Earned Income Tax Credit (United States, 1975 to 2025)

Congress created the Earned Income Tax Credit in 1975 as a temporary work bonus for parents, made it permanent in 1978, and expanded it several times; it now reaches about 23 million filers and lifts millions above poverty, but about one in four to one in three dollars paid has been judged improper, about one in five eligible filers does not claim it, and workers without children get very little.

The record

What the credit is and how it began

• The Earned Income Tax Credit (EITC) is a refundable credit for workers with low and moderate earnings. A refundable credit can pay out more than the tax owed. It is claimed on the federal income tax return, so a person must file to get it. (Congressional Research Service, "The Earned Income Tax Credit (EITC): A Brief Legislative History," R44825; American Action Forum, program design and history; both opened)
• The Tax Reduction Act of 1975 (P.L. 94-12) created it as a temporary credit for tax year 1975 only. It paid 10 percent of the first $4,000 of earnings, a maximum of $400, and phased out between $4,000 and $8,000 of income. It targeted low-income working families with children. Senator Russell Long proposed it as a "work bonus" meant to encourage work and reduce welfare rolls. (CRS R44825; opened)
• The Revenue Act of 1978 (P.L. 95-600) made it permanent and raised the maximum to $500. (CRS R44825)
• The Tax Reform Act of 1986 (P.L. 99-514) raised the maximum to $800 and began annual inflation indexing. (CRS R44825)
• The 1990 budget law (P.L. 101-508) began varying the credit by family size from 1991. The 1993 budget law (P.L. 103-66) expanded that variation sharply and created the first credit for workers without children, with a maximum of $323 in 1996, limited to adults aged 25 to 64 who are not claimed as dependents. CRS says the childless credit was meant partly to offset a gasoline tax increase in the same law, and that some lawmakers opposed it because they saw the EITC as a program for families with children. (CRS R44825)
• Later changes: marriage penalty relief in 2001 (P.L. 107-16); more relief and a 45 percent rate for families with three or more children in 2009 (P.L. 111-5); both made permanent by the PATH Act of December 2015 (Division Q of P.L. 114-113). (CRS R44825)
• I did not open the text of the statutes themselves. The enactment details above come from the CRS history. See Section 5.

Timeline

Size and current rules

• For tax year 2025 the maximum credit is $4,328 with one child, $7,152 with two, $8,046 with three or more, and $649 with no children. The no-child credit phases out above $19,104 of income for a single filer. (Center on Budget and Policy Priorities, "Policy Basics: The Earned Income Tax Credit"; opened)
• CBPP reports that 23 million working families and individuals received the credit for tax year 2023, and that in tax year 2022 the average credit for a family with children was $3,338 and for a filer without children was $383. (CBPP)
• Thirty-one states, plus the District of Columbia and Puerto Rico, have their own EITCs. (CBPP)
• The American Action Forum reports $59.2 billion of EITC claimed in 2020, of which $51.7 billion (87 percent) was paid as refunds. (AAF)

Poverty effects

• CBPP reports that, using the Supplemental Poverty Measure for 2024, the EITC lifted about 4.4 million people above the poverty line, including 2.3 million children. (CBPP)
• A 2013 Economic Policy Institute brief, using 2011 data, found the EITC and Child Tax Credit together lifted almost six million people above the line, over half of them children. (EPI, "The Earned Income Tax Credit"; opened)
• These counts depend on the measure used and the year. They are not directly comparable.

Employment effects: contested

• Nada Eissa and Jeffrey Liebman studied the 1986 expansion. Single women with children raised labor force participation by 1.4 percentage points, from a base of 73.1 percent, relative to single women without children. They found no detectable effect on hours for those already working. (NBER Working Paper 5158, 1995; opened)
• Austin Nichols and Jesse Rothstein reviewed later research (2015). They report positive labor supply effects for single mothers, smaller negative effects for married mothers, essentially none for men, and that employers capture some of the benefit through lower wages. (NBER Working Paper 21211; abstract opened)
• Henrik Kleven reviewed every state and federal reform since 1975. He reports that only the 1993 expansion clearly raised employment, and that the large 1990s gains appear to reflect welfare reform and a strong economy. (Described in a Marginal Revolution post; opened. The paper itself was not opened.)

