Getting Paid to Reduce Risk: The Low-Risk Anomaly

New research on the low-risk anomaly – the fact that less risky stocks have had higher risk-adjusted returns – reveals exactly which types of stocks are likely to perform poorly over time, especially in a bear market. If the funds and ETFs you own lack construction rules to screen out those stocks, you will be exposed to unnecessary risks.

In addition to momentum, one of the great challenges to the efficient markets hypothesis is the low-risk anomaly – the highest risk stocks have lower returns, not just lower risk-adjusted returns, than the lowest risk stocks. This is true whether we consider volatility as the measure of risk, market beta, or broader measures such as quality (the QMJ factor). This is a real problem, as some investors seem able to enjoy a free lunch – higher return with less risk.

Nishad Kapadia, Barbara Ostdiek, James Weston and Morad Zekhnini (KOWZ) contribute to the literature on the low-risk anomaly with their study “Getting Paid to Hedge: Why Don’t Investors Pay a Premium to Hedge Downturns?” which was published in the June 2019 issue of the Journal of Financial and Quantitative Analysis. They began by noting: “At the heart of most rational asset pricing models is a concept of ‘bad times,’ when the marginal utility of consumption is high. Stocks that do well in bad times should have low expected returns because of the insurance they provide.” They tested this hypothesis using an intuitive measure of bad times: bear markets (periods from peak to trough in Standard & Poor’s (S&P) 500 levels at the business-cycle frequency).

To perform their test, they first identified stocks that performed particularly poorly in bear markets. They then constructed tradable portfolios that hedge bear market risk (long safe stocks and short risky stocks, creating an SMR, or safe minus risky, factor) and tested whether they do, in fact, earn low average returns. They identified nine bear markets between 1966 and 2015, with an average duration of 14 months and an average loss of about 30%. They excluded financials and microcap stocks.

Following is a summary of their findings:

  • It is possible to identify stocks ex-ante with differing sensitivities to bear markets.
  • The risky (the R in SMR) small-growth stocks with high short-term debt, high capital expenditures and low dividend yields suffer the most during downturns. Investors seeking to avoid the worst outcomes during bear markets should shun stocks with those characteristics.
  • Large, profitable, dividend-paying value stocks that make smaller investments with limited debt (the S in SMR) suffer the least during downturns.
  • Safe stocks are in the highest decile of predicted bear market returns, and risky stocks are in the lowest decile.
  • SMR has out-of-sample average monthly returns of 3.6% in bear markets – risky stocks lose 5.5% a month versus the 1.9% a month loss for safe stocks.
  • The unconditional average returns of safe stocks are much greater than those of risky stocks – the safe portfolio outperforms the market (by about 0.2% per month), and the risky portfolio underperforms U.S. Treasury securities. The results are not due to small illiquid stocks, but are pervasive patterns across the market.
  • From May 1967 through December 2015, SMR generated a four-factor (beta, size, value and momentum) alpha of 0.85% per month. It’s an anomaly for asset pricing models.
  • The high returns for SMR were robust to a battery of tests (such as value weighting and equal weighting, including financials and microcaps) and cannot be explained by co-skewness, idiosyncratic skewness or the betting-against-beta anomaly.