7

Potential Labor Supply and Employment Effects of Tax Credits

Understanding how the American Rescue Plan Act of 2021 (ARPA) Earned Income Tax Credit (EITC) and Child Tax Credit (CTC) Policies in 2021 affected child poverty depends, in part, on whether and how families changed their employment in response to the policy changes.¹ Large declines in employment following the ARPA-induced changes to these credits, for instance, could offset some of their poverty-reducing effects.² Similar labor supply responses could also offset some of the poverty-reducing effects of future policies, for example, if the introduction of a permanent child allowance altered the financial incentives associated with working. These behavioral responses would likely differ from those resulting from changes to the EITC and CTC under ARPA, for reasons discussed below. This chapter summarizes the relevant literature that examines how the EITC and CTC have historically affected labor supply and provides a rationale for accounting for labor supply effects presented in Chapter 8, as well as

___________________

¹ As discussed throughout the report, the committee focused on estimating the child poverty impacts of the provisions of the CTC under ARPA relative to both those under the Tax Cuts and Jobs Act (TCJA) and to the absence of the CTC and the impact of the EITC under ARPA and relative to no EITC.

² Parents may choose to reduce their employment in response to these policy changes for a variety of reasons that have implications for their economic futures, such as going back to school, searching for a better job, or to care for an ill family member. While these behavioral responses may have both short- and long-term implications for child poverty, explicitly modeling these responses is beyond the scope of this report. These trade-offs are discussed in more detail in Chapter 1.

for the committee’s policy options presented in Chapter 9.³ Main messages from this chapter are highlighted in Box 7-1.

A BRIEF OVERVIEW OF LABOR SUPPLY EFFECTS

A large body of economic research literature examines how individuals respond to changes in incentives to work—which economists call labor supply. Work incentives do not only result from rising or falling labor market wage rates (or compensation more broadly); they also occur when transfer programs provide income to individuals.

Work incentives from transfer programs can increase or decrease depending on program design. Programs that provide benefits only when individuals are working raise the return to work—the additional income gained from working compared with not working. In contrast, programs that provide benefits when individuals are not working reduce the return to work. An example of how a cash transfer program would be expected to affect work incentives is given in Box 7-2.

___________________

³ Throughout the chapter, the discussion is focused primarily on labor supply responses among families with income below $80,000 because higher-income households are less likely to experience changes in child poverty as a result of the policy changes evaluated.

Work incentives also vary across the benefit schedules of transfer programs. For instance, historically, both the EITC and CTC benefit schedules include phase-in and phase-out regions, where benefits increase or decrease with changes in earnings. Individuals face different work incentives depending on where their earnings place them along the benefit schedule, as noted in the Summary of this report.

Two types of labor supply effects are distinguished in the literature, namely income effects and substitution effects. Income effects refer to changes in labor supply driven by changes in income due to changes in labor earnings. Substitution effects refer to changes in labor supply driven by changes in what is called the return to work. Income effects are generally expected to be negative, indicating that individuals work less as income transfers increase. On the other hand, substitution effects are generally expected to be positive, as individuals tend to work more as the economic gains from work increase.

In many contexts, such as historical reforms to the EITC and CTC, policy changes can generate both income and substitution effects, which can move in opposite directions, thereby creating ambiguity as to whether employment will increase or decrease as a result. In addition, workers can respond to changes in work incentives by deciding whether to work at all and/or how many hours to devote to working if they do work—which economists refer to as the “extensive” and “intensive” margins of labor supply, respectively.

Historically, labor supply effects have been estimated using changes in tax rates that generate shocks to income, the return to work, or both. Descriptive studies have also measured labor supply effects simply by measuring the changes in work coinciding with changes in the wage rate. A broader literature estimates labor supply effects outside the U.S. context.

This report’s review focuses on literature that studies the labor supply effects of the EITC and CTC. The broader literature is probably less relevant, as it includes groups whose labor supply decisions are less likely to be significantly affected by the EITC or CTC, such as those without children or those with higher incomes. For a broader discussion of labor supply effects, see Chetty et al. (2013) and McClelland and Mok (2012).

