Expiration of the Expanded Child Tax Credit and Energy Insecurity in US Households With Children, 2021–2022

Abstract

Objectives. To assess the relationship between the expiration of the expanded child tax credit (ECTC) and changes in energy insecurity among US households with children from 2021 to 2022.

Methods. We used nationally representative survey data to evaluate changes in energy insecurity among credit-eligible versus -ineligible households during and after the ECTC expansion. We performed difference-in-differences analyses to estimate changes in energy insecurity in the 2 groups and conducted stratified analyses by percentage of the federal poverty line.

Results. There was a 0.95 percentage point increase in inability to pay energy bills (95% confidence interval = 0.07, 1.85) among households with children but no difference in other measures of energy insecurity. Stratified analyses showed that households at 200% to 399% and 400% to 599% of the federal poverty line experienced increases in energy insecurity after expiration of the ECTC. We saw no differences in the lowest-income groups (< 200% federal poverty line).

Conclusions. The expiration of the ECTC was associated with increased inability to pay energy bills, suggesting that the credit prevented some forms of energy insecurity in households with children. Associations were limited to middle-income groups, indicating benefits were limited to this income stratum. (Am J Public Health. 2025;115(8):1312–1321. https://doi.org/10.2105/AJPH.2025.308105)

Energy insecurity, defined as the inability to adequately meet basic household energy needs, is a pervasive health-related social need in the United States.¹ Although more than 1 in 4 US households experience some form of energy insecurity,² only 22% of income-eligible households receive federal Low Income Home Energy Assistance Program (LIHEAP) assistance annually.^3,4 Increasingly extreme temperatures have created a context in which the share of households experiencing energy insecurity may continue to grow.⁵ Energy insecurity is higher among households with children and lower-income households.^2,6,7

Energy insecurity harms children’s health and may contribute to health disparities. Its immediate effects are experienced as extreme home temperatures, heat- and cold-related illnesses, alternative and hazardous heating strategies, trade-offs between competing necessities, and parental stress.^1,8 When income is insufficient to cover the cost of living expenses, families are forced to choose between food and energy. This so-called “heat or eat” dilemma has dire consequences for children’s health: children are more often diagnosed with failure to thrive in wintertime, and families with threatened utility shutoff are more likely to report that their children are hungry.^9,10 If families prioritize other needs over utilities, children are subjected to unsafe indoor temperatures; nationwide, 11% of households with children reported keeping the home at an unsafe temperature.² Infants and toddlers in energy-insecure homes are also at increased risk for developmental delays.¹¹ A child living in a cold home is twice as likely to have respiratory problems and up to 5 times more likely to have mental health problems.^12,13

Tax credits may prevent energy insecurity and related adverse health effects. Tax credits are a form of nondirected cash benefit that gives families the autonomy to spend flexibly where their need is greatest. Under the American Rescue Plan Act of 2021 (Pub L No. 117–2), the child tax credit was expanded from $2000 to up to $3000 to $3600 per child annually (depending on household income and child age) and made fully refundable. Half the value of the credit was distributed in monthly advance payments from July to December 2021, with the remainder of the credit distributed upon tax filing.¹⁴ All 2-parent households with incomes less than $150 000 and single-parent households with incomes less than $112 500 qualified to receive $300 per month for each child aged 5 years and younger and $250 per month for children aged 6 to 17 years during this 6-month period.

The expanded child tax credit (ECTC) lifted 2.1 million children above the poverty line, resulting in a historic low of 5.2% children in poverty at the end of 2021; its expiration increased child poverty to 12.4%.^15,16 Previous studies have found that, by supplementing families’ budgets, implementation of the ECTC was associated with a 26% relative reduction in food insufficiency and a 1.9 percentage point (pp) absolute reduction in food insecurity.^17,18 Conversely, the expiration of the ECTC in December 2021 was associated with a 17% to 23% relative increase in food insufficiency.¹⁹ We hypothesized that the ECTC may have similarly protected against energy insecurity, another form of material hardship.

