Federal Reserve Economic Data

The Federal Reserve Bank of Philadelphia has a new data sandbox for credit enthusiasts to dig into: the FR Y-14M program. These 75 new mortgage and credit card data series reflect the aggregated portfolio of the largest financial institutions in the U.S., offering novel insights into some of the largest credit portfolios in the market. Save this quarterly dataset to your FRED account for a trove of unique intelligence on credit cards, first-lien mortgages, first-lien home equity loans, consumer habits, credit quality, and percentile indicators.

The Philadelphia Fed also provides a report on the data: Insights: Large Bank Credit Card and Mortgage Report.

Mortgage data: The most recent report notes that large bank mortgage origination volumes were flat in the second quarter of 2022 after declining sharply in the prior quarter. This contrasts with 2021, which saw the largest annual volume of originations since 2012 accompanied by rapid house price increases.

The FRED graph above shows mortgage origination loan-to-values (LTVs), which are increasing to pre-pandemic levels, rising from 68% in the fourth quarter 2021 to 75% in the second quarter of 2022. As rising mortgage rates reduce refinance demand and as originations fall, purchase loans constitute a greater share of new originations, which has pushed LTVs higher.

Credit data: Credit card originations have fully recovered to historic norms, and higher credit limits are being made available to new accounts compared with a year ago. The second FRED graph shows credit card balances: Coupled with an overall increase in consumer spending, credit card balances rose 16% on a year-over-year basis in the second quarter of 2022, bouncing back from near pandemic lows in the second quarter of 2021. That was the fastest yearly card balance growth since the FR Y-14M data collection began in 2012 and far greater than the typical 4% annual growth from 2013 through 2019.

With stronger consumer spending, card utilization has also begun to recover, hitting 18.5% in the second quarter of 2022. Credit card delinquency rates have increased modestly over the past year, though rates remain near their lowest level in the past 10 years, with fewer than 1% of balances reaching 90 days past due.

Details behind this large bank dataset: The credit card data are largely reflective of the total U.S. credit card market, representing roughly 3/4 of total U.S. bank card balances. The mortgage data provide new information on large bank lending and represent roughly 1/10 of the U.S. residential mortgage market.

The respondent panel for this dataset comprises U.S. bank holding companies, U.S. intermediate holding companies of foreign banking organizations, and covered savings and loan holding companies with $100 billion or more in total consolidated assets. These institutions are required to report credit card or first-lien mortgage data if portfolio balances exceed $5 billion or are material relative to Tier 1 capital. Firms with over $100 billion in total consolidated assets that do not meet these thresholds may also voluntarily provide FR Y-14M data.

For more information, see the data methodology.

How these graphs were created: Search FRED for “large bank mortgage 50th percentile” and pick the first series. Click on “Edit Graph,” open the “Add Line” tab, and search for the second series. For the second graph, proceed similarly by searching for “large bank credit card balances.”

Suggested by Jeremy Cohn.

The FRED graph above plots the consumer price index for motor vehicles: Clearly, the prices of both new and used cars have risen over the course of the pandemic. What’s been driving this rise in prices?

On the one hand, the generous fiscal stimulus to mitigate the impact of the pandemic likely increased demand and put upward pressure on car prices. On the other hand, as the graph also shows, total domestic auto production has been and remains substantially below January 2020 levels, which suggests that supply-side factors have also likely played a role in the rise of car prices. With the demand for cars outpacing production during this period, car inventories were gradually depleted from early 2020 to the present, as also shown in the graph.

The outbreak of COVID-19 and its impact on workplace operations may account for the drop in production over the first few months of the pandemic, but it’s unlikely that these factors account for the persistent decline of domestic car production.

A popular narrative attributes this decline to semiconductor shortages. Semiconductor production indeed experienced significant pressure over the past couple of years, with demand spiking and manufacturers unable to scale up production fast enough to meet that demand. Because semiconductors have become a critical input for the production of cars, an input with few close substitutes, shortages of semiconductors can severely curb the production of new cars.

So, semiconductor shortages are likely playing a significant role. But the FRED graph below shows that the recent decline of car production falls broadly within the long-term downward trend over recent decades: Car production has generally been declining since the mid 1990s. This trend suggests that, even without the significant increase in demand following COVID-19, car production may still have continued to decline; moreover, car production may not revert to pre-pandemic levels even after the semiconductor shortages end and demand eases up.

How these graphs were created: Search FRED for “Consumer Price Index for All Urban Consumers: New and Used Motor Vehicles in U.S. City Average.” From the graph, click on “Edit Graph,” open the “Add Line” tab, and search for and select “Domestic Auto Production” and “Domestic Auto Inventories” Use the “Edit Lines” tab to change units to “Index (Scale value to 100 for chosen date).” Enter the base year date as “2020-01-01.” Select the “Copy to All” button to apply to all series. Set the earliest date in the window to “2020-01-01.” For the second graph, search FRED for “Domestic Auto Production.”

Suggested by Jason Dunn and Fernando Leibovici.

Next Monday is a holiday for the St. Louis Fed: since 1937, Columbus Day, and since last year’s presidential proclamation, Indigenous Peoples’ Day. This post, like so many others, offers timely examples of the data you can find in FRED—today, it’s specific to places in the U.S. bearing the name Columbus.

For these 3 metropolitan statistical areas and 1 county, we focus on the unemployment rate because it’s a good gauge of the health of a region and has units that don’t depend on the region’s size. And, as it turns out, these four regions are quite different from each other:

• Columbus, Ohio, is a major city with substantial public employment, led by Ohio’s state government and The Ohio State University.
• Columbus, Indiana, concentrates more on manufacturing.
• Columbus, Georgia (plus a county in Alabama), is remarkable for its military focus.
• Columbus County, North Carolina, is rural with a declining population.

The FRED graph above shows that the unemployment rate in these four regions varies quite a bit over time but, remarkably, consistently maintains the ranking of these four regions. (The brief exception is when the rate in Columbus, Indiana, shot up at the height of the pandemic: The many factories there didn’t have a work-from-home option and closed temporarily.)

How this graph was created: Search FRED for “Columbus unemployment” and select one of the four, making sure not to take the smoothed series. From the “Edit Graph” panel, open the “Add Line” tab and repeat the search until the graph is complete.

Suggested by Christian Zimmermann.