- Worst Day in 2 years leads to: Lower Volatility?(!)
- Sector and style exposures explain the discrepancy
- The short-horizon model risk changes behave more as expected
Worst Day in 2 years leads to: Lower Volatility?(!)
We lead today with this graph from the US Equity Risk Monitor: STOXX USA 900 as of 16 September 2022:
The chart shows vectors of change in risk and return over important trailing time periods. The lines show the change in risk on the x-axis and the return over the same period on the y-axis. The dot represents a one standard deviation change, based on the risk model forecast at the beginning of the period. One could be forgiven for thinking that the worst was over looking at the trailing 3 month change (the gray line) vs. the 12 month change (in red), but in the shorter term, returns have taken a turn for the worse, as over the last month and the last week in particular many of the summer gains have been wrestled back by continued inflation surprises. It is notable that in the US at least, the volatility forecast for the broad market is lower, despite Tuesday the 13th of September being the worst day for US markets since early 2020 with the STOXX USA 900 down well over 4%. How can this be?
Sector and style exposures explain the discrepancy
The most likely explanation is that this is being driven by changes in exposures to factors with strong negative or positive forecast correlations to the Market Intercept Factor: namely Market Sensitivity, Oil &Gas, Pharmaceuticals & Biotech. The STOXX USA 900 now has an increasingly negative exposure to Market Sensitivity. The STOXX USA 900 is not style-neutral, because it is a large-cap tilted index; the overall US market is generally Style factor neutral:
The Risk Contribution from this factor Exposure as of the market close on 9/16 is:
Because the Market Sensitivity factor is so highly correlated with the Market Intercept and the Index’s exposure is negative, we get a big diversification effect from it. The same goes for the Industries that are gaining in index weight by moving in the opposite direction of the overall market:
See graph from the STOXX USA 900 Equity Risk Monitor as of 16 September 2022:
The risk contributions in the chart above are bottom up, meaning the stock-level risk contributions are summed over each sector. If we look at the risk model’s factor decomposition where the factor effects are decomposed cleanly, we see the following:
So, these four industry factors actually diversify our factor risk because of their negative correlation to the market, and reduce the overall risk forecast by 37 bps, or 1.7% of the total index risk. In addition, the Health Care and Energy industries also have negative exposures to Market Sensitivity, although the Automobile industry does not (mainly because of Tesla).
See table from STOXX USA 900 Equity Risk Monitor as of 16 September 2022.
The short-horizon model risk changes behave more as expected
It should be noted here that while the new Axioma US5-TH trading model shows similar diversification benefits from the same group of factors, the risk forecast from that model has gone up over the past month (albeit from much lower levels):
The new Trading Model has a volatility half-life of 20 days, compared with 60 days for the Short Horizon Model and 125 days for the Medium-Horizon model. It also has a market risk forecast that is adjusted by the observed level of the VIX implied volatility index (see US5-TH model factsheet). All three models appear to have converged at around 22% annualized, but the trading model better reflects the experience of rising volatility from recent lows during the summer.