Schild et al., Spending Response to the Expanded Child Tax Credit: An Analysis Using U.S. Consumer Expenditure Survey Data (BLS Working Paper 577, 2024-11-05)

Summary

BLS Working Paper 577, dated November 5, 2024, titled Spending Response to the Expanded Child Tax Credit: An Analysis Using U.S. Consumer Expenditure Survey Data, by Jake Schild, Sophie M. Collyer, Thesia Garner, Neeraj Kaushal, Jiwan Lee, Jane Waldfogel and Christopher T. Wimer. The abstract states that the authors use data from the Consumer Expenditure Interview Survey and a parameterized difference-in-difference design to examine whether the expanded CTC increased household spending. It reports that for each $100 of CTC payment, households spent $44, mainly on housing ($28) and food ($12), and spent $16 per $100 in a subset of child-related expenditures. It also states that the increase in child-related spending was larger for Black- and Hispanic-headed households than for White-headed households.

Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used

Cite this document

“Schild et al., Spending Response to the Expanded Child Tax Credit: An Analysis Using U.S. Consumer Expenditure Survey Data (BLS Working Paper 577, 2024-11-05).” Original at www.bls.gov. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/schild-et-al-spending-response-to-the-expanded-child-tax-credit-an-ana-e3885b19ba/, SHA-256 d9a583e904f951e20802f80aee75fbea4a69e26d036c9b2ff7dd64903c3f61b4.

Full text

BLS WORKING PAPERS

U.S. Department of Labor
U.S. Bureau of Labor Statistics
Office of Prices and Living Conditions

Spending Response to the Expanded Child Tax Credit:
An Analysis Using U.S. Consumer Expenditure Survey Data

Jake Schild, U.S. Bureau of Labor Statistics
Sophie M. Collyer, Columbia University
Thesia Garner, U.S. Bureau of Labor Statistics
Neeraj Kaushal, Columbia University
Jiwan Lee, Columbia University
Jane Waldfogel, Columbia University
Christopher T. Wimer, Columbia University

Working Paper 577
November 5, 2024
Spending Response to the Expanded Child Tax Credit:
An Analysis Using U.S. Consumer Expenditure Survey Data

Key Words: Child Tax Credit, Spending Response, Fiscal Policy
JEL Codes: D04, I38
November 5, 2024

Abstract
The Child Tax Credit (CTC) was substantially expanded through the American Rescue Plan Act of 2021.
Early studies documented that the expanded CTC reduced poverty and food insufficiency, but there is
little research on its impact on household spending, particularly child-related spending. We use data from
the Consumer Expenditure Interview Survey and a parameterized difference-in-difference design to
examine whether the expanded CTC increased spending overall, in major categories, and on specific
items related to children's education and development. Our findings indicate that households used the
CTC payments to enhance the well-being of both their children and the entire household. For each $100
of CTC payment, our models show that households spent $44, mainly on housing ($28) and food ($12).
In a subset of child-related expenditures, households spent $16 per $100. We also find that the increase in
child-related spending was larger for Black-and Hispanic-headed households than for White-headed
households.

1
1. Introduction
The American Rescue Plan Act of 2021 (ARP), enacted in March 2021, significantly expanded
the Child Tax Credit (CTC) in three major ways. First, it made the benefit more generous, increasing the
maximum benefit size from $2,000 per child to $3,000 per child for children aged 6 to 17 years and
$3,600 per child for children aged 0 to 5.1 Second, it made the benefit “fully refundable,” meaning that
tax filers were able to receive the full credit regardless of their tax obligation. Third, it allowed
households to receive up to half of their full credit in monthly installments delivered from July to
December of 2021 with the remainder delivered at tax time in 2022. As a result of the expansions, from
July through December 2021, most low- and middle-income households with children in the United
States (U.S.) received monthly cash payments of $300 per child under age six and $250 per child between
ages of 6 and 17.2
Though now expired, the ARP CTC broadened the CTC’s reach and role in addressing the
economic hardship of low-income households with children. Early studies analyzing the ARP CTC
established that the expansion drove the child poverty rate to the lowest rate on record (Creamer et al.,
2022; Marr et al., 2021), reduced food insufficiency (Parolin, Ananat, et al., 2023; D. J. Perez-Lopez,
2021), and lowered rates of material hardships (Collyer et al., 2022; Pilkauskas et al., 2022). Findings
related to the effect of the 2021 ARP CTC expansion on parental employment are mixed, with some
studies not finding evidence of an effect (Ananat et al., 2022; Roll et al., 2022a), and others finding a
small decline in employment among certain subgroups of parents and caregivers (Pac & Berger, 2024;
Schanzenbach & Strain, 2023).3 Using data from the Consumer Expenditure Interview Survey (CE), this
study adds a new set of outcomes to this body of literature by examining the effects of the CTC expansion
on household spending across several domains, including child-related spending.4
The family investment model posits that income affects children’s well-being and long-term
outcomes. As parents receive more income they are able to purchase items and services that enhance their
children’s development (e.g., book and enriching activities) as well as meet basic needs, for example,
alleviating food insufficiency and housing insecurity (Wimer & Wolf, 2020; Yeung et al., 2002).
Therefore, determining whether and how households spent the additional income received from the
monthly CTC payments is pertinent to evaluating the CTC expansion and its potential effects on

1 Before the ARP, eligible children included those under age 17 with a Social Security Number (SSN) who could be claimed as a

dependent. Under the ARP, eligible children included those under age 18 with an SSN who could be claimed as a dependent.
2 Households with qualifying dependents who had previously filed taxes were automatically enrolled to receive the credit while

non-filers (typically the lowest income households) had to sign up to receive it using a portal managed by the Internal Revenue
Service (IRS). The first monthly payment was distributed to the households of 59.3 million children in July 2021, while the last
payment reached 61.2 million children in December 2021 (United States Department of Treasury, 2021).
3 Curran et al. (2024) provide a comprehensive overview of research looking at the impact of the ARP CTC.
4 While the CE collects data from consumer units,4 we use household and consumer unit interchangeably in this study.

2
children’s well-being and development. Policymakers have also raised questions as to how households
spent the monthly payments and if CTC payments were spent in ways that did or did not promote
children’s well-being. There have been investigations into how parents spent the CTC income using
survey reports (Karpman et al., 2021; D. Perez-Lopez & Mayol-García, 2021; Pilkauskas et al., 2022;
Pilkauskas & Cooney, 2021; Rachidi, 2021; RAPID-EC, 2021; Roll, Chun, et al., 2021; Zippel, 2021),
transactional-level data (Wheat et al., 2022; Parolin et al., 2022), and geo-location data tracking visits to
different establishments (Parolin et al., 2022). However, to our knowledge, no study has used nationally
representative item-level 5 expenditure data to analyze changes in spending on items across a host of
different domains, including on food, housing, or child-related spending.
Comprehensive spending data from the CE allow us to examine the spending response to
expanded CTC payments across several domains. We harness the CE data to examine the household
spending response to modeled, as opposed to reported, monthly payments on (1) major expenditure
categories, such as food, housing, alcohol and tobacco, leisure, etc., and (2) child-related spending
including that for books, clothing, childcare, computers and tablets, and enrichment activities.
Following many prior studies of policy effects (Currie & Gruber, 1996; Hoynes & Patel, 2018,
2018; Jones et al., 2019; Michelmore & Pilkauskas, 2021; Pilkauskas et al., 2022), we predict how much
different households would likely benefit from the policy change based on their characteristics prior to the
policy change and incorporate this predicted CTC value into a difference-in-differences framework to
identify the expansion’s effect on spending across domains.6 Our identification strategy relies on variation
in spending between households who could have potentially received the payments and those who could
not and on variation in payment size. Our estimates draw on CE data collected in reference to spending in
2021 and 2019. To avoid potential biases arising from differences in spending patterns across household
composition and income levels, we restrict our sample to households who have qualifying children and
are income eligible for the CTC. We designate all households reporting on spending in 2021 as
“potentially treated,” while the “non-treated” or “control group” were selected from interviews conducted
in reference to 2019. This choice allowed us to control for seasonal spending variation correlated with the
period monthly CTC payments were delivered.7 Our models estimate the difference in spending levels
based on payment size for potentially treated households in the treatment period (i.e., the second half of
2021). Our main source of variation in payment sizes comes from the reference quarter for which

5 Note that when we refer to “items,” we also are referring to child-related services, such as enrichment activities.
6 It was necessary to impute the value of the CTC because the CE Survey did not collect data on the amount of CTC received.
7 We considered a model that relies only on the variation in spending between recipients and non-recipients (i.e., a binary

treatment), as opposed to one exploiting variation in payment size, but this strategy drops a significant portion of the variation
needed to identify the treatment effect across spending subdomains and demographic groups. However, we analyzed total
spending for all households in our sample using the binary treatment model and found the results to be consistent with the
continuous treatment model.

3
respondents reported their expenditures (they could have received 1, 2, or 3 months of payments in the
quarter), and the fact that the monthly payments for children under age 6 were $50 greater than those for
children of ages 6-17.
Our findings suggest that households spent the monthly CTC payments on goods and services
known to improve child and household well-being. We find that households spent $44 of each $100 of
imputed CTC payment, with the largest portions devoted to housing ($28) and food ($12). This result
indicates that households spent most of their benefits to meet basic needs and thereby promote material
well-being. We also conduct a separate analysis on a subset of child-related goods and services and find
evidence that supports the family investment model, with households spending $16 of each $100 CTC
payments on child-related goods and services that directly enhance child development and well-being. We
validate our model using a placebo test in which we change the treatment year to 2018—a year when
there was no treatment—and keep 2019 as the comparison year. The results of this test show coefficients
close to zero and statistically insignificant, reinforcing the conclusion that our main findings are driven by
the expanded, monthly CTC, rather than by a spurious correlation or general trend of increased spending
in the latter half of the year.
Prior work suggests that non-Hispanic Black children and Hispanic households with children
benefitted more from the ARPA CTC expansions than White children (Collyer et al., 2019; Goldin &
Michelmore, 2022). We also find that the increase in spending on child-related goods and services was
larger for non-Hispanic Black- and Hispanic-headed households, as well as households where the head
identified as an ‘other’ racial group, compared to White-headed households. We estimate a statistically
significant $47 increase in child-related spending per $100 in CTC for households with a head identified
as non-Hispanic ‘other’; a $27 increase for non-Hispanic Black-headed households; a $12 increase for
Hispanic-headed households; and an insignificant $10 increase for White-headed households. Taken
together, these results suggest that households that benefitted disproportionately from the ARPA CTC
expansion allocated a greater share of their CTC income to child-related spending.
2. Background

The 2021 Child Tax Credit Expansion
Prior to the passage of the ARP, the CTC’s most recent parameters were established by the Tax
Cuts and Jobs Act of 2017 (TCJA). Under the TCJA, tax filers received a maximum CTC of $2,000 per
child per year.8 However, the credit was not fully refundable. It was phased in with earnings, and tax
filers claiming dependent children needed to earn a minimum amount to qualify for any benefit, and

8 For additional information on the history of the Child Tax Credit see Crandall-Hollick (2021), Crandall-Hollick (2018), and

Garfinkel et al. (2016).

4
higher earnings to qualify for the maximum credit.9 As a result, one in three children were ineligible for
the full benefit value because their parents did not earn enough. Children with single parents, those in
rural areas, non-Hispanic Black and Hispanic children, and those in larger households were
disproportionately ineligible for the full credit (Collyer et al., 2019; Curran & Collyer, 2020; Goldin &
Michelmore, 2022).
The ARP temporarily transformed the CTC into a nearly universal child allowance through three
fundamental changes:10 (1) expanded eligibility, (2) higher credit amounts, and (3) delivery of credit in
monthly installments for the second half of 2021. Eligibility was expanded to almost all children,
including children in low-income households who were previously excluded, by removing the earnings
requirement, and by making the credit fully refundable. Figures 1a and 1b plot the credit amount a
household qualified for as a function of adjusted gross income (AGI) according to the TCJA and ARP
legislation, respectively. The figures shows that the maximum annual credit amounts were increased from
$2,000 per child under the TCJA parameters to $3,000 for children ages 6-17 and $3,600 for children
under age 6 under the ARP (Figure 1, panel b).11 Households with incomes below $112,500 (heads of
household) or $150,000 (joint filers) were eligible for the higher ARP maximum credits; above these
levels, the credit began to phase out at a rate of 5% per dollar of AGI until reaching the credit amount that
households were eligible for under prior law. That is, no household received a smaller credit than they
would have under the TCJA, but the highest income households were not eligible for the higher
maximum credits ($3,000 or $3,600 per child depending on age). This created a two-step phase out
structure depicted in Figure 1b.

9 See Curran and Collyer (2020) for details on how much a filer needed to earn to qualify for the full credit.
10 The changes to the CTC in the ARP follow those outlined in the American Family Act (a bill first introduced in both the Senate

and House of Representatives in 2017 and reintroduced in 2019) with one exception: in the AFA, the credit would begin to phase
out for heads of household with earnings above $120,000 or and joint filers with Adjusted Gross Incomes (AGI) over $180,000.
In the ARP, the credit began to phase out for households with AGIs above $112,500 or $150,000 per year, depending on filing
status, but it only phased out until matching the credit values that a household would receive under prior law. This alteration was
made because the Biden administration committed to not raising taxes for those with incomes below $400,000 per year
11 Included within this change is an increase in the qualifying maximum age from 16 to 17.

