Federal Reserve Economic Data

Each week, the Federal Reserve’s Board of Governors publishes the H.8 Release, which details aggregate balance sheet data (assets and liabilities) for all commercial banks in the United States. And each week, the data can be found in FRED. In this first of three related posts, we examine recent trends in selected assets and liabilities of large and small domestically chartered commercial banks.

Although we aren’t providing a comprehensive examination of recent developments in the US banking sector, the FRED graph above does plot total bank assets for large and small commercial banks since the week ending January 7, 2015. This 8-year period encompasses the previous FOMC tightening cycle (2015-2018), the pandemic recession in early 2020, and the current economic recovery and current FOMC tightening period since March 2022. The values below were calculated as of March 15, 2023:

• Assets at large commercial banks increased from $9.1 trillion to $13.2 trillion, or by 44.8%.
• Assets at small commercial banks increased from $3.6 trillion to $7 trillion, or by 94.5%.
• On January 7, 2015, the ratio of total assets at large banks vs. small banks was 2.56. By March 15, 2023, that ratio had declined to 1.90.

So, there’s been a modest shift in total assets away from the 25 largest commercial banks to small commercial banks. (The H.8 Release also includes foreign-related institutions active in the United States and shows their assets increased from $2.5 trillion to $3.1 trillion, or by 22.7%, over this period.)

Large commercial banks are defined in the H.8 as the 25 largest commercial banks in terms of domestic assets, and small commercial banks are those not in the top 25 but with at least $300 million in assets. The bulleted details below are based on the commercial bank call report data ending December 31, 2022:

The 25 largest banks:

• These banks had total domestic assets of $13.43 trillion, or a little more than 51% of nominal GDP.
• The largest bank, JPMorgan Chase, had assets of $2.48 trillion.
• The 25th largest bank had assets of $155.4 billion.
• Before it failed on March 10, 2023, Silicon Valley Bank was the 16th largest bank, with assets of $194.51 billion.

The 2,099 other (small) banks:

• Signature Bank, which failed on March 12, was the 29th largest bank and had assets of $110.36 billion.
• Domestic assets of these small banks totaled $6.4 trillion, with an average of $3.05 billion.
• A majority of these banks (1,312, or 61.8%) had assets of less than $1 billion.

The full list of commercial banks and their assets as of December 31, 2022, can be found here,

Recently, many financial market participants have worried about the financial health of the US banking sector. This had led to sharp declines in stock prices of many “regional banks.” In response, the Federal Reserve, the FDIC, and US Treasury have taken actions to calm the fears of both depositors and investors. Nevertheless, while Federal Reserve and Treasury officials continue to stress that the commercial banking sector is “sound and resilient, with strong capital and liquidity…events of the last few weeks raise questions about evolving risks.” (Governor Michael Barr, Congressional testimony)

Our next related blog post will look at recent trends in securities held on commercial bank balance sheets, and the related post after that will look at trends in loans and leases and deposits, which has garnered attention lately. FRED users interested in monitoring the US commercial banking sector can do so by analyzing weekly trends in commercial banks’ balance sheets found in these H.8 data.

How this graph was created: Search FRED for and select “total assets large banks.” Click on “Edit Graph,” open the “Add Line” tab, and search for and select “total assets small banks.” Start the graph on 2015-01-01.

Suggested by Kevin Kliesen and Cassie Marks.

The FRED graph above depicts the St. Louis Fed’s Financial Stress Index (STLFSI). This data series in FRED was created in 2010 to measure changes in U.S. financial market conditions in response to a broad array of macroeconomic and financial developments. In particular, the STLFSI is designed to quantify financial market stress. There’s no specific definition for financial market stress, but periods of stress have historically been characterized by increased volatility of asset prices, reduced market liquidity conditions, or the narrowing or widening of key interest rate spreads. The STLFSI is constructed using 18 key indicators of financial market conditions—7 interest rates, 6 yield spreads, and 5 other indicators.

In late 2021, some Federal Reserve officials encouraged financial market participants and others to consider using an alternative short-term interest rate benchmark because of concerns about the eventual retirement of the London interbank offered rate (LIBOR). Since the STLFSI had two yield spreads based on the LIBOR, we replaced the LIBOR rate with the secured overnight financing rate (SOFR). Specifically, we shifted to the 90-day average SOFR. This rate measures the compounded average of the SOFR over a rolling 90-day period. In other words, it’s a backward-looking measure. We showed that the correlation between the previous version (STLFSI2) and the new version (STLFSI3) was 0.99 over the sample period dating back to December 1993. Click here for details and more information about this switch.

