By Samuel Lee
This article was published in the July 2014 issue of Morningstar ETFInvestor. Download a complimentary copy of ETFInvestor here.
It’s been a little over three years since iShares Core High Dividend (NYSEARCA:HDV) launched. Now that iShares has slashed its fee to 0.12% from a rich 0.40%, this fund is worth another look. As a rough rule, I don’t like to buy U.S. stock exchange-traded funds with expense ratios higher than 0.25%. Thanks to bone-crunching competition among the titans of the ETF industry, there are plenty of good products that charge less than that.
It’s hard to eke out an edge in the U.S. stock market, the major leagues of investing. Even with the best factor funds, the best you can reasonably hope for is an extra percentage point or two of return--risk-adjusted, of course. The fund’s new expense ratio makes it competitive with many passive options and lowers the hurdle for outperformance.
This fund tracks the Morningstar Dividend Yield Focus Index, a basket of 75 high-yield U.S. stocks weighted by their aggregate dividends, a scheme pioneered by WisdomTree. (The ETF research team had nothing to do with its construction.) It is unusual in several respects, the foremost being that some of its key inputs require Morningstar equity analysts’ qualitative judgments.
The main qualitative screen relies on Warren Buffett’s notion of the economic moat: a sustainable advantage that makes life difficult if not impossible for a business’ competitors. A moat allows a business to reap above-market returns on its invested capital. Coca-Cola (NYSE:KO), with its iconic brand, globe-spanning distribution network, and economies of scale, is a classic example. Morningstar equity analysts assign all stocks they cover a Morningstar Economic Moat Rating: wide, narrow, or none. Wide indicates that a business can reap above-average returns on invested capital for 20 years. None is self-explanatory, and narrow is in between. No-moat stocks get the boot. This reduces the field to quality firms, which tend to have fat, steady profits and low leverage.
A secondary qualitative screen relies on the analyst-assigned Morningstar Uncertainty Rating, which can take on values of low, medium, high, very high, or, in rare circumstances, extreme. Firms with steady, predictable sales in slow-changing industries and low financial and operating leverage are assigned low uncertainty ratings. Firms that don’t have these qualities are assigned higher ratings. The rating was introduced in early 2008 to formalize the intuition that any forecast of a business’s prospects should include information about its potential dispersion. Stocks with very high or extreme ratings are excluded from the index.
There’s some active management here. Humans are weeding out stocks, but with a light touch. Almost 90% of the S&P 500’s assets are in wide- or narrow-moat stocks. Of the 960 or so U.S.-listed stocks with ratings, around 680 pass these screens. Many of the excluded stocks don’t pay dividends in the first place. A few do sport high yields, but they’re hardly durable firms. Shares of no-moat Seadrill (NYSE:SDRL), for example, yield almost 10%, but analyst Stephen Ellis writes that the firm “is almost recklessly aggressive with its leveraged balance sheet.”
Human judgment is supplemented with an algorithmic screen. Each firm is assigned a quantitative Morningstar Distance to Default score that estimates the likelihood of default. It’s based on financial economist Robert Merton’s insight that a firm’s equity can be modeled as a call option on its assets with a strike price equal to the face value of the firm’s debt. The irony is Buffett would scoff at the Merton model’s assumptions, particularly its use of price volatility to value an ultra-long-dated option (and rightly so, in my opinion). In practice, the distance-to-default model seems harmless: It penalizes stocks with lots of debt and volatile stock prices, weeding out firms that wouldn’t have passed the qualitative moat and uncertainty ratings in the first place. If a stock is analyst-rated, then it has to have a distance-to-default score in the top 50% of its sector to make the cut. If it isn’t, it needs a distance-to-default score in the top 30%. While this rule allows stocks not covered by analysts to make it into the index, in practice the majority of the index’s assets will be in stocks that have passed the qualitative moat and uncertainty screens.
Stocks that pass these hurdles are what you’d expect: stable, profitable firms in boring businesses or quality firms. However, as an investor, you shouldn’t care about quality for its own sake. Wonderful firms for the most part trade at wonderful prices. And, of course, investors want yield. The index sorts all the eligible stocks by their indicated dividend yield (calculated by annualizing the latest dividend and dividing by current price) and buys the top 75. The stocks are weighted by the aggregate share of indicated dividends. This procedure boosts yield and overweights cheap stocks.
Most dividend funds reconstitute annually. HDV does it quarterly. While its index has a buffer rule that allows current holdings to stay in if they’re in the top 100 stocks by yield after running the aforementioned gauntlet, its turnover can be high. Last year, HDV turned over 43% of its assets, and the year before that it turned over 28%.
That the buttoned-up iShares licensed this index is somewhat out of character for it. A look at HDV’s holdings shows that it is unusually concentrated. Its top 10 stocks make up a little more than 58% of assets. This is the second-highest value of all large-cap U.S.-equity ETFs, excluding funds of funds, and it’s in the top 2.5% of all actively managed U.S. large-value mutual funds. This fund isn’t going to behave like others.
