Interpreting the St. Louis Fed Stress Index

by Jeff Miller

While corporate earnings are the ultimate story, there are other important questions:

• Is there systemic risk?
• Does the current selling presage a 2008-style economic collapse?

I have encouraged the use of objective, quantitative indicators to answer these questions. In sharp contrast, many others make simple analogies and rely upon anecdotal evidence.

To underscore the difficulty, here is the opinion of Jeffrey Gundlach, one of the very top fixed income managers:

…the time is ripe [for another AIG or Lehman-level collapse] based upon the growing lack of confidence in the growing debt of Spain, Italy, Greece, Portugal, Ireland and ultimately of France.

Adding to the danger is the fact that European financial institutions—particularly French banks own these toxic assets. The result will be a “restructured default”—unless the Germans are “going to pay the debts of everybody in Europe,” something Gundlach says is extremely unlikely.

Meanwhile, Larry Kudlow assembled a nice panel of experts addressing the specific question of how European sovereign debt affected European banks and the US banks. I watch and read many sources, highlighting only the most important for readers. This video segment is worth watching.

• None of the experts (David Malpass, William Rhodes and Bill Isaacs) see US systemic risk.
• Malpass explains how interest rate differentials have caused foreign banks to reduce holdings at the Fed.
• The entire panel points to strong balance sheets and other differences from 2008.

No wonder it is so easy to be confused!

A Data-Based Approach

In an effort to help investors avoid a 2008 scenario, our team has been doing research on various indicators of systemic risk. A leading candidate was the St. Louis Fed Stress Index, since it was developed with this particular question in mind. I have been enthusiastic about the methodology used, and the general value of this measure.

As background, here is a helpful short paper on the development of the index, and a chart illustrating the SLFSI level at various important times in history.

The level announced this past week, reflecting data as of last Friday (August 19), was 0.73. This is significantly below all of the listed events in the financial crisis of 2007-08.

What Level is Important?

A focus for our research has been to identify objectively-defined risk. Since everyone would like to predict future market moves, we hoped that objective identification of risk would also be an early warning for potential market declines.

We have promising results which we can eventually publish. The recent market decline has encouraged me to reveal some findings a bit early — before we can do the polishing for a final presentation. Let me just say that the results I am citing are accurate, but the terminology and presentation is still a little “nerdy.”

The Findings

Let me start with our research objective:

Discover reality-based predictors of market drawdowns.

There are many methods of forecasting the market with technical criteria, including several employed by our team. This question is a bit different. We are looking to fundamental, not technical indicators.

One of our approaches involved using the SLFSI. Out of many different analyses, I am choosing one result that I believe best reflects our research.

This chart reflects the use of the SLFSI to predict the maximum drawdown in the S&P 500 over the upcoming three months. The orange points represent predicted values and the yellow points the actual values. In general, the prediciton is pretty good.

To interpret the scales, you can see that SLFSI scores of below 1 are usually not very important. (The predictions here reflect the score, and the short-term change in the score). The vertical scale is the amount of the maximum drawdown over the upcoming three months. (e.g., -.2 means a 20% drawd0wn).

Preliminary Conclusions

There are several interesting conclusions, a lot of room for discussion, and the need for more research.

1. Increased stress implies stock market danger, as reflected by future drawdowns.
2. Drawdowns frequently occur even when not “justified” by actual increases in stress, including some of the very largest. It is incorrect to conclude that a decline market accurately reflects actual systemic risk.
3. The current stress levels are much lower than the events of the 2007-08 crisis.
4. The SLFSI did not provide an “early warning” signal for the current market decline.

Briefly put, there is a difference between predicting financial stress and forecasting market reaction. The SLFSI does the former. Sometimes it also helps with the latter, but we do not see that so far in 2011.

We continue to monitor the the SLFSI with a “trigger” in the 1.1 – 1.5 area. We also monitor the underlying 18 components (insofar as is possible) on a real time basis to avoid the four-day lag in the published results. Put another way, our research reflects the lag (as it should) but as traders we seek an additional edge.

A Final Thought

It is always helpful to have sophisticated measures of economic and market indicators. Sometimes the investor is left wondering how to interpret the information. It is not the job of the St. Louis Fed to predict the stock market. Those of us interested in fundamental analysis of risk can use their data to gain additiona insight into actual systemic risk.

This work may help us distinguish between the anecdotal risk that we hear about every day and something that can actually be measured.

Related Articles

Investing Blog articles by Jeff Miller

About the Author

Jeff Miller has been a partner in New Arc Investments since 1997, managing investment partnerships and individual accounts. He has worked for market makers at the Chicago Board Options Exchange, where he found anomalies in the standard option pricing models and developed new forecasting techniques. Jeff is a Public Policy analyst and formerly taught advanced research methods at the University of Wisconsin. He analyzed many issues related to state tax policy and provided quantitative modeling which helped inform state and local officials in Wisconsin for more than a decade. Jeff writes at his blog, A Dash of Insight.