In 2022, we received numerous inquiries about the behavior of the STLFSI during the year. Most asked why the STLFSI was continuing to indicate lower-than-average levels of financial market stress, while other measures showed a “tightening” in financial market conditions. The divergence between the STLFSI and other indexes occurred more or less at the time when the Federal Open Market Committee (FOMC) began to signal its intent to raise its federal funds rate target in March 2022 and, importantly, subsequently signaled that further increases in the policy rate were likely in 2022—and perhaps in 2023.

Our analysis showed that instead of using the 90-day backward-looking SOFR rate, we should have used the 90-day forward-looking SOFR rate. In our view, using the forward-looking SOFR better captures financial market expectations in response to expected changes in the federal funds rate and its attendant effects on other asset prices and yields.

The second FRED graph plots the STLFSI4 and the STLFSI3 since early January 2020—just prior to the financial market turmoil and deep recession spawned by business and government actions designed to counteract the COVID-19 virus. In the graph, the two versions track each other closely over most of this period. But the close comovement began to erode in early February 2022, as it became clear that the FOMC was poised to begin raising its policy rate to combat an inflation rate that was the highest in 40 years. For example, the correlation between STLFSI3 and STLFSI4 was 0.993 from the week ending December 31, 1993, to the week ending January 28, 2022. Since the week ending February 4, 2022, the correlation has declined to 0.526.

A final takeaway from this second graph is that the new measure of the STLFSI shows that financial market stresses during the current Fed tightening episode are moderately higher compared with the previous version. Still, levels of financial market stress are currently near their historical levels. (In the index, zero is designed to be an “average” level of stress.) Moreover, the current Fed tightening episode has not triggered the kind of financial market stress seen during the heights of the pandemic-spawned shutdowns in the economy.

How these graphs were created: Search FRED for “Financial Stress Index” and make sure to take version 4. For the second graph, take the first, click on “edit graph,” open the “add line” tab, and search for “Financial Stress Index,” making sure to take version 3.

Suggested by Cassandra Marks, Kevin Kliesen, and Michael McCracken.

The FRED graph above shows that real gross domestic product (GDP) has declined over the first two quarters of 2022, after increasing by an average of 5.3% over the previous five quarters. In the eyes of some economists and financial market participants, two consecutive quarters of negative real GDP growth is sufficient evidence to declare a recession.

In the 75-year history of quarterly estimates of real GDP growth, there has been only one episode when two consecutive quarters of negative real GDP growth was not associated with a recession episode: the second and third quarters of 1947. So, from a historical standpoint, two consecutive quarters of negative real GDP growth is a pretty consistent signal for dating recessions. But what do the arbiters of dating business cycles have to say?

The dating committee

The National Bureau of Economic Research Business Cycle Dating Committee (hereafter NBER) maintains a chronology of monthly and quarterly dates of the peaks and troughs (i.e., turning points) of the business cycle. Rather than the popular two-quarter definition, the NBER employs a more comprehensive approach to dating the beginnings and ends of recessions. Specifically, they determine both the months and the quarters when economic activity peaked and troughed. Typically, the peak month occurs in the same quarter—but not always. For example, the NBER’s monthly peak of the pandemic-spawned recession occurred in February 2020, but their quarterly peak occurred in the fourth quarter of 2019.

The indicators

To determine the months of peaks and troughs, the NBER looks at several data series, such as industrial production, nonfarm payroll employment, civilian employment, and real personal income less transfer payments. The NBER also considers two other monthly series: real personal consumption expenditures and civilian employment. Civilian employment is measured using the household survey (Current Population Survey), while nonfarm payroll employment counts the number of jobs and is measured using the establishment survey (Current Employment Statistics).

The NBER also looks at estimates of the expenditure- and income-side measures of aggregate economic activity—otherwise known, respectively, as real GDP and real gross domestic income (GDI). Theoretically, GDP and GDI should equal each other in dollar terms, but they rarely do. (This difference between the two series is known as the statistical discrepancy.) The NBER also examines average GDP and GDI.

A new resource in FRED

This is a lot of information to gather, so FRED now offers some help navigating the ebbs and flows of these key data series with a new dashboard that compiles all these series on one page.

As with any user-created FRED dashboard, it updates automatically. Now, there won’t be any commentary on the current or prospective trends in the dashboard. But FRED users can make their own determination as to the likelihood of a turning point in the business cycle. Users can also use the dashboard as a starting point for creating their own variations.

Suggested by Kevin Kliesen.