Tax credits may benefit families who do not qualify for other benefit programs because of the “benefit cliff” phenomenon. Federal benefit programs such as the Supplemental Nutrition Assistance Program and LIHEAP, which helps families pay energy bills, are available only to families with incomes up to 130% and 150% of the federal poverty level (FPL; as determined by these respective programs), respectively.^20,21 Low-income families above these thresholds may not qualify for benefits despite household needs. Thus, the child tax credit presents an approach to addressing family health-related social needs that are above the benefit cliff threshold and may offset loss of benefits because of employment income exceeding a threshold.²²

To our knowledge, there are no studies evaluating the impact of tax credit programs on energy insecurity. We aimed to (1) assess whether expiration of the ECTC was associated with changes in the prevalence of energy insecurity among those eligible for the credit, and (2) identify disparate consequences of ECTC expiration between income subgroups. Findings can inform future iterations of the child tax credit, in particular for households who may exceed income limits for existing benefit programs.

METHODS

We used nationally representative, cross-sectional data from the Household Pulse Survey (HPS). The HPS is an online survey conducted by the US Census Bureau to collect real-time data regarding the impact of COVID-19 on the US population that began in April 2020 and is continuing, with new collection cycles starting every 4 weeks. HPS wave 34 was the first to include items on energy insecurity. We used data from waves 34 to 48 of the HPS (July 2021– August 2022: response rate = 4.4%–7.9%), which included approximately 6 months during ECTC payments and 6.5 months after the ECTC expired. One adult (aged ≥ 18 years) per household responded to the HPS and provided responses for themselves and their household.

We applied several exclusions to the HPS to arrive at our final analytic sample. Survey waves 34 to 48 included 971 836 respondents. To increase comparability between eligibility groups by focusing our analysis on households with working-aged adults, we excluded respondents older than 65 years (n = 397 072; 40.8%). We additionally excluded individuals in the top 1% of income, as income eligibility for the credit was capped at $400 000 (∼1500% of the FPL for a family of 4 in 2021) for joint tax filing (n = 3283; 3.3%).²³ Finally, we excluded respondents who did not have complete covariate or outcome data, apart from those missing income data (n = 9274; 1.0%). Our main analytic sample was made up of 562 207 participants (Table 1). We used a subsample of 88 864 participants who self-reported that they had received a monthly ECTC payment from survey waves 34 to 42 (during disbursement of the credit) to better understand spending patterns among those who received the ECTC.

TABLE 1—

Sample Characteristics by Expanded Child Tax Credit (ECTC) Eligibility: Household Pulse Survey, United States, July 2021–August 2022

Characteristic	Comparison Groups, No. (%) or Mean ±SD		P b
	ECTC Ineligible (Without Children; n = 337 055a)	ECTC Eligible (With Children; n = 225 152)
Age, y			< .001
18–24	15 340 (8.8)	4 723 (5.1)
25–44	106 406 (38.0)	131 171 (63.4)
45–64	215 309 (53.2)	89 258 (31.5)
Mean age caregiver, y	44.5 ±0.05	40.2 ±0.04
Sex at birth			< .001
Male	136 687 (51.5)	81 321 (44.2)
Female	200 368 (48.5)	143 831 (55.8)
Race/ethnicity			< .001
Non-Hispanic White	254 896 (67.7)	155 623 (57.7)
Hispanic	25 932 (11.7)	23 643 (17.1)
Non-Hispanic Black	23 160 (10.4)	18 527 (13.2)
Non-Hispanic Asian	16 723 (4.8)	14 163 (5.8)
Otherc	16 344 (5.3)	13 196 (6.3)
Education level			< .001
Less than high school graduate	4 868 (4.8)	5 469 (8.2)
High school graduate or general equivalency diploma/equivalent	36 295 (27.2)	23 127 (26.2)
Some college or 2-yr degree	105 739 (31.1)	68 027 (30.5)
College graduate	106 618 (21.7)	66 432 (18.4)
Graduate degree	83 535 (15.3)	63 097 (16.8)
Household income, % of the FPL			< .001
0–199	65 728 (19.5)	57 327 (25.5)
200–399	80 699 (23.9)	54 045 (24.0)
400–599	63 881 (19.0)	52 165 (23.2)
≥ 600	116 131 (34.5)	52 413 (23.3)
Missing	10 616 (3.2)	9 202 (4.1)
Marital status			< .001
Married	167 429 (43.6)	167 521 (67.7)
Not married	169 626 (56.4)	57 631(32.3)
Number of people in household			< .001
1	86 623 (26.3)	0
2	172 316 (48.6)	15 917 (7.3)
≥ 3	78 116 (25.0)	209 235 (92.7)
Housing ownership			< .001
Own	228 160 (60.1)	170 226 (66.2)
Rent	104 354 (38.0)	52 207 (32.1)
Occupied/nonrented	4 541 (1.9)	2 719 (1.7)
Housing type			< .001
Apartment	86 252 (29.8)	27 744 (17.0)
Single-family house	237 162 (64.9)	188 945 (77.4)
Mobile home	11 528 (4.5)	7 815 (5.2)
Other	2 113 (0.8)	648 (0.4)
Public benefits: household receives SNAP/food stamps	22 358 (9.3)	26 871 (18.9)	< .001

Note. FPL = federal poverty level (determined by Health and Human Services for 2021 and 2022, matched with survey data collected in each of these years); SNAP = Supplemental Nutrition Assistance Program.