For this report, labor supply effects were analyzed in response to changes in the EITC and CTC, holding the rest of the tax and transfer system constant. However, a broader literature examines labor supply effects of other means-tested transfer and social safety net programs, such as the Supplemental Nutrition Assistance Program (SNAP), Temporary Assistance for Needy Families, Medicaid, and housing benefits. While the effects of any single program may be small, the combined impact can create high marginal tax rates (i.e., the fraction of an additional dollar of earnings that goes to taxes or reduced public benefits). A Congressional Budget Office study estimated an average marginal tax rate for low- and moderate-income taxpayers of 31% across all income tax programs, with the highest rates affecting those near the poverty line (Congressional Budget Office, 2015).⁴

LABOR SUPPLY EFFECTS OF THE EITC AND CTC

Because the EITC and CTC are currently available only to those who have positive earnings, the expected effect of these policies is unambiguously to incentivize working over not working. However, the expected effect of the EITC and CTC (or expansions in their generosity) on the amount of hours one works among those who are working is less clear because of potentially offsetting income and substitution effects, as discussed above. Furthermore, the expected effect on hours is likely to differ at different levels of earnings because, historically, both the EITC and the CTC include phase-in and phase-out regions to the benefit schedule, where benefits increase or decrease with changes in earnings. On the other hand, the provisions of the CTC under the TCJA, i.e., of the TCJA CTC Only,⁵ provided a benefit even for those with no earnings, thereby reducing the return to work relative to the pre-ARPA CTC.

The EITC has been in place since 1975, with several expansions of this federal credit over time—especially in the late 1980s and early 1990s—as well as the adoption of state EITCs by over 30 states (see Chapter 2 for

___________________

⁴ Higher or lower marginal tax rates have ambiguous effects on the intensive margin of work because they change the composition of workers. For instance, lower marginal tax rates might incentivize new workers to enter the labor market (extensive margin), but if these new employees work few hours, average hours of work (intensive margin) might decline.

⁵ See Box 1-2 for a description on this policy.

a detailed history of the EITC). These changes have been used to estimate labor supply effects. Many studies compare changes in the outcome of interest before and after the policy changed for individuals heavily affected by the EITC—like low-income unmarried mothers—to changes for groups that were not affected or less affected (which is known as a “difference-in-differences” analysis).⁶ The impact of the EITC is then determined by comparing the employment rates of unmarried mothers to single women without children, or families with two or more children to those with none or only one child (e.g., Eissa & Liebman, 1996; Hoynes & Patel, 2018), before and after policy changes. Other studies used a simulated benefit approach, using federal and/or state changes to the EITC over time to estimate changes in average EITC benefits. This simulated benefit is then used to study how changes in the generosity of the EITC affect labor supply (e.g., Michelmore & Pilkauskas, 2021; Schanzenbach & Strain, 2021; see Nichols & Rothstein, 2015, for a review).

The federal CTC has been in place since the late 1990s, and federal expansions to the credit have occurred over time, including under ARPA in 2021, which significantly expanded the credit (see Chapter 2 for additional background on the CTC). Compared to the EITC, far less empirical evidence exists illustrating how the CTC affects parental labor supply. However, in recent years, a few working papers have attempted to evaluate the effect of expansions to the CTC on labor supply, and several have estimated the effects of the expansion of the CTC under ARPA in 2021 on parental labor supply. Literature on pre-ARPA reforms typically employs a simulated benefit approach, using federal changes to the CTC over time to estimate changes in average CTC benefits. This simulated benefit is then used to study how changes in the generosity of the CTC affect labor supply.

Much of the EITC and CTC literature focuses on the labor supply effects of whether to work at all, because evidence suggests that effects on how much people work (if they do so) are very small for the groups most likely to be affected by these credits (Bastian, 2024; Eissa & Hoynes, 2004, 2006; Eissa & Liebman, 1996; Meyer, 2002; Meyer & Rosenbaum, 2001; Saez, 2002). Studies on the EITC and CTC used various approaches to present “extensive margin” labor supply effects. Many studies do not report an elasticity—a measure of the percentage change in employment relative to a change in the EITC or CTC. Among those that do, methods for calculating

___________________

⁶ Previous work varies in how unmarried mothers are defined. Most of the literature restricts the sample to unmarried mothers with less education, defined as either those without a college degree or, in some older work, those without a high school diploma. Recent summaries of the labor supply response of the EITC have focused on unmarried mothers with incomes below an income threshold that coincides with the upper income threshold for the EITC (Bastian, 2024). Previous work also evaluated how the EITC has impacted married mothers and fathers, but such work is much more limited (Eissa & Hoynes, 2004).

elasticities vary. For example, some studies examining the labor supply effects of increasing the generosity of the EITC express employment elasticity as a percentage change in employment relative to the percentage change in the maximum EITC benefit, while others use percentage change in average EITC benefits.

To enable consistent comparisons, the analyses included in this report calculate elasticities using a standard approach across all studies reviewed. For the substitution effect, elasticities are defined as the percent change in employment divided by the percent change in the return to work, with the return to work calculated based on after-tax and transfer income.⁷ This approach ensures that labor supply elasticities are comparable across studies. (Further details on the calculations for each study are provided in Appendix F.)

