A recent study explores how Fama-French factors perform across various stages of recession and recovery.
Arnav Sheth and Tee Lim contribute to the literature on the performance of factor-based investment strategies through their December 2017 study, “Fama-French Factors and Business Cycles.”
In it, they examined the behavior of six Fama-French factors—market beta (MKT), size (SMB), value (HML), momentum (MOM), investment (CMA) and profitability (RMW)—across business cycles, splitting the business cycles into four separate stages: recession, early-stage recovery, late-stage recovery and very-late-stage recovery.
The recession stage used in the study is the same as deﬁned by the National Bureau of Economic Research (NBER); early-stage recovery is deﬁned as up to 24 months after the recession stage; late-stage recovery is deﬁned as up to 24 months after early-stage recovery; and very-late-stage recovery is the entire period after late stage recovery but before the next recession.
The authors then examined the factors’ behavior in each of these periods by looking at their cumulative returns. They also looked at the factors’ behavior after a yield-curve inversion starts and ends, as the relationship between yield-curve inversions and recessions has been well-explored. Finally, they tested the predictive power of the term spread on the NBER recession indicator.
For the economic stages part of their analysis, Sheth and Lim’s data set covers the period April 1953 through September 2015. For their analysis of the Fama-French factors’ behavior following a yield-curve inversion, the authors’ data starts in 1966. Based on the average lead time established in prior research, they estimated a lag of 12 months as the time it takes for mainstream markets to price in yield-curve inversions.
There have been 10 NBER-designated recessions of varying durations since April 1953. The shortest recession, which lasted six months, started in January 1980, and the longest recession, which lasted 18 months, started in December 2007. The median recession length was 10 months. Thus, the authors measured the cumulative returns for 10 months following the start of an NBER-designated recession, and then took the average of the cumulative returns for each factor across the 10 recessions.
The term spread between three-month Treasury bills and 10-year Treasury notes is used to determine when an inversion starts—e.g., when the three-month rate exceeds the 10-year rate. There have been eight inversions and seven recessions since 1966. An inverted yield curve has successfully forecasted, within six quarters, six of those seven recessions. A false positive occurred in 1966 when an inversion was not followed by a recession within six quarters.
The most recent inversion, in January 2006, also resulted in another false positive. It was followed by a recession—but in December 2007, more than six quarters later. While both false positives did not successfully predict a recession within six quarters, it was nevertheless true that, in both cases, the economy slowed.
|State Of Economy
|September 1966 – February 1967
|Economic slowdown (1967)
|December 1968 – February 1970
|Recession (December 1969)
|June 1973 – November 1974
|Recession (November 1973)
|November 1978 – May 1980
|Recession (January 1980)
|October 1980 – September 1981
|Recession (July 1981)
|May 1989 – August 1989
|Recession (July 1990)
|July 2000 – January 2001
|Recession (March 2001)
|January 2006 – August 2007
|Recession (December 2007)
The authors’ analysis showed an increase in the probability of a recession occurring 12 months later, as the term spread becomes less than zero.
They next examined the cumulative factor returns across economic stages.
Cumulative Returns for 6 Factors Across Economic Stages (%)
As you can see, the best performer in a recession is CMA, the investment factor, which delivered an 18.3% cumulative return. Firms that invest conservatively outperform ﬁrms that invest aggressively in a recession, a logical finding. This performance did not last, however, as the cumulative returns for these ﬁrms deteriorate moving through the stages.
The second-best performer in a recession was HML, the value factor, which delivered a 12.5% cumulative return. Value ﬁrms outperform growth ﬁrms, perhaps a surprising finding (although markets are forward-looking). The value factor’s performance in recessions is exceeded by its results in early stage recovery, where it had a cumulative return of 17.1%, and then its performance tapers off.
The third-best performer in a recession is MOM, the momentum factor, which delivered an 11.2% cumulative return. In fact, momentum, though not the highest performer, remains consistently positive across all four stages. The market-beta factor underperforms in a recession, but far outperforms all the other factors in each of the other three stages, delivering outsized cumulative returns in comparison.
The most important takeaway should be that, because different factors outperform at different stages, diversification across factors—not concentrating risk in a single factor—is the prudent strategy. We can see the benefits of diversifying across factors in the following table, which shows the correlation of returns among the six factors across all economic stages.
Correlation Across Economic Stages
With the exception of the correlation between the value and investment factors, the correlations are all low-to-negative. However, correlations are not static. They change depending on the economic regime.
For example, during a recession, the correlation between HML and RMW switches from 0.1 to -0.3, indicating that value and proﬁtability have a small positive correlation, on average, but are negatively correlated in a recession. This demonstrates the benefits to value investors of adding exposure to the profitability factor. Similarly, the correlation between MOM and CMA switches from 0.0 to 0.4, indicating a fairly strong positive correlation between momentum and investment during a recession.
