In a previous set of posts, we looked at how to think about and define asset classes – as they relate to bonds.
Now let’s apply the idea of defining an asset class in terms of characteristics to stocks.
From the point of view of theoretical finance, the main difference between stocks (equity) and bonds (debt) is that the scheduled cash flows from bonds are known and usually fixed, while there are usually no scheduled cash flows from stocks.
While we can use the scheduled cash flows of bonds as a good guide to putting a particular bond in a particular asset class, how to classify stocks into asset classes is a bit more subjective.
One method of making the classification for stocks is to look at so-called return factors.
Factor Models
In academic finance, a great deal of attention has been paid to identifying the factors that historically have explained the returns to stocks. The assumption, and the question, is whether the identified factors will continue to explain stock returns.
These factors, if they exist and if they are relatively stable, will help us define asset classes among stocks.
Identifying Factors
Equity factors are not necessarily easy to identify. In the academic world, the most widely held view is probably the Fama-French three factor model.
In this model, the three factors are:
- Market return
- Size
- “Value”
Fama and French published their findings of these three factors in a 1992 paper[1] that has become highly influential. The “Three Factor Model” is widely taught as fact in business schools across the US and around the world.
Of course the fact that it is widely taught and widely believed doesn’t mean it is correct. Or perhaps more significantly for investors, even if the paper was correct in 1992, the world may have changed.
Market Return
In “explaining” the return on a stock, Fama and French found that, on average, the best explanatory variable was the return on the overall market. You might be forgiven for thinking that this is a non-explanation.
Let me briefly explain what the Fama and French (and most similar research) findings represent.
In their Three Factor Model (which is what we’re talking about here), they started with the Capital Asset Pricing Model. This model assumes that on average, over time, the return to owning an asset is a function of the asset’s risk level. That risk level, in turn, is represented by a number—called “beta” and often written using the Greek letter b— which is calculated from the covariance of the asset and the variance of the market return.
Thus, the Capital Asset Pricing Model is a linear equation. A simplified version is:
- Expected Return = Risk-Free Rate + b x Market Return.
The b here is the famous “Beta” and represents the riskiness of the stock, or volatility compared to market volatility.
In the Three Factor Model, the first factor is the market return, just as in the Capital Asset Pricing Model.
Beta
Historically, b has done a reasonably good job of explaining returns. An illustration is provided by comparing the performance of the Nasdaq 100 against the “market” represented by the S&P 500.
Many studies have looked at b, and historically b has explained something like 70% of the return variation in US market. The inability of b alone to explain returns was one motive behind Fama and French introducing the three factor model.
Size
For example, the Fama-French “size” factor meant that the size of a company, measured by market capitalization. They found in the historical data a positive return to small size as compared to large size in the US markets.
The supposed reality of the size effect has given rise to an entire sub-industry of “small cap” funds and strategies. And the existence of that sub-industry, through marketing, appearances, and media, reinforces the common wisdom that size is a meaningful factor.
However, since the publication of the Fama-French paper, there has been no measurable size factor in the US.
In a 2018 paper, the authors observe:
“Size has never been a very strong effect… Given its pedigree, you would be forgiven for thinking that the size factor is one of the strongest and most robust anomalies in finance. In fact, it is one of the weakest.”[2]
And there is serious doubt about whether there is a meaningful size effect in international markets. Alquist, Israel and Moskowitz also looked at 24 non-US markets. They report:
“None of the t-statistics for the regional portfolios is close to being reliably positive, and most are, in fact, negative. Thus, we see no consistent evidence of a positive size premium in these other markets.”
Value
The third Fama-French factor is generally referred to as “value.” In their 1992 paper, Fama and French used a ratio of book value to market value to represent the value factor.
Somewhat confusingly, Fama and French calculated the value using the ratio book/market. In their world, a “value” stock had a high book to market ratio.
But in everyday language, most people talk about “value” stocks as those stocks with a low ratio of market value (or price) to book value. The information is exactly the same, it’s only a question of which number is on top when you write it as a fraction.
Thus, in common language, a “value” stock is one with a low ratio of price to book value.[3]
For example, as of this writing, Exxon has a price to book ratio of about 1.68, while Apple’s price to book ratio is about 44.9.