5
Figure 1: Child Tax Credit as a Function of AGI and Household Size

Note: All values are the authors’ calculations for households where Adjusted Gross Income is equivalent to earned income. The
ARP increased the qualifying age for children from less than 17 years-old to less than 18 years-old. Additionally, the ARP
provides a larger credit for children under the age of 6, $3,600 instead of $3,000. Figure (a) also applies to households with
children less than 6 years old, and Figure (b) also applies to households with children 17 years old.

Under the ARP, households also received the CTC through advanced, monthly payments.
Beginning on July 15, 2021, the IRS delivered the credit in advanced, monthly installments of up to $250
per child 6 to 17 years old or up to $300 per child under the age of 6 years old, for a period of six
months.12 Note that although the ARP expanded eligibility for the CTC, monthly payments did not reach
all eligible households. Payments were distributed automatically based on prior-year tax information.
Households who did not file taxes in the prior year, presumably due to having an income below the tax-
filing threshold, generally needed to register with the Internal Revenue Service (IRS) in order to receive
their monthly credit. Data from the U.S. Department of the Treasury (2021) indicate that payments were
distributed to 59.3 million children in July and increased steadily until it reached 61.2 children in
December.13 Although data on the number of children eligible for the monthly payments were not
publicly available at the time of the study, estimates suggest that the monthly payments reached between
roughly 88.5 percent and 91 percent of eligible children (Curran, 2022; Parolin, Collyer, et al., 2021).

12 Because the payments began halfway through the year, households received half of the full amount of their credit in 2021

through monthly advanced payments. The remainder of the credit was received when households filed taxes in 2022. Note,
children born in 2021 were not eligible for monthly payments. Their households could claim the entirety of the credit at tax time.
13 U.S. Department of the Treasury issued monthly reports about the disbursement of the advanced CTC payments by state. Use

the following link to see the July report: https://home.treasury.gov/system/files/131/Advance-CTC-Payments-Disbursed-July-
2021-by-State-07142021.pdf

6
The link between spending and child outcomes
Additional income provided through cash transfers like the CTC can be characterized as
improving children’s long-term outcomes through two channels: family stress and family investments
(Wimer & Wolf, 2020). The family stress model posits that economic hardship impairs family
functioning and increases parents’ stress, undermining parents’ mental health and ultimately children’s
development (Conger & Conger, 2002). Increased income can thus reduce family stress and improve
child outcomes. The family investment model posits that increased income allows parents to purchase or
invest in various things that enhance child development and well-being and meet children’s basic needs
(e.g., books, toys, enriching activities, high-quality childcare, or nutritious food) (Duncan et al., 2011;
Yeung et al., 2002). Thus, parents’ spending on children represents an important pathway to promote
children’s development and well-being (Jackson & Schneider, 2022; Kaushal et al., 2011; Kornrich &
Furstenberg, 2013; Schneider et al., 2018). This study focuses on the family investment model by
investigating whether the expanded CTC affected household expenditures, overall, in major categories,
such as food and housing, as well as expenditures on children.
Related research
There are several streams of research that document spending responses to changes in income.
Some examine changes in response to tax refunds and rebates (Johnson et al., 2006; Parker et al., 2022;
Souleles, 1999), while others focus specifically on the receipt of the Earned Income Tax Credit (Barrow
& McGranahan, 2000; Gao et al., 2009; Goodman-Bacon & McGranahan, 2008; Halpern-Meekin et al.,
2015), in-kind benefits (Hastings & Shapiro, 2018; Smith et al., 2016; Beatty & Tuttle, 2015), cash
transfers (Jones et al., 2019; Najjarrezaparast & Pendakur, 2021; Gregg et al., 2006), the Alaska
Permanent Income Dividend (Amorim, 2021), and payments received from the Baby’s First Years
randomized control trial (Gennetian et al., 2022).
Research suggests that households with higher liquidity constraints spend a larger share of their
tax refunds (Souleles, 1999), and spending from lump-sum payments like tax refunds and the EITC are
disproportionately put towards durable goods (Halpern-Meekin et al., 2015; Gao et al., 2009; Goodman-
Bacon & McGranahan, 2008; Barrow & McGranahan, 2000; Souleles, 1999), though there is also
evidence that these payments increase spending on nondurables like food (McGranahan & Schanzenbach,
2013; Gao et al., 2009; Johnson et al., 2006). While most of these studies do not examine spending
specifically on child-related items, Gao et al. (2009) find evidence suggesting that EITC payments
increase spending on children’s clothing. A narrower body of literature that examines how households
spent more regular cash payments found that increases in regular cash payments lead to increased
spending on basic needs, childcare and children’s goods such as books, toys, and activities for children
(Gennetian et al., 2022; Gennetian et al., 2024; Gregg et al., 2006; Jones et al., 2019).

7
A subset of these previous studies also addresses the question, what share of the additional
income transferred to households did they spend in the period when they received it – that is, what was
their marginal propensity to consume (more accurately referred to as the marginal propensity to spend),
Jones et al. (2009) found that low-income Canadian Child Benefit recipients spent 45% of their benefit on
non-durable goods, and Johnson et al. (2006) found that 2001 Income Tax Rebate recipients spent 37% of
their rebate on non-durable goods. Looking at the Alaska Permanent Fund, Kueng (2018) found that
households spent roughly 25% of the additional receipt on non-durable goods in the quarter when
payments were made. More recently, a study analyzing the effects of the 2020 CARES Act found that
households spent approximately 46% of their April 2020 stimulus payments within two weeks of
receipt, with substantial differences by income: low-income households with low liquidity spent 62% of
their payment within two weeks, while higher-income households spent only 35% of their payment
during the same period (Karger and Rajan, 2020). A survey of stimulus check recipients by Coibion et al.
(2020) found that they spent 40% of their April 2020 stimulus payments by July 2020. In contrast, Parker
et al. (2022) found slightly lower spending responses, about 25%, for the first two rounds of stimulus
payments within the three-month reference period a payment was received. Altogether, these studies
suggest that the marginal propensity to spend is within the range of 25% to 46%.
Several studies have examined how parents and caregivers reported spending the CTC, but to our
knowledge, none of these studies has yet documented a causal relationship between CTC payments and
households’ item-level spending decisions. Studies relying on parents’ and caregivers’ reports of how
they spent their CTC largely find parents reporting spending their monthly CTC payments on food, bills,
and other basic household necessities (Karpman et al., 2021; D. Perez-Lopez & Mayol-García, 2021;
Pilkauskas et al., 2022; Pilkauskas & Cooney, 2021; Rachidi, 2021; RAPID-EC, 2021; Roll, Chun, et al.,
2021; Zippel, 2021). Perez-Lopez & Mayol-García (2021) found households were more likely to report
spending their monthly payments on school-related items in September and October of 2021, and those
with younger children were more likely to report using the payments to cover childcare costs. These self-
report studies, however, may be subject to social desirability bias or other general response biases, which
necessitate more causal research designs.
A handful of innovative studies aimed to surpass self-reported information by utilizing novel
data. For example, Parolin et al. (2024) used anonymized mobile-location data and debit/credit card data
and found that counties benefiting most from the CTC expansion show a greater increase in visits to
childcare centers and transactions in grocery stores and health- and personal- care establishments. Wheat
et al. (2022) used transaction-level data from JPMorgan Chase account holders and examined spending on
durable goods, services, and debt payments. The study found that recipients spent 40% of their July
advanced CTC payment (within the range of the MPCs from previous studies), transferred 18% to other

8
accounts, and used 1% for debt payments. However, these studies are limited in their ability to provide a
comprehensive picture of the specific types of goods and services households purchase with the CTC
income due to the specific types of spending data utilized.
This study is the first to examine changes in spending patterns across a host of spending
categories, and specifically on child-related items, using a causal identification strategy in a population-
representative dataset for the full period when the advanced CTC payments were in effect.14 Further, the
detailed CE spending categories are constructed based on item-level spending (as opposed to credit card
data or location-level spending which show total, but not item-level spending). This feature allows us to
examine changes in spending on specific types of items as opposed to spending in specific places or just
spending overall. We are also able to look at heterogenous treatment responses by household income level
and the race/ethnicity of the household head, making a unique contribution to the literature of the effects
of the CTC expansion.
To build on the previous analysis, this paper refines the findings presented in an earlier working
paper (Schild et al., 2023). In that prior work, the main model controlled for variation in expenditures by
CU size using the number of adults. In contrast, the principal model in this paper controls for the total
number of individuals in the CU. The estimated spending response when using the total number of
individuals in the CU is smaller than when using the number of adults. Our previous analysis also
included estimates based on both non-inflation-adjusted and inflation-adjusted expenditures, allowing for
an exploration of how the CTC helped households maintain a stable standard of living during a period of
high inflation. This paper, however, presents results solely using inflation-adjusted expenditures. Finally,
the Appendix of this paper includes estimates of the treatment-on-treated spending response.
3. Data and estimation strategy

Consumer Expenditure Interview Survey
This study uses internal, micro-level data from the CE collected between January 2019 and
March 2022.15 The CE is a nationally representative survey sponsored by the U.S. Bureau of Labor
Statistics that collects spending, demographics, and other financial information for households living in
the U.S.16 While the CE is composed of two data collections with their own independent samples, an

14 Our prior work (Collyer et al., 2022) uses a similar methodology, but was restricted to the first two months, July and August, of

when the advanced CTC payments were in effect, which is why the findings are described as “preliminary.”
15 CE Public Use Microdata (PUMD) can be used to reproduce our analysis (U.S. Bureau of Labor Statistics, 2023). The results

produced will be similar but will not exactly match due edits made to the PUMD to protect the confidentiality of respondents. In
the CE PUMD, income and expenditures are top and bottom coded and the state code maybe recoded or suppressed. Within the
data files are flag variables which indicate if a variable is top coded, suppressed, or adjusted in some other way.
16 As part of the U.S. Bureau of Labor Statistics commitment to producing high quality data, the Division of Consumer

Expenditure Surveys conducts a data quality analysis and comparison. The latest report can be found using the following link:
https://www.bls.gov/cex/cecomparison.htm.

9
Interview and a Diary,17 this study uses data from the Interview only. The CE is a rotating panel of
consumer units (CU), which we refer to interchangeably as households, with new CUs added to the
survey each month. These CUs are interviewed up to four times at three-month intervals and asked about
their demographic characteristics at the time of the interview but spending over the previous three
months; we refer to these months as a reference quarter. Given the structure of the survey, we observe
expenditures at overlapping three-month intervals (or quarters). For example, data collected in August
2021 refer to expenditures from May to July 2021; data collected in September 2021 refer to expenditures
from June to August 2021, and so on.

Spending categories
We analyze changes in outlays, which we refer to as “spending,” across nineteen different
categories. We use outlays (spending) instead of BLS defined expenditures because outlays are a better
reflection of a household’s out-of-pocket spending. For categories like food and clothing, a household’s
outlays are equivalent to their expenditures. In contrast, outlays for categories like housing and
transportation are not the same as expenditures. Housing outlays include payments towards mortgage
principal, which are not included in expenditures because they are considered an investment.
Additionally, the net purchase price of a vehicle is included in expenditures, whereas vehicle outlays
include any down payment and loan payments.18
Ten of the categories we look at are the major spending categories provided in the CE data: (1)
housing and utilities; (2) food; (3) alcohol and tobacco; (4) clothing19; (5) transportation; (6) health; (7)
leisure; (8) personal care; (9) education and reading; and (10) miscellaneous items. We also use detailed
item-level spending reports to construct seven item-specific spending categories that we identify as child-
related goods and services: (1) children’s clothes20; (2) books and toys; (3) computers and tablets; (4)
school items; (5) sports items; (6) childcare; and (7) enrichment activities. Note, the categories that make
up total child-related spending are not mutually exclusive but overlap with the categories that make up

17
See the BLS Consumer Expenditure Surveys website for details: https://stats.bls.gov/cex/
18 Expenditures consist of transactions costs, including excise and sales taxes, of goods and services acquired during the interview

period. Expenditure estimates include expenditures for gifts but exclude purchases or portions of purchases directly assignable to
business purposes. Periodic credit or installment payments on goods or services already acquired are also excluded. The full cost
of each purchase is recorded, even though full payment may not have been made at the date of purchase. See Glossary:
https://www.bls.gov/cex/csxgloss.htm. More information about the differences between expenditures and outlays can be found on
the BLS Frequently Ask Questions page (https://www.bls.gov/cex/csxfaqs.htm).
19 What we refer to as “clothing” is equivalent to what the CE refers to as “apparel and services,” which includes more than just

expenditures on clothing.
20 Starting in 2021Q2, the CE Interview Survey stopped asking about clothing expenditures for separate items and implemented a

“global” clothing question. An internal review revealed the change in question led to an increase in the mean value of clothing
expenditures. We include a year fixed effect in our model specification which will control for any change in spending as a result
of the change in question that is correlated with the effect of the advanced CTC payments.

