Accessible Version
Debt Payments and Spending: Evidence from the 2023 Student Loan Payment Restart, Accessible Data
Figure 1. Student Loan Balances and Demographic Relationships
Figure 1a: College Share and Average Income
Figure 1a, left panel:
This panel shows a bin-scatter of the share of U.S. adults with college degrees from the American Community Survey (y-axis) across deciles of ZIP codes by student loans per capita (x-axis). The share of college-educated adults is strictly increasing by deciles starting from around a share of 0.2 in the first decile to a little more than 0.35 in the tenth decile.
Figure 1a, right panel:
This panel shows a bin-scatter of the average income in tens of thousands of dollars from the IRS Statistics of Income (y-axis) across deciles of ZIP codes by student loans per capita (x-axis). The average income of ZIP codes is increasing in deciles although with some nonlinearities visible between the second and ninth deciles. The average income in the first decile is about $65,000 while the average income in the tenth decile is a little more than $90,000.
Figure 1b: Aged 25–34 and Subprime Share
Figure 1b, left panel:
This panel shows a bin-scatter of the share of U.S. adults between the ages of 25 through 34 from the ACS (y-axis) across deciles of ZIP codes by student loans per capita (x-axis). The share of adults between the ages of 25 through 34 is increasing by decile with the first decile at a share of around 0.11 and the tenth decile at a share of around 0.17. There is strong nonlinearity present moving up in deciles, with a large jump between the ninth and tenth decile. More specifically, the share in the ninth decile is around 0.135 while the tenth decile is around 0.17; however, the eighth decile is 0.13.
Figure 1b, right panel:
This panel shows a bin-scatter of the share of U.S. subprime borrowers from the FRBNY/Equifax Consumer Credit Panel (y-axis) across deciles of ZIP codes by student loans per capita (x-axis). The share of subprime borrowers is decreasing by decile with the first decile at a share of around 0.28 while the tenth decile is around 0.23. The minimum share of subprime borrowers occurs in the ninth decile, at a share less than 0.22.
Note: This figure shows the relationship at a ZIP code level of student loan per capita deciles against mean income, the share of college degrees, share of individuals aged 25-34, and share of subprime borrowers.
Source: FRBNY/Equifax Consumer Credit Panel, American Community Survey, and IRS Statistics of Income.
Figure 2. Dynamic Response of Student Debt on Spending
This figure shows a line chart displaying the estimated weekly spending response of $10,000 of student loan debt (y-axis) over weeks in the period January 2022 through April 2024 (x-axis). The dots connected by the solid line show the point estimates of responses, and the dashed lines show 95 percent confidence bands around these estimates. Between January 2022 and June 2023, the pre-announcement period, the point estimates hover close to zero, going from $2 in the first week of January 2022 to -$6 in the first week of June 2023 with a maximum of $6 occurring in the first week of January 2023. The 95 percent confidence bands over the same timeframe range from about $15 to -$15 and strictly contain zero. Between June 2023 and October 15, 2023, the point estimates are strictly negative ranging from -$1 to -$10. The 95 percent confidence bands on the same time frame range from around $8 to around -$17. After October 15, 2023 and through April 2024, the point estimates steadily decrease from -$5 to a minimum of -$19 in the last week of March before ending at -$14 in the last week of April. The 95 percent confidence bands decrease in a similar fashion as upper band decreases from $5 to a minimum of -$5 in the last week of March and settles around -$2 in the last week of April. The lower band starts at -$15 and decreases to a minimum of -$30.
Note: This figure shows the evolution of spending following the resumption of student loan repayment for all 18,178 ZIP codes in the model sample, which is obtained from running the model specified in Equation 1. The dotted lines signify 95 percent confidence bands using standard errors clustered at the ZIP code-level. The thin solid burnt orange line signifies June 2, 2023—the date at which Congress passed legislation blocking President Biden’s plan for student loan relief. The thick solid blue line signifies September 1, 2023, which is the date at which interest began to accrue on student loan balances. The purple dashed line is October 15, 2023, the period in which monthly payments for student loan balances restarted following the end of student loan forbearance.
Source: Authors’ calculations using Verisk Commerce Signals Spend Tracker, FRBNY/Equifax Consumer Credit Panel, American Community Survey, and IRS Statistics of Income.
Figure 3. Average Response of Student Debt on Spending
This figure shows a dot and whisker plot of the average response per $10,000 of student loan debt (y-axis) over three different time periods (on x-axis): 1) “Pre-Announcement,” defined as June 2, 2023 when Congress passed legislation blocking President Biden’s plan for student loan relief; (2) “Post-Announcement,” defined as after June 2, 2023 and up until payment resumption on October 15, 2023; (3) “Payment Resumption,” defined as after October 15, 2023. Dots represent point estimates, and whiskers show the 95 percent confidence interval. The “Pre-Announcement” period is set as the base period and is zero. The “Post-Announcement” period is estimated at -$6.20 with 95 percent confidence bands ranging from -$11 to -$2. The “Payment Resumption” period is estimated at -$12.20 with 95 percent confidence bands ranging from -$17 to -$7.
Note: This figure shows the evolution of spending over for 18,178 ZIP codes over three time periods: (1) “Pre-Announcement”, defined as June 2, 2023 when Congress passed legislation blocking President Biden’s plan for student loan relief; (2) “Post-Announcement”, defined as after June 2, 2023 and up until payment resumption on October 15, 2023; (3) “Payment Resumption”, defined as after October 15, 2023. We obtain the estimates by running the model specified in Equation 1 using “Pre-Announcement” as the base period. We include 95 percent confidence bands for the “Post-Announcement” and “Payment Resumption” periods using standard errors clustered at the ZIP code-level.
Source: Authors’ calculations using Verisk Commerce Signals Spend Tracker, FRBNY/Equifax Consumer Credit Panel, American Community Survey, and IRS Statistics of Income.