Primer: the factor drivers of investment returns
Many of the concepts behind factor investing are nearly as old as investing itself. Much newer is the idea of bringing them together systematically. This is transforming the way investors think of portfolio construction, and the reasons are clear: factor investing offers diversification, transparency and economy. We review the underlying concepts and show how they underpin a number of apparently disparate developments from smart beta to alternative risk premia. Factor investing provides important new tools for the investor, but, like all tools, understanding remains critical to results.
Why factor investing?
Traditionally, investors have thought of the world in terms of asset classes, such as equities and bonds. Factor investing looks beneath asset classes to allocate according to the factors that drive risk and return. Some of these so-called “risk premia” may correspond more or less directly to an asset class, such as the equity risk premium. Other asset classes represent a combination of factors. For instance, corporate bonds combine long-term interest rate risk (itself divisible into inflation and duration risk) and credit risk, and these two sources of risk are separable, at least in principle1. The aim here is to achieve more stable diversification than asset class allocation alone. The risk premia underlying assets may have a more stable relationship than that between asset classes, since the mix of risks and expected returns in an asset class can vary. In October 2008, for instance, corporate bond risks and expected returns were dominated by credit risk, while by 2016 long-term interest rates (duration) were more important (Figure 1). These variations affect the risk, return and diversification of portfolios considerably. Factor investing seeks more direct exposure to the drivers of return, which have shown more stable relationships with each other and therefore more consistent diversification.
What are the factors?
The factors we have discussed so far are traditional elements of portfolios. But much of the recent excitement comes from less traditional factors, which are increasingly used as portfolio building blocks. The best known are equity strategies such as value, size, quality, low volatility and momentum. These sorts of “dynamic” factors, widely marketed as “smart beta”, require active management to maintain exposure, as a stock that was good value last month, for example, may be overpriced this month. These differ from traditional factors in another way: not everyone can hold them. If one investor holds a value strategy, all the other equity investors in the world must be collectively underweight value, relative to the market.
So here is our first note of caution:
If factors are well known, either they may not continue to outperform, or there must be another reason why some choose not to use them. At their most basic, factors are rulebased strategies that can generate outperformance over lengthy periods. A more helpful way to analyse them is to look at how the basic rule can be turned into different strategies of varying sophistication and risks (Figure 2). Suppose we have a rule that ranks the desirability of stocks and then use it to build a portfolio: we have created a form of “smart beta”. The term derives from the idea that we have captured something systematic about stock returns over and above general market moves. “Alpha”, by contrast, is taken to represent above-market returns deriving purely from stock selection.
A key point about this portfolio is that the single largest driver of its returns is still the broad equity market as a whole. We hope that our rule tilts the risk and return of the portfolio in our favour, but if equity markets fall, for example, our portfolio will also tend to fall. If we are a bit more sophisticated, and use our factor not only to buy the desirable stocks, but to short the undesirable stocks, we have entered the world of long-short equity hedge funds or, more generically, of “alternative risk premia”. Of course, building a portfolio that includes shorts is operationally more difficult than building a long-only portfolio. And if the factor underperforms, the portfolio will not merely underperform the market, but actually lose money.
Hence our second cautionary note about factor investing:
Since all known factors experience periods of underperformance, seeking a diversified portfolio of factor exposures is a much better idea than focusing on just one or two.
Bringing these arguments together, we can say that factor investing aims to:
1. enhance investment returns through historically demonstrated systematic strategies,
2. improve diversification by breaking down risk into its underlying components, and
3. reduce fees, since strategies can be implemented more efficiently. While the concept and intentions are straightforward, the execution is not quite so simple.
One of the canons of practical finance is that there are four or five fundamental factors, discovered in stock markets, that can be applied equally well in other markets, such as those for bonds and currencies. These factors are value, momentum, low volatility, and often size or quality. Conventional wisdom is, however, wrong in two important ways. First, there are many more than four or five plausible factors. Second, factor investing is just as relevant to other asset classes as to equities, but often involves quite different factors.