Disputed numbers

| Item | Figure A | Figure B | Notes | |---|---|---| | Improper payment rate, 2013 | 22.1% to 25.9% of payments (Tax Policy Center, projected from tax year 2009 returns) | 20% to 25% (EPI, citing TIGTA 2013) | Similar range | | Improper payment rate, FY2024 | 27.3%, $15.9 billion of $58.4 billion (TIGTA report, May 9, 2025) | 27% of amount claimed, $15.9 billion, tax year 2022 (McCubbin, Brookings, 2025) | Same source data, consistent | | Improper payment rate, FY2025 | 32.7% on $64.7 billion outlays (paymentaccuracy.gov; the site lists accuracy at 67.3%, so the improper rate is derived) | none other found | Rate swings year to year: 27.8, 31.6, 33.5, 27.3, 32.7 for FY2021 to FY2025 | | Participation rate, tax year 2021 | 78% of taxpayers, 81% of dollars (IRS and Census linked study) | 80.8% (IRS state table, which uses a different eligibility model) | Methods differ | | Participation rate, tax year 2022 | 80.8% (IRS state table) | 78% (IRS cites a Current Population Survey estimate) | Same page gives both | | Poverty effect | 4.4 million lifted above poverty, 2024 (CBPP) | almost 6 million, with Child Tax Credit, 2011 (EPI) | Different years and credits |

Errors and who they affect

• Treasury's payment accuracy site lists the improper payment rate for FY2021 through FY2025 as 27.8, 31.6, 33.5, 27.3 and 32.7 percent (derived from the listed accuracy rates). For FY2025 it classes all of the $21.1 billion as overpayments outside agency control. It names as root causes the lack of a pre-approval process, reliance on self-certification, difficulty verifying residency and relationship tests, misreported self-employment income, and first-time claimants misunderstanding the rules. (paymentaccuracy.gov; opened)
• The Treasury Inspector General for Tax Administration (TIGTA) reported in May 2025 a 27.3 percent rate for FY2024. It says the IRS cannot verify taxpayer information before issuing refunds, has asked for "correctable error" authority that it has not received, and that a meaningful reduction is unlikely without independent data to verify claims. (TIGTA report 2025-400-025; opened)
• Janet McCubbin of the Brookings Institution (June 2025) summarizes a study of tax years 2006 to 2008. Qualifying child errors were 42 to 56 percent of the overclaimed amount, income errors 24 to 32 percent (self-employment alone about 19 percent), and filing status errors 9 to 17 percent. She notes the error rate and participation rate have stayed roughly flat for decades despite enforcement. Her slides do not say how much is intentional versus unintentional. (Brookings, "Improper EITC payments"; opened)
• The Tax Policy Center says most qualifying child errors fail the residency test, that the IRS has no administrative data to check where a child lived, and that only a minority of improper payments appear to be fraud. (Tax Policy Center, error rates; opened)

Who does not get it

• An IRS and Census Bureau study of tax year 2021 found 78 percent of eligible taxpayers (and 81 percent of dollars) claimed the credit. About 5.7 million eligible units did not: 4.6 million non-filers and 1.1 million filers who did not claim. About $8.2 billion went unclaimed. Taxpayers with qualifying children participated at 86 percent; those without at 71 percent, a gap the paper ties to the temporary 2021 expansion. (IRS Statistics of Income paper on tax year 2021 participation; Census working paper CES-WP-24-75; both opened)
• The American Action Forum reports that childless workers were nearly a third of claimants but received about 4 percent of the credit value in 2020, and that the 2023 maximum was $600 for them against $3,995 to $7,430 for parents. (AAF)
• People with no earnings get nothing. Only about half of qualifying children live in married-couple households, which AAF says complicates eligibility. (AAF)