10
total spending. The remaining two categories are “total spending” and “total child spending,” which are
simply the sum of spending reported across the ten major spending categories and of spending in the
seven child-related spending categories, respectively. Details on the specific purchases included in each
spending category are presented in Appendix Tables 1 (major categories) and 2 (child-related categories).
Measure of the Child Tax Credit Advanced Payment
Our method for estimating the spending response to the expanded, monthly CTC payments relies
on variation in the total amount of monthly CTC income that households could have received in the
reference quarter for which they are reporting spending as part of the CE. We discuss our estimation
strategy further in the next section, but here we describe how we impute the CTC payments that
households could have received.21
CTC payments were determined by the IRS at the tax-unit level. The CE data includes tax unit
identifiers; thus, using the number and age of dependents claimed by the tax unit, as well as the tax unit’s
AGI, we can calculate the annual 2021 CTC for which tax units are eligible.22 To get the value of the
monthly advanced payment, we divide the imputed annual CTC by twelve. If a household contained
multiple tax units, the monthly CTC payments for each tax unit were aggregated to yield a household
level monthly CTC payment.
Recall that the reference period for the CE Interview is the previous three-month period (or
quarter); therefore, we need to convert the monthly advanced payment into a three-month value to get the
total CTC income a household could have received during the reference quarter. Importantly, depending
on the interview month, households could have received 0, 1, 2, or 3 months of CTC payments during the
reference quarter (see Table 1). Therefore, when determining quarterly CTC payments, we multiplied the
monthly payment amount by the number of months in the reference quarter during which households
could have received payments. For example, July 2021 interviews have a reference quarter covering April
to June 2021, and thus contained no months during which advanced payments could have been delivered.
The reference period for August 2021 interviews was May through July 2021, thus containing one month
when advanced payments were delivered, so the quarterly CTC amount for households interviewed in
August was equivalent to one month of monthly CTC payments. The reference quarter for September
2021 interviews contained two months where payments were delivered, so quarterly CTC payment for

21 The BLS included two questions in the CE regarding the receipt and use of the CTC between October 2021 and June 2022.

These questions were similar to the ones used by the BLS to collect data on Economic Impact Payments, but respondents were
not asked the value of advanced CTC payment. See Appendix A for the specific survey questions and a discussion of the
responses.
22 Tax unit level data included in the CE were generated using the National Bureau of Economic Research’s TAXSIM model. For

more information about the TAXSIM model see Feenberg and Coutts (1993), and for more information about how TAXSIM is
used by the BLS see Paulin and Hawk (2015). Although the version of TAXSIM currently used by the BLS (TAXSIM32)
imputes annual 2021 CTC payments for tax units in the CE, we chose to calculate these values ourselves because it does not
appear that this version of TAXSIM accounts for the increase in qualifying age, from 16 to 17 years old.

11
households interviewed in September amounted to two monthly payments, and so on. Figure 2 shows the
value of the quarterly CTC payment across interview months for a household with one child or two
children of varying ages. The depiction shows how the value of the quarterly CTC payment varied both
by interview month, the number of children in the household, and the ages of those children (as younger
children under age 6 were eligible for larger payments of $300 per month instead of the $250 older
children received).

Table 1: Treatment months in reference quarter by interview month
Interview month
Jul-21 Aug-21 Sep-21 Oct-21 Nov-21 Dec-21 Jan-22 Feb-22 Mar-22
Months in Apr-21 May-21 Jun-21 Jul-21 Aug-21 Sep-21 Oct-21 Nov-21 Dec-21
reference May-21 Jun-21 Jul-21 Aug-21 Sep-21 Oct-21 Nov-21 Dec-21 Jan-22
quarter June-21 Jul-21 Aug-21 Sep-21 Oct-21 Nov-21 Dec-21 Jan-22 Feb-22
Note: The CE Interview asks respondents to report on spending across a three-month reference period (the reference quarter).
Table 1 displays the months comprising the reference quarter for CE interviews conducted between July 2021 and March 2022.
Months in the reference quarters during which households could have received monthly CTC payments are highlighted in blue.

Figure 2: Quarterly CTC Income for Different Household Types by Child Age and Interview Month

Note: Figure 2 shows the total CTC payment a one-child and two-child household could receive in the reference quarter for
which they are reporting spending in the CE based on CE interview month. All values are the authors’ calculations.

12
Our imputations yield an average monthly advanced CTC payment among qualifying households of $485
and median payment is $500. These translate to an average quarterly advanced CTC payment of $1,099
and median payment of $900 within the reference period. Table 2 presents the average and median
payments by household income and race/ethnicity of household head. It should be noted these values
represent averages based on the amount households qualify for across all households, regardless of
whether they reported receiving the payments.23

Table 2: Imputed Value of the Advanced Child Tax Credit Payment Eligibility for Qualifying
Households
Three-Month
Monthly Value
Reference Quarter
Num
Mean Median Mean Median Obs.
All Households $485 $500 $1,099 $900 3,547
Income Category
Under $50,000 $530 $500 $1,176 $900 980
$50,000 to $100,000 $509 $500 $1,175 $900 1,067
$100,000 to $150,000 $472 $497 $1,079 $900 1,101
$200,000+ $325 $333 $729 $500 399
Race
Asian $439 $470 $968 $800 255
Black $507 $500 $1,139 $900 375
Hispanic $520 $500 $1,179 $900 785
Other $545 $550 $1,241 $1,000 115
White $467 $431 $1,065 $817 2,017
Note: Values are calculated using only data for households who qualified for the CTC (i.e., imputed CTC > 0) from
August 2021 through March 2022 interviews and weighted using the household population weight, FINLWT21.
Reference quarter values are not equivalent to 3 monthly payments because there are some reference quarters during
which households could have received only one or two payments based on the months in that quarter. See Table 1
for additional details.

Estimation strategy
We use a parameterized difference-in-differences, following many prior studies of policy effects
(Currie & Gruber, 1996; Hoynes & Patel, 2018, 2018; Jones et al., 2019; Michelmore & Pilkauskas,
2021; Pilkauskas et al., 2022). We use an imputed measure of how much households could have benefited
from the policy expansion (described in the preceding section) and a difference-in-difference framework
to estimate the intent-to-treat spending response to the expanded, monthly CTC payments.24

23 If we restrict the sample to only those who report receipt in the CE, the average and median imputed three-month payment

increase to $1,336 and $1,100, respectively.
24 Although there is no formal test for the parallel trends assumption, we conduct an “eyeball” test be looking at the average

outlays by interview month and year for households in our sample. See Figure C1 in Appendix C. Until May, total outlays for
2019 and 2021 appear to follow similar trends. Starting in June the two series begin to diverge, which is two months before we
would expect to see any impact of the monthly CTC payments, suggesting there might be some anticipation effects. As the

13
To estimate the spending response, we use CE data from January 2019 through March 2020 and
January 2021 through March 2022. Our sample is limited to households with children who could be
considered income-eligible for the 2021 monthly CTC payments.25 In order to identify “income-eligible”
households, we calculate the CTC following the ARPA legislation for all households in our sample
(2019/2020 interviews as well as the 2021/2022 interviews). All households with a non-zero CTC value
based on these calculations are considered “income-eligible.” Under these sample restrictions, households
interviewed in 2021 and 2022 represent the potentially treated units, with the pre-treatment period being
January 2021 through July 2021 and the treatment period being August 2021 through March 2022.
Although an ARPA CTC is calculated for all households in the sample, only households interviewed
between August 2021 and March 2022 are considered treated. The control group is represented by
households interviewed between January 2019 and March 2020. By using households interviewed in
2019 and early 2020 as our control group, we are also able to control for any seasonal variation in
spending that may be correlated with the timing of monthly CTC delivery (e.g., higher levels of spending
on back-to-school related items in the fall months.) We use the following equation to estimate the effect
of the advanced CTC payments on spending:
𝐸𝑖𝑚𝑦 = 𝛽𝑜 + 𝛽1 𝑄𝑢𝑎𝑟𝑡𝐶𝑇𝐶𝑖𝑚𝑦 + 𝛽2 𝑋𝑖 + 𝛿𝑠 + 𝜕𝑚 + 𝛼𝑦 + 𝜌𝑎 + 𝜀𝑖𝑚𝑦 (1)
𝐸𝑖𝑚𝑦 represents quarterly spending for household 𝑖 interviewed in month 𝑚 and year 𝑦. 𝑄𝑢𝑎𝑟𝑡𝐶𝑇𝐶𝑖𝑚𝑦 is
the quarterly advanced CTC payment that the potentially treated household 𝑖 could have received during
the reference period, which covers spending during the three months prior to the interview month 𝑚, as
discussed in the preceding section. 𝑄𝑢𝑎𝑟𝑡𝐶𝑇𝐶𝑖𝑚𝑦 is set to $0 for all households in the control group. For
greater ease of interpretation, 𝑄𝑢𝑎𝑟𝑡𝐶𝑇𝐶𝑖𝑚𝑦 is measured in units of $100. The parameter of interest is 𝛽1 ,
which represents the effect of a $100 increase in imputed CTC benefits on spending. 𝑋𝑖 is a vector of
household characteristics, including the unit’s before-tax income and demographic characteristics of the
reference person (age group,26 race/ethnicity, sex, education). We recognize our study period also
coincides with the distribution of the Economic Impact Payments (EIP), specifically the second and third
rounds of EIP. Although Parker et al. (2023) shows households spend a small portion of their EIP during
the period it is received, there could be lagged spending responses that were out of the scope of their
paper. Therefore, included in the vector of household level characteristics is a set of controls, with lags,

monthly payments begin to phase out at the end of the treatment period, spending for 2021 appears to converge with 2019
spending.
25 We excluded households without children because their spending patterns differ from those of households with children, which

makes them a poor comparison group. Additionally, households that do not qualify for the CTC due to their high income are also
excluded from the sample because they may have sufficiently different spending patterns than lower income households that
qualify for the CTC, which would make them a poor comparison group as well. 1.8 percent of the sample is dropped due to the
income restriction.
26 Age group defined as 18 to 35, 36 to 50, 51 to 65, and 66 and up.

14
for the three rounds of EIP.27 State fixed effects (𝛿𝑠 ) are included to control for state level characteristics
or policies that might affect spending. We also include interview month fixed effects (𝜕𝑚 ) to control for
seasonal variation as well as interview reference year fixed effects (𝛼𝑦 ) to control for variation in
spending between 2019 and 2021, including annual changes in spending patterns. To control for the
variation in household spending that is correlated with the size of the unit, and in particular the number of
children, fixed effects for the size of the household (𝜌𝛼 ) are also included in the model.
Figure 3 shows the Consumer Price Index for 2019 and 2021 rebased to January of the respective
year. Prices for these two years do not follow the same trend. Furthermore, the divergence in the price
trends from 2019 and 2021 increased during the same period the advanced CTC payments were delivered.
Both higher prices and the additional income from the CTC could lead households to increase their
spending. To account for the different trends in prices between 2019 and 2021, we convert the spending
data to constant January 2021 dollars using the BLS’ Consumer Price Index for All Urban Consumers
(CPI-U).
Figure 3: Consumer Price Index Rebased to January of the Respective Year

Note: Price index data used to produce this figure comes from the U.S. Bureau of Labor Statistics. The Consumer Price Index for
All Urban Consumers (CPI-U) for 2019 and 2021 are rebased January of the respective year.

Further, it is possible that the spending response to the CTC payments varies across households
based on their liquidity and baseline income levels, as was found by Wheat et al. (2022). Reports from

27 The CE collected data on the receipt, amount, and use of the EIP. See Parker et al. (2023) for more information about the CE

EIP questions. A separate dummy variable was created for each of the three rounds of EIP. The dummy variable took a value of
one if the household reported receiving an EIP during the reference period, and zero otherwise. One- and two-period lags were
created by using responses to the receipt questions from previous interviews in which the household participated. If the
household did not have a previous interview for which responses could be drawn, the lags were assumed to be zero.

15
other survey data (Rachidi, 2021), as well as responses to the CE suggest higher-income households were
more likely to save income received from the monthly CTC payments while lower-income households
were more likely to spend the payments or use it to pay down debt.28 For this reason, we separately
estimate the spending response by household income level. Since income and liquidity are correlated with
race, we also estimate separate spending response by race and ethnicity. We define the following
racial/ethnic categories: non-Hispanic Asian, non-Hispanic Black, Hispanic, non-Hispanic other race; and
non-Hispanic White.29
Our primary estimate is the intent-to-treat effect (ITT), as it captures the effect of the treatment on
all eligible households, regardless of actual CTC receipt. Although the CE included questions on CTC
payment receipts for part of the study period, we did not use this data in our main analysis because they
were first introduced in October 2021, after the CTC payments had already begun. Using these data
would require excluding households with a reference period that included CTC payments but did not
collect data on receipt (i.e., interviews conducted in August and September). Additionally, there is a
notable discrepancy between the CTC take-up rates reported by the CE and the IRS, with the CE
estimates being significantly lower.30 This gap may be attributed to underreporting bias in the survey data,
which is a common issue. Relying on this data could introduce endogeneity if the underreporting is
correlated with factors that also influence our outcomes. Acknowledging these limitations, we include
estimates of the treatment effect on the treated (TOT), or the local average treatment effect, using self-
reported information on the CTC receipt with the full sample in the Appendix Tables C3 and C4.

4. Results

Overall
Our analysis reveals that the CTC payments are associated with a notable increase in total
spending. A $100 increase in CTC payment received during the reference quarter is associated with a $44
increase in total spending among potential recipients. Figure 4 presents the estimated spending response
to the CTC payments across major expenditure categories based on the estimation strategy described in
Equation 1.31 Among the different expenditure categories, the greatest increases in spending are seen in
food, particularly food at home, housing, and clothing. For every $100 increase in income, there is an

28 See Appendix Table A3 for a discussion of responses to CE questions on use of the advanced monthly CTC payments.
29 The race and ethnicity of a household was assigned based on responses to the CE in relation to the reference person. See the

CE Frequently Asked Questions page for more information about how the reference person is defined
(https://www.bls.gov/cex/csxfaqs.htm).
30 The IRS estimated that by December the CTC reached 61.2 million children (Parolin et al., 2022), whereas the CE receipt

variables suggest the CTC only reached 41.5 million children.
31 For a set of descriptive results showing average changes in spending in 2021 for households with children before and after the

rollout of the monthly CTC payments (i.e., before and after the August 2021 interview) and compare these changes to between
the same periods in 2019, see Appendix B.