HDV bets big on individual stocks and bets frequently. This may raise concerns about transaction costs and tax efficiency. Because HDV is so concentrated, some of its trades may take up a big chunk of a stock’s daily trading volume. The biggest a stock can get is 10%. As of this writing, HDV has $4 billion in assets, meaning it can in theory trade $400 million worth of stock in a single day--meaningful enough to cause significant price impact in all but the most liquid stocks. If the fund were to grow much bigger, I’d be worried about market impact and would hesitate to own it.
Believe it or not, all this trading is unlikely to generate capital gains. High-turnover equity ETFs can be tax-efficient with some optimizations and use of the in-kind creation/redemption mechanism. iShares is particularly good at this. iShares Morningstar Small-Cap (NYSEARCA:JKJ), one of iShares’ highest-turnover broad equity funds, has averaged 63% turnover without distributing any capital gains since its 2004 inception.
HDV is a quantitative trading strategy. Its prospective returns don’t just come from the valuation of its current holdings but also how effectively HDV trades stocks in the future. The same could be said for all factor strategies, including size and value tilts. Investors often think of small-cap and value stocks as static groups. They’re not. Since inception more than a decade ago, Vanguard Small-Cap Value ETF (NYSEARCA:VBR) has averaged turnover of 30% as firms were dropped and added. Eugene Fama and Kenneth French studied how the movement of stocks across style buckets contributed to the value and size premiums. The size effect was mostly driven by tiny firms that grew into giants. The value effect was largely driven by growth stocks that lagged and dropped into blend or value buckets and value stocks that did well and migrated to the blend and growth buckets. A small portion of the value effect was driven by stocks that stayed in the value bucket doing better than stocks that stayed in the growth bucket.
When assessing simple quant strategies, or factor strategies, don’t look at recent performance. The funds with the best recent performance often hold the most overvalued assets. At the very least, one should assess performance over a full market cycle, but even that isn’t great information. Oakmark Select’s (OAKLX) Bill Nygren posted strong, positive returns in the post-dot-com downturn but fell harder than the market during the global financial crisis. Careful studies show that historical performance, naively used, is an almost useless predictor of future performance. Day-to-day, month-to-month, and even year-to-year stock returns are largely random.
I couldn’t tell you the three-, five-, or 10-year returns of any of the funds I own. I don’t look at them. What I can tell you is their factor loadings. The factor approach offers the best hope for predicting the behavior of quant strategies.
If you’ve read "Understanding Factor Models," the following will be mostly review. If not, don’t worry if you don’t grasp the details. I summarize the practical implications later on.
The Fama-French three-factor model describes stock returns as combinations of three factors: market (Mkt-RF, or market minus risk-free-rate), value (HML, or high book/market minus low book/market), and size (SMB, small minus big). The Carhart model adds a momentum factor (UMD, or up minus down). The factors in these models are the periodic returns of factor-mimicking portfolios.
The market factor is simply the return of the U.S. stock market minus the return of the one-month Treasury bill. Value is a portfolio that goes long high book/market stocks and shorts low book/market stocks, reconstituted yearly. Momentum is a portfolio that goes long stocks with the highest 12-month returns, excluding the most recent month, and shorts the stocks with the lowest returns, reconstituted monthly. Size is a portfolio that goes long small-cap stocks and shorts large-cap stocks, reconstituted yearly.
The chart below shows the cumulative returns of the momentum, value, and size factors in the United States plotted on a logarithmic scale. As you can see, momentum has the highest returns, with the rare but brutal drawdown; value has done OK; and size has been weak and inconsistent.
Value and momentum have been found to work almost everywhere. Size was thought to be a distinct factor, but it turns out most of its returns can be explained by its greater exposure to value and market factors, and it doesn’t seem to exist in foreign markets. It is the weakest and least interesting of the factors, but it hangs around because of inertia and smart marketing by Dimensional Fund Advisors. Lots of quant hedge funds employ value and momentum, but few bother with size.
Running a fund’s historical returns through a factor model produces several key data points: how well the factors explain the fund’s returns; the fund’s factor loadings and the likelihoods they came about by chance; and any unexplained return, or alpha, and the likelihood it came about by chance. If you’ve taken introductory statistics in college, the Fama-French and Carhart models are simply multiple linear regressions, where the factors are the independent variables that seek to explain the target portfolio’s periodic returns.
When predicting a fund’s behavior, we care mostly about its factor loadings and their stability.
If you run the Morningstar Dividend Yield Focus Index’s historical monthly returns, which start in 2005, through the Carhart model, you get the output below.
- source: Morningstar Analysts
Here’s how you interpret it.
The Mkt-RF coefficient, or loading, of 0.67 means that for each percentage-point change in the market, the index’s return is expected to change 0.67 percentage points in the same direction, holding all other factors constant. In other words, the index’s market beta is 0.67.
