AAEAAQAAAAAAAAZZAAAAJDUyM2Q1NmI5LTRiOWYtNGU0NS1hMmEzLWRkZWM3YmE5ZjE2Nw

Is the World About to End?

In our last update, we pointed out the risks in the emerging markets. That theme has continued to play out. By now, investors would’ve experienced maximum losses of -44% in China and -32% in EM equities. US equities is still doing better than other markets – currently at -12% maximum loss.

In this environment, every online robo-advisor is telling their investors to keep calm and carry on. We here at Cassia prefer to be more proactive when it comes to managing risk. We have been performing emergency rebalances in all of our client accounts since August 22. And we continue to monitor the situation in real-time to ensure accounts can move quickly as new data comes in.

Quantifying the risk

The best decisions are data-driven. So what is the data telling us about the current markets? Let’s talk about risk. No I’m not talking about the flawed risk measures that everyone can calculate in Excel. I’m talking about a GARCH volatility forecast that accounts for nuances like clustering and mean reversion that we wrote about in our white paper.

Typically, the S&P 500 has a volatility of 12%. In October 2008, it had a forecasted volatility of 51%. Today, our advanced risk forecasting system sees short-term volatility at 18% and 3-month volatility at 13%[1].

This tells us two things about US Equities:
1. The current level of volatility isn’t crazy high by historical standards.
2. The forecasted volatility in the short-term is higher than the long-term (i.e. backwardation).

Volatility spike – what does it mean?

The second point warrants some investigation. So we ranked the volatility forecasts into four buckets (quartiles). We ask ourselves: “how volatile is the market relative to the long-term volatility?” Do spikes in the short-term volatility have any implications for future returns?

Lo and behold. During the bull market from 2009 to present, the market appears to yield higher returns after volatility spikes. Q4 represents the days where the volatility is the highest, and Q1 represents volatility being the lowest, relative to forecasted long-term volatility.

AAEAAQAAAAAAAAY7AAAAJDRkMjA2MTA0LTNhYjYtNDE5NS04NzZhLTFkYmM0YWEwYTA3NQ

The results hold for returns 1, 5, 10, 20, 60 days into the future. This makes intuitive sense because volatility spikes tend to coincide with short-term lows in the market. Buying the dips tend to yield higher returns – in a normal environment.

But look what happens when we include the highly abnormal period of 2008. Buying when volatility spikes becomes a terrible idea. Instead of yielding higher return, buying the dips in a financial crisis, yields negative returns.

AAEAAQAAAAAAAAUuAAAAJGE5MDJlNGEzLWIzMGItNDA3Zi04MzMzLWEwYWZlMDEzY2NmNg

Another 2008? Let’s look at systemic risk

So the key question now is. Is this a 2008 environment or not? Rising correlation could indicate contagion between markets, indicating increased systemic risk and potentially a 2008 environment. Here’s what the data says as of the end of August 2015. The chart on the left shows the correlation between the S&P 500 and other major asset classes such as real estate, bonds, commodities, and emerging equities. Do you see any signs of an increase in correlation?

 

AAEAAQAAAAAAAAQcAAAAJDg0YWM2MzY5LTdlMDMtNDIxOS1iYTQ5LWQ5NDBmMDBhMWNiMgCorrelation AAEAAQAAAAAAAAQhAAAAJDQwMjU1MDIzLWZhMjUtNDkxNS1iN2Y4LTZmNjE3NzJmYzI4Ng Standardized Change

 

We don’t. The current environment does not look similar to 2008.

To make it easier to see, the chart on the right is the standardized change. If anything, correlation between equities and other assets is experiencing the largest drop in 10 years – by almost 2 standard deviations! In contrast, correlation increased by 1.5 standard deviations at the start of 2008.

For robustness we also consulted another measure of systemic risk called the Absorption Ratio and ran it on 9 US sector funds (Kritzman 2010). The authors note that a +1 standard deviation increase in the Absorption Ratio is an indication of increased systemic risk. The current reading is at -1 standard deviation, nowhere close to systemic risk territory.

 

AAEAAQAAAAAAAAWKAAAAJGRkY2Y0Y2QwLTE4ZDAtNDlhNy1iY2U0LTIxNzY2MWJmYzQwYQAbsorption Ratio AAEAAQAAAAAAAAMdAAAAJDNhMDcxNDg5LWUxYTYtNDE1Zi04Y2Q4LTdhMjdmMzA2YmQ1OA Standardized Change

 

We also repeated our test on days where systemic risk is heightened (absorption ratio greater than +1 SD). Our findings are consistent. In periods where volatility spikes are accompanied with heightened systemic risk, future 1-month returns tend to be low. On the other hand, volatility spikes during normal periods gives higher future returns (absorption ratio below +1 SD).

GARCH volatility term structure vs. next month return (annualized)

Volatility spikes during increased systemic risk: 0%
Volatility spikes during normal periods: 38%

Conclusion

The data suggests that a) the current environment is not similar to 2008, b) forecasted volatility isn’t high compared to history, and c) volatility spikes in the absence of systemic risk have historically offered higher future returns. What can change this? We would put a September rate hike by the Fed at the top of the list.

Based on the current data, we feel that there’s no overwhelming case to scale back on equities allocation. Should forecasted volatility or correlation increase dramatically, Cassia’s systems stand ready to step in and adjust allocations for its clients, faster than a human manager can react.

References

Kritzman, M., Li, Y., Page, S., & Rigobon, R. (2010). Principal components as a measure of systemic risk.

[1] Data as at the end of August 2015



Build your practice on the backs of 3 Nobel prizes — Prospect Theory, MPT, and GARCH. 

See what GARCH can do for you.
DOWNLOAD NOW
See what GARCH can do for you.
Build your practice on the backs of 3 Nobel prizes — Prospect Theory, MPT, and GARCH.