We no longer support this browser. Using a supported browser will provide a better experience.

Please update your browser.

Close browser message

RESEARCH How families used the advanced Child Tax Credit

On March 11, 2021, President Biden signed the American Rescue Plan Act of 2021 (ARPA) into law. The stimulus bill aimed to deliver direct relief to families and workers continuing to struggle financially during the ongoing COVID-19 pandemic. ARPA included provisions for an additional round of economic impact payments (EIP), state and local fiscal recovery funds, homeowner and rental assistance funds, and programs to help small businesses and unemployed workers 1. A notable component of ARPA was the expansion of the existing Child Tax Credit (CTC) program. Among other things, this expansion introduced automatically disbursed cash payments to the overwhelming majority of families with children under 17. It was both novel in its structure and unprecedented in its scale—characterized by its designers as “the largest child tax credit ever and historic relief to the most working families ever” 2. Accordingly, the program provides a unique opportunity for policymakers to understand how effective future policies like the CTC might be.

Other studies have tried to measure the impact of the CTC expansion on household finances, particularly the impact of the expansion on household consumption. Studies leveraging data from the Census Bureau’s Household Pulse Survey show that recipients report using these advanced payments to pay for bills and other living expenses and establish differences in spending patterns by income and race (Pilkauskas and Cooney 2021; Karpman et al. 2021; Pilkauskas et al. 2022). However, these studies are not able to quantify how much of the advanced CTC payments households spent—a critical question for policymakers concerned with balancing the goal of increasing the welfare of families with children against concerns about potential adverse inflationary or labor market effects of expansions to CTC. 3

This report seeks to fill this gap by using transaction-level data to estimate the impact of these accelerated payments on household spending. We explore a range of spending categories, including spending on durable goods, services, and debt payments. We also analyze whether spending responses differed by type of household—whether changes were more pronounced for lower liquidity households, or households of different races. Given the unprecedented nature of this program, ongoing assessments of the credits’ impacts on household finances will be important in understanding the program’s success and planning similar programs in future 4.

The Advanced Child Tax Credit

ARPA introduced several specific updates to the Child Tax Credit. The credit amount was increased from $2,000 to $3,600 for children under age 6 and to $3,000 for other eligible children; eligibility was expanded to include 17 year-olds; the credit became fully refundable; and 50 percent of the credit was scheduled for automatic disbursal via monthly payments between July and December 2021. Families were automatically enrolled in the advanced CTC payments and the IRS provided a channel to opt out of advanced CTC payments for families that preferred to receive the full tax credit when filing their 2021 tax returns 5. Married couples with income up to $150,000 (or $112,500 for single parents) were eligible for the full credit; married couples with income under $400,000 (or $200,000 for single parents) qualified for at least $2,000 of the CTC, with $166 per child each month disbursed via advanced payments 6.

Advanced CTC payments substantially increased the income of families that received them—particularly lower income families. Among the recipients in our sample, the median household received $300 per month for households earning less than $73,000 in annual take-home income (the lowest three income quartiles; refer to the appendix for income quartile details), and $500 per month for households earning above that threshold (Figure 1). This difference is due to differing family composition across the income groups. Only 38 percent of the lowest-earning CTC recipients had more than one dependent child, compared to 61 percent of the highest-earning recipients (46 and 53 percent for the middle quartiles, respectively) 7. Despite receiving, on average, greater monthly amounts, the overall income impact was lowest for the highest-earning households, whose monthly income increased by only 4 percent with the CTC payments. By contrast, the advanced CTC payments represented a 10 percent increase in monthly income for the lowest-earning households (those with annual take-home income less than $31,000). So while lower-earning households typically received lower payment amounts (due to the presence of fewer children per household), that amount had a greater impact on monthly income and may have therefore been felt more keenly.

Figure 1: Advanced CTC payments boosted monthly income by 10 percent for the lowest-earning recipients.

Finding One: Households spent 40 percent of their July advanced Child Tax Credit payments within one week, transferred 18 percent to other accounts, and held 41 percent as cash in their checking accounts.

To understand the impact of advanced CTC payments on household consumption, we begin by comparing the spending behavior of CTC recipient households to the behavior of non-recipient households. To isolate the effect of the CTC payments themselves, we control for household-level differences in spending levels, as well as trends in spending levels over time that are common to both groups (see appendix for details). After removing these trends, we suppose that the spending behavior of non-recipients is similar to what recipients would have done had they not received the CTC payments.