The notion that there are just four or five simply-described equity factors reflects the seminal findings of Eugene Fama and Kenneth French, then of the University of Chicago, starting in the early 1990s. In a series of research articles2, Fama and French showed that the returns from many different equity portfolios could be mostly explained using returns from the market, along with size and value. This was a major development in finance, as the significance of size and value shredded the previous academic view that market returns were the only systematically important and consistently rewarded risk in equity markets. (This was the basis of the famous capital asset pricing model developed in the early to mid 1960s.)3
Fama and French influenced a generation of students and practitioners, and their work laid the foundation for some of the most successful strategies of recent decades. But the students they influenced understandably were not content with accepting that two academics had discovered every conceivably important factor in equity markets. Their search for additional factors was spectacularly successful. Over 300 different equity “factors” have now been identified in the academic literature, with around one new factor having been discovered each month on average over the past 10 years.
The academic quest now is to bring order to the “factor zoo” (a term coined by John Cochrane, also formerly of the University of Chicago).4 In time, a new taxonomy may well emerge, but there is no particular reason to expect that it will resemble the remarkable simplicity of the model Fama and French propounded almost a quarter of a century ago.
It is no accident that equities have been a hotbed for new factors. Research is facilitated by freely available data, relatively straightforward return modelling and highly dispersed returns from many different stocks, allowing for the empirical examination of many different factor ideas. Achieving the same results for government bonds, commodities, currencies or, indeed, any other asset class is much harder. Few should be surprised, therefore, that nothing like 300 factors able to generate outper formance have been found anywhere beyond equities.
Nevertheless, systematic strategies have been established across all asset classes. One type of factor that is common across many is “carry” – roughly speaking, the amount an investor is paid to hold an asset, independently of price changes. This idea was originally applied to currency investing, where carry corresponds to a given currency’s shortterm interest rate. In equities, dividend yield is often identified as “carry”. Similarly, momentum – the idea that recent winners go on winning – seems to be an empirical regularity across many asset classes and geographies. Other types of factors seem asset-class specific – for example, roll-down is the profit from holding a bond if an upward sloping yield curve remains fixed.
The existence of factors across many asset classes is important because it increases the possible opportunities for factor investing. And because of the low historical correlation betweenfactors across asset classes, a multiasset approach to factor investing should provide investors with greater opportunities for diversification benefits than focusing solely on equities.
Are factors generic?
The concept of value investing dates back at least to Benjamin Graham and David Dodd’s classic work in 19345. Fama and French chose to formalise the idea by measuring value as the ratio of the book value of a company’s assets to its market value. There are, however, equally useful ways of measuring “value”, including, for example, earnings relative to a company’s market value. In one sense, the choice of measurement does not matter much – groups of stocks measured by different ideas of value tend to move up and down in similar ways; that is, portfolios formed from different types of value measurements tend to be quite highly correlated. In a much more important way, however, the choice of measurement is critical: cumulative returns of these portfolios can be very different.
Performance can differ even across portfolios of stocks selected from the same universe, by the same firm, using similar methodologies. Figure 3 shows cumulative returns from three different value indices created by MSCI, a leading index provider. The monthly returns from these indices are more than 90 % correlated, and their “active” returns (those that differ from the capitalization-weighted benchmark) are around 60 % correlated. But over 19 years, the Value Weighted index has added only around 10 % to the performance of the benchmark, while the Enhanced Value index has added over 110 %. So even small differences in index construction and portfolio management can lead to enormous differences in performance.
Investors may be wondering: if there can be such widely different returns produced by the best known factor from the same index provider, working on the same set of stocks, is factor investing truly systematic and reliable? The answer is that it can be, but it depends on the implementation: some factor implementations are mo e efficient than others. Because factor investing is quantitative, it is possible to study the different building blocks that go into factor construction and find out which approaches are more likely to produce acceptable returns in the long run.
An important example of this is how the stocks in a portfolio are weighted.If, for example, the selected stocks are capitalization-weighted, the resulting index is likely to have little effective exposure to the factor. This is largely because returns from many factors are more pronounced in smaller stocks, which are likely to be less well represented in the portfolio.
