Federal Reserve Economic Data

The FRED® Blog

Posts tagged with: "STLFSI"

View this series on FRED

The St. Louis Fed’s Financial Stress Index, Version 2.0

Economists, banking regulators, policymakers, and financial market analysts use a variety of indicators to monitor financial market conditions. Many indicators are constructed from market-based prices, since information about the health of the economy, a bank, or a firm is often reflected in equity and debt markets. So market prices are forward-looking indicators of potential changes in economic and financial conditions.

The best known macroeconomic measures are interest rate spreads between so-called “risk-free” and “risky” securities. For example, the spread between long-term and short-term Treasury yields—often termed the yield curve—tends to be a reliable forecaster of future economic growth. (See McCracken, 2018, and Owyang and Shell, 2016.)

To help the public monitor financial market conditions on a weekly basis, the St. Louis Fed unveiled a financial market stress index (FSI) in 2010. (See Kliesen and Smith, 2010.) Similar to other FSIs, the St. Louis Fed’s (STLFSI) measures different types of financial market stress. Falling prices of financial market assets—such as stock prices—is an obvious example, since it could signal expectations of lower corporate profits due to slower growth of aggregate economic activity. Other types of stress include changing market perceptions of “risk” in its different forms. As noted above, risk is often measured by examining interest rate spreads: Default risk is regularly measured as the difference between yields on a “risky” asset (e.g., corporate bonds) and a “risk-free” asset (e.g., U.S. Treasury securities). But financial market stress can arise in other dimensions, too.

One type of risk prominent in the 2008-2009 financial crisis is once again present—in the current COVID-19 (novel coronavirus) crisis. It is the inability of many financial institutions to secure funding to finance their short-term liabilities, such as repurchase agreements (repos). This type of risk is known as “liquidity risk.” Yet another type of risk is uncertainty about the future direction of inflation, termed inflation risk.

The STLFSI, as with all other FSIs, attempts to measure financial market stress by combining many indicators into a single index number. This index number then becomes a collective measure of financial market stress. How is this accomplished?

The STLFSI is calculated using principal component analysis (PCA), which is a statistical method of extracting a small number of factors responsible for the co-movement of a larger group of variables. Specifically, the STLFSI is the first principal component of 18 distinct measures of financial stress and is thus a measure of overall financial market stress. The STLFSI used weekly data beginning in late 1993 from 18 data series: 7 interest rates, 6 yield spreads, and 5 other indicators. Values of the STLFSI above 0 indicated higher-than-average levels of financial market stress, while values below 0 indicated lower-than-average levels of stress.

Over time, we understood that the original construction of the STLFSI wasn’t adequately capturing some stresses that had developed in the financial markets. This was easy to spot visually on a graph: Despite several economic and financial market developments, the index fell below 0 in mid-2010 and has indicated below-average levels of financial market stress ever since. So we’ve unveiled an improved version—STLFSI 2.0—that makes a few simple, but necessary, changes to the original version unveiled in 2010.

Our plan is to publish a more thorough analysis later in the year,  documenting our motivation and methodological changes that we believe make STLFSI 2.0 an improvement.

The key difference between versions 1.0 and 2.0 is that 2.0 uses daily changes in interest rates and stock prices, rather than the levels of interest rates and stock prices in the PCA calculation. Why is this important? The full answer is more complicated, but the primary reason is that interest rates have trended lower and stock prices have trended higher, on average, over the period the STLFSI calculation covers. This has introduced a subtle  but nevertheless important statistical bias in the calculation of the STLFSI. This table summarizes the transformation applied to the data series used to construct the STLFSI.

The figure plots the new version of the STLFSI along with the original version we’re retiring. Here are two examples why we believe the STLFSI 2.0 more accurately measures financial stress:

On August 5, 2011, Standard and Poor’s reduced the long-term sovereign credit rating on the United States from AAA to AA+. Although other rating agencies maintained the AAA-rating on U.S. Treasury debt, this action severely rattled equity markets, as the Dow Jones Industrial Average fell nearly 2,000 points over the next two weeks. However, the original version of the STLFSI continued to report below-average levels of financial market stress (values less than 0). The revised STLFSI, however, moved decisively above 0, indicating above-average levels of financial market stress, peaking at 1.2.

The second example is the ongoing turmoil in financial markets stemming from the fear and uncertainty associated with the COVID-19 pandemic. Since mid-February 2020, and continuing through the week ending March 20, 2020, COVID-19 uncertainty has triggered a massive sell-off in stocks and consequent plunge in stock prices, sharp declines in interest rates, and stunning increases in financial market volatility. Still, the original version of the STLFSI continued to report financial stress slightly below average (-0.1). The revised STLFSI, however, has increased sharply—reminiscent of the worst of the financial market turmoil during the Great Recession in 2008-2009—registering a value close to 5.8.