^a Total weighted participants = 70 053 457. All percentages reflect weighted values.

^b P values represent weighted χ² test.

^c Other indicates respondent selected “any other race alone or race in combination.”

Expanded Child Tax Credit Comparison Groups

We defined ECTC eligibility for the purpose of this study based on responses to “How many people under 18 years old currently live in your household?” Households with children were eligible for the ECTC (treatment group), and those without children were not eligible for the ECTC (control group). Previous studies of associations between the ECTC and social needs have compared these 2 groups, adjusting for potential confounders.^17–19 By comparing change over time (before vs after ECTC expiration) in outcomes between ECTC-eligible versus -ineligible groups (the treatment vs. control group, respectively), we were able to derive an intent-to-treat estimate of the ECTC’s impact on families with children. The intent-to-treat approach is more valid and policy relevant than a per protocol analysis comparing outcomes between families who did versus did not receive the ECTC because (1) it does not rely on potentially biased self-report of ECTC receipt, and (2) it reflects a real-world scenario in which not all eligible families received the tax credit (i.e., noncompliance).

Outcome

We examined 3 measures representing distinct dimensions of energy insecurity: inability to pay energy bills, foregoing basic necessities to pay energy bills (henceforth referred to as “trade-offs”), and keeping the home at an unsafe temperature. The questions in the HPS that describe these components of energy insecurity are (1) “In the last 12 months, how many times was your household unable to pay an energy bill or unable to pay the full bill amount?”; (2) “In the last 12 months, how many months did your household reduce or forego expenses for basic household necessities, such as medicine or food, in order to pay an energy bill?”; and (3) “In the last 12 months, how many months did your household keep your home at a temperature that you felt was unsafe or unhealthy?” Response options were “almost every month,” “some months,” “1 or 2 months,” and “never.”

We dichotomized responses, assigning a value of 0 for “never” and 1 for all other responses. This was based on our understanding that any degree of energy insecurity could be detrimental to one’s health and is consistent with the US Energy Information Administration’s reporting approach.² The Cronbach’s alpha indicated that these measures had low internal consistency, leading us to treat them as separate outcomes representing distinct, and possibly increasingly severe, forms of energy insecurity.

Statistical Analyses

In all analyses, we used HPS household survey weights to produce nationally representative estimates. First, we described characteristics of our ECTC-eligible and ECTC-ineligible groups, using χ² statistics to compare frequencies of sociodemographic characteristics between groups. Second, we conducted a difference-in-differences (DiD) analysis with repeated cross-sections to assess associations between ECTC eligibility and changes in energy insecurity following expiration of the ECTC. Those surveyed from July 21, 2021, through February 7, 2022 (survey waves 34–42), during the disbursement period of the ECTC, constituted the “preexpiration” group; those surveyed March 2, 2022, through August 8, 2022 (survey waves 43–48), after expiration of the ECTC, constituted the “postexpiration” group. Although the last monthly ECTC payment was reportedly sent in December 2021, we decided to conservatively group respondents in wave 42 (January 26, 2022–February 7, 2022) in the “preexpiration” period because many respondents to the HPS were still reporting receipt of the ECTC during that period (n = 7265, or 38.8% of eligible households, reported receiving ECTC during survey wave 42).

We adjusted our DiD models for covariates, including respondent age (18–24, 25–44, or 45–64 years), sex at birth (male or female), self-reported race and ethnicity, education level, income as a percentage of FPL, marital status, number of people in household, homeowner versus renter status, housing type, receipt of Supplemental Nutrition Assistance Program benefits, and state-level energy use and cost. We adjusted for race and ethnicity as proxies for experiences of structural racism, which decrease opportunities for wealth and homeownership and increase risk of energy insecurity.^1,24 We calculated percentage of FPL by assigning respondents the midpoint values of their reported income range, relative to household size–specific federal poverty guidelines (i.e., the FPL) from 2020 and 2021.²⁵

We used percentage of the FPL as opposed to income brackets, as it better reflects household-level poverty and is more useful for understanding implications of the ECTC relative to other FPL-based federal benefit programs. Additionally, we used US Energy Information Administration data to create a measure of customer average energy costs by state–month (data in cents/kilowatt hours/month) to account for state-specific and seasonal variation in temperature and utility costs.²³ Additionally, our model included state fixed effects.

