The US market soared in November, producing one of the highest monthly returns since at least 1982. With regard to factor returns, the month started out fairly slowly. But things changed on November 9, when it started to look like the pandemic could end someday. We wrote about the impact the news from Pfizer on Momentum’s return Pfizer Vaccine Announcement Puts Momentum in the ICU and the impact on active risk Active Risk Pops on the Heels of a Big Day for Factors.
Now that the month has ended, we revisited what happened to US factors during the full month and find that the whole month was extremely unusual. In general, we find that most factors’ saw some of the best — or some of the worst — monthly returns in the history of the US model back to 1982. For many, the higher-than-average magnitude of returns meant an increase in volatility as well. And a few factors saw sizable changes in correlation, which along with the higher volatility is likely to have impacted active risk.
Dividend Yield, Exchange Rate Sensitivity, Liquidity, Market Sensitivity, Value and Volatility ALL experienced monthly returns that were among the highest 10% in their almost-40-year history, and Leverage and Size came close. In contrast, Earnings Yield, Medium-Term Momentum and Profitability returns were all within the lowest 5% relative to history. Only Growth had an “average” return. In addition, Volatility had its highest monthly return ever while Market Sensitivity fell just short of that benchmark.
The preponderance of unusual magnitude returns in the month is in itself unusual. There were only 3 months since 1982 in which 11 of 12 factors saw returns in the top or bottom 15% (the other two were January 2001 and August 1998), and only 10 months in which 10 of the factors breached those limits.
Exhibit 1. Factor Returns and Historical Rankings, US4 Model, November 2020
For Momentum, Size and Profitability, much of the shortfall came in the early part of the month mentioned earlier. Volatility and Market Sensitivity, on the other hand, climbed steadily for most of the month, only giving back some of those gains in the last few days.
Exhibit 2. Cumulative Factor Returns, US4 Model, November 2020
The impact on factor volatility and correlation was a little less stark than the return rankings. Short-horizon volatility for Medium-Term Momentum and Dividend Yield jumped by more than 30% from the end of October to the end of November, and Growth and Market Sensitivity saw increases of more than 20%. Most other factors’ volatility rose somewhat less, although many had high volatility to begin with, in the 90th percentile or above relative to history.
Finally, risk model correlation changes were not of sufficient magnitude to be considered statistically significant, but they may still have an impact on active risk. Most notably, Profitability became more correlated with Growth and Medium-Term Momentum as all three factors produced negative returns in the month, and its correlation with the better-performing Dividend Yield, Market Sensitivity and Value factors fell.
Exhibit 3. Changes in Volatility and Correlation, US 4 (Short-Horizon) Model, Oct 30 – Nov 30, 2020
Conclusion
November was an unusual month for factor returns, with some performing unusually well and others producing higher-magnitude-than-expected negative returns. In some cases, these high returns were probably welcome for investors who bet on them, such as those for Dividend Yield and Value. For others, the returns were likely to have been extremely disappointing, such as Medium-Term Momentum and Profitability. And it was a particularly bad month for investors who have a low-risk or small-cap stance, as high Market Sensitivity, high Volatility and larger stocks soared relative to their lower-risk or smaller-cap counterparts. Finally, some of the factor returns may have just confounded investors, as any unintentional bets would have had a large contribution. Exchange Rate Sensitivity, Leverage and Liquidity, factors that generally do not offer a long-term risk premium, all experienced highly positive returns.
Note: We often discuss “normalized” returns, which we define as the actual return minus the long-term average return, divided by the standard deviation. We usually use the factor risk forecast at the beginning of the period for the divisor. Although November returns were large in magnitude, only Dividend Yield, Market Sensitivity and Volatility fell outside of a two-standard-deviation range. As noted above, factor volatilities started the month at an usually high level, and therefore the returns, although large, were generally still not “outsized”.