Again, we can use value as an example. Figure 4 shows the differences in returns from investing in different-sized stocks in four different value portfolios over the past 19 years. We divided each portfolio by size and compared the performance of the smallest 25 % of stocks with the largest 25 %. In each case, buying “value” stocks and selling expensive stocks was handsomely rewarded among the smallest stocks, but generated almost no return at all mong the largest stocks. This is evident across many factors, where returns are often more pronounced in smaller stocks than larger ones.
Clearly, just as with any other strategy, investors must pay close attention to implementation. We have found plenty of evidence of persistent systematic strategies, but that does not mean that any strategy labelled “systematic” is as good as any other. Fortunately for investors, careful analysis can illuminate differences among strategies in advance.
Who should hold factors?
One other note of caution on factors relates to their d namic nature. As suggested earlier, as a particular factor becomes widely used, either its previous outperformance is likely to tail off or there must be some other reason why non-users adopt an opposite strategy. Does this imply that factors become less attractive as their popularity increases? Here we can draw lessons from the past. One study considered 100 published equity factors and found longer-term evidence for about a quarter. So while the majority were not reliable, dozens of publicly-available equity factors were still found to produce significant positive returns, even after they became well known.
Some of these factors are likely to be of little use for most investors. Others, however, may continue to provide attractive returns. Many of the latter have probably been known about, albeit with different labels, for a long time. After all, investors were seeking “cheap” stocks long before Fama and French appeared in the early 1990s. So even if investment flows into “value” factors are new, flows to managers seeking to exploit value as an idea certainly are not. This is another reason to think that such well-known factors will probably not be competed away.
If a well-known factor can persist, why wouldn’t everyone invest in it? In truth, investors have different goals and needs. Consider those saving for retirement. Some are at the start of their working lives and will not need retirement savings for years; others are close to retirement, with little opportunity to make up for investment losses. Suppose a factor performs well in most markets, but does badly when prices plunge. This risk may be acceptable for young savers, who have a long investment horizon, but older investors should avoid it, given the possible impact of losses. This points younger investors towards factors like size and value, and older ones towards low volatility and quality.
Similarly, most contributors to defined contribution pension schemes probably want relatively straightforward investments, implemented cheaply, with easily-understood risks. For these investors, long-only factor investing is a sensible way to try to provide extra return. At the other end of the spectrum, sophisticated institutions or family offices may be willing to take on more opaque risks, particularly if they help to diversify their existing portfolio more effectively. For those investors, taking on intensified factor exposure using alternative risk premia may make more sense.
Factor investing is a powerful tool for managing investments. By breaking down assets into risk premia, it can provide greater transparency of portfolio construction and greater control over the drivers of risk and return. Dynamic risk premia, such as value or momentum, can diversify the sources of return in a portfolio beyond traditional assets. But successful implementation of factor investing requires care and skill. Creating a portfolio of dynamic risk premia means sorting those with lasting value from those that are ephemeral or illusory, concentrating factor exposure on securities where it is rewarded, broadening exposure across asset classes, and staying on the leading edge of financial research. Meeting these challenges should enable investors of all types – from the least engaged to the most sophisticated – to gain access to new sources of return in the form that best fits their investment needs.
1 For instance, by hedging out the duration risk, or investing in credit default swaps. For more on this subject, see “Putting a premium on risk”, Investment Horizons, issue 1, 2014.
2 E.g. “The Cross-Section of Expected Stock Returns”, Eugene F. Fama and Kenneth R. French, Journal of Finance, vol. XLVII, June 1992; “Common risk factors in the returns on stocks and bonds”, Journal of Financial Economics, vol. 33, February 1993; and “Size and Book-to-Market Factors in Earnings and Returns”, Journal of Finance, vol. L, March 1995.
3 For a more recent discussion of this see “The Capital Asset Pricing Model: Theory and Evidence”, EF Fama and KR French, Journal of Economic Perspectives, vol. 18, summer 2004.
4 “Presidential Address: Discount Rates”, John H Cochrane, President of American Finance Association 2010, Journal of Finance, vol. LXVI, August 2011.
5 Security Analysis, Benjamin Graham and David Dodd, Whittlesey House (McGraw-Hill), 1934.