To better understand our overall estimates and identify subgroups particularly affected by ECTC expiration, we conducted stratified DiD analyses by income (using percentage of the FPL). To further understand the results of stratified analyses, we conducted post hoc analyses describing ECTC spending patterns by income in a subsample of respondents who were ECTC recipients (88 864 adults who reported receiving the ECTC during the preexpiration period, July 2021–February 2022).

Finally, we conducted several sensitivity analyses. First, we applied an event study specification to our main DiD models. An event study is a more flexible form of DiD approach that allows treatment effects to vary over time relative to treatment.²⁶ The model included dummy variables for HPS waves (with survey wave 42 set as the reference), interacted with the treatment indicator (i.e., ECTC eligibility). By graphing event study coefficients, we were able to check for potential violations to the parallel trends assumption underlying DiD analyses. We also performed event studies for all significant findings in the income subgroups.

We conducted 2 additional analyses to check our results’ sensitivity to model specification. First, we ran a model that included controls for pandemic programs at the state–month level that might have a meaningful effect on energy insecurity. These included utility shutoff moratoria (1 = moratorium in place, 0 = no moratorium; coded based on dates in Benfer and Koehler’s Eviction Moratoria & Housing Policy database)²⁷ and emergency rental assistance (cumulative dollars per capita distributed, based on data from US Treasury Department reports).²⁸ We chose not to include these controls in our main analysis because we did not expect them to affect energy insecurity differently between households with and households without children. In a second sensitivity analysis, we used HPS wave 41, fielded December 29, 2021, through January 10, 2022, as the reference wave.

We conducted a final sensitivity analysis to estimate “as treated” associations. Because we collected data on whether participants received the ECTC only during the preexpiration period (and not the postexpiration period), we could not use ECTC recipients as our intervention group; we instead used families with children as a proxy. To understand whether the experience of households that received the ECTC would have been comparable to the overall sample of households with children, we compared adjusted energy insecurity during the preexpiration period in (1) households without children, (2) those with children who received the ECTC, and (3) those with children who did not receive the ECTC. We used Stata/BE version 17.0 (StataCorp LP, College Station, TX) for all analyses.

RESULTS

Our sample of 562 207 adults represented 70 053 457 households nationally and 28 602 826 households with children. As seen in Table 1, compared with ECTC-ineligible adults, ECTC-eligible adults were younger (mean age 40.2 vs 44.5 years), were more likely to be women (55.8% vs 48.5%), were more often Hispanic (17.1% vs 11.7%) or non-Hispanic Black (13.2% vs 10.4%) individuals, and had lower incomes. ECTC-eligible individuals also reported lower educational attainment (e.g., 18.4% vs 21.7% with a college degree), were more likely to be married (67.7% vs 43.6%), and had a larger household size (i.e., households with 3 or more people; 92.7% vs 25.0%). They were more likely to be homeowners (66.2% vs 60.1%), live in a single-family home (77.4% vs 64.9%), and receive Supplemental Nutrition Assistance Program benefits (18.9% vs 9.3%).

Figure 1 shows marginal estimates, that is, the adjusted prevalence of each energy insecurity outcome, in ECTC-eligible and -ineligible groups before versus after expiration of the ECTC, as well as the pre–post change by eligibility group. Both households with and those without children experienced an increase in all 3 measures of energy insecurity when the ECTC expired. The magnitude of these increases differed by outcome: they were larger among ECTC-eligible versus -ineligible households for inability to pay energy bills and trade-offs but smaller for unsafe temperatures. Baseline inability to pay energy bills and trade-offs were more common among households with children (25.2% vs 21.4% and 33.1% vs 32.2%, respectively), whereas keeping the home at an unsafe temperature was more common among households without children (21.3% vs 17.9%).