16
associated increase of $28 in spending on housing, $12 on food, and $5 on clothing. The sum of the
spending response to these three categories is slightly more than the spending response for total spending
because the some of the categories have a negative spending response. We did not find any significant
effect on spending in the other categories we examined, which include alcohol and tobacco,
transportation, health, leisure, personal care, education, and miscellaneous spending. These results
suggest a majority of the spending response is allocated to necessities.

Figure 4: Estimated effects on spending of a $100 increase in CTC payments during the reference quarter
(3-month period) for major expenditure categories among households with children

Note: All results derived from the CE fielded from January 2019 to March 2020 and from January 2021 to March 2022. The
sample is limited to households with children who would qualify for a monthly CTC payment (i.e., imputed CTC > 0), regardless
of interview month. The estimated effect, coefficient from the regression model, represents the change in spending by domain
associated with a $100 increase in quarterly income assumed to be from the CTC. All models include state, year, and month
fixed effects, and a fixed effect for the household size, and demographic controls for the household reference person, including
age group, race/ethnicity, sex, and education level. We also control for contemporaneous and lagged spending responses to the
EIP payments. Estimates are weighted to the national level using the household population weight FINLWT21. Confidence
intervals calculated with robust standard errors are represented by the horizontal lines in the chart.

In Figure 5, we present the estimated association between an increase of $100 in CTC payments
and changes in spending on child-related items. For every $100 increase in income from the CTC
payments during a reference quarter, there is a corresponding $16 increase in spending on child-related
items. Notably, spending on childcare ($6) and children’s clothing ($5) contribute the most to the

17
spending response, accounting for 39 and 30 percent of the total child-related spending response,
respectively.

Figure 5: Estimated effects on spending of a $100 increase in CTC payments during the reference quarter
(3-month period) across child-related expenditure categories among households with children

Note: All results derived from the CE fielded from January 2019 to March 2020 and from January 2021 to March 2022. The
sample is limited to households with children who would qualify for a monthly CTC payment (i.e., imputed CTC > 0), regardless
of interview month. The estimated effect, coefficient from the regression model, represents the change in spending by domain
associated with a $100 increase in quarterly income assumed to be from the CTC. All models include state, year, and month fixed
effects, a fixed effect for the household size, and demographic controls for the household reference person, including age group,
race/ethnicity, sex, and education level. We also control for contemporaneous and lagged spending responses to the EIP
payments. Estimates are weighted to the national level using the household population weight FINLWT21. Confidence intervals
calculated with robust standard errors are represented by the horizontal lines in the chart.

Results stratified by income level and race/ethnicity
In Table 3, we provide estimates of the spending response across the major CE spending
categories for the overall sample as well as estimates stratified by household’s before-tax Adjusted Gross
Income (AGI), which we refer to as “income” going forward. We split the income distribution into four
groups: (1) below $50,000, (2) $50,000 to $100,000, (3) $100,000 to $200,000, and (4) $200,000 or
higher. We also include a fifth income group, income below $24,000, which represents the households

18
that benefited the most from the changes in CTC legislation.32 The spending response for total outlays
appears to increase across the income distribution, with the end points being the exception. However, only
the coefficient for households with income under $24,000 is statistically significant. For the lowest-
income households, an additional $100 of income from the CTC results in a $52 increase in spending.
Looking at the other spending categories in Table 3, clothing is the category with the strongest spending
response across the income distribution. Households with income below $100,000 spend about $6 out of
every $100 of CTC on clothing. Units with income below $24,000 show a slightly higher spending
response, $7 out of every $100 of CTC, but the spending responses across these income groups are not
statistically different.

Table 3: Estimated effects on spending of a $100 increase in CTC payments during the reference quarter
(3-month period) across major expenditure categories among households with children
Before tax income
Income under Income under $50,000 - $100,000 -
Overall $200,000 +
$24,000 $50,000 $100,000 $200,000
Total Outlays 43.619* 52.086* 33.213 52.925 56.189 -7.728
(17.268) (25.469) (20.225) (28.137) (34.421) (136.592)
Housing 28.028** 20.745 16.380 30.100 37.021 24.613
(10.142) (14.024) (10.536) (18.612) (19.823) (79.698)
Food 12.309** 9.845 13.212 15.651* 9.287 3.009
(3.992) (7.036) (7.174) (6.501) (6.897) (20.995)
Food, home 7.406* 7.967 9.864 8.448 4.900 -8.341
(3.001) (5.851) (5.838) (4.758) (4.963) (16.623)
Food, away 4.902* 1.878 3.348 7.203 4.387 11.35
(1.961) (3.211) (2.467) (3.736) (3.678) (10.292)
Alcohol and
-0.255 0.850 -0.575 0.393 0.041 -2.380
Tobacco
(0.855) (1.802) (1.25) (1.678) (1.512) (3.299)
Clothing 5.083** 6.740** 5.503* 5.612* 1.307 10.256
(1.985) (2.497) (2.663) (2.392) (5.442) (7.998)
Transport -2.697 5.334 -3.762 -4.730 5.353 1.311
(6.246) (7.890) (5.767) (9.721) (15.481) (37.493)
Health -1.474 2.030 -0.602 -1.436 3.030 -33.885*
(3.801) (5.170) (3.652) (5.224) (10.459) (16.836)
Leisure 5.082 5.187 2.879 -1.619 6.093 53.958
(3.215) (4.052) (3.285) (3.065) (5.463) (40.063)
Personal Care 0.589 1.249 0.731 0.161 1.002 -0.329
(0.418) (0.785) (0.595) (0.594) (0.886) (3.074)
Education -5.016 -0.575 0.207 2.684 -6.48 -66.143*
(3.006) (2.166) (1.83) (3.464) (6.431) (29.799)
Miscellaneous 1.919 0.682 -0.759 6.026 -0.466 1.861
(1.439) (1.885) (1.633) (3.886) (1.462) (6.749)
N 14,365 2,259 4,248 4,439 4,196 1,482
Note: All results derived from the CE Interview fielded from January 2019 to March 2020 and from January 2021 to March 2022.
The sample is limited to households with children who would qualify for a monthly CTC payment (i.e., imputed CTC > 0),
regardless of interview month. The estimated effect, coefficient from the regression model, represents the change in spending by
domain associated with a $100 increase in quarterly income assumed to be from the CTC. All models include state, year, and

32 Under the TCJA, a household needed to have income above at least $24,000 in AGI to receive the full benefit of the CTC –

households with more than one child needed more income to qualify for the full CTC. Households with income between $2,500
and $24,000 could receive a partial benefit, and households with income below $2,500 did not qualify for any CTC benefit. As
part of the ARPA, the minimum income requirement was removed, allowing the lowest-income households to now receive the
full benefit of the CTC. See Figure 1.

19
month fixed effects, a fixed effect for the household size, and demographic controls for the household reference person, including
age group, race/ethnicity, sex, and education level. We also control for contemporaneous and lagged spending responses to the
EIP payments. Estimates are weighted to the national level using the household weight FINLWT21. Robust standard errors
reported in parentheses. ***p<0.001 **p<0.01 *p<0.05.

Table 4 presents estimates of the spending response across child-related spending categories for
the overall sample as well as estimates stratified by before-tax income levels. Similar to the response for
total outlays, the effect on child-related total spending appears to increase across the income distribution
with the endpoints being the exception. However, unlike the estimates for total spending, the effect on
child-related total spending is statistically significant for all income groups except for households with
income above $200,000. Households with income below $100,000 increase their child-related spending
between $10 and $23 for every $100 of quarterly CTC payments. However, these difference in
coefficients are not statistically significant.
Among the subcategories of child-related spending, only children’s clothing and childcare show a
significant effect. Children’s clothing shows the most consistent spending response across the income
distribution. All income groups, except households with $100,000 to $200,000, have statistically
significant spending response between $5 and $8 for every $100 of quarterly CTC. For households with
income below $100,000, spending on child-related clothing appears to be the largest contributor to their
overall spending response on child-related outlays. For households with $100,000 to $200,000 in income,
childcare appears to be the largest contributor to their overall spending response. For every $100 of
additional CTC payment, these households increase their spending on childcare by $13. Overall, the
CTC’s effects on childcare spending is only significant among the lowest-income group ($5 per $100 in
CTC) and those with income between $100,000 and $200,000 ($13 per $100 in CTC). This might reflect
the fact that middle-income households are in a uniquely challenging position and face high childcare cost
burden as their income is too high to be eligible for public childcare programs (e.g. childcare subsidy,
Head Start), but still too low to afford the high-quality childcare with the market price (Hardy & Park,
2022).

20
Table 4: Estimated effects on spending of a $100 increase in CTC payments during the reference quarter
(3-month period) across child-related goods and services among households with children
Before tax income
Income under Income under $50,000 - $100,000 -
Overall $200,000 +
$24,000 $50,000 $100,000 $200,000
Total Child Outlays 15.661*** 16.066*** 10.422** 13.567** 22.840** 20.909
(3.656) (4.660) (3.367) (4.740) (8.608) (32.211)
Children's Clothes 4.758*** 5.795*** 4.936*** 6.263*** 2.200 7.876***
(0.957) (1.759) (1.492) (1.622) (2.088) (3.590)
Books and Toys 0.473 0.737 1.102 0.447 -0.166 1.849
(0.557) (1.457) (0.985) (0.890) (1.066) (2.430)
Computers and
0.242 0.859 0.949 -0.871 1.203 -0.697
Tablets
(0.684) (0.644) (0.843) (1.066) (1.709) (3.676)
School Items 1.499 2.219 1.474 4.179 -0.720 -2.501
(2.121) (1.314) (0.977) (2.804) (4.859) (22.729)
Sports Items 0.396 0.494 0.176 0.923 0.480 0.164
(0.607) (1.006) (0.552) (0.895) (1.421) (4.391)
Childcare 6.086*** 5.180** 2.185 1.800 12.680** 17.632
(1.890) (2.060) (1.385) (2.381) (4.668) (13.899)
Enrichment 2.205 0.784 -0.399 0.826 7.163 -3.414
(1.544) (1.498) (1.719) (1.330) (4.162) (10.294)
N 14,365 2,259 4,248 4,439 4,196 1,482
Note: All results derived from the CE fielded from January 2019 to March 2020 and from January 2021 to March 2022. The
sample is limited to households with children who would qualify for a monthly CTC payment (i.e., imputed CTC > 0), regardless
of interview month. The estimated effect, coefficient from the regression model, represents the change in spending by domain
associated with a $100 increase in quarterly income assumed to be from the CTC. All models include state, year, and month fixed
effects, a fixed effect for the household size, and demographic controls for the household reference person, including age group,
race/ethnicity, sex, and education level. We also control for contemporaneous and lagged spending responses to the EIP
payments. Estimates are weighted to the national level using the household population weight FINLWT21. Robust standard
errors reported in parentheses. ***p<0.001 **p<0.01 *p<0.05.

Next, we examine how the spending response on general and child-specific categories varies
across households based on the race/ethnicity of the household reference person (Tables 5 and 6,
respectively). Table 5 shows that the spending response for total outlays is only statistically significant
among households with a reference person who identifies as Hispanic. It is possible the low precision of
the estimated spending response for units where the head identifies as non-Hispanic Asian, non-Hispanic
Black, and non-Hispanic ‘other’ is a result of the small sample sizes for these groups. This limitation of
the data should be considered when interpreting the findings by race/ethnicity.
For every $100 in quarterly CTC payments, Hispanic-headed households increased their total
outlays by $69. Like the overall sample, their spending response appears to be largest for necessities,
specifically housing, food, and clothing. For every $100 in potential CTC payments among Hispanic-
headed units, spending on housing increased by $29; spending on food increased by $18, driven primarily
by food away from home ($10); and spending on clothing increased by $10. The only other significant

21
effect identified when examining results by race/ethnicity was on miscellaneous spending among units
with a reference person who identified as non-Hispanic Black ($5).