The literature is summarized in two parts: evidence on the pre-ARPA reforms to the EITC and the CTC, and the evidence on the expansion of the CTC under ARPA. The discussion is organized this way for several reasons. Before ARPA, the EITC and CTC had similar benefit schedules—particularly in the phase-in region—and both credits included a work requirement. As a result, benefit increases generally increased the return to work. Also, the labor supply responses estimated from pre-ARPA reforms are likely to better inform potential labor supply responses to permanent changes to the EITC and CTC than is the evidence from the changes to the CTC under ARPA, which were temporary. In contrast to prior reforms, the latter changes to the CTC effectively transformed the credit into a universal child benefit, eliminating the minimum income requirement and removing the phase-in portion of the benefit schedule. This reform introduced a potential new work disincentive for most recipients through both the income effect—by increasing the generosity of the benefit—and the substitution effect—by decreasing the return to work. As noted above, the changes to the CTC under ARPA were temporary, and payments were distributed on a monthly basis for the first time in the credit’s history. While many families may not have known that the policy change was temporary, it often takes several years before individuals fully respond to changes in the tax code. All these factors might lead to employment responses that differ from those due to prior EITC and CTC reforms. Therefore, the results of the literature on the CTC under ARPA are discussed in a separate section and the estimates from this literature are used to determine how to account for short-term labor supply responses to the CTC under ARPA.

___________________

⁷ For the ARPA CTC Expansion literature, when this method was not employed in the paper in question, the percent change in return to work estimated by Corinth et al. (2023) for the 2021 ARPA reform was applied.

There are some drawbacks to combining the discussion of labor supply responses from the EITC and pre-ARPA CTC literatures. Much EITC literature focuses on reforms that took place in the 1990s; in contrast, the CTC has only been in place since the late 1990s, and much of the literature exploits variation in benefit generosity that occurred over the course of the 2000s and 2010s. Labor supply elasticities estimated from CTC literature may therefore be smaller than those found in EITC literature if these elasticities have declined over time. Debate exists in the literature regarding the extent to which this is true, especially for women (e.g., Bastian, 2024; Heim, 2007; Winship, 2022). Another concern with discussing EITC and pre-ARPA CTC literatures together using the same estimates for the effects of both is that the various expansions of the credits likely affected different populations. For instance, prior to the expansion of the CTC under ARPA, the CTC had a minimum earnings threshold required for tax filers to claim any credit. Prior to 2009, this threshold was set at $10,000, which meant that households with less than $10,000 of earnings were completely ineligible for the credit. In contrast, the EITC begins phasing in at the first dollar of earnings (see Chapter 2 for details). This means that the labor supply elasticities generated from the various expansions of the EITC and CTC may also differ due to differences in the populations impacted by the respective reforms, with those at the bottom of the possible earnings distribution likely responding the most to EITC changes. Finally, current literature on the labor supply effects of the pre-ARPA CTC is quite limited, with no peer-reviewed publications on the topic as of this writing (the committee relied on a few working papers on the topic); so there is considerable uncertainty about the estimates.⁸ For these reasons, the non-ARPA EITC and CTC literatures are discussed together, and a single set of labor supply elasticities is used when modeling the employment effects of permanent reforms to the EITC or CTC.

REVIEW OF EITC AND PRE-ARPA CTC LABOR SUPPLY LITERATURE

Within the literature on the effects of tax credits that primarily target households with children, estimated labor supply elasticities exhibit considerable variation. However, there is some consensus on a few points. First, most studies find or assume that the effect on the intensive margin (i.e., the effect on hours worked for those who work) is very small for unmarried mothers or low-income households (Bastian, 2024; Eissa & Hoynes, 2004,

___________________

⁸ While only a few studies directly examine the effect of the pre-ARPA provisions of the CTC on employment, recent studies of unconditional income transfers have found evidence of employment effects for low-income individuals (Balakrishnan et al., 2024; Vivalt et al., 2025).

2006; Eissa & Liebman, 1996; Meyer, 2002; Meyer & Rosenbaum, 2001; Nichols & Rothstein, 2015; Saez, 2002).

