Believe those who are seeking the truth. Doubt those who find it. Andre Gide

Wednesday, October 5, 2011

The St. Louis Fed Financial Stress Index

My colleague Kevin Kliesen (and his coauthor, Doug Smith) have recently introduced a new financial market stress index; see Measuring Financial Market Stress. Here's a link to the Appendix, which describes its construction and the data series used: appendix. A brief description:
The St. Louis Fed’s Financial Stress Index is constructed using principal components analysis. Briefly, principal components analysis is a statistical method of extracting factors responsible for the comovement of a group of variables. We assume that financial stress is the primary factor influencing this comovement, and by extracting this factor (the first principal component) we are able to create an index with a useful economic interpretation. We construct the STLFSI using 18 weekly data series beginning December 31, 1993.  Prior to the principal components analysis, each of the data series are de-meaned and then divided by their respective sample standard deviations.
The index is available on FRED here. Let me reproduce some of this data here. Here is what the FSI looks like since December 2009.

Stress is on the rise, but looks like we're still a long way from 2008...

The big question, however, is whether the temperature will keep rising. 


  1. When a government messes with spreads, should this type of behaviour (taking some proper English from Dr. J) enter the index with the same weight as it did in the past? Their index includes quite a few treasury yields/spreads. Im guessing that this type of policy change doesn't capture the same financial stress as `natural' market movements; maybe its actually *more* informative (?) and should be weighted more, or maybe this type of *unexpected and exogenous* shock should be censored.

    Doesnt seem like this is an issue for them:

    "First, each of the data series is de-meaned. The de-meaned series are then divided by their respective sample standard
    deviations (SDs). With the variables now expressed in the same units, we use the method of principal components to calculate
    the coefficients of the variables in the financial stress index (FSI). We then scale these coefficients so that the SD
    of the index is 1. Finally, each data series is multiplied by its respective adjusted coefficient, and the observation of the
    FSI for time t is the sum of each series multiplied by its respective adjusted coefficient."


  2. At the risk of sounding like a hater, the private sector already had a similar risk index at least seven years ago.

  3. @Kyle, I'll let Kevin reply to your question (if he is able and willing to).

    @Anonymous, yes, that's right. We are currently at where we were in 2003. Of course, as always, the concern is whether things might get rapidly worse.

  4. Historical lock-in. I think the index was constructed in 2007, months before QE and all that were even conjectured. They then computed it retroactively. Kyle makes a good point

  5. "Stress is on the rise, but looks like we're still a long way from 2008..."

    Said without a trace of irony? The level today on that index coincides with that prevailing just before the Lehman failure. At that time, anyone would have said, "stress is on the rise, but looks like we're sill a long way from...", only to be proven wrong the following week.

  6. True enough. I stress forecastable?

  7. Stress is not forecastable; but fragility is identifiable.

  8. If policy makers focused on fragility rather than "stress", they would have asked Dexia to raise capital a few months ago, instead of pretending that it passed the stress test with flying colors. It seems policy makers are persistently determined to confuse positive expectations-setting with macroprudential policy.

  9. This graph accurately represents my financial situation. I enjoy high risk high return economics, I apply this strategy every time I make an investment. When I really find an investment that I second guess I turn to my advisers to calculate the exact risk.