Table 5: Estimated effects on spending across major expenditure categories of a $100
increase in the CTC payments by race/ethnicity of reference person
Race/ethnicity of household reference person
Overall Asian Black Hispanic Other White
Total Outlays 43.619* 72.544 43.476 69.556** 39.196 29.925
(17.268) (84.808) (37.610) (24.106) (80.869) (26.790)
Housing 28.028** 55.640 18.564 34.128** 19.383 28.892
(10.142) (56.852) (20.582) (12.308) (38.088) (16.423)
Food 12.309** 31.242 11.332 17.155* -18.619 8.668
(3.992) (16.613) (8.647) (8.113) (18.908) (5.873)
Food, home 7.406* 14.300 7.406 6.323 -10.961 6.265
(3.001) (12.983) (6.416) (5.801) (14.126) (4.492)
Food, away 4.902* 16.942 3.926 10.832* -7.658 2.403
(1.961) (8.901) (4.713) (4.560) (9.139) (2.609)
Alcohol and Tobacco -0.255 2.020 0.719 1.212 -3.291 -1.865
(0.855) (2.801) (1.516) (1.174) (6.469) (1.385)
Clothing 5.083** 10.382 6.617 10.176** -3.107 2.500
(1.985) (6.679) (3.850) (3.709) (7.811) (3.126)
Transport -2.697 -27.324 -0.293 5.726 24.579 -7.750
(6.246) (30.487) (9.774) (10.626) (34.931) (10.027)
Health -1.474 -9.791 -5.171 2.212 -0.545 0.447
(3.801) (11.539) (6.316) (4.180) (19.343) (6.680)
Leisure 5.082 16.361 8.439 4.389 7.462 3.214
(3.215) (10.922) (5.703) (3.153) (9.171) (5.513)
Personal Care 0.589 1.148 1.196 1.063 -1.617 0.096
(0.418) (0.979) (1.647) (0.695) (1.802) (0.471)
Education -5.016 -10.273 -3.491 -5.107* -3.174 -5.883
(3.006) (17.107) (3.533) (2.127) (12.812) (5.163)
Miscellaneous 1.919 3.141 5.303* -1.399 18.126 1.606
(1.439) (8.665) (2.491) (1.818) (15.560) (2.484)
N 14,365 1,073 1,530 3,132 395 8,235
Note: All results derived from the CE fielded from January 2019 to March 2020 and from January 2021 to March 2022. The
sample is limited to households with children who would qualify for a monthly CTC payment (i.e., imputed CTC > 0), regardless
of interview month. The estimated effect, coefficient from the regression model, represents the change in spending by domain
associated with a $100 increase in quarterly income assumed to be from the CTC. All models include state, year, and month
fixed effects, a fixed effect for the household size, and demographic controls for the household reference person, including age
group, race/ethnicity (in the case of model 1), sex, and education level. We also control for contemporaneous and lagged
spending responses to the EIP payments. Estimates are weighted to the national level using the household population weight
FINLWT21 Robust standard errors reported in parentheses. ***p<0.001 **p<0.01 *p<0.05.

Table 6 presents the estimated spending response on child-related outlays by race and ethnicity of
the household head. Unlike the general spending categories, the effect of the CTC payments on child-
related spending appears to be more diverse. The Hispanic, non-Hispanic Black, and non-Hispanic ‘other’
groups all show a statistically significant spending response with respect total child outlays, $13, $27, and
$41, respectively. Looking at the subcategories among child related spending, the spending response

22
again seems to be restricted to children’s clothing and childcare, as well as enrichment in the case of non-
Hispanic Black-headed units. Non-Hispanic Asian households had the lowest spending response on
children’s clothing, $5, and non-Hispanic Black and Hispanic households increased their spending by $7
and $8, respectively, although these differences are not statistically significant. Both non-Hispanic Black
and non-Hispanic ‘other’ units exhibited statistically significant, but not statistically different, spending
responses on childcare, $9 and $27, respectively. While we find a significant increase in child-related
outlays for Hispanic-headed households, this result does not appear to be driven by a significant increase
in childcare spending. This can be potentially explained by variation in types of childcare use, as Hispanic
children are less likely to use center-based care and more likely to rely on parental or relative care than
children of any other race and ethnicity (Datta et al., 2022). Non-Hispanic Black households also were
shown to increase spending on enrichment by $10 for every $100 in quarterly CTC payments.

Table 6: Estimated effects on child-related spending of a $100 increase in CTC payments during the
reference quarter (3-month period) by race/ethnicity of reference person
Race/ethnicity of household head
Overall Asian Black Hispanic Other White
Total Child Outlays 15.661*** 38.397 27.326*** 12.797** 47.184** 9.997
(3.656) (19.893) (6.686) (4.064) (18.187) (6.063)
Children's Clothes 4.758*** 4.898* 7.977*** 8.615*** 2.680 2.022
(0.957) (2.065) (2.477) (1.955) (4.505) (1.298)
Books and Toys 0.473 3.664 0.145 1.733 0.969 -0.313
(0.557) (2.343) (1.694) (1.264) (2.044) (0.700)
Computers and
0.242 -1.458 1.142 0.212 -1.676 0.268
Tablets
(0.684) (3.333) (1.714) (1.067) (5.845) (1.029)
School Items 1.499 0.067 -1.335 -0.242 13.668 2.567
(2.121) (5.128) (2.165) (1.505) (9.422) (3.824)
Sports Items 0.396 7.048 0.005 -0.625 1.971 0.494
(0.607) (4.946) (0.916) (0.774) (2.115) (0.983)
Childcare 6.086*** 15.867 9.083** 2.366 27.456** 5.464
(1.890) (13.936) (3.545) (1.936) (10.872) (3.095)
Enrichment 2.205 8.310 10.31* 0.737 2.116 -0.505
(1.544) (8.524) (5.038) (1.016) (3.164) (2.134)
N 14,365 1,073 1,530 3,132 395 8,235
Note: All results derived from the CE fielded from January 2019 to March 2020 and from January 2021 to March 2022. The
sample is limited to households with children who would qualify for a monthly CTC payment (i.e., imputed CTC > 0), regardless
of interview month. The estimated effect, coefficient from the regression model, represents the change in spending by domain
associated with a $100 increase in quarterly income assumed to be from the CTC. All models include state, year, and month fixed
effects a fixed effect for the household size, and demographic controls for the household reference person, including age group,
race/ethnicity (in the case of model 1), sex, and education level. We also control for contemporaneous and lagged spending
responses to the EIP payments. Estimates are weighted to the national level using the household population weight FINLWT21.
Robust standard errors reported in parentheses. ***p<0.001 **p<0.01 *p<0.05.

Finding statistically significant spending responses for only non-Hispanic Other, Hispanic, and non-
Hispanic Black households makes sense when considering the correlation between income and

23
race/ethnicity. Among the five race/ethnicity groups analyzed, non-Hispanic Other, Hispanic, and non-
Hispanic Black households have the lowest average before tax income, $87,000, $71,000, and $68,00,
respectively. These average incomes are statistically different from those of the other two groups.33 As
such, non-Hispanic Other, Hispanic, and non-Hispanic Black households are disproportionately
represented in the lower income range, where we observed the strongest spending response in results
disaggregated by income (Table 4).

Placebo test
To validate our model specification, we ran a series of placebo tests where we estimate our main
model in CE data covering a period when households did not benefit from the ARP expansion — 2018
and 2019. Following our main specification, interviews from 2019 were used as the control group, and
interviews from 2018 were used as the placebo treatment group. We expect to find no significant
spending response by the treatment group because they were not actually subjected to the policy change.
We simulated a quarterly CTC payment that households interviewed in August through December of
2018 would have been eligible for if the ARP rule had been in effect (i.e., the placebo). As in our main
specification, the pre-treatment period was January through July, and the treatment period was from
August through December.34 The placebo was set to zero for all other households in the sample. If our
main results capture a causal effect of the policy change, we expect null and insignificant estimates in
spending associated with the placebo CTC payments simulated for households interviewed in the second
half of 2018.
Results from the placebo test are presented in Figure 6. Overall, the coefficients are close to zero and
statistically insignificant, lending support to the interpretation that our main findings are attributable to
the expanded, monthly CTC rather than a spurious correlation or general pattern of increased spending in
the second half of the year. The only exception is children’s clothing, which shows a statistically
significant increase of $2.70 in expenditures over a three-month period during the latter half of 2018. The
statistically significant spending response on children’s clothing suggests that the effect found using our
main specification might not be entirely attributable to the treatment, requiring caution in interpretation.
In Appendix C, we also show results from these sensitivity tests by income level (Table C1) and

33 The average income (standard error) for the five race/ethnicity groups are: $127,546 ($3,034) for non-Hispanic Asian, $68,424

($1,534) for non-Hispanic Black, $70,583 ($1,066) for Hispanic, $87,109 ($3,834) for non-Hispanic Other, and $114,420 ($969)
for non-Hispanic White. Averages were calculated using households interviewed in 2019/2020 and 2021/2022 with children that
are income-eligible for the expanded CTC, weighted using household population weight FINLWT21, and the standard errors
were calculated using the 44 replicate weights provided in the CE.
34
We were unable to include January through March of the following year as part of the treatment period, as we did
in our main specification, because it would require treating interviews from January through March of 2019 as both
control and placebo.

24
race/ethnicity (Table C2). Our findings related to these subgroups – that there was a larger spending
response among lower-income households, and that Black and Latino-headed households spent a larger
share of their payment on child-related goods and services than White-headed households – are not
undermined by the results from the placebo tests, which show no significant effects on these outcomes
associated with the placebo treatment at any income level or by the race/ethnicity of the household head.

Figure 6: Results from placebo tests: Estimated effects of a $100 increase in CTC payments during the
reference quarter (3-month period) in 2019

Note: All results derived from the CE Interview fielded from January 2018 to December 2019. The sample is limited to
households with children who would qualify for a monthly CTC payment (i.e., imputed CTC > 0), regardless of interview month.
The estimated effect, coefficient from the regression model, represents the change in spending by domain associated with a $100
increase in quarterly income assumed to be from the CTC. All models include state, year, and month fixed effects, and a fixed
effect for household size, and demographic controls for household reference person, including age group, race/ethnicity (in the
case of model 1), sex, and education level. Estimates are weighted to the national level using the household population weight
FINLWT21. Confidence intervals calculated with robust standard errors are represented by the horizontal lines in the chart.

5. Conclusion

The 2021 expansion to the federal Child Tax Credit was a historic change to social policy in the
United States. For 6 months in 2021, households with children in the United States received monthly

25
payments akin to child allowances distributed in many other high-income nations (e.g., Canada, Ireland,
and the United Kingdom). In this paper, we use a difference-in-difference framework to model how these
households might have spent these monthly payments. We exploit the variation in income gains that
households received from the monthly payments by age of the children in the household and by time
(interview month). We estimated the effect on household spending overall and across several spending
domains using data from the Consumer Expenditure Interview Survey with data collected between
January 2019 and March 2022.
Our results show a strong positive spending response to the estimated 2021 monthly CTC
payments. We find that for each $100 of CTC payments received during the reference quarter, households
spent $44, mainly on housing ($28), food ($12), and child-related goods and services ($16). Our
estimated overall marginal propensity to spend (also referred to in the literature as marginal propensities
to consume or MPC), (44%) is in line with previous finding from the literature that suggests 25-50% of an
overall MPC (Coibion et al., 2020; Johnson et al., 2006; Karger and Rajan, 2020; Kueng, 2018; Wheat et
al., 2022). Further, for the average monthly CTC payments of $485, households increased monthly
spending on total by 4%, housing by 6.2%, food by 6.5%, and child-related goods and services by 20%,
on average. Our results are also robust to placebo tests. When testing the validity our model using data
from 2018 as a placebo, we find no significant effects of modeled CTC on spending associated with the
exception of spending on children’s clothing. Taken together, results from our main model and our
placebo test suggest households increased spending on necessities like housing and food, as well as child-
related goods and service in response to the 2021 ARPA CTC expansion.
We also found the spending response was most prominent for low-income households and
households whose reference person identifies as Hispanic or non-Hispanic Black. Notably, these groups
were historically more likely to be excluded from the full CTC credit under pre-ARPA structure. Our
findings add to the prior studies that show the ARP-CTC disproportionately improved the material and
psychological well-being of households with low-income, and Hispanic or non-Hispanic Black-head
households (Batra et al., 2023; Cha et al., 2023; Nam and Kwon, 2024; Parolin et al., 2021; Pilkauskas et
al., 2022).
Our study contributes primarily to two bodies of literature. First, by using a causal framework,
our study complements and builds on many previous studies of the CTC that use direct-reports on how
households used the payments (Karpman et al., 2021; D. Perez-Lopez & Mayol-García, 2021; Pilkauskas
et al., 2022; Pilkauskas & Cooney, 2021; Rachidi, 2021; RAPID-EC, 2021; Roll, Chun, et al., 2021;
Zippel, 2021). Second, much of the other causal evidence on the effects of similar transfers is either based
on other countries (Gregg et al., 2006; Jones et al., 2019) or on lump-sum payments to a narrow
population, like those from the Alaska Permanent Dividend (Amorim, 2021; Kueng, 2018) or the EITC

26
(Barrow & McGranahan, 2000; Gao et al., 2009; Goodman-Bacon & McGranahan, 2008; Halpern-
Meekin et al., 2015). We add to this body of literature by providing an analysis of the spending response
to a child allowance-like policy that reached most households with children the United States, though we
note that 2021 was also an atypical year due to the pandemic and responses could be different in more
typical economic circumstances.
That said, in the precarious circumstances of 2021, households increased spending on basic needs
that promote household wellbeing (food and housing), as well as on child-related items, in response to the
expanded ARPA CTC. Our results are in accordance with reports of reduced food insecurity and material
hardship in the period when households received the monthly CTC payments (Parolin et al., 2023;
Pilkauskas et al., 2022; 2023). The findings on increased child-related spending are in line with the family
investment model, where increases in income, in this case from the monthly CTC payments, lead to
investment in goods and services associated with child development and household wellbeing. This
implies that the monthly CTC likely improved child health and well-being in the short and long run,
potentially with greater impacts for low-income and Hispanic and non-Hispanic Black children.
Future avenues of research include studying how household spending changed when the
payments were rolled back and if these decreases in spending explain, for example, the higher rates of
food insufficiency in 2022 relative to 2021 (Zereyesus et al., 2022). Additionally, it would be of interest
to compare changes in spending from the monthly CTC payments to the Economic Impact Payments, and
to determine if labeling the CTC payments as specifically payments for children led households to
allocate a larger portion of them to child-related goods and services.