Second, previous studies have generally treated the income effect on employment as small but not zero. Empirical studies of the EITC have not separately estimated an income effect, but studies relying on other variations in income have found that the income effect is small (McClelland & Mok, 2012). Bishop et al. (2009), for example, used variation in after-tax wages to estimate an income elasticity of −0.02, but this estimate is for all single women ages 25 to 55, not just unmarried mothers, who are the primary beneficiaries of the EITC. Perhaps the most relevant estimate is from Jacob and Ludwig (2012), who studied housing voucher applicants, the majority of whom were unmarried mothers. They estimated an income effect of −0.09. More recent evidence from unconditional cash transfer programs that target low-income populations have estimated even larger participation and hours elasticities due to an income effect, although these results were not for a sample of just unmarried mothers (Balakrishnan et al., 2024; Golosov et al., 2024; Vivalt et al., 2025).

Third, there is general consensus in the EITC/CTC literature regarding a positive extensive margin substitution effect (i.e., employment increases with an increase in the return to work). Debate primarily focuses on the magnitude of this substitution effect. Some of the confusion in this debate, particularly for the EITC literature, is because different studies produce estimates for different samples or generate estimates that are not comparable across studies.

Fourth, while there is a lack of consensus on the magnitude of the substitution elasticity within subgroups, there is some agreement that the magnitude is largest for low-income or low-educated (i.e., EITC-eligible) unmarried mothers, and the magnitude is smallest for men and for women without children.

Fifth, while there are far fewer studies on the labor supply effects of the CTC than the EITC, there is a bit more consistency in the estimated effects across CTC studies.

Finally, none of these studies explored whether labor supply responses to incentives were muted during economic downturns, as there were fewer jobs available for workers to take. For instance, during the COVID-19 pandemic (hereafter, “pandemic”), there is some evidence that expanded Unemployment Insurance had little or no disincentive effects in 2020, when unemployment was very high, while there is more evidence of negative effects on employment in 2021 (Holzer et al., 2024).

Employment Effects of the EITC

Many studies spanning several decades have examined the effect of the EITC on employment. This research has predominantly focused on unmarried mothers, as the majority of EITC dollars go to this group, although a smaller set of studies have examined married couples and single individuals. Several studies also provide summaries of this literature (Chetty et al., 2013; Hotz & Scholz, 2003; Nichols & Rothstein, 2015). Assessment of employment responses to the EITC for unmarried mothers focuses on studies that meet the following criteria:

• Studies use quasi-experimental variation in the EITC (i.e., across demographic groups and over time) to identify the causal effect of changes in these tax credits on employment (or labor force participation) and use at least some variation from 1986 or afterward.⁹
• Studies report sufficient information to derive an employment elasticity.
• Studies only include those that provide estimates for unmarried mothers or samples of predominantly unmarried mothers.

Table F-1 in Appendix F presents employment elasticity estimates for the studies included in the employment response assessment. In the EITC literature, the sample used for most of these estimates is unmarried mothers or some subset of unmarried mothers, although one of the estimates is for a slightly different sample, namely welfare recipients. When a study provides estimates for all unmarried mothers and low-educated unmarried mothers, both estimates are included. As noted above, types of reported elasticities vary across studies. To allow for cross-study comparisons, elasticity estimates are, when possible, expressed in terms of how changes in the EITC affect the return to work. Additional details on the calculations used for each study are provided in Appendix F.

Three key takeaways are evident from employment elasticity estimates for samples of unmarried mothers, shown in Table F-1 of Appendix F. First, a strong consensus exists within this literature that the EITC has led to greater employment for unmarried mothers.¹⁰ All nine studies estimated a positive response, with an elasticity of at least 0.42. Second, the estimated employment response appeared to be larger for particularly disadvantaged unmarried mothers, such as those with low education or who are welfare recipients. Third, the range of these estimates is quite wide, extending from

___________________

⁹ This variation includes tax changes from the 1986 Tax Reform Act and other reforms.

¹⁰ An exception is Kleven et al. (2024), which is not included in the set of studies examined. See Appendix F for more details.

0.42 to 2.05. This wide range is consistent with three recent reviews that summarize the EITC’s employment effects for unmarried mothers in the range of 0.3 to 1: 0.7 to 1.0 (Nichols & Rothstein, 2015); 0.75 (Corinth et al., 2023); and 0.3 to 0.45 (Goldin et al., 2022a). Translating these elasticities into employment effects, studies have generally found that a $1,000 increase in average EITC benefits among unmarried mothers increases employment by 4 to 8 percentage points.

