When a Raise Costs You: The Welfare Cliff and the Case for Making Work Pay
A raise can shrink a working family's help almost as fast as it grows the paycheck. The statutes and the federal research show where, and they point to a fix both sides can live with.
By Keith Gross, Legal & Policy Correspondent
Published · 9 min read
The Overview
The facts, straight
Most of us grew up believing a raise was good news. More hours, a better title, a bigger check. For a lot of working parents, the law doesn't see it so simply. Under rules Congress and the states have written, a raise can shrink a family's help almost as fast as it grows the paycheck. At a few points on the income ladder, the family actually ends up behind.
That's a serious charge against our own laws, and it deserves a careful look. So I went to the statutes and to the government's own research.
Economists have a name for this. The Congressional Budget Office defines the effective marginal tax rate as "the percentage of an additional dollar of earnings that is unavailable to an individual because it is paid in taxes or offset by reduced benefits from government programs." A rate of 30 percent means you keep 70 cents of the next dollar. A rate over 100 percent means the raise leaves you worse off. HHS's research office calls that a "cliff effect."
As a lawyer, I want to start with the text. Each program sets its own phase-out rule in law:
- SNAP. The food benefit is the cost of the thrifty food plan "reduced by an amount equal to 30 per centum of the household's income" (7 U.S.C. § 2017(a)). That income is counted after a 20 percent deduction for earnings (7 U.S.C. § 2014(e)(2)(B)). In the simplest case, then, each new dollar of wages trims about 24 cents of food aid.
- Medicaid. For adults covered through the Affordable Care Act expansion, the line is income that "does not exceed 133 percent of the poverty line" (42 U.S.C. § 1396a(a)(10)(A)(i)(VIII)). A 5-percentage-point disregard (§ 1396a(e)(14)(I)) makes it 138 percent in practice. There's no gradual phase-out here. You're in or you're out, though CBO notes a parent who crosses 138 percent can move to subsidized marketplace coverage.
- Public housing and vouchers. Tenant rent is the highest of "30 per centum of the family's monthly adjusted income," 10 percent of monthly income, or a welfare-based amount (42 U.S.C. § 1437a(a)(1)). Roughly 30 cents of each new dollar of adjusted income goes to rent.
- Child care (CCDBG). States set their own starting income limits. Once a family is in, federal law keeps a child eligible for at least 12 months as long as family income stays at or below 85 percent of the state median (42 U.S.C. § 9858c(c)(2)(N)). States must also describe a "graduated phaseout" for families above the starting limit (§ 9858c(c)(2)(N)(iv)). Past 85 percent of the state median, a child is no longer eligible (42 U.S.C. § 9858n(4)(B)).
- EITC. The credit phases out at 15.98 percent for one child, 21.06 percent for two or more, and 7.65 percent with no children (26 U.S.C. § 32(b)(1)). The Congressional Research Service has noted that in 2016 that phase-out alone raised a one-child filer's effective rate by 15.98 points, from $18,190 of earnings up to $39,296.
Add a few of those together on paper. Take a parent with two kids in the EITC phase-out range who also gets SNAP and lives in assisted housing. About 75 cents of a new dollar is spoken for by those three rules alone, before payroll and income taxes. Programs count income differently and affect one another, so the real number moves around. That's why the federal simulations matter.
What the Left Says
The progressive and Democratic case
Those who want a stronger safety net read the same research and draw a different lesson. In their view, the trouble lies in how the phase-outs are designed. The federal numbers give them real support.
Most low-income workers aren't stacking every program. CBO estimated that 62 percent of lower-income families did not receive the means-tested transfers in its example in 2013, and "of those who received transfers, most participated in only one program." Across all workers below 450 percent of the poverty line, CBO put the average effective marginal tax rate at 31 percent for 2016. Federal payroll and income taxes were the biggest pieces. SNAP's phase-out added 14 percentage points for recipients but only about 3 points to the overall average, because only 21 percent of the sample got SNAP.
HHS's Office of the Assistant Secretary for Planning and Evaluation (ASPE) found that true cliffs are real but uncommon. After a $2,000 raise, about 0.6 million households with children, or 2.5 percent, faced rates above 100 percent. About 2.6 million, or 11 percent, faced negative rates, meaning their benefits actually rose with the extra earnings.
The fixes ASPE laid out are mostly about how the rules are built. Programs "could adjust program phase-out schedules to make benefit reductions more gradual," extend recertification periods, or "allow a grace period before eligibility is reassessed." Policymakers "could coordinate across programs so that families phase out of programs sequentially and not simultaneously."
The District of Columbia is testing that approach. Its Career Mobility Action Plan, run by the D.C. Department of Human Services, serves 500 families who have experienced homelessness. It offers rent discounts to offset lost SNAP or cash aid, plus a cash fund of up to $10,000 a year to cover lost medical and child care benefits as income rises. Atlanta Fed researchers modeled the plan and found its "hold harmless" payments keep participants' median rates at or below 60 percent. That's a model, and the pilot runs through 2027, so results on real families are still to come.
Where Reality Likely Sits
Radio Purple's analysis
Both sides are looking at the same map, and the map is accurate.
The worst cases sit at the hard lines written into law. A single dollar over Medicaid's 138 percent line, or a state's child care limit, can end eligibility for the whole benefit. HHS says that kind of cliff "arises when income increases by a modest amount and families lose eligibility for programs such as child care vouchers, housing assistance, or WIC." At those narrow points, a small raise can cost more than it pays. Across the broad middle, though, the research shows rates in the 30s for typical workers and the 40s to 70s for families combining programs. A small share sits above 100 percent.
That's still too high. A family keeping 27 cents of a raise is facing a steeper rate than any statutory federal income tax bracket. The Atlanta Fed researchers noted that a 56 percent rate is 19 points above the 37 percent top federal bracket. We would never ask a surgeon or a business owner