27
Reference
Abdelwahed, L., Robbins, J., Campbell, C., Ohannessian, S., & Czurylo, T. (2022). Measuring Spending

Inequality in the United States in Real Time. https://equitablegrowth.org/measuring-spending-

inequality-in-the-united-states-in-real-time/

Amorim, M. (2021). Socioeconomic Disparities in Parental Spending after Universal Cash Transfers: The

Case of the Alaska Dividend. Social Forces. https://doi.org/10.1093/sf/soab119

Ananat, E., Glasner, B., Hamilton, C., & Parolin, Z. (2022). Effects of the Expanded Child Tax Credit on

Employment Outcomes: Evidence from Real-World Data from April to December 2021 (Working

Paper 29823; Working Paper Series). National Bureau of Economic Research.

https://doi.org/10.3386/w29823

Barrow, L., & McGranahan, L. (2000). The Effects of the Earned Income Credit on the Seasonality of

Household Expenditures. National Tax Journal, 53(4.2), 1211–1243.

https://doi.org/10.17310/ntj.2000.4S1.08

Beatty, T. K. M., & Tuttle, C. J. (2015). Expenditure Response to Increases in In-Kind Transfers:

Evidence from the Supplemental Nutrition Assistance Program. American Journal of Agricultural

Economics, 97(2), 390–404. https://doi.org/10.1093/ajae/aau097

Bullinger, L. R., & Boy, A. (2023). Association of Expanded Child Tax Credit Payments With Child

Abuse and Neglect Emergency Department Visits. JAMA Network Open, 6(2), e2255639.

https://doi.org/10.1001/jamanetworkopen.2022.55639

Cha, E., Lee, J., & Tao, S. (2023). Impact of the expanded child tax credit and its expiration on adult

psychological well-being. Social Science & Medicine, 332, 116101.

Collyer, S., Gandhi, J., Garfinkel, I., Ross, S., Waldfogel, J., & Wimer, C. (2022). The Effects of the 2021

Monthly Child Tax Credit on Child and Family Well-Being: Evidence from New York City.

Socius: Sociological Research for a Dynamic World, 8, 237802312211411.

https://doi.org/10.1177/23780231221141165

28
Collyer, S., Harris, D., & Wimer, C. (2019). Left Behind: The One-Third of Children in Families Who

Earn Too Little to Get the Full Child Tax Credit. Poverty and Social Policy Brief, 3(6).

https://www.povertycenter.columbia.edu/s/Who-Is-Left-Behind-in-the-Federal-CTC-CPSP-

2019.pdf

Coibion, O., Y. Gorodnichenko, and M. Weber. 2020. "How Did US Consumers Use Their Stimulus

Payments?" National Bureau of Economic Research Working Paper No. 27693.

Conger, R. D., & Conger, K. J. (2002). Resilience in Midwestern Families: Selected Findings from the

First Decade of a Prospective, Longitudinal Study. Journal of Marriage and Family, 64(2), 361–

373. https://doi.org/10.1111/j.1741-3737.2002.00361.x

Creamer, J., Shrider, E., Burns, K., & Chen, F. (2022). Poverty in the United States: 2021.

Curran, M. (2022). Research Roundup of the Expanded Child Tax Credit: One Year On. Center on

Poverty and Social Policy at Columbia University.

https://static1.squarespace.com/static/610831a16c95260dbd68934a/t/63732dd8efcf0e5c76aea26e/

1668492763484/Child-Tax-Credit-Research-Roundup-One-Year-On-CPSP-2022.pdf

Curran, M., & Collyer, S. (2020). Children Left Behind in Larger Families: The Uneven Receipt of the

Federal Child Tax Credit by Children’s Family Size. Poverty and Social Policy Brief.

https://www.povertycenter.columbia.edu/s/Child-Tax-Credit-Larger-Families-CPSP-2020-

8rsg.pdf

Currie, J., & Gruber, J. (1996). Saving Babies: The Efficacy and Cost of Recent Changes in the Medicaid

Eligibility of Pregnant Women. Journal of Political Economy, 104(6), 1263–1296.

https://doi.org/10.1086/262059

Duncan, G. J., Morris, P. A., & Rodrigues, C. (2011). Does Money Really Matter? Estimating Impacts of

Family Income on Young Children’s Achievement With Data From Random-Assignment

Experiments. Developmental Psychology, 47(5), 1263–1279. https://doi.org/10.1037/a0023875

29
Gao, Q., Kaushal, N., & Waldfogel, J. (2009). How Have Expansions in the Earned Income Tax Credit

Affected Family Expenditures? Welfare Reform and Its Long-Term Consequences for America’s

Poor, 104–139. https://doi.org/10.1017/CBO9780511605383.005

Gennetian, L. A., Duncan, G. J., Fox, N. A., Halpern-Meekin, S., Magnuson, K., Noble, K. G., &

Yoshikawa, H. (2024). Effects of a monthly unconditional cash transfer starting at birth on family

investments among US families with low income. Nature Human Behaviour, 1-16.

Goldin, J., & Michelmore, K. (2022). Who Benefits from the Child Tax Credit? National Tax Journal,

75(1), 123–147. https://doi.org/10.1086/717919

Goodman-Bacon, A., & McGranahan, L. (2008). How do EITC recipients spend their refunds? Economic

Perspectives, 32(2), 17+. Gale General OneFile.

Gregg, P., Waldfogel, J., & Washbrook, E. (2006). Family expenditures post-welfare reform in the UK:

Are low-income families starting to catch up? Labour Economics, 13(6), 721–746.

https://doi.org/10.1016/j.labeco.2005.10.002

Halpern-Meekin, S., Edin, K., Tach, L., & Sykes, J. (2015). It’s not like I’m poor: How working families

make ends meet in a post-welfare world. University of California Press.

Hastings, J., & Shapiro, J. M. (2018). How Are SNAP Benefits Spent? Evidence from a Retail Panel.

American Economic Review, 108(12), 3493–3540. https://doi.org/10.1257/aer.20170866

Hoynes, H. W., & Patel, A. J. (2018). Effective Policy for Reducing Poverty and Inequality? The Earned

Income Tax Credit and the Distribution of Income. Journal of Human Resources, 53(4), 859–890.

https://doi.org/10.3368/jhr.53.4.1115.7494R1

Jackson, M. I., & Schneider, D. (2022). Public Investments and Class Gaps in Parents’ Developmental

Expenditures. American Sociological Review, 87(1), 105–142.

https://doi.org/10.1177/00031224211069975

Johnson, D. S., Parker, J. A., & Souleles, N. S. (2006). Household Expenditure and the Income Tax

Rebates of 2001. The American Economic Review, 96(5), 1589–1610.

30
Jones, L. E., Milligan, K., Jones, L. E., Milligan, K., & Stabile, M. (2019). Child cash benefits and family

expenditures: Evidence from the National Child Benefit. Canadian Journal of Economic.

Karger, E., and A. Rajan. 2021. "Heterogeneity in the Marginal Propensity to Consume: Evidence from

Covid-19 Stimulus Payments (Revised February 2021)." Working Paper 202015.

Karpman, M., Maag, E., Kenney, G. M., & Wissoker, D. A. (2021). Who Has Received Advance Child

Tax Credit Payments, and How Were the Payments Used?

Kaushal, N., Magnuson, K., & Waldfogel, J. (2011). How is family income related to investments in

children’s learning? (G. J. Duncan & R. Murnane, Eds.; pp. 187–206). Russell Sage Foundation.

https://www.russellsage.org/

Kornrich, S., & Furstenberg, F. (2013). Investing in Children: Changes in Parental Spending on Children,

1972–2007. Demography, 50(1), 1–23. https://doi.org/10.1007/s13524-012-0146-4

Kueng, L. (2018). Excess Sensitivity of High-Income Consumers*. The Quarterly Journal of Economics,

133(4), 1693–1751. https://doi.org/10.1093/qje/qjy014

Nam, J., & Kwon, S. J. (2024). Expansion of Child Tax Credits and Mental Health of Parents With Low

Income in 2021. JAMA Network Open, 7(2), e2356419-e2356419.

Marr, C., Cox, K., & Sherman, A. (2021). Build Back Better’s Child Tax Credit Changes Would Protect

Millions From Poverty—Permanently. 19.

McGranahan, L., & Schanzenbach, D. W. (2013). The Earned Income Tax Credit and Food Consumption

Patterns (SSRN Scholarly Paper 2366846). https://doi.org/10.2139/ssrn.2366846

Michelmore, K., & Pilkauskas, N. (2021). Tots and Teens: How Does Child’s Age Influence Maternal

Labor Supply and Child Care Response to the Earned Income Tax Credit? Journal of Labor

Economics, 39(4), 895–929. https://doi.org/10.1086/711383

Najjarrezaparast, P., & Pendakur, K. (2021). How Did the Canada Child Benefit Affect Household

Spending? Canadian Public Policy, 47(4), 479–496. https://doi.org/10.3138/cpp.2020-137

Pac, J., & Berger, L. M. (2024). Quasi‐experimental evidence on the employment effects of the 2021 fully

refundable monthly child tax credit. Journal of Policy Analysis and Management, 43(1), 192-213.

31
Parker, J., Schild, J., Erhard, L., Johnson, D. (2022). Economics Impact Payments and household

spending during the pandemic. The Brookings Institution.

https://www.brookings.edu/articles/economic-impact-payments-and-household-spending-during-

the-pandemic/

Parolin, Z., Ananat, E., Collyer, S., Curran, M., & Wimer, C. (2023, May). The effects of the monthly and

lump-sum Child Tax Credit payments on food and housing hardship. In AEA Papers and

Proceedings (Vol. 113, pp. 406-412). American Economic Association.

Parolin, Z., Collyer, S., & Curran, M. A. (2022). Sixth Child Tax Credit Payment Kept 3.7 Million

Children Out of Poverty in December. Poverty and Social Policy Brief. Vol. 6, No. 1. Center on

Poverty and Social Policy, Columbia University.

https://www.povertycenter.columbia.edu/publication/montly-poverty-december-2021

Parolin, Z., Collyer, S., Curran, M. A., & Wimer, C. (2021). Monthly Poverty Rates among Children after

the Expansion of the Child Tax Credit. 14.

Parolin, Z., Giupponi, G., Lee, E. K., & Collyer, S. (2024). Consumption responses to an unconditional

child allowance in the United States. Nature Human Behaviour, 8(4), 657-667.

Parolin, Z., Hoynes, H., & Curran, M. (forthcoming). Evaluating the Effects of the 2021 Expansion of the

Child Tax Credit. The ANNALS of the American Academy of Political and Social Science.

Perez-Lopez, D. J. (2021). Household Pulse Survey Collected Responses Just Before and Just After the

Arrival of the First CTC Checks. Census.Gov.

https://www.census.gov/library/stories/2021/08/economic-hardship-declined-in-households-with-

children-as-child-tax-credit-payments-arrived.html

Perez-Lopez, D., & Mayol-García, Y. (2021). Nearly a third of parents spent Child Tax Credt on school

expenses, parents with young children used Child Tax Credit payments for childcare. U.S.

Census Bureau. https://www.census.gov/library/stories/2021/10/nearly-a-third-of-parents-spent-

child-tax-credit-on-school-expenses.html

32
Pilkauskas, N., & Cooney, P. (2021). Receipt and Usage of Child Tax Credit Payments among Low-

Income Families: What We Know. https://poverty.umich.edu/publications/receipt-and-usage-of-

child-tax-credit-payments-among-low-income-families-what-we-know/

Pilkauskas, N., Michelmore, K., Kovski, N., & Shaefer, H. L. (2022). The Effects of Income on the

Economic Wellbeing of Families with Low Incomes: Evidence from the 2021 Expanded Child Tax

Credit (w30533; p. w30533). National Bureau of Economic Research.

https://doi.org/10.3386/w30533

Rachidi, A. (2021). New Survey Data Raises Questions About the Expanded Child Tax Credit.

https://www.aei.org/poverty-studies/new-survey-data-raises-questions-about-the-expanded-child-

tax-credit/

RAPID-EC. (2021). The Child Tax Credit is Buffering Families from Financial Hardship.

Roll, S., Chun, Y., Brugger, L., & Hamilton, L. (2021). How are Mississippi families using their Child

Tax Credit payments? Evidence from Census Data.

Roll, S., Hamilton, L., & Chun, Y. (2021). Expanded Child Tax Credit Payments Have Not Reduced

Employment. 4.

Schanzenbach, D. W., & Strain, M. R. (2023). Employment and labor supply responses to the Child Tax

Credit expansion: Theory and evidence. The ANNALS of the American Academy of Political and

Social Science, 710(1), 141-156.

Schild, J., Collyer, S., Garner, T. I., Kaushai, N., Lee, J., Waldfogel, J., & Wimer, C. (2023). Effects of

the Expanded Child Tax Credit on Household Spending: Estimates Based on U.S. Consumer

Expenditure Survey Data. BLS Working Paper No. 601.