The data and information services of the St. Louis Fed’s Research Division are intended to illuminate economic and financial concepts, educate, and enhance decisionmaking. In this vein, it is our view that STLFSI 2.0 better captures evolving stresses in financial markets. Our new version suggests that the tectonic upheaval in financial market conditions today has, thus far, been surpassed only by the upheaval registered in 2008-2009.

Suggested by Kevin Kliesen and Michael McCracken, with the research assistance of Kathryn Bokun and Aaron Amburgey.

View on FRED, series used in this post: STLFSI, STLFSI2

Move along. Nothing to see here. (Seriously.)

Searching for financial stress

You may be relaxing over the holidays, but Team FRED Blog feels a little contrarian, like that uncle you can never agree with. So let’s talk about stress.

FRED offers three series from different regional Federal Reserve Banks that are intended to alert us to financial stress. All three indices use available data from the financial sector to try to establish an aggregate that highlights the level of risk in that sector, with higher values showing more stress.

The good news? Today, things are looking pretty steady. You could even say that there’s nothing to see here. At least as far as financial stress goes.

But the data overall show a couple of things quite clearly: The Great Recession was definitely financial in nature, with great financial stress, whereas the preceding recession was not. And all three indices show the same course: As early as July 2007, conditions were getting worrisome. Still, it’s good to be careful when reading indicators like these, as increasing stress doesn’t always signal an impending recession.

How this graph was created: Search FRED for “stress,” check the two series, and click “Add to Graph.” From the “Edit Graph” panel, use the “Add Line” tab to search for “Chicago Financial Conditions” and add that line to the graph.

Suggested by Christian Zimmermann.

View on FRED, series used in this post: KCFSI, NFCI, STLFSI

What’s normal for financial data?

"Norming" indicators such as the St. Louis Fed Financial Stress Index

Financial data are useful for many reasons. One (perhaps subtle) reason is that they are never revised. Markets determine the prices and quantities of assets at the time of the transaction and that’s that. As such, once you observe the value of a particular financial variable at a particular point in time, you know it will remain at that value forever.

One might assume, then, that the St. Louis Fed Financial Stress Index, which includes 18 series of financial data to measure stress in the markets, would also remain the same forever. Well, the graph shows us something different: It plots 10 distinct vintages of the index starting with the first, from March 2010, and then one from every year since then. Despite the fact that this index is composed entirely of unrevised financial data, the lines of past data are not exact replicas of each other, so the index clearly changes over time. But the reason for this variation lies in the construction of the index and not in the underlying data.

The index is calculated as a weighted average of the underlying financial series. So, what changes are the weights used in this calculation when new data are released. These weights are functions of the means, variances, and correlations of those 18 financial data series used in the index; and these means, variances, and correlations are recalculated every time another observation is added. In March 2010, the first index was constructed based on data ranging from 1993 through early 2010. In the most recent vintage, from April 2019, the index was constructed based on data ranging from 1993 through early 2019. Since the means, variances, and correlations differ across the two samples, the weights differ as well, and therefore we have different paths for the index.

Why should we care about this? The entire premise of the index is that it is useful for identifying periods of financial stress that are higher or lower than normal. But what’s normal? Take a look at the “Notes” section of the series page for this index: Stress is considered normal when the index takes the value zero. So an index value greater than zero indicates above-normal levels of financial stress and a value less than zero indicates below-normal levels.

Unfortunately, the index’s position above or below zero changes across vintages. For example, consider the index’s value in 1996: The first (March 2010) vintage of the index indicated that financial stress was less than normal in 1996. But this isn’t the case in more-recent vintages, which indicate that financial stress was higher than normal in 1996.

The bottom line is that “normal” is relative to the total sample available. It’s not an absolute statement about financial markets at any given time and shouldn’t be interpreted as such.

How this graph was created: From ALFRED, search for “St. Louis Fed Financial Stress Index.” Check the appropriate series in the search results and click “Add to Graph.” By default, ALFRED creates a bar chart; to change to a line graph, use the “Graph type” menu under the “Format” tab. To include the earliest vintage, click on the “Edit Graph” button, go to the “Edit Line 1” tab, and select vintage 2010-03-05. To add the next vintage, go to the “Add Line” tab, search for and select the St. Louis Fed Financial Stress Index, click “Add data series,” go to the “Edit Line 2” tab, and select vintage 2011-03-03. Repeat this process until all desired vintages have been added. Finally, adjust the range of the plot so the end date is 2010-02-28.

Suggested by Michael McCracken and Joseph McGillicuddy.

View on FRED, series used in this post: STLFSI


Subscribe to the FRED newsletter


Follow us

Back to Top