Only limited evidence exists on the EITC effects on employment for other groups. A few studies have found the employment effect for men to be small. Gelber and Mitchell (2012) indicated little labor force participation response among single men, but they did not break down their samples by presence of children. Results from Bastian and Lochner (2022) suggested little impact of the EITC on employment for married fathers (also see Eissa & Hoynes, 2004, discussed below). A randomized controlled trial of Paycheck Plus, an expanded childless EITC in New York City and Atlanta, found no employment effects associated with increasing the EITC amount in Atlanta and small effects for women and more disadvantaged men in New York City (Miller et al., 2018; Yang et al., 2022). The broader literature examining how male labor supply responds to changes in the return to work (beyond just studies considering EITC effects) generally found small elasticities (Heim, 2007; Juhn et al., 2002; McClelland & Mok, 2012).¹¹

Eissa and Hoynes (2004) examined the effect of expansions in the EITC during the 1980s and 1990s on employment for married couples. They noted that although expansions in the EITC increase the return to employment for many low-wage workers, it could raise both the tax rate on the first dollar of earnings and average tax rates for secondary earners within a family. They found that expansions in the EITC reduced overall family labor supply, but increased labor supply for fathers and decreased it for mothers. This negative response for married mothers, however, is small compared to the positive response for unmarried mothers (Nichols & Rothstein, 2015).

CTC Prior to ARPA

There is much less empirical evidence on the impact of the CTC prior to ARPA. Three working papers on the topic used similar criteria: Kang (2022), Lippold (2019), and Zheng (2023). Both Kang (2022) and Zheng (2023) employed a simulated benefit strategy, using policy changes to

___________________

¹¹ While the focus of this report is on families with children, there is also evidence on the labor supply effects for childless adults. Evidence from this literature suggests the employment elasticities are small. For example, Meer and Witter (2023) examined the effect of the EITC for young, low-educated, childless adults and found little effect on employment.

evaluate how changes in the generosity of the CTC over time affected labor supply. Lippold (2019) employed a regression discontinuity strategy, evaluating the labor supply effects of having a child age out of CTC eligibility at age 17 (controlling for other child-age-related changes). Because no other credits or policies typically eliminate eligibility at age 17, this regression discontinuity approach likely isolates the impact of the loss of CTC benefits on labor supply. However, it is important to exercise caution in extrapolating the findings from this approach, since estimates were generated from parents with teenage children—it is unclear whether these estimates apply more broadly to individuals with children of all ages. Despite this caveat, these working papers report a relatively narrow range of estimates on labor supply responses to changes in the return to work, with elasticities ranging between 0.07 and 0.50 among unmarried mothers and the higher elasticities in this range found among low-income unmarried mothers.

Translating these elasticities into labor supply effects, Kang (2022) and Zheng (2023) found similar point estimates using a simulated benefit approach: a $1,000 increase in average CTC benefits was associated with a 1 to 2 percentage point increase in employment among unmarried mothers. Lippold (2019) found a much larger response, but his estimation strategy was quite different from the other two papers: he found a reduction in employment of 8.4 percentage points among low-income parents whose youngest child aged out of the CTC eligibility window, which he estimated to correspond roughly to a $1,000 decline in benefits. However, Lippold (2019) cautioned that these labor supply responses are likely temporary, driven by the sudden change in credit eligibility generated when a child ages out.¹²

SUMMARY OF ARPA LABOR SUPPLY LITERATURE

Following the 2021 ARPA expansion of the CTC, several empirical papers examined how the expansion impacted labor supply among families with children (Ananat et al., 2022; Enriquez et al., 2023; Han et al., 2022; Lippold & Luczywek, 2023; Pac & Berger, 2024; Pilkauskas et al., 2023;

___________________

¹² See Feldman et al. (2016) for more evidence on this type of temporary response.

Schanzenbach & Strain, 2024). For the most part, these studies took a difference-in-differences approach, comparing the labor supply of families with children and those without children in the months when the expansion was in place to the labor supply in months when the expansion was not in place.¹³ Much of this literature relies on the Current Population Survey monthly employment records, though some evidence comes from a national sample of families receiving SNAP benefits (Pilkauskas et al., 2023) and from administrative tax records (Lippold & Luczywek, 2023).

In some cases, authors leveraged changes in the benefit amounts received among various groups, using a continuous treatment measure representing the dollar gain in benefits during the reform months. One study (Lippold & Luczywek, 2023) used administrative tax data and a regression discontinuity approach to evaluate the impact of the expanded credit on labor supply. The paper compared the annual labor supply of individuals with 5-year-olds to those with 6-year-olds. Families with 5-year-olds received $50 more per month in benefits compared to families with 6-year-olds; thus, this study isolated an income effect. Like the regression discontinuity study discussed above, this work can truly inform only the labor supply responses of parents with children around the age 5 cutoff; it is unclear whether these responses apply more broadly to parents with older children. Further, the study only captured the overall effect on annual income and may have missed important differences within the year, which is particularly relevant in 2021 since the policy change was announced in the middle of the year and was only in place for one year.

