliance with fund industry regulations that prohibit trading in individual stocks or bonds. However, investment in actively-managed mutual funds is not prohibited, which is evidenced by one-fifth of professionals having only non-passive products in their private portfolios. It really looks like most fund professionals tout “alpha-generating” products to others by day, but personally shun them by night. How revealing.