FIGURE 1—

Adjusted Prevalence of Energy Insecurity Before and After the Expanded Child Tax Credit (ECTC) Expiration by (a) Inability to Pay Bills, (b) Trade-offs, and (c) Unsafe Household Temperatures: Household Pulse Survey, United States, July 2021–August 2022

Note. The graphs display marginal estimates (i.e., adjusted prevalence of energy insecurity before and after ECTC expiration based on ECTC eligibility). The left y-axis represents pre–post prevalence. The right y-axis and black diamonds reflect pre–post difference estimates. Whiskers indicate 95% confidence intervals. Analyses are adjusted and weighted.

Adjusted DiD analyses (Figure 2) demonstrated that ECTC expiration was associated with a 0.95 pp (95% confidence interval [CI] = 0.06, 1.85) increase in inability to pay energy bills among ECTC-eligible households compared to ECTC-ineligible households. Because the baseline prevalence of inability to pay bills was 22.9% in our sample, this 0.95 pp increase equates to a 4.2% relative increase associated with the expiration of ECTC. There was no significant overall difference in trade-offs or unsafe temperatures after expiration of the ECTC. There were mixed results in subgroup analyses by FPL. ECTC expiration was associated with an increase in inability to pay energy bills (1.94 pp; 95% CI = 0.11, 3.78) and an increase in trade-offs (2.82 pp; 95% CI = 0.85, 4.80) among households at 200% to 399% of the FPL. It was also associated with an increase in inability to pay bills of 2.34 pp (95% CI = 0.90, 3.80) among households at 400% to 599% of the FPL. There was a significant decrease of 1.9 pp (95% CI = –2.93, –0.86) in unsafe household temperatures among households at 600% or higher FPL. There was no change in any measure of energy insecurity for households at lower income levels.

FIGURE 2—

Difference-in-Difference Estimates of Energy Insecurity Among Households With Children After Expiration of the Expanded Child Tax Credit (ECTC): Household Pulse Survey, United States, July 2021–August 2022

Note. CI = confidence interval; FPL = federal poverty level (determined by Health and Human Services for 2021 and 2022, matched with survey data collected in each of these years). This figure represents difference-in-difference estimates of percentage point change in energy insecurity among ECTC-eligible households with children after expiration of the ECTC. Overall estimates are followed by stratified estimates by percentage of the federal poverty level (determined by Health and Human Services for 2021 and 2022, matched with survey data collected in each of these years). Error bars represent 95% confidence intervals.

Spending data (Figure 3) suggested that in each progressively lower stratum of the FPL, there was progressively higher spending on all competing necessities, including food, housing, utilities, and debt.

FIGURE 3—

Competing Expanded Child Tax Credit (ECTC) Spending Needs by Household Income: Household Pulse Survey, United States, July 2021–August 2022

Note. FPL = federal poverty level (determined by Health and Human Services for 2021 and 2022, matched with survey data collected in each of these years). This figure depicts spending of ECTC recipients on competing basic necessities. It represents a subsample of 88 864 adults who reported receiving the ECTC during the disbursement period (survey waves 38–42; July 2021–February 2022). Respondents could select more than 1 category, so total spending does not equal 100%. The category of food includes groceries, eating out, and takeout. The category of utilities was specified to indicate both utilities and telecommunications, including natural gas, electricity, cable, internet, and cell phone. The category of housing includes spending on rent or mortgage. The category of debt indicates respondents who mostly spent their ECTC on debt. The categories of spending selected reflect categories commonly understood as basic necessities, but they do not encompass all possible spending categories.

Visual inspection of event study coefficients (Figure A, available as a supplement to the online version of this article at http://www.ajph.org) showed a relatively flat pretreatment (i.e., during ECTC) trend, indicating that energy insecurity did not increase significantly among ECTC-eligible families relative to ineligible families before ECTC expiration and suggesting that the parallel trends assumption was satisfied. One notable exception to this was the event study for unsafe temperatures among households at or above 600% of the FPL: in this group, report of unsafe temperatures was declining among ECTC-eligible households, relative to ineligible households, before expiration. Therefore, the DiD result showing a significant decrease associated with expiration may be spurious, possibly driven by preexpiration trends.

Our sensitivity analysis incorporating controls for state utility moratorium shutoffs and emergency rental assistance produced results that were qualitatively similar to our main results (Table A, available as a supplement to the online version of this article at http://www.ajph.org). Our analysis using wave 41 as a cutoff was broadly similar to our main results, although the coefficient for inability to pay energy bills was slightly attenuated (0.82 rather than 0.95; P = .06; Table A).