Table F-1 in Appendix F shows the implied employment elasticity from five studies that reported estimates for samples of unmarried mothers or predominantly unmarried mothers, and from four studies that reported estimates for other groups. For the full population of families with children (including both two-parent and single-parent families), there is largely a consensus that the expansion of the CTC under ARPA in 2021 had little to no effect on labor supply relative to families without children. Based on point estimates from several studies, elasticities ranged from 0 to 0.34 with respect to changes in the return to work, with a median of 0.06 and a midpoint of 0.17.

This very modest response may be entirely expected given that the reform was temporary and occurred amidst several unique circumstances—a global pandemic, rising inflation, and three economic stimulus payments. Additionally, even if parents expected the reform to be made permanent,

___________________

¹³ Some studies tested robustness using the announcement date rather than the implementation date of the ARPA reform (e.g., Enriquez et al., 2023); others also incorporated labor supply patterns from the 2018 and 2019 calendar years in the comparison (e.g., Pac & Berger, 2024).

it may take several years for individuals to fully understand and respond to policy changes, particularly given the complexity of the tax code. For these reasons, it is unclear from this literature what the short- or long-term employment responses would have been if the expansions were made permanent.

Several of the studies also looked at subgroups to test whether particular groups were more responsive to the expansions. Across all the studies examined, small, negative employment effects were found among families with young children and unmarried, low-educated mothers with young children in particular. No study found significant declines in employment among any other group. For families with young children and unmarried mothers in particular, the estimated elasticities with respect to changes in the return to work ranged from 0.27 to 0.50, with a median of 0.45 and a midpoint of 0.38. Given that the ARPA CTC Expansion reduced the return to work by 7% to 10% (Corinth et al., 2023), this implies employment reductions ranging from 1.9% to 5% among unmarried mothers with young children. On the other hand, a few studies also found suggestive evidence of positive effects of the expansion of the CTC under ARPA in 2021 on labor supply; these were concentrated among married fathers with young children and unmarried mothers with children older than 5 (Pac & Berger, 2024).

LIMITATIONS OF EXISTING LITERATURE

Several caveats are important to consider when interpreting findings from the summarized EITC and CTC literature. First, studies on the EITC and pre-ARPA CTC all evaluate the impacts of making credits more generous by increasing the return to work. The analysis assumed that individuals respond symmetrically to decreases in these incentives—that is, that the long-term labor supply response to an increase in the return to work is equal in magnitude and opposite in sign to the labor supply response to a decrease in the return to work. This might not be the case, particularly if the salience of one policy change is different from another, or if work requirements are removed entirely. No empirical studies have been identified explicitly testing the hypothesis of symmetric responses to changes in the return to work.

Second, all evidence identified on the labor supply responses to the EITC is based on samples of lower-income households because the income eligibility range for the EITC is fairly narrow. To date, no work has been identified that explicitly examines how higher-income households respond to the incentives embedded in the CTC structure, so all labor supply elasticities used in this analysis are based on either nationally representative samples of households or on samples of lower-income/lower-educated households in particular. Estimates from the broader literature (i.e., beyond studies of the EITC or CTC) are used to determine how to account for labor supply responses for higher-income households (Chetty et al., 2013; McClelland & Mok, 2012).

Third, some disagreement exists in the literature regarding how labor supply elasticities have changed over time. Some have argued that more recent studies indicate that labor supply elasticities for unmarried mothers have declined (Bastian, 2024). While there is some agreement that labor supply elasticities have declined for married women (Blau & Kahn, 2005; Heim, 2007), the evidence for unmarried mothers is mixed. Recent studies examining the effects of the pre-ARPA CTC provide elasticity estimates for unmarried mothers that are on the low end of the distribution from the broader literature summarized above. However, this distribution includes high and low estimates from both older and more recent studies. For example, one of the lower estimates comes from an older paper (Eissa & Liebman, 1996), while the high end of the distribution includes evidence from some of the most recent studies (e.g., Schanzenbach & Strain, 2021). McClelland and Mok (2012) highlighted evidence that labor supply elasticities for women have fallen over time, but this evidence is not for unmarried mothers specifically. In fact, their proposed explanations for the decline—a decline in marriage, the rising average age of married women, and falling fertility—do not rely on changes in elasticities for unmarried women. Bishop et al. (2009) examined how female labor supply responded to changes in their after-tax wages (or their potential after-tax wages) for the years 1979 to 2003. For unmarried mothers, they found that the elasticity in 2003 was lower than that in the early 1990s. However, they did not find a clear downward trend in elasticities for this group. The 2003 estimate is similar to the estimate in the late 1990s and greater than the estimate in 2001. See Winship (2022) for further discussion of changes in elasticities over time.