Figure 2 shows how total spending of CTC recipients evolved week-to-week, relative to non-recipients. This can be thought of as “excess” spending among CTC recipients relative to non-recipients and is measured as a fraction of the household’s CTC payment: if a household received a $200 CTC payment and then spent $100 more than it would have otherwise, its excess spending in Figure 2 would be 0.5. Our measure of spending here is the sum of all payments made from each household’s accounts via debit card, paper checks, cash withdrawals, electronic payments, and Chase credit cards, excluding debt payments 8.

We see a clear spending response after every advanced CTC payment (Figure 2). We also observe a downward sloping trend, indicating that apart from the spending spikes after each CTC payment, recipient households generally decreased their spending relative to non-recipient households in the months after the first CTC payment. This might be because CTC-recipients are more likely to be households with children who received larger EIP amounts in March 2021 and their spending is decelerating after that influx of cash, or some other behavior specific to households with children. It may also be that the large spending response after each payment is pulling some future spending forward in time. In the appendix, we present evidence that suggests that the former explanation—differing trends between households with and without children—is more likely.

Figure 2: CTC recipients’ spending spiked in each week of advanced CTC payments.

We can also use Figure 2 to calculate the marginal propensity to consume (MPC) out of CTC payments. This is how much of the average CTC payment was spent within the first week after receipt. Because of the observed downward trend in spending among CTC households, we measure the MPC for each payment by comparing the CTC recipients’ excess spending in the week the payment was received to their excess spending in the prior week. That is, the MPC out of the July 14 payment is the difference between the dot in Figure 2 for July 14 and the dot for July 7.

Using this approach, we calculate a 40 percent MPC for the first advanced CTC payment in July (Figure 3). In other words, within the first week of receipt, households consumed 40 percent of their CTC payment via spending on credit and debit cards, checks, cash, and electronic payments.

Figure 3: Recipients spent 40 percent of their July advanced CTC payments in the first week.

Beyond the July payments, we observe spending MPC values ranging between 21 percent (November) and 33 percent (December) for the advanced CTC payments 9. However, it is difficult to conceptualize marginal propensity to consume in a situation of monthly cash disbursals, particularly beyond the first month of the series. CTC payments land every month and are not (on average) fully consumed each month, meaning that household cash-on-hand in a given month is impacted by the amount of CTC payment and associated MPC from the previous month. It is therefore difficult to assess a marginal propensity to consume in the latter months of the advanced CTC payment series. For clarity of concept and measurement, this report will focus results on MPC measurements out of the July advanced CTC payments.

Like with spending, we also observe that advanced CTC recipient households respond to the CTC payments by transferring more cash to other accounts than non-recipient households do. The accounts receiving these transfers could be owned by the household—such as savings accounts or non-Chase accounts—or accounts owned by other households, such as family members. CTC recipient households in our sample transferred 18 percent of their July CTC payments to other accounts (Figure 4a). Debt payments—to non-Chase credit cards, auto loans, mortgages, student loans, or other loans—were much less responsive to advanced CTC payments than spending or transfers. While debt payments exhibit clear response patterns during CTC weeks (Figure 4b), only 1 percent of July CTC dollars go toward debt payments.

Figure 4: Recipients transferred a notable portion of advanced CTC payments to other accounts (18 percent in July) and used some of the funds for debt payments (1 percent in July).

Finally, we differentiate types of spending within the total spending category (Figure 5). Most spending out of CTC payments goes toward non-durable goods, which accounts for more than three quarters of the total spending response (31 percent of the CTC payment) 10. Of that non-durable spending, roughly 12 percent goes toward groceries and fuel (or 4 percent of the CTC payment). Durable goods also exhibit a clear spending response during CTC payment weeks, but at notably lower levels (4 percent of CTC amount) 11. Healthcare spending also increases with CTC receipt, but while the increase is statistically significant, it small in magnitude (approximately 0.5 percent of CTC amount) 12.

Overall, households consumed 40 percent of July CTC payments via spending within one week of receipt, with the majority of that spending on non-durable goods; an additional 18 percent went toward transfers to other accounts, with most of the remainder held as cash savings in households’ checking accounts.

Figure 5: Recipients spent 40 percent of July CTC payments in the first week, transferred 18 percent to other accounts, and maintained most of the remainder as cash savings.

Finding Two: Liquidity was a stronger predictor of spending response than income; for all income quartiles, spending response roughly doubled for low-liquidity households compared to high-liquidity households.