There is a great deal of historical evidence in the US that over long periods of time, value stocks, that is stocks with relatively low price to book ratios, have outperformed so-called growth stocks. For the purposes of most of these studies, growth stocks are stocks that have higher price to book ratios, and are therefore classified as not value stocks.
In international markets, the evidence is similar. In a 2017 paper, Elroy Dimson and Paul Marsh found that in 19 of the 23 European stock markets they looked at, there had been a “value premium” (i.e. value stocks outperformed growth stocks) from 1975 to 2016.[4]
Additional Factors
Notwithstanding the popularity of the three factor model, academics and practitioners continue to hunt for additional, new, or previously unknown factors that might explain past returns, and even predict future returns. We’ll look at a few of these other factors.
Momentum in Stocks
Momentum, which is tendency of a stock that has been going up to continue to go up, or vice-versa, has been widely claimed to be a positive, measurable “factor” in the Fama-French factor-model view of stock returns.
While investors have probably been investing using an implicit theory of momentum forever, the first major paper on momentum was not published until 1993.[5]
Looking at historical data, the authors, Sheridan Titman and Narasimhan Jegadeesh, found that stocks that had done the best over the past 3 to 12 month periods tended to outperform other stocks in the following 3 to 12 months. They believed that profitable trading strategies could be built on that insight (ignoring taxes).
Probably at least a thousand papers since then have looked at the issue. And there are probably hundreds of mutual funds that incorporate momentum strategies.
There are even now momentum indexes. MSCI, for example, lists 22 such indexes.[6]
Does Momentum Work?
Assuming that momentum in stocks has worked in the past, does it still work, and should we expect it to continue to work?
In a 2021 interview,[7] Titman expressed his view that momentum is no longer as robust as it was before he published his results. There might be several reasons for that, including the fact that as more and more quant investors (i.e. professional investors and funds that explicitly use quantitative and statistical tools to make investment decisions) put more capital into exploiting momentum, they collectively make the markets more efficient. In other words, the more people there are to buy what’s been going up, and sell what’s been going down, the faster the mispricing gets corrected, and the harder it is to profit.
Titman also observed that momentum hasn’t worked well since about 2000. He said, “We do have weaker momentum after 2000… a big part of the weaker momentum after 2000 is kind of a really bad period for momentum, between 2007 and 2009.”
For most investors, a strategy that did “really bad” between 2007 and 2009, when the overall market lost 50% of its value, would be “really, really bad.” When the overall market is getting crushed is exactly the worst time to be doing significantly worse than the market.
The Factor Zoo
One issue with defining asset classes by factors is that there are a tremendous number of factors that have been proposed in the finance literature. By some counts, over 300 distinct factors have been proposed.
In a 2020 paper,[8] the authors examined 99 proposed factors. They conclude:
“While most of these new factors are shown to be redundant relative to the existing factors, a few have statistically significant explanatory power beyond the hundreds of factors proposed in the past.”
In case you’re interested, email us at [email protected] to receive the full list of 99 factors discussed in the paper.
Statistical Significance
Statistical significance is not necessarily the same as real world significance. Statistical significance is the result of a calculation of the probability that some result occurred by chance.
The conventional standard is that if the probability of a result occurring by chance is less than 5%, the result is said to be “statistically significant” or “significant at the 5% level.”
One consequence of this approach to statistical significance is that if you examine enough data, even if the data is random, by definition about 5% of the time you will find “statistically significant” results.
Given certain reasonable assumptions, this statistical fact means that if you selected 100 factors randomly, and even if none of those factors had true explanatory power, purely because of the meaning of randomness and the way the statistical analysis works, you would expect about five of these “factors” to show statistical significance.
In their paper, Feng et. al. started with 150 factors. They used sophisticated econometric techniques to eliminate about 50 of those, and then further sophisticated techniques to fit their model and identify “significant” factors. They identified 14 significant factors.
This result is unlikely to arise purely by chance. However, that does not mean that any or all of the 14 factors are meaningful in a real world investing sense. To be useful, the factors must be investable and profitable after taking into account transactions costs, and if applicable taxes. For most investors, that eliminates factors that produce small effects and factors that require high trading activity.
Which Factors are Useful to Investors?
For most factors that are believed to have some degree of investability, funds exist.
A complete discussion of factors would fill a book (and has filled many). So our discussion here will be very limited.
