Continue active refreshing of this index's data?

Continue active refreshing of this index's data?

Blog Posts — April 26, 2019

Is Investor Sentiment Subject to Seasonality?

Does the risk tolerance of investors ebb and flow with the seasons? Are there certain times during the year when investors tend to be distinctly more or less bold than others? While the subject of seasonality has been explored by a number of researchers, we decided to see if greater clarity could be obtained by looking at this phenomenon through the lens of a metric based on Axioma risk models.

Last month Axioma introduced a market sentiment Indicator built on eight factor-based metrics from our short-horizon fundamental risk models designed to quantify the supply and demand for risk assets in the market by quantifying the balance between risk-tolerant and risk-averse investors[1]. We have subsequently augmented that methodology with two additional factors—Leverage and Profitability—to create what we call the ROOF™ indicator (for risk-on/risk-off). The rationale for adding those factors is as follows.

  • If Leverage is being ‘rewarded’ (i.e. the factor return is positive), we take this as indication that there are more risk-tolerant investors in the market, thinking that either interest rates will decline (a bullish signal for equities) or that earnings growth will surpass interest-payment growth. If loading on leverage is being ‘punished’, then we infer that there are more risk-averse investors in the market, thinking that interest rates are likely going to rise, or earning growth to stall, making interest payments more difficult.
  • If Profitability is being ‘rewarded’ (i.e. the factor return is positive), we take this as indication that there are more risk-averse investors in the market, thinking that as economic conditions ahead worsen, profitable companies will be under less pressure than non-profitable ones. Conversely, if loading on Profitability and seeking the safety of strong fundamentals is being ‘punished’, then we infer that there are more risk-tolerant investors in the market, who are prepared to forgo the current safety of a positive income statement for the risk and leveraged return that a negative one brings, as they bet that economic conditions ahead will provide the earnings growth to put that company into the black (i.e., think of this as buying an out-of-the-money call option).

In this post we use the ROOF™ Indicator to seek evidence of seasonality in the risk tolerance of investors. The pivot table below shows the quarterly aggregated ROOF™ indicator values from January 2010 to April 2019 for each of our eight benchmarks. This table aims to capture the aggregate sentiment for that quarter over the last decade (e.g. if you only invested in Q1 for the last 10 years, what kind of cumulative market sentiment would you have faced over that period?). Observations over this period include a number of highlights.

  • With the exception of Asia ex-Japan, investors seem to start the year being very risk-tolerant, and, with the exception of Japan, end the year being very risk-averse.
  • Since January 2010, investors in Japan seem to be permanently risk-tolerant.[2]
  • Emerging market investors are more risk-averse, more of the time, than any other (i.e., 3 out of 4 quarters).
  • Investors in Australia and Emerging markets appear to experience the most anguish during the second quarter.
  • Investors in China seem to get progressively more risk-averse as the year progresses.

The old adage of “sell in May and go away” seems to forecast higher risk-aversion beyond that month. Looking at the monthly aggregates in the table below, this seems good advice for most markets with the exception of Japan and the US. Across all other markets, risk-aversion seems to exceed risk-tolerance in June and July. In those markets, selling in May and going away for the summer seems good advice.

October is a flash point for the US and (by association) the Global Developed markets, and moderately so for developed Europe and Asia ex-Japan. December seems to bring considerable risk-aversion in Australia, China, and global emerging markets. 

Looking at just the US market over the last 10 years, we see that October has been a risk-averse month in the last seven years out of nine. “Selling in May and going away” would really only have been aligned with market sentiment for the rest of the year in 2011, and mildly in 2014 and 2015. By comparison, becoming risk-averse in September and going away for Q4 would have earned an investor the distinction of sentiment guru in seven out of nine years.

Focusing on China and the periods of 2015 and 2018 in the table below, we see that sentiment turned very risk-averse in that market as early as April 2015 and remained so in May, before a short rebound in sentiment in June, then a resurgence of risk-aversion in July, followed by the China crash of that summer. In 2018, we see sentiment turning in favor of risk-aversion in May and becoming increasingly negative as the trade war worsened that summer, only to recover in November with the Trump-Kim summit and the announced cessation of tariff hostilities until March 2019. 

The imbalance between risk-averse investors and risk-tolerant ones in the market during that summer was so large that it reached levels greater than two standard deviations on multiple occasions. In concert with this imbalance, the cumulative return for the CSI 300 for the June-August 2018 period was a negative 12%. The upper chart below shows the cumulative return for the CSI 300 index in black, the risk-aversion level in red and the risk-tolerance level in green. The lower chart shows the balance between risk-tolerance and risk-aversion in the market, with the green areas showing periods when there is more risk-tolerance in the market than risk-aversion, and the red areas indicating periods when risk-aversion is greater than risk-tolerance. The black trend lines show the long-term one and two standard deviation imbalance levels. Sentiment fell below the two standard deviation trend line on four occasions during this three-months period in China. It wasn’t until late August when this imbalance between risk-tolerance and risk-aversion in the market normalized closer to equilibrium that the market settled down and started its recovery.

To put the above analysis in perspective, there is a difference between the definitions of volatility and risk. Volatility is the mathematical measure of the dispersion in returns investors might experience around their forecasts, while risk is simply the probability of losing some or all of their investments by being wrong. The latter includes a behavioral component of risk that is driven mostly by uncertainty-aversion. Investing is forecasting, and investors accept a certain amount of deviation between their forecast versus realized outcomes. This acceptance, or tolerance, is based on their confidence level in these forecasts. When an event sharply broadens the confidence intervals around their initial forecasts, uncertainty-aversion kicks in.

The Axioma ROOF™ Indicator was designed to capture the behavior that follows a rise or fall in uncertainty-aversion by focusing on the reward given by the market for risk-taking. Volatility measures alone will capture a rise or fall in future return dispersion, but will not capture the kind of behavior or sentiment that causes it. Was the rise in volatility caused by investors fleeing towards risk, or away from it? In supply and demand parlance, was the rise in volatility caused by the higher premiums willingly paid by risk-tolerant investors for the risk assets of their far fewer risk-averse peers, or by the bigger discounts they obtained for taking on the risk-assets of a larger population of risk-averse investors wanting to sell them? In other words, was volatility caused by this being a seller’s market or a buyer’s one? This is the question the Axioma ROOF™ Indicator was designed to answer, as a complement to the predicted volatility forecasts from the Axioma Risk Models.

[1] The paper titled “Risk-on/Risk-off and the Schrödinger Quadrant” can be found here. It includes the rationale for the existing eight factors selected previously.

[2] This includes the outlier month of March 2011, with a ROOF™ indicator score of -8.0 on March 17, when the market reopened after the earthquake and tsunami disaster.