Compared to households with substantial liquidity, cash-constrained households are more likely to cut spending upon job loss, or increase spending after receiving a tax refund (e.g. Farrell et al. 2019). To assess the sensitivity of the CTC spending response to liquidity, we compare the marginal propensity to consume of high liquidity families to that of low families. We measure the liquidity of a household by computing its cash buffer—the amount of cash it has on hand divided by its typical spending. By this metric, families with smaller cash buffers are more liquidity constrained (see the appendix for details on the construction of our cash buffer metric, as well as how cash buffer compares with income).

We find that the spending response to advanced CTC payments among low-liquidity households is significantly stronger than it is among high-liquidity households. We observe clear spending responses during CTC payment weeks across all cash buffer quartiles, though the level becomes more muted for higher buffer groups (Figure 6). Households in the lowest cash buffer quartile (with cash-on-hand covering less than 2 weeks of spending) consumed 73 percent of their July CTC payments via spending in the first week (Figure 7). By contrast, households in the highest cash buffer quartile (with cash-on-hand covering over 4 months of spending) consumed only 19 percent of their CTC payments via spending in the first week. Although the spending response decreases for each increase in cash buffer quartile, the households with the most cash-on-hand still spent a non-trivial fraction of the CTC payment in the first week.

Figure 6: CTC recipients’ spending spiked in each week of advanced CTC payments, much more so for low-liquidity recipients than their high-liquidity counterparts.

Figure 7: Recipients with the lowest liquidity spent 73 percent of their July CTC payments in the first week, compared to only 19 percent spent by the highest liquidity recipients.

Households’ cash-on-hand remains a strong predictor of spending response even when we account for household income (Figure 8). Among households with below-median cash buffers, the spending response was strongest for the lowest earners, but still notable for high-earning households, indicating that cash liquidity was a stronger predictor of spending response than income. The lowest earners (take-home income under $31,000) in this group spent a substantial share—80 percent—of July advanced CTC payments within the first week. The highest earners (take-home incomes greater than $73,000) still spent nearly 30 percent within a week.

Figure 8: Of recipients with below-median liquidity, the lowest-income households spent 80 percent of their July CTC payments in the first week, and the highest-income households still spent nearly 30 percent.

As with our analysis of the overall sample, we again break down consumption of advanced CTC payments across a set of categories, and assess how cash buffer differences contribute to trends in how households distribute their consumption across these categories (Figure 9). Within each cash buffer quartile, the relative size of the spending response by consumption category mirrors the overall results in Figure 5. Among active uses of CTC money—the categories listed above “other savings”—the majority went toward spending on goods and services, especially non-durable goods, and the next most common use was transfers to other accounts.

Households with lower cash buffers actively used a greater share of their CTC money, relative to households with higher cash buffers. That is, households with low cash buffers used a larger portion of their CTC payments via spending and debt payments, maintaining very little “unused” CTC money (2 percent, vs. 70 percent for high-buffer households). While this “other savings” bucket represents additional cash savings in the household’s accounts, it also demonstrates a lack of active engagement with the received funds. High-liquidity households appear especially unresponsive to advanced CTC payments. They not only consume less of their CTC via spending relative to low-buffer households (19 percent vs. 73 percent), but also make proportionally smaller debt payments and transfers to other accounts.

Figure 9: Low-liquidity recipients consumed a larger portion of their July CTC payments via spending and debt payments, and maintained very little as cash savings (2 percent vs. 70 percent for high-liquidity recipients).

Finding Three: Differences in CTC consumption response by race are substantially smaller after accounting for differences in liquidity.

Previous JPMorgan Chase Institute research has documented racial gaps in various financial outcomes. In particular, the Institute observed large racial differences not only in income, but also in liquid assets, which we have just shown to be a key driver of CTC consumption response. While existing research has assessed racial differences in CTC response (e.g., Karpman et al. 2021), our data uniquely positions us to add to this conversation given our ability to observe both race and liquidity to further contextualize observed differences by race.

Analyzing the subset of households for whom we have self-reported race information (see appendix), Black households have the highest spending response, spending about half of CTC payments within the first week (Figure 11). The response was lower for Hispanic and White households, with White households having the smallest response. However, while the difference in these responses is large in an economic sense, we cannot statistically distinguish the spending responses of Hispanic households and White households (the p-value of a Wald test of these parameters is 0.33). 13

Figure 10: CTC recipients’ spending spiked in each week of advanced CTC payments, with higher July spending responses for Black and Hispanic recipients than their White counterparts.