If you are going to invest using funds, separately managed accounts, or other commercially provided account management, your choice of factors will be determined by the available funds. The main factors represented by funds are:
- Size
- Value
- Momentum (e.g. MTUM, PDP)
- Low Volatility (e.g. SPLV, USMV)
- Quality (QUAL, SPHQ)
- Dividend Yield (VIG, DVY)
Size
There are hundreds of funds positioned by size. Unfortunately, the evidence for size being a systematically exploitable factor is very limited.
Value
There are also hundreds of value funds. Value has an excellent historical track record. It also has regular and long periods of severe underperformance.
Momentum
Momentum seems to work better in theory than in practice. In the world of stocks, momentum has not produced outstanding results. One of the best performers seems to be the MTUM fund, which for the ten years ending 12/31/24 produced returns of 13.16%.[9]
This ten year return compares to the return on Vanguard’s S&P 500 ETF (ticker VOO) over the same period of 13.16%. However, MUTM produced significantly higher volatility. And it is likely that MUTM would have produced lower after-tax returns for taxable investors.
For reasons that I have never understood, momentum does appear to have “worked” in a wide range of futures markets over the past fifty years. We intend to address that in a future post.
Low Volatility
Many “factors” disappear after they are “discovered.” This appears to have happened with the low volatility factor. For the ten years ending 12/31/24, the Invesco Low Volatility S&P 500 ETF (ticker SPLV) trailed the S&P 500 badly, producing a return of 9.24%, over 3% per year less than the S&P 500.[10]
Quality
The factor called “quality” may be defined various way. MSCI, in its USA Sector-Neutral Quality Index, screens the S&P 500 companies for “return on equity”, “earnings consistency” and “debt-to-equity.” The iShares ETF with ticker QUAL follows this MSCI quality index. That fund returned 12.9% per year for the ten years ending 12/31/24, just slightly lagging the S&P 500.
Dividend Yield
Though it may surprise many people, dividend yield is not an independent factor affecting stock returns. Here’s why: Everything else equal, when a company pays a dollar per share dividend, the value of each share drops by one dollar. The dividend itself does not create value.[11]
A great deal of empirical research has found that there is no “dividend” factor separate from the value factor. That is, high dividend stocks also tend to be value stocks, and when the value factor is taken into account, dividends provide no additional explanatory power.
For taxable investors planning to reinvest the dividends, dividends create a tax drag that is unambiguously worse than if the company had reinvested the earnings directly instead of paying them out.
Despite this fact, it is estimated that there are over 3000 dividend oriented funds.
Which Factors Matter?
You will have to determine whether you believe any of these factors, or others, are likely to provide a significant performance edge going forward.
Based on my reading of the research, of these many factors, perhaps only value has both stood the test of time, and has a relatively easy-to-understand rationale.
Next Steps
To speak with us further, call us at 703 437 9720, email us back at [email protected] to set up a time to talk, or click here to request an advisor guide.
[1] The Cross-Section of Expected Stock Returns, The Journal of Finance, Volume 47, Issue 2, Jun 1992
[2] Alquist, Israel & Moskowitz, Fact, Fiction, and the Size Effect, Journal of Portfolio Management, Fall, 2018.
[3] Price to book usually looks at the price of a share of stock and compares it to the book value of a share. This saves the effort of recomputing the market cap of the company every time the share price moves.
[4] Factor-Based Investing: The Long-Term Evidence, Journal of Portfolio Management, 2017
[5] Jegadeesh, Narasimhan, and Sheridan Titman (1993). “Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency.” Journal of Finance, Vol. 48, No. 1
[6] https://www.msci.com/indexes/group/momentum-indexes
[7] https://www.youtube.com/watch?v=dfGuhlQ-CLQ
[8] Taming the Factor Zoo, Feng, Giglio & Xiu, Journal of Finance, June 2020
[9] iShares MSCI USA Momentum Factor ETF, https://www.ishares.com/us/products/251614/ishares-msci-usa-momentum-factor-etf
[10] https://www.invesco.com/us/financial-products/etfs/product-detail?audienceType=Investor&ticker=SPLV
[11] This point was the topic of a 1961 paper Dividend Policy, Growth, and The Valuation of Shares by Franco Modigliani and Merton Miller, in the Journal of Business, Oct, 1961
Leave a Reply