What this case shows

1. A program ordinary people could not easily read. To claim the credit, a filer must meet 6 rules that apply to everyone, 5 more with a qualifying child, or 3 more without (McCubbin slides). Many claimants use paid preparers. EPI reports that in 2010 about 60 percent of filers used a paid or volunteer preparer. The people the program is meant to help rely on someone else to interpret it.
2. Paid out first, checked later. TIGTA and Treasury describe a system with no pre-approval step and a 45-day refund requirement. The result is that taxpayers get money quickly, and the cost of errors falls on the federal budget and on later audits of claimants. Voters rarely see an error rate of this size attached to a program they hear praised.
3. Two kinds of exclusion at once. About one in five eligible filers does not claim it, and childless workers receive a small fraction of the amount that parents do. Both are visible in federal data, yet the public debate usually centers on one at a time.
4. Congress changed the rules in open votes. The credit was set by statute at least ten times between 1975 and 2015. Each step is on the record and can be read. The 1993 childless credit was partly a trade for a gas tax rise, which shows how a design choice can come from an unrelated bargain.
5. Evidence was built outside the agency. The employment effect was tested by outside economists (Eissa and Liebman in 1995, Nichols and Rothstein in 2015, Kleven later) and they do not agree on its size. A voter can ask what the credit is meant to do, and the record shows even experts disagree.

The strongest objections, and our answers

1. Objection: The error rate shows fraud and waste. Mitigation: The sources say most errors are likely unintentional and tied to complexity, and that only a minority look like fraud (Tax Policy Center). Brookings says intent is not quantified. We report the rates and do not call them fraud. The Tax Policy Center also notes the 2004 Taxpayer Advocate finding that 43 percent of taxpayers who sought reconsideration of audits were owed all or nearly all of the credit they claimed, so audits can themselves err.
2. Objection: The error rate is overstated or misleading. Mitigation: Possibly. Rates swing from 27.3 to 33.5 percent across recent years, and the measure counts overpayments and underpayments without netting and excludes some enforcement. We show the range and the source for each figure.
3. Objection: The EITC is the most effective work-support program. Mitigation: The evidence of poverty reduction is strong (CBPP, EPI). The evidence of employment gains is mixed: positive for single mothers in Eissa and Liebman and Nichols and Rothstein, weak in Kleven. We report both.
4. Objection: The childless worker gap is a policy choice, not a flaw. Mitigation: Agreed that it is a choice. CRS records that some lawmakers wanted the credit limited to families with children. We describe the gap and do not say whether it should be closed.
5. Objection: Think-tank sources lean one way. Mitigation: CBPP and EPI are advocacy-oriented research groups; the American Action Forum is a center-right policy group. We cross-checked their figures against Treasury, TIGTA, the IRS and Census, and CRS.

Open questions: what would change our minds

• The text of P.L. 94-12 and later statutes was not opened. Enactment details rest on the Congressional Research Service history. Unverified against the statutes.
• The CRS report R44057 could not be fetched because the site blocked the request. The IRS EITC statistics page also could not be opened, so the count of returns and dollars comes from CBPP and AAF, not the IRS.
• The two participation figures for tax year 2021 (78 percent, 80.8 percent) are not reconciled. The IRS state table's description came through a summarizing fetch tool and should be rechecked on the page.
• The FY2025 improper rate of 32.7 percent is derived from the listed accuracy rate, not stated on the page. Check the Agency Financial Report.
• Kleven's paper was not read, only a blog summary of it. The claim about only the 1993 expansion raising employment needs the original.
• Nichols and Rothstein and Eissa and Liebman were read at abstract level only.
• The CBPP and AAF numbers for poverty effects are from advocacy-oriented sources. A Census or Treasury poverty table would be a better primary source.
• What would change our minds: new IRS random audit data showing a very different error rate, or a pre-approval system that cuts errors without cutting participation.

Sources

Found an error or a better source? We keep every correction on the record with its date and reason. Nothing here is quietly rewritten.