Figure 11: Black recipients spent 53 percent of their July CTC payments in the first week, compared to 34 percent for Hispanic and 25 percent for White recipients (though Hispanic-White differences are not significant).

We next assess how much of the differences in spending response across race groups may be driven by other underlying differences across the groups. For example, households’ cash buffers differ significantly by race: 30 percent of Black households have a cash buffer in the lowest quartile of the population, compared to only 23 percent of White and 20 percent of Hispanic households (see appendix, Figure A4). 

The differences in spending response by race are partially explained by differences in cash buffers (Figure 12). For households with above-median cash buffers, differences in spending response by race are no longer statistically significant. Large standard errors result in wide confidence intervals, and an MPC that is not statistically distinguishable from zero for White households. For households with below-median cash buffers, Black households spent the largest share of their July CTC payments within one week (81 percent) followed by Hispanic and White families (61 and 48 percent, respectively). As in the aggregate results above, differences between White and Hispanic households are not statistically significant. However, the response of Black households is still significantly different from that of White households (p-value 0.009) and marginally significant compared to Hispanic households (p-value 0.073). 

This remaining difference is likely driven by several factors. First, even among households with below-median cash buffers, there are still differences across race groups in cash-on-hand. Appendix Figure A4 shows that among below-median households, Black households are relatively more likely to be in the first quartile. Second, other financial characteristics affect the CTC response as well; as we show in Figure 8, income still affects the spending response even after conditioning on liquidity. Finally, there may be other race-specific factors not captured by income or cash buffer that affect the CTC spending response, such as differences in familial or intergeneration wealth (Farrell et al. 2020).

Figure 12: Racial differences in spending responses to July CTC payments are substantially smaller among households with similar liquidity.

Finding Four: Implications

Families with children may spend substantial shares of payments from programs like the Advanced Child Tax Credit. Across groups, we found that advanced CTC recipients spent 40 percent of their July CTC payments, the majority of which went to non-durable goods. Notably, many of these families were already holding unusually high cash balances due to Economic Impact Payments. Given the sensitivity of the consumption response to cash liquidity, it is possible that families would spend an even higher share of future payments in the context of household liquidity closer to historical levels. 

To reach families that are most likely to utilize the payments, policymakers might consider programs that target families that are low-liquidity in addition to being low-income. We found that low-liquidity families were significantly more likely to spend out of advanced CTC payments than were high-liquidity families. Cash liquidity was a very strong predictor of spending response. Recipients with the lowest liquidity—deposit account balances totaling 2 weeks of spending or less—spent 72 percent of their July CTC payments. Moreover, they actively engaged with most of their July CTC payments, maintaining only 2 percent as cash in the checking account. Income did differentiate consumption responses among low-liquidity families. Low-income recipients with below-median liquidity had the strongest response of any group, spending 80 percent of their July CTC payments within one week. While high-income recipients exhibited a notably strong spending response when they had below-median liquidity, they spent just under 30 percent of this first payments over the same week. 

High-liquidity families were notably unresponsive to advanced CTC payments. In contrast to low-liquidity families, recipients with the highest liquidity—who had nearly 5 months of typical spending in their deposit accounts—actively engaged with only 30 percent of July payments, leaving the remaining 70 percent of the payment untouched in their checking accounts. These families only transferred 2 percent of July CTC payments to other accounts, while low-liquidity families transferred 18 percent. While we characterize both transfers and cash left in checking accounts as forms of savings, transfers reflect a more active response to a payment.

The spending response to a similar future program is likely to be larger than what we find here because cash balances were historically high in 2021 due to the pandemic. As recent Institute work has shown, cash balances were elevated in the second half of 2021 when advanced CTC payments landed. This was especially true for lower-income families and younger families. Given our finding that families respond less to CTC payments when they have high cash buffers, similar initiatives are likely to see higher spending responses if they occur when cash buffers are no longer generally elevated.

Underlying differences in liquidity are also largely responsible for observed aggregate differences in consumption response by race. The consumption response to advanced CTC payments differed by race and ethnicity, with Black families spending a larger proportion of their July payments within the first week than Hispanic and White families (the latter two of which did not significantly differ). However, after accounting for differences in household liquidity, racial differences in spending responses to July CTC payments for high-liquidity households were substantially smaller. This again underscores the importance of accounting for liquidity in targeting future cash disbursal initiatives.


Chris Wheat


Erica Deadman

Consumer Research Lead

Daniel M. Sullivan

Consumer Research Director