ELASTICITIES USED TO ACCOUNT FOR AN EMPLOYMENT RESPONSE

This report’s assessment of the effects of the CTC and other policies on child poverty incorporates a behavioral labor supply response (as well as reporting results with no labor supply response). Based on the summary of existing literature, several decisions were made regarding estimating the employment effects of the EITC and CTC. The resulting labor supply elasticities from these decisions are summarized in Table 7-1.

First, labor supply estimates from both the EITC and pre-ARPA CTC literatures are used to simulate employment responses to permanent changes in either or both programs. Second, given the strong consensus from the literature that employment elasticities vary across demographic groups, different labor supply elasticities are applied to different groups. Prior research has shown that employment responses differ by marital status, gender, presence of children, and income. In particular, studies of the effects of the EITC and CTC have found the employment response for unmarried mothers to be larger than that for other groups. Informed by this literature, labor supply responses are distinguished for the following groups:

• Unmarried mothers with incomes below $80,000
• All other mothers with incomes below $80,000
• All others with children and incomes below $80,000
• Those with incomes above $80,000

Categorizing labor supply responses into these groups follows past work that applied behavioral adjustments in response to a policy that affected the return to work. Specifically, this breakdown—both the separation by family type and focusing on families with incomes below $80,000—mostly follows the approach taken by Bastian (2024) and used by the Budget Lab at Yale (2024). Other recent studies such as Corinth et al. (2023) have considered similar groups.

Third, no adjustments are made to earnings for those already working and who would continue to work (i.e., no adjustment is made on the intensive margin). As noted above, some consensus exists within the EITC literature that, for those who are working, the effect of the EITC on hours worked is very small. Recent studies applying behavioral effects have followed this approach (Corinth et al., 2023). Some recent studies have made intensive margin adjustments. For example, the National Academies of Sciences, Engineering, and Medicine (National Academies) report Roadmap to

Reducing Poverty made a small adjustment for an income effect on hours worked, applying the following elasticities for those who are working: −0.05 for fathers, −0.09 for married mothers, and −0.07 for unmarried mothers (National Academies, 2019, Table CA1).

Fourth, two sets of adjustments for behavioral labor supply effects are provided: one for the committee’s policy options (e.g., making the provisions of the CTC and EITC under ARPA permanent) and one for the ARPA CTC. The labor supply response to the ARPA CTC is likely to differ significantly from the response to other policies that change the return to work.

Fifth, for policy scenarios involving permanent changes—such as permanent expansions to the EITC or CTC—a modest income effect is applied, varying by demographic group. While much less evidence exists on the income effect, particularly within the EITC/CTC literature, studies that applied behavioral response adjustments in similar contexts typically included a modest, negative income effect. Corinth et al. (2023) used an income elasticity of −0.085 for unmarried mothers and −0.05 for all other units. The National Academies’ Roadmap to Reducing Poverty used 0 for men, −0.12 for married women, and −0.085 for unmarried women (National Academies, 2019). Bastian (2024) made no adjustment for the income effect. Informed by the literature, and in attempt to maintain similar groups when possible, the following income elasticities are used to adjust employment in response to policy scenarios that involve a permanent change that affects income, such as a permanent expansion to the EITC or CTC:

• Unmarried mothers with incomes below $80,000: −0.085
• All other mothers with incomes below $80,000: −0.05
• All others with children and incomes below $80,000: 0
• Those with incomes above $80,000: 0

Given the unique nature of the ARPA CTC, particularly the fact that it was temporary, no income effects are assumed for this policy change.

Finally, for policy scenarios that entail a permanent change, such as the policy options considered in Chapter 9, a range of adjustments (i.e., high, medium, low) are used to account for changes in the return to work, with different adjustments used for unmarried mothers than are used for all other demographic groups. One clear takeaway from literature on the labor supply effects of the EITC/pre-ARPA CTC is that, while clear evidence of an employment response exists, there is considerable uncertainty about the magnitude of the substitution effect for unmarried mothers. As discussed above, the literature includes a wide range of estimates and, to date, substitution effects for the pre-ARPA CTC are smaller than those for the EITC. For this reason, a range of elasticities is applied to adjust for the employment response for unmarried mothers—the group on which much

TABLE 7-1 Substitution and Income Elasticities Used to Account for Employment Effects

NOTES: For both sets of elasticities, the medium elasticity was derived from the median of the elasticities in the prior literature. For the ARPA CTC elasticities, the minimum and maximum elasticities in the literature were used to determine the low and high elasticities since there are, to date, only five estimates of labor supply elasticities specifically for low-educated single mothers. For the permanent elasticities, the 25th and 75th percentiles of the range of estimates in the prior literature were used. ARPA = American Rescue Plan Act of 2021, CTC = Child Tax Credit, EITC = Earned Income Tax Credit, SPM = Supplemental Poverty Measure.