Schneider, D., Hastings, O. P., & LaBriola, J. (2018). Income Inequality and Class Divides in Parental

Investments. https://doi.org/10.1177/0003122418772034

Smith, T. A., Berning, J. P., Yang, X., Colson, G., & Dorfman, J. H. (2016). The Effects of Benefit

Timing and Income Fungibility on Food Purchasing Decisions among Supplemental Nutrition

33
Assistance Program Households. American Journal of Agricultural Economics, 98(2), 564–580.

https://doi.org/10.1093/ajae/aav072

Souleles, N. S. (1999). The Response of Household Consumption to Income Tax Refunds. American

Economic Review, 89(4), 947–958. https://doi.org/10.1257/aer.89.4.947

United States Department of Treasury. (2021). Treasury and IRS Announce Families of Nearly 60 Million

Children Receive $15 Billion in First Payments of Expanded and Newly Advanceable Child Tax

Credit. https://home.treasury.gov/news/press-releases/Treasury-and-IRS-Announce-Families-of-

Nearly-60-Million-Children-Receive-%2415-Billion-Dollars-in-First-Payments-of-Expanded-

and-Newly-Advanceable-Child-Tax-Credit

U.S. Bureau of Labor Statistics. (2015, February). Consumer Expenditure Surveys: Glossary.

https://www.bls.gov/cex/csxgloss.htm#:~:text=Consumer%20unit%20%2D%20A%20consumer

%20unit,in%20permanent%20living%20quarters%20in

U.S. Bureau of Labor Statistics. (2023, May). Consumer Expenditure Surveys: Public Use Microdata.

https://www.bls.gov/cex/pumd.htm

Wheat, C., Deadman, E., & Sullivan, D. (2022). How families used the advanced Child Tax Credit.

JPMorgan Chase Institute. https://www.jpmorganchase.com/institute/research/household-income-

spending/how-families-used-advanced-CTC

Wimer, C., & Wolf, S. (2020). Family Income and Young Children’s Development. The Future of

Children, 30(2). www.futureofchildren.org

Yeung, W. J., Linver, M. R., & Brooks–Gunn, J. (2002). How Money Matters for Young Children’s

Development: Parental Investment and Family Processes. Child Development, 73(6), 1861–1879.

https://doi.org/10.1111/1467-8624.t01-1-00511

Zippel, C. (2021). 9 in 10 Families With Low Incomes Are Using Child Tax Credits to Pay for

Necessities, Education. https://www.cbpp.org/blog/9-in-10-families-with-low-incomes-are-using-

child-tax-credits-to-pay-for-necessities-education

34
Appendix A. Spending Category Definitions and CE Interview Questions
Appendix Table A1: Major Expenditure Categories
Category Description
Housing and utility Shelter cost; utility cost; household operations; house furnishings and equipment
Food Food at home and away from home (including meals as, and not as, pay)
Alcohol and tobacco Alcoholic beverages and tobacco and smoking supplies
Clothing Clothing and footwear for men, women, boys, and girls, and other apparel
products and services
Transportation Cars and trucks, other vehicles, gasoline and motor oil, maintenance and repairs,
vehicle insurance, rental, leases, licenses, and public transportations
Health Health insurance, medical services, prescription drugs, and medical supplies
Leisure Fees and admissions to entertainment activities, televisions, radios, and sound
equipment, pets, toys, and playground equipment, and other entertainment
Personal care Wigs, hairpieces, or toupees, electric personal care appliances, and personal care
services for males and females, including haircuts
Education and reading Tuition, schoolbooks, supplies, and equipment for college, elementary and high
school, day care center, and other schools, and other school-related expenses;
newspapers and magazines and books
Miscellaneous Miscellaneous expenditures including funeral, cash contributions, insurance and
pension
Total Expenditure Total of above

Appendix Table 2: Child-related Expenditure Categories
Category Description
Children’s clothes Infant and children clothing and footwear
Books and toys Books and toys including infant furniture and equipment
Computers and tablets Computers, tablets, digital book readers, and other related software and
accessories including CD
School items School related items including tuition, schoolbooks, supplies, and school lunch
Sports items Sports item including athletic gear, bicycles, and camping items
Childcare Babysitting and childcare and day care
Enrichment activities Enrichment activities and outings including trips club membership tickets to
events fees for lessons musical instruments
Total Child Expenditure Total of above

Child Tax Credit Receipt
In response to the passage of the ARP, the BLS included a module of questions in the Consumer
Expenditure Interview Survey asking about the CTC starting with the October 2021 interviews. BLS did
not include questions about the expanded CTC in the Diary Survey. The BLS developed CTC questions
based on the questions used in earlier modules that were designed to ask about Economic Impact
Payments, except no question about the amount of CTC received was included. The CTC questions asked
are presented below and focus on receipt and general use:

35
CHDTXP. The Federal government’s 2021 American Rescue Plan included an advanced
payment of the Child Tax Credit. Since the first of the (reference month), have (you/you or any
members of your household) received a MONTHLY Child Tax Credit payment from the Federal
government?
• Yes (go to CHDTXPTH)
• No (go to S20B_END)

CHDTXPTH. How did or will (you/you or any members of your household) use the Child Tax
Credit payment?
• Mostly spend it (go to S20B_END)
• Mostly save it (go to S20B_END)
• Mostly use it to pay off debt (go to S20B_END)

Appendix Table A3 shows responses to the two CTC questions for three different cross-sections:
by interview month, by income, and by race and ethnicity. The frequencies are derived from the CE
fielded from August 2021 through March 2022. The sample for the column reporting the rate of receipt is
restricted to only households with at least one child under the age of 18 years old and who does not have a
missing value for CHDTXP. The sample for the columns reporting the rates of use is restricted to
households who reported receiving an advanced CTC payment (CHDTXP = 1) and who does not have a
missing value of CHDTXPH.
Panel A of Table 3A shows the rate of reported receipt and reported use by interview month. The
first row of Panel A shows the overall rate of receipt, 59 percent. Rates for August and September are
missing because the CTC questions were not included until October 2021 interviews. The rate of receipt
in October was 63 percent and grew to 66 percent in January. Reported receipt fell to 58 percent in
January and then to 38 percent in March. The overall rate of receipt in the CE is lower than the 79 percent
Hamilton et al. (2022) report; however, this receipt rate was restricted to eligible households, rather than
all households with children. Data from the Census Bureau Household Pulse Survey (HPS) show about
57 percent of households with children report receiving an advanced CTC payment (Karpman et al.,
2021).
Most households that reported receipt reported mostly spending the advanced CTC payment (54
percent) with the remaining households about evenly split between “mostly save” (24 percent) and
“mostly use it to pay off debt” (22 percent). The distribution of reported use appears to hold fairly steady
across the interview months until the last few interview months. In February, there is a slight increase in
the reported rate of using the CTC for spending that is intensified in March. The increase in spending
comes from a decrease in both savings and using the CTC to pay off debt, but the rate of savings saw a

36
bigger decline. These results are in line with what other studies have found (Hamilton et al., 2022;
Karpman et al. 2021; Pilkauskas and Cooney, 2021).
Panels B show the rates of reported receipt and use by income. Households with income between
$50,000 and $100,000 were slightly more likely to report receipt than households with income below
$50,000. Households with income above $200,000 were the least likely to report receipt, which is in line
with the phaseout rules for the CTC. Reported rates of using the CTC payment for spending were higher
for households with lower income. 62 percent of households with incomes under $50,000 report mostly
spending the CTC and only 12 percent reported mostly saving it. In contrast, 56 percent of households
with incomes between $50,000 and $100,000 report mostly spending the CTC and the rate of mostly
saving it increases to 20 percent. The discrepancy in report usage is even more stark when looking at
households with income above $200,000. Only 40 percent report mostly spending the CTC and 47
percent report mostly saving it.
Finally, Panel C shows the rates of receipt and use by race and ethnicity. Non-Hispanic, White
households were the most likely to report receiving a CTC payment (61 percent), but this is only slightly
more likely than the rate of receipt reported by Hispanic households (60 percent). Non-Hispanic, Black
households were the least likely to report receiving a CTC payment, with only 55 percent of these
households reporting receipt. Non-Hispanic, Asian and Non-Hispanic, White households were the least
likely to report mostly spending the CTC payment, 49 percent and 53 percent respectively. Non-Hispanic,
Other and Hispanic households were the most likely to report mostly spending the CTC, 60 percent and
56 percent respectively. Non-Hispanic, Black households were the most likely to report using the CTC
payment to pay off debt, 27 percent.

37
Appendix Table A3: Rates of reported receipt and use of the advanced CTC payment
A. Reported receipt and use by interview month
Use
Reported Mostly use it
Mostly spend it Mostly save it
receipt to pay off debt
Overall 59.72% 53.73% 23.68% 22.59%
August ‘21 - - - -
September ‘21 - - - -
October ‘21 63.06% 52.20% 25.71% 22.09%
November ‘21 63.50% 53.74% 25.27% 20.99%
December ‘21 66.84% 53.28% 22.88% 23.84%
January ‘22 66.21% 51.37% 24.62% 24.01%
February ‘22 58.64% 54.69% 22.72% 22.59%
March ‘22 38.76% 60.44% 18.53% 21.03%

B. Reported receipt and use by income
Use
Reported Mostly use it
Mostly spend it Mostly save it
receipt to pay off debt
Overall 59.72% 53.73% 23.68% 22.59%
Less than $24,000 51.57% 57.93% 12.52% 29.55%
Less than $50,000 57.17% 62.20% 11.74% 26.06%
$50,000 - $100,000 62.35% 55.92% 20.15% 23.93%
$100,000 - $200,000 61.53% 47.91% 31.39% 20.71%
$200,000 + 54.14% 40.04% 46.00% 13.95%
C. Reported receipt and use by race of reference person
Use
Reported Mostly use it
Mostly spend it Mostly save it
receipt to pay off debt
Overall 59.72% 53.73% 23.68% 22.59%
Asian 51.89% 49.26% 36.91% 13.83%
Black 55.22% 55.00% 17.68% 27.32%
Hispanic 59.65% 56.01% 18.62% 25.37%
Other 56.28% 59.87% 20.16% 19.97%
White 61.87% 52.74% 25.86% 21.40%
Note: Results derived from the CE fielded from August 2021 to March 2022. Estimates are weighted to the national
level using household population weight FINLWT21. Values for August and September are not reported because
the questions were not asked during these interview months. The sample used for the “Reported receipt” column is
limited to households with children under the age of 18 years old and whose value was not missing. About 4 percent
of the sample across all interview months have a missing value for receipt, not including August or September. The
sample used for the “Use” columns is limited to households who reported receiving the CTC (CHDTXP = 1) and
whose value was not missing. About 1 percent of the sample across all interview months have a missing value for
reported use, not including August or September.

38
Appendix B. Descriptive results
Our descriptive findings show that spending among households with children was higher in the
period when they could have received the monthly CTC payments compared to prior periods. Table B1
shows the average three-month expenditures for consumer units with children under age 18 across our
expenditure categories of focus. We present results for 2021 and 2019, and we disaggregate average
expenditures within each year into two periods. ‘2021’ is defined as interviews conducted in January 2021
through March 2022, and ‘2019’ is defined as interviews conducted in January 2019 through March 2020.
Period 1 includes data from CE interviews conducted before August and period 2 includes data from
interviews conducted in August or later. In 2021, period 2 is period during which households could
receive their monthly CTC payments. Data from 2019 serves as a helpful reference for what might have
happened between these periods in the absence of the CTC payments. Overall, the results show increases
in spending among consumer units with children in 2021 between period 1 and 2 across all expenditure
categories, and our “naïve diff-in-diff (final column of Table B1), shows that most of these increases were
larger than what was observed in 2019. These results show a difference in spending between these two
periods in 2021 that differs from 2019 patterns. However, they do not control for several other factors that
could affect spending, which we deal address by estimating effects with our main model, and in our
sensitivity analysis (i.e., the role of inflation).

39
Table B1: Change in spending between pre and post for households with children (naïve first difference)
2019 2021
Period 1 Period 2 Period 1 Period 2 Naïve
Jan '19 to Aug '19 to Jan '21to Aug '21to diff-in-
Diff Diff
Jul '19 Mar '20 July '21 Mar '22 diff
Interviews Interviews Interviews Interviews
Major categories
Total outlays $15,270 $15,237 -$33 $16,340 $17,309 $969 $1,002
Food $6,525 $6,404 -$120 $7,014 $7,299 $286 $406
Food at home $2,753 $2,763 $10 $3,034 $3,285 $251 $241
Housing $1,911 $1,911 $0 $2,213 $2,283 $70 $70
Alcohol & tobacco $842 $852 $10 $820 $1,002 $181 $171
Clothing $190 $207 $18 $203 $234 $31 $13
Transportation $429 $470 $41 $493 $611 $118 $77
Health $2,853 $2,827 -$26 $2,828 $3,065 $237 $263
Leisure $1,196 $1,189 -$8 $1,296 $1,265 -$31 -$23
Personal care $770 $764 -$5 $873 $909 $37 $42
Education $109 $105 -$4 $108 $120 $11 $15
Misc. $311 $390 $79 $343 $374 $31 -$48
Child-rel. spending
Total child spending $1,126 $1,186 $60 $1,165 $1,308 $143 $83
Children's clothes $165 $197 $32 $210 $285 $75 $43
Books and toys $66 $67 $0 $87 $95 $8 $7
Computers and tablets $64 $64 $0 $87 $102 $15 $15
School items $186 $196 $10 $193 $202 $8 -$2
Sports items $50 $53 $4 $90 $85 -$4 -$8
Childcare $326 $341 $15 $253 $283 $29 $15
Enrichment $269 $268 -$1 $245 $256 $11 $12
N 3,606 3,876 3,336 3,547
Note: Average spending amounts derived from the CE fielded from January 2019 to March 2020 and from January 2021 to
March 2022. The sample is limited to households with children who would qualify for a monthly CTC payment (i.e., imputed
CTC > 0), regardless of interview month, and weighted using household population weight FINLWT21.