Comparison of adjusted energy insecurity in (1) households without children, (2) those that have children and reported receiving the ECTC, and (3) those that have children and did not report receiving the ECTC (Table B, available as a supplement to the online version of this article at http://www.ajph.org) revealed higher levels of inability to pay bills in both households that reported ECTC receipt and those that did not report receipt compared to ineligible households. Trade-offs were significantly higher in households that reported ECTC receipt compared to ineligible households. As stated in the “Methods” section, this was not our primary comparison for 2 reasons: (1) we find the intent-to-treat estimates more policy relevant, and (2) data on ECTC receipt were not available after ECTC expiration, making the DiD analysis unfeasible.

DISCUSSION

Ours is the first study, to our knowledge, to examine the association between expiration of the ECTC and household measures of energy insecurity. We found a statistically significant increase (0.95 pp) in inability to pay bills among ECTC-eligible households. Based on the 2022 estimates of households with children, this translates to an additional 308 560 households that have children and became energy insecure after ECTC payments ended.²⁹ We did not find statistically significant changes in overall measures of trade-offs or keeping households at unsafe temperatures. The latter is consistent with previous studies of energy insecurity, which have found low prevalence of unsafe temperatures²⁰—either because it is a coping mechanism of last resort or owing to social desirability bias (parents of children may be hesitant to report “unhealthy” temperatures). These findings suggest that after ECTC expiration, households with children had greater difficulty paying energy bills, although we detected no difference in other, potentially more severe forms of energy insecurity.

Income-stratified results showed an increase in 2 measures of energy insecurity following expiration of the ECTC among eligible households at 200% to 399% of the FPL. We henceforth describe this group as “middle income,” consistent with previous studies.³⁰ Respondents at 400% to 599% of the FPL had increased inability to pay energy bills after expiration but did not demonstrate any change in foregoing basic necessities. Taken together, these findings suggest that ECTC prevented milder forms of energy insecurity overall, in particular for middle-income families.

We propose 3 potential mechanisms through which the ECTC had the greatest impact on middle-income households. First, these households would not have been federally eligible for the LIHEAP or Supplemental Nutrition Assistance Program.²¹ The protective association of the ECTC seen in the 200% to 399% of the FPL group may reflect benefit cliff effects, whereby households who earn just above eligibility thresholds for benefit programs are ineligible despite existing needs.²² Second, ECTC may not have affected energy insecurity among the lowest-income groups (0%–199% of the FPL) because of a greater number of competing needs, leading families to prioritize ECTC spending on necessities such as food. Third, lowest-income households may have had a lower uptake of the ECTC, as Karpman et al. have demonstrated, ³¹ which may explain the varying associations between ECTC expiration and energy insecurity by FPL.

Strengths and Limitations

Our study builds on previous work that examined the association between ECTC and other forms of material hardship. Shafer et al.,¹⁷ Bovell-Ammon et al.,³² and Rook et al.¹⁸ have contributed to evidence of decreased food insecurity and insufficiency during the period of ECTC receipt. Our results are comparable to, albeit lower than, the 3.5 pp increase in food insufficiency identified by Bouchelle et al. in low-income households after ECTC withdrawal.¹⁹ Our study demonstrates that the reduction in poverty may not be sufficient to address all forms of material hardship for the lowest-income households, given inequitable burdens of debt and competing necessities.

Our study had several limitations. Although we used nationally representative data from during and after ECTC disbursements, our data were limited by their serial cross-sectional nature. Additionally, although our sample size was large, stratified analyses may have been underpowered to make comparisons. Response rate for the HPS is low, and data are subject to nonresponse bias, although we partially accounted for this with survey weighting. Although we controlled for several potential confounders, households with and those without children and periods before and periods after ECTC expiration may differ in ways that are unmeasured. Our outcome variable asked participants to reflect back over 12 months and thus might not be sensitive to acute changes in household finances. Additionally, our intent-to-treat approach to measuring exposure based on presumed eligibility provided policy-relevant estimates of ECTC impact but likely biased results toward the null.

Public Health Implications

Our findings suggest that the ECTC may have prevented some forms of energy insecurity in households with children, particularly middle-income households; however, the ECTC did not address potentially more severe measures of energy insecurity or the experience of energy insecurity in the lowest-income households. Given the rise of energy insecurity, and with the knowledge that social needs lead to negative and costly health outcomes, federal and state governments should consider reinstating more generous child tax credit policies and distributing credits according to household need.