The SMB loading of negative 0.33 means that for each percentage point that small-cap stocks beat large-cap stocks, the index is expected to lag by 0.33
The positive HML loading of 0.24 means ... well, you get the point. The index is moderately tilted to value stocks.
The close-to-zero UMD loading means the index doesn’t have an appreciable tilt to momentum or anti-momentum stocks.
The annualized alpha value of 4.43% is the annual outperformance of the index after controlling for the above factors. It is statistically significant, meaning it’s unlikely to have arisen from chance.
The adjusted R-squared of 0.73 means the model can explain 73% of the index return’s variance. This is not a high R-squared by the standards of equity portfolios. It’s common to see R-squared values of 0.90 or more. The relatively low value of R-squared and the statistically and economically significant alpha suggest the index is exposed to other factors or taking on lots of stock-specific risk, or both.
Recent research has uncovered more factors that explain stock returns. The most exciting is quality or profitability. Firms that have fat and stable margins, low debt, high return on equity, and high payout yields have been found to outperform companies without those characteristics. Like value and momentum, quality has generated excess returns almost everywhere. The quality factor can be defined in several ways, but I find Cliff Asness, Andrea Frazzini, and Lasse Pedersen’s most compelling. I’ve written before about their quality-minus-junk factor, or QMJ. It’s defined as going long on stocks that score high on a composite index composed of profitability, growth, payout, and safety, and shorting stocks that score low on that composite index.
The chart shows the cumulated returns of QMJ. The factor’s returns are eerily consistent, suggesting we’ve either struck gold or the back-test is data-mined. I think the truth falls somewhere in between. Keep in mind hedge funds have been trading on quality signals for a long time, so it’s possible--almost certain, in fact--that the strategy’s future returns will be lower than its past returns.
Once we throw QMJ into our factor model, R-squared goes up and alpha disappears. The index’s historical back-tested outperformance is entirely due to its massive loading to the quality factor. [Note: Since this article was written, Frazzini updated the publicly available QMJ series; HDV's live factor loadings can be seen here and are consistent with the back-tested loadings.]
- source: Morningstar Analysts
These results make sense. QMJ’s construction favors high-yield stocks with low leverage, low earnings variability, and fat margins--the kind of stocks HDV’s methodology favors. The high loadings come at the cost of high portfolio concentration and frequent turnover.
In order to check for factor stability, I’ve plotted rolling three-year loadings from May-end 2008 to December-end 2012. As you can see, the index’s quality and value characteristics weakened in recent years and its momentum loading picked up, likely because of investors’ grasping for yield. Unfortunately, the QMJ data series end in 2012, so we can’t extend the analysis to more recent periods. Despite the recent weakening in QMJ and HML loadings, there’s no reason to believe the fund won’t maintain respectable loadings.
Overall, the results are largely reassuring. Because HDV’s back-tested index goes back only to 2005, it’s hard to be confident that the strategy is the real deal. The fact that HDV exploits QMJ means that it targets the same kind of stocks that have produced excess returns over many decades in the U.S. and abroad.
Now we can make a forecast of how the fund can be expected to perform in the future. We start with the factor’s historical returns, or premiums. Unfortunately, U.S. stock valuations are high, so I’ve reduced the expected equity premium to 3%. I’ve also cut all the other factor premiums because it’s unrealistic to think that these strategies will continue to work as well as they have in the past.
Lots of funds are targeting these factors. Besides, it’s better to be conservative and pleasantly surprised than the reverse.
If you believe value and quality strategies will continue to pay off, then a reasonable forecast for HDV’s before-fee returns above the market can range from 1% to 3%. However, HDV turns over its portfolio frequently and may suffer from significant market impact. If the fund gets big, its expected excess return will shrink.
iShares Core High Dividend uses Morningstar analyst discretion to weed out stocks without sustainable competitive advantages and with murky futures.
It also uses a quantitative scoring model that penalizes firms with lots of debt and volatile stock prices.
HDV is among the most concentrated funds out there, and it can turn over its portfolio quite a bit. It’s idiosyncratic and won’t track conventional indexes.
The payoff for the complexity and concentration is among the deepest exposures to value and quality factors I’ve seen among ETFs.
I estimate HDV can be reasonably expected to return 1%-3% more than the market over a full market cycle, but there’s a lot of uncertainty to this forecast and it may not realize an excess return if its size grows significantly.
Disclosure: Morningstar, Inc. licenses its indexes to institutions for a variety of reasons, including the creation of investment products and the benchmarking of existing products. When licensing indexes for the creation or benchmarking of investment products, Morningstar receives fees that are mainly based on fund assets under management. As of Sept. 30, 2012, AlphaPro Management, BlackRock Asset Management, First Asset, First Trust, Invesco, Merrill Lynch, Northern Trust, Nuveen, and Van Eck license one or more Morningstar indexes for this purpose. These investment products are not sponsored, issued, marketed, or sold by Morningstar. Morningstar does not make any representation regarding the advisability of investing in any investment product based on or benchmarked against a Morningstar index.