SOURCE: These estimates are from the committee’s synthesis of the literature as explained in the text and in Appendix F.

of the literature focuses. Reporting results using a single adjustment would convey a greater sense of certainty about the actual employment response than is warranted.

The range of employment responses for unmarried mothers is specified using 16 estimates from the EITC or pre-ARPA CTC literature, reported in Table F-1 in Appendix F. Estimates for this group range from 0.07 to 2.05, with a median of 0.66 and an average of 0.78.¹⁴ While the labor supply elasticities at the higher end of this distribution largely come from the EITC literature and, therefore, may not reflect present day labor supply elasticities, given the lack of peer-reviewed, rigorous, causal evidence on more recent labor supply elasticities, this range of elasticities is used to capture a range of potential labor supply responses. The magnitude of employment elasticities has been less debated for other groups, or at least there is broader consensus that these elasticities are small. For married women, Chetty et al. (2013, Table 1) reported participation elasticities ranging from 0.15 to 0.17. McClelland and Mok (2012, Table 2) reported a range of 0 to 0.3. Taking the midpoint of this latter range, a participation elasticity of 0.15 for married mothers is applied.

For others with children—including unmarried and married fathers and other families with children—a smaller elasticity is applied, based on recent reviews of the labor supply literature for men. Chetty et al. (2013) estimated an employment elasticity (regarding the extensive margin) of 0.25, but this estimate is an average over many studies that focused on many demographic groups, not just men. Focusing on studies that included just men in the United States, the range of elasticities was 0.13 to 0.15. McClelland and Mok (2012, Table 3) reported a range of 0 to 0.13 for the employment elasticity for men. A midpoint of this range of 0.065 was applied.¹⁵

In summary, participation elasticities (or a range of elasticities) are proposed for each subgroup in policy scenarios that involve a permanent change to the return to work, such as a permanent expansion to the EITC or CTC:

• Unmarried mothers with incomes below $80,000: high = 0.97, medium = 0.66, low = 0.43

___________________

¹⁴ When restricting the set of studies further to include only those published since 2000, or only those that report estimates for unmarried mothers, distributions of elasticity estimates have midpoint, median, and average estimates qualitatively similar to those for the broader set of papers. For example, restricting to papers published since 2000, the median estimate is 0.7 and the average is 0.81.

¹⁵ Elasticities for married mothers and married and single fathers are similar to those used by Bastian (2024) and The Budget Lab at Yale (2024), and slightly lower than those used by Corinth et al. (2023), who applied an elasticity of 0.25 for all groups other than unmarried mothers.

• All other mothers with incomes below $80,000: 0.15
• All others with children and incomes below $80,000: 0.065
• Those with incomes above $80,000: 0

Appendix F describes the process used to derive elasticities for each paper in the literature, the criteria for selecting the range of elasticities, and the methods for incorporating these elasticities to account for an employment response when calculating poverty estimates.

CONCLUSION

The antipoverty effects of the EITC and CTC depend on the extent to which receiving the credits affects families’ work-related decisions. Published studies indicate that receiving the EITC and CTC largely affects the decision of whether to work or not work. For those already working, receiving the EITC and CTC does not substantially affect the choice of how much to work.

Several points of uncertainty exist, based on published studies. There is no consensus in the literature as to whether the effects of the EITC and CTC on labor supply are symmetric with respect to increases versus

decreases in the return to work; whether changes in high-income households can be inferred from representative samples or those focusing on low-income households; and whether elasticities have been changing over time.

Evidence from permanent changes in either the EITC or the CTC indicates that credit changes induce changes in employment, especially for disadvantaged unmarried mothers. However, the literature indicates that the exact magnitudes of these effects are uncertain. The committee concluded that the temporary changes to the EITC and CTC that took effect under ARPA in 2021 had very small—although not zero—effects on labor supply. This conclusion is consistent with the temporary nature of this reform and the fact that it occurred during the pandemic, when other policy responses and rising inflation were also at play.

The committee chose to use a range of low, medium, and high elasticities for modeling employment effects of both the 2021 reforms (discussed in Chapter 8) as well as alternative policy options (discussed in Chapter 9). These choices, along with the implementation strategy used, are intended to adequately reflect the range of existing evidence on employment effects of the EITC and CTC, which the report turns to next.