In early-stage recovery, the negative correlation of -0.2 that exists between SMB and HML changes to 0.0, indicating that although size and value are negatively correlated overall, during early-stage recovery, they are independent of each other.
Similarly, during that stage, the low correlation of 0.1 between HML and RMW becomes negatively correlated at -0.3. Finally, the -0.2 correlation between SMB and CMA reverses to 0.2, indicating that size and investment are positively correlated in early-stage recovery.
In late-stage recovery, HML changes its relationship with both MOM and RMW. HML and MOM are correlated negatively at -0.2 overall, but in late-stage recovery, they are correlated at 0.2. Similarly, HML and RMW have a small positive correlation of 0.1 overall, but it changes to a sizable negative correlation of -0.4 in late-stage recovery.
Finally, in very-late-stage recovery, the momentum factor has three of its relationships switch signs. First, size and momentum, which are basically uncorrelated for the entire period, have a fairly strong positive relationship at 0.3. Similarly, market and momentum are negatively correlated at -0.1 overall, but, in very-late-stage recovery, this turns positive to 0.1. Lastly, the proﬁtability and investment factors have a weak negative correlation of -0.1 overall, but switch to a correlation of 0.2 in very-late-stage recovery.
After Yield-Curve Inversion
Following the beginning of an inversion, the market-beta factor tended to be relatively ﬂat until about a year, followed by a steep decline and a subsequent strong recovery. The momentum factor also tended to be relatively ﬂat until about a year, followed by a strong recovery. The SMB factor tended to be strong for about a year, then ﬂattened, fell and rose again. The HML factor tended to be strong for about 10 months and then ﬂattened for a few months before rising again. The RMW factor tended to be strong for about a year, dropped for about six months, and then rose again. The CMA factor tended to be strong for about nine months, remained ﬂat for about six months thereafter, and then rose again.
Following the end of an inversion, the market-beta factor tended to be strongly up for about a year, ﬂat for a few months, and continue its climb thereafter. The momentum factor tended to be strong for about a year, ﬂattened, rose and then fell. The SMB factor tended to be strong during the entire period. The HML factor tended to be ﬂat for about a year and then rose. The RMW factor tended to be strong for about a year and a half and then declined. The CMA factor tended to be strong during the entire period.
Based on these findings, the authors concluded there are differences in returns across each of the economic stages, as well as around yield curve inversions, and that factors perform differently in each stage. Additionally, because of the predictive power of the yield curve, investors can exploit those differences.
Before you jump to any conclusions, there are a few things to consider:
- These findings neither take into account any implementation costs nor consider taxes.
- Despite the well-known predictive power of yield-curve inversions, there is little-to-no evidence that active managers have been successful at timing factors.
- We have very few data points, which may explain why the authors failed to provide any t-statistics to demonstrate the statistical significance of the data.
- We have no out-of-sample (non-U.S.) data to evaluate.
In addition, at least based on relative factor valuations, research from Wei Dei of Dimensional Fund Advisors found that “there is no compelling evidence supporting a robust relation between interest rate changes and the size, value and profitability premiums.”
Furthermore, in the study “Contrarian Factor Timing Is Deceptively Difficult,” which appeared in the 2017 special issue of The Journal of Portfolio Management, Cliff Asness, Swati Chandra, Antti Ilmanen and Ronen Israel found “lackluster results” when investigating the impact of value timing (in other words, whether dynamic allocations can improve the performance of a diversified, multistyle portfolio). They write: “Strategic diversification turns out to be a tough benchmark to beat.”
As tempting as the proposition might be, there doesn’t seem to be enough convincing evidence that a style-timing strategy can be expected to be profitable going forward. With that said, if you are going to “sin” by trying to time factors, I’d recommend following Asness’s advice to “sin a little.”
The bottom line is that the most prudent strategy is for investors to build portfolios that are strategically (as opposed to tactically) diversified across factors that show persistence in their premiums, have low correlation to other factors, are pervasive around the globe and across asset classes, have intuitive reasons to believe the premiums should persist (whether behavioral-based or risk-based) and are implementable (meaning they survive transaction costs).
This commentary originally appeared February 2 on ETF.com
By clicking on any of the links above, you acknowledge that they are solely for your convenience, and do not necessarily imply any affiliations, sponsorships, endorsements or representations whatsoever by us regarding third-party Web sites. We are not responsible for the content, availability or privacy policies of these sites, and shall not be responsible or liable for any information, opinions, advice, products or services available on or through them.
The opinions expressed by featured authors are their own and may not accurately reflect those of the BAM ALLIANCE. This article is for general information only and is not intended to serve as specific financial, accounting or tax advice.
© 2018, The BAM ALLIANCE