40
Appendix C. Supplementary results
Figure C1. Average three-month total expenditures by interview year and month among households with
children who would be eligible for the ARPA CTC

Note: The data used to produce Figure C1 come from the Consumer Expenditure Interview Survey. Values represent the average
three-month total spending among households with children who would be eligible for the ARPA CTC and are weighted to the
national level using household population weight FINLWT21. Spending data is inflation adjusted to January 2021 dollars. All
values are authors’ calculations.

41
Table C1: Placebo Test, Estimated effects on spending of a $100 increase in CTC payments during the
reference quarter across expenditure categories among households with children, by household income
Before tax income
Income under Income under $50,000 - $100,000 -
Overall $200,000 +
$24,000 $50,000 $100,000 $200,000
Total 3.049 26.562 11.077 -8.939 13.124 161.265
(15.840) (23.170) (18.663) (21.212) (39.245) (217.909)
Housing 6.711 -0.179 3.105 -12.578 20.643 242.903
(7.569) (8.663) (7.935) (8.835) (19.268) (125.278)
Food 3.368 10.316 0.310 9.714 2.935 8.404
(3.404) (8.115) (4.271) (5.958) (6.867) (26.281)
Food, home 5.132 10.070 1.937 9.761 5.833 -0.438
(2.801) (7.631) (3.536) (5.372) (5.031) (16.849)
Food, away -1.764 0.246 -1.627 -0.047 -2.898 8.842
(1.548) (2.090) (1.709) (2.567) (3.688) (16.770)
Alcohol and Tobacco -0.668 1.838 1.818 -0.758 -2.086 -11.484*
(0.787) (1.960) (1.496) (1.222) (1.208) (5.046)
Clothing 0.251 4.100 1.764 3.422 -6.950 8.163
(1.416) (2.288) (1.970) (1.969) (3.574) (15.065)
Transport -2.191 13.959 3.694 -6.028 -2.144 10.577
(7.544) (16.345) (11.995) (10.940) (12.976) (96.737)
Health 0.555 -0.957 1.451 1.217 -0.330 -10.481
(2.794) (3.031) (2.623) (4.124) (7.646) (22.612)
Leisure -1.410 1.079 -0.145 -0.151 -5.788 3.205
(1.800) (1.717) (1.568) (2.416) (4.673) (24.393)
Personal Care 0.015 0.254 -0.256 -0.181 1.345 -2.869
(0.383) (0.440) (0.498) (0.532) (0.963) (2.704)
Education -3.186 -4.226 -4.984** -2.232 9.493 -87.108
(4.943) (2.307) (1.872) (6.560) (14.724) (62.239)
Miscellaneous -0.396 0.378 4.320 -1.364 -3.993** -0.044
(1.398) (1.093) (3.579) (1.163) (1.521) (15.515)
Total Child 6.722 8.426* 4.675 5.347 11.797 -24.788
(5.407) (3.419) (3.083) (6.188) (17.046) (64.754)
Children's Clothes 2.701*** 3.590* 2.996* 4.051*** -0.572 7.607
(0.752) (1.792) (1.470) (1.092) (1.378) (4.534)
Books and Toys -0.347 -0.808 0.067 0.114 -1.333 0.105
(0.292) (0.418) (0.403) (0.414) (0.750) (2.169)
Computers and Tablets 0.083 0.008 0.219 0.625 -1.687 5.805
(0.463) (0.493) (0.415) (0.890) (0.930) (5.207)
School Items 3.779 0.963 -0.540 3.444 11.234 -32.195
(4.539) (1.016) (1.009) (5.457) (14.438) (55.363)
Sports Items -0.223 -0.600 -0.921* 0.316 -1.079 2.915
(0.376) (0.523) (0.370) (0.526) (0.930) (4.497)
Childcare 1.143 4.860 3.914* -4.058* 5.810 -3.829
(1.857) (2.507) (1.987) (1.880) (5.425) (22.733)
Enrichment -0.414 0.412 -1.060 0.856 -0.575 -5.196
(1.195) (0.655) (0.572) (1.465) (3.338) (16.63)
N 12,618 2,238 4,155 4,033 3,354 1,076

Note: All results derived from the CE fielded from January 2018 to December 2019. The sample is limited to households with
children who would qualify for a monthly CTC payment (i.e., imputed CTC > 0), regardless of interview month. The estimated
effect, coefficient from the regression model, represents the change in spending by domain associated with a $100 increase in
quarterly income assumed to be from the CTC. All models include state, year, and month fixed effects, and a fixed effect for
number of people in the household, and demographic controls for household reference person, including age group, race/ethnicity
(in the case of model 1), sex, and education level. Estimates are weighted to the national level using household population weight
FINLWT21 Robust standard errors reported in parentheses. ***p<0.001 **p<0.01 *p<0.05.

42
Table C2: Placebo Test, Estimated effects on spending of a $100 increase in CTC payments during the
reference quarter across expenditure categories among households with children by race/ethnicity of
household head
Race/ethnicity of household reference person
Overall Asian Black Hispanic Other White
Total 3.049 -228.372* 9.236 9.977 298.249* -1.928
(15.840) (115.515) (30.915) (21.974) (139.025) (24.510)
Housing 6.711 -37.610 23.011 2.202 80.928 4.675
(7.569) (56.039) (15.904) (9.435) (44.177) (11.872)
Food 3.368 -28.517 -9.073 3.480 81.799* 4.880
(3.404) (16.989) (7.812) (7.925) (41.38) (4.228)
Food, home 5.132 -15.875 -5.780 6.378 40.757 5.744
(2.801) (11.496) (6.284) (7.385) (21.534) (3.188)
Food, away -1.764 -12.642 -3.293 -2.898 41.042 -0.865
(1.548) (9.558) (3.099) (2.671) (22.560) (2.284)
Alcohol and Tobacco -0.668 2.837 -2.061 -0.667 -0.115 -0.376
(0.787) (2.390) (1.292) (0.781) (8.001) (1.33)
Clothing 0.251 8.880 4.623 1.608 3.908 -2.008
(1.416) (8.491) (5.309) (2.047) (14.651) (1.847)
Transport -2.191 -123.054 0.873 9.440 24.31 -1.655
(7.544) (80.700) (10.198) (16.452) (51.168) (9.766)
Health 0.555 6.543 -3.838 -1.160 33.889 0.520
(2.794) (10.541) (5.399) (3.425) (21.197) (4.481)
Leisure -1.410 -15.826 -3.565 0.071 36.490** -2.265
(1.800) (16.241) (2.741) (2.133) (13.991) (2.848)
Personal Care 0.015 -0.837 -0.075 0.123 0.191 0.126
(0.383) (1.279) (1.715) (0.474) (1.924) (0.447)
Education -3.186 -57.904* -2.394 -4.970* 19.560 -2.688
(4.943) (23.712) (4.821) (2.124) (21.488) (8.876)
Miscellaneous -0.396 17.117 1.735 -0.150 17.290 -3.136**
(1.398) (19.519) (5.110) (0.947) (19.695) (1.181)
Total Child 6.722 -27.241 10.044 1.189 38.931 6.578
(5.407) (23.192) (7.511) (3.579) (32.755) (9.489)
Children's Clothes 2.701*** 4.293 5.532 1.886 7.319 2.014*
(0.752) (2.923) (3.328) (1.195) (9.035) (0.803)
Books and Toys -0.347 0.963 -1.420* -0.302 4.671 -0.302
(0.292) (1.206) (0.692) (0.531) (3.607) (0.423)
Computers and Tablets 0.083 0.068 1.010 0.589 -0.427 -0.392
(0.463) (2.617) (1.285) (0.949) (1.894) (0.595)
School Items 3.779 -23.631* -1.253 -0.142 16.818 6.774
(4.539) (11.766) (2.798) (1.009) (18.731) (8.326)
Sports Items -0.223 0.608 -0.728 -0.512 5.413 -0.381
(0.376) (1.460) (0.729) (0.353) (3.822) (0.629)
Childcare 1.143 2.339 7.734 0.801 -6.502 -1.081
(1.857) (11.516) (4.59) (2.166) (13.878) (2.903)
Enrichment -0.414 -11.881 -0.831 -1.132 11.638 -0.053
(1.195) (14.583) (1.204) (1.340) (8.426) (1.873)
N 12,618 848 1,385 2,862 285 7,238

Note: All results derived from the CE fielded from January 2018 to December 2019. The sample is limited to households with
children who would qualify for a monthly CTC payment (i.e., imputed CTC > 0), regardless of interview month. The estimated
effect, coefficient from the regression model, represents the change in spending by domain associated with a $100 increase in
quarterly income assumed to be from the CTC. All models include state, year, and month fixed effects, and a fixed effect for
number of people in the household, and demographic controls for household reference person, including age group, race/ethnicity
(in the case of model 1), sex, and education level. Estimates are weighted to the national level using FINLWT21 Robust standard
errors reported in parentheses. ***p<0.001 **p<0.01 *p<0.05.

43
Table C3 shows the treatment effects for major and child-related expenditure categories using four
different model specifications. The first column provides the results from our main analysis, which are
discussed in the main text of the paper. The second column presents results from the same model as our
main analysis but excludes observations from August and September – interviews conducted during the
period when CTC payments were distributed but before the receipt question was added to the CE. The
third column presents estimates for the treatment-on-treated effect, where the imputed CTC payment is
interacted with an indicator for payment receipt. Therefore, only CUs that reported receiving a CTC
payment during the reference period are considered treated. Finally, the fourth column presents results
from a two-stage least squares model. In the first stage, the imputed CTC payment is regressed on
reported receipt, and in the second stage, the predicted CTC payments from the first stage are used to
estimate the treatment effect.

44
Table C3: Treatment on treated results, Estimated effects on spending of a $100 increase in CTC
payments during the reference quarter across expenditure categories among households with children, by
household income
Overall Restricted Sample Received Received 2SLS
Total 43.619* 34.190 33.558* 47.623
(17.268) (17.804) (16.614) (24.635)
Housing 28.028** 25.934** 14.351 36.124*
(10.142) (10.435) (9.183) (14.534)
Food 12.309** 10.551** 7.651 14.697*
(3.992) (4.158) (4.118) (5.79)
Food, home 7.406* 5.056 4.871 7.042
(3.001) (3.097) (3.183) (4.288)
Food, away 4.902* 5.495*** 2.780 7.654**
(1.961) (2.038) (1.86) (2.866)
Alcohol and Tobacco -0.255 -0.587 -0.578 -0.817
(0.855) (0.883) (0.872) (1.226)
Clothing 5.083** 4.342* 2.167 6.048*
(1.985) (2.132) (2.194) (2.953)
Transport -2.697 -2.160 -1.028 -1.823
(6.246) (6.529) (0.288) (0.597)
Health -1.474 -0.970 -2.265 -4.745
(3.801) (4.047) (0.203) (0.469)
Leisure 5.082 0.798 -1.103 -2.502
(3.215) (2.541) (0.39) (0.941)
Personal Care 0.589 0.494 1.158 -3.009
(0.418) (0.433) (6.297) (9.069)
Education -5.016 -5.602 -0.966 -3.196
(3.006) (3.167) (5.39) (7.761)
Miscellaneous 1.919 1.342 2.124 0.188
(1.439) (1.507) (2.784) (4.143)
Total Child 15.661*** 13.926*** 15.077*** 19.398***
(3.656) (3.831) (3.802) (5.249)
Children's Clothes 4.758*** 4.243*** 3.241** 5.910***
(0.957) (1.009) (1.04) (1.407)
Books and Toys 0.473 0.483 0.612 0.672
(0.557) (0.588) (0.608) (0.816)
Computers and Tablets 0.242 0.082 0.069 0.114
(0.684) (0.7) (0.674) (0.972)
School Items 1.499 1.143 2.861 1.593
(2.121) (2.237) (2.184) (3.104)
Sports Items 0.396 0.559 0.757 0.779
(0.607) (0.616) (0.657) (0.855)
Childcare 6.086*** 5.393*** 3.837* 7.511**
(1.89) (1.941) (1.842) (2.697)
Enrichment 2.205 2.024 3.700* 2.820
(1.544) (1.66) (1.695) (2.285)
N 14,365 12,486 12,486 12,486
Note: All results derived from the CE fielded from January 2018 to December 2019. The sample is limited to households with
children who would qualify for a monthly CTC payment (i.e., imputed CTC > 0), regardless of interview month. The estimated
effect, coefficient from the regression model, represents the change in spending by domain associated with a $100 increase in
quarterly income assumed to be from the CTC. All models include state, year, and month fixed effects, and a fixed effect for
number of people in the household, and demographic controls for household reference person, including age group, race/ethnicity
(in the case of model 1), sex, and education level. Estimates are weighted to the national level using FINLWT21 Robust standard
errors reported in parentheses. ***p<0.001 **p<0.01 *p<0.05.

File and source

File: bls-ec240070.pdf
Size: 1,249,906 bytes
SHA-256: d9a583e904f951e20802f80aee75fbea4a69e26d036c9b2ff7dd64903c3f61b4
Our copy: bls-ec240070.pdf
Original: www.bls.gov