The first paper of Fama & French on equity risk factors came out 26 years ago: The Cross-Section of Expected Stock Returns, June 1992.This document, which criticised single-factor SLB models (Sharpe, Lintner, Black), was the starting point for intensive research on the determining factors of stock returns.
The multifactorial approach is a criticism of the notion of beta, but it is globally an attack on the hypothesis of market efficiency, because the defined factors (size and book-to-price in the case of Fama and French) are considered as anomalies, as market inefficiencies that allow the passive extraction of excess returns.
Today, many risk factors have been identified, and even exist listed future contracts that deal with the most known: size, value, quality, risk and momentum. Here, roughly, is the evolution of risk factors over time:
It is difficult to justify why these factors present excess returns, and the precise topic of our discussion is: do these factors really present positive excess returns?
Of course, the answer is in the question. The new generation of investors lap up these strategies while considering that such anomalies are an acquired fact. However, this is not the case. There is no fundamental justification for the fact that grouping securities on a common criterion would allow passive, positive returns. This is simply an extrapolation from the past.
Fama and French were working on American data for the 1963-90 period, and because the value style outperformed the growth style, they concluded that the book-to-price ratio was a significant risk factor.
We know today what has happened since: the growth style has kept outperforming. We have no data from before 1975, but we show on the previous graph the performance of the MSCI USA indexes over the period from January 1975 to May 2018. We have the proof here of the fact that “past performance is no guide to the future”…
A further remark concerns the reason for the supposed “outperformance” of value (that is, low price-to-book stocks). Today, it is acknowledged that low-valued securities carry an abnormally high risk premium, and that it is this higher risk that supports better performance. This analysis – which is today very consensual – is not found in Fama and French: they did not mention it in their 1992-93 papers. They remained at the time very factual and only mentioned the difference in profitability:
“[…] book-to-market equity is related to relative profitability. On average, low-BE/ME firms have persistently high earnings and high-BE/ME firms have persistently poor earnings. The evidence here then suggests that HML, the difference between the returns on high- and low-BE/ME stocks, captures variation through time in a risk factor that is related to relative earnings performance.”
The factors that are popular today are simply the factors that have performed the best in the past years.
All of this seems to us very fragile. We can also sound intelligent by inventing new factors. Let’s do this! What about alphabetical ranking? We can invent a factor that groups companies according to their initials. We show below the performance of securities which first letter of their social denomination is comprised between A and I, against those of securities from I to Z. In the United States, the excess return of the first group is 1.1% per year since 2010.
All that remains now is to find a story to explain why we should buy securities from the first letters of the alphabet. For this, there are experts in the financial sector. “Rationalisation” is the official sport of our industry; many claim to be able to explain everything. This is a very well-known psychological bias: humans want to understand and refuse the element of chance (this element was once called the divine). Everything must be explainable.
Everyone can recognise something in the graph below (the euro, real rates, a slope of the bond yield curve, etc. …), but the truth is that there is no reason why this series should be correlated with an economic series.
What we are trying to say is that there is no reason that, in the long run, a static way of grouping securities ought to generate alpha. The book-to-price factor of Fama and French turned out to be only a rationalisation of a past period. Indeed, the securities with a weak price-to-book ratio do not out-perform because of a higher implicit risk premium, or depending on the economic cycle. No. The securities with a weak price-to-book ratio have scattered performance, and if we group them together, we notice that sometimes they outperform, and sometimes they underperform.
Clearly, we believe in market cycles, which is to say that we believe in the instability of factors. A factor that has worked well in one period tends to work less well in another.
The best-known groupings are country and sector groupings. From this point of view, we can also identify “anomalies” and most notably: 1) the near-structural outperformance of US equity indices and 2) that of the technological sector. Are they risk factors?
Some would answer that these two assets exhibit the growth factor. Contrary to Fama and French’s observation, it is the growth style that has outperformed; there is then no surprise in seeing such leadership from the US and from Tech.
This is not quite right. We show below – the graph on the left – the performance of US Tech stocks labelled growth compared with those labelled value (Russell methodology). The leadership of growth isn’t obvious. The graph on the right keeps the same Y-axis and shows the performance of US Tech stocks labelled value relative to that of the S&P500. The leadership is much more pronounced, and we can then conclude, the Tech sector has been in itself a very powerful factor, more so than the growth/value distinction.
What to make of this? These days are characterised by the constant outperformance of American technological securities, no matter the level of interest rates, of the dollar, or of growth. This trade - which is, it seems, consensual – is an anomaly since its performance adjusted for volatility seems so exceptional. Is this a structural phenomenon (innovation is a key factor of endogenous growth models), or simply just a cycle that is longer and stronger than the others?
In other words, should we buy risk factors with the idea that these “anomalies” are perennial, or should we sell them with the idea that the market regimes are too unstable to present reliable excess returns?
The “tech” factor is casting serious doubts. The monopoly situation acquired by the American giants from the sectors of information and technology is creating an unprecedented situation: is it really different this time?
The comments and opinions expressed in this article are those of the authors and not necessarily those of Candriam.