Notes on The Meltdown

The Russell 1000 fell more than 5% from July 31 to August 5, and many other indices around the world were down by a similar order of magnitude. Owing to those market moves, the risk environment abruptly changed. Below we offer some notes and observations.

Note 1: Current level of volatility rose, but remains low relative to history

The sudden and sharp decline in stock prices led to a similarly substantial jump in short-horizon risk. As of this writing, risk remains low relative to historical levels, however, and the recent jump looks more like a small blip.

Figure 1. Russell 1000 year-to-date return and risk, and longer-term risk level

Source: FTSE Russell, Axioma

Note 2: Risk may be relatively low, but the jump was huge

While risk remains low, the increase in index volatility in such a short period of time was huge, and is likely fueling perceptions that volatility is quite high—and it may be moving in that direction. Risk climbed more than 20% from July 29 through August 5, one of the biggest five-day increases in the history of our model. In the chart below we show that In the more than 14,000 five-day periods since the inception of our US model in 1982, the increase has only been larger than this 86 times. So even if the level remains fairly low, investors likely percieve that volatility is much higher than it had been just recently.

Figure 2. Historical changes in five-day risk, Russell 1000

Source: FTSE Russell, Axioma

Note 3: Decomposition of the change in benchmark risk shows that higher asset correlations and higher stock and factor volatility all contributed to the overall increase in benchmark risk

The decomposition of the change in risk—which breaks down the change from the factor model and from the standpoint of a full asset-asset covariance matrix—shows that, for most single-country benchmarks, higher factor volatility drove risk higher. The UK saw an increase in factor correlations as well. Factor correlations for multi-country benchmarks also added a boost to risk (possibly the result of higher currency correlations), for all but FTSE Developed. The asset-asset matrix breakdown shows that higher asset correlations contributed at least as much as higher asset volatility in most cases. For more information on the philosophy and mechanics of this change-in-risk breakdown, see our quarterly Insight report (here).

Figure 3. Decomposition of the change in benchmark risk, single-country benchmarks

Source: FTSE Russell, Standard and Poor’s, Axioma

Figure 4. Decomposition of the change in benchmark risk, multi-country benchmarks

Source: FTSE Russell, Axioma

Note 4: Factor Returns were mixed, but at least a few held up well

The move did not appear to be a flight to cheapness, at least in the US. Over the recent period, both Value and Earnings Yield saw returns that were more than two standard deviations below the long-term four-day average, based on the expected volatility at the beginning of the period. It was a flight to lower risk, however, as Volatility and Market Sensitivity posted highly negative returns that were also more than two standard deviations below their (already negative) long-term averages. Larger stocks lagged smaller ones, as the return to the Size factor was more than -1%. The good news for many factor-based investors is that Medium-Term Momentum, Profitability and especially Growth held up well.

Figure 5. US4 model factor returns

Note: Returns are for the period July 31 through August 5. An * indicates the return was more than two standard deviations below (or above in the case of Growth) the long-term average, using predicted volatility of the factor at the beginning of the period.

Source: Axioma

Note 5: Realized active risk may have jumped in some cases, but not across-the-board

We calculated realized active risk for several of our sample factor portfolios. The target tracking error for all the portfolios is 3%. We found that over this period some portfolios, such as Unconstrained Value, far exceeded the target, whereas others were right in line or well below. Admittedly this is quite a short period of time, so it is difficult to make any conclusions from this small and short data set, but it does suggest that, at least for now, the heightened market volatility did not uniformly result in higher realized active risk. In addition, constraints meant to add a layer of insurance that the portfolio would stay within its targets did not necessarily mean that risk did not exceed its target (as for the Constrained Multifactor portfolio), but did seem to help in some cases (such as Constrained Value). See our quarterly Insight report (here) for more detail on these portfolios.

Figure 6. Sample factor portfolios realized risk

*Based on return from July 30 through August 5

Source: FTSE Russell, Axioma

Melissa R. Brown, CFA

As Managing Director of Applied Research, Melissa Brown generates unique insights into risk trends by consolidating and analyzing the vast amount of data on market and portfolio risk maintained by Axioma. Brown’s perspectives help both clients and prospects to better understand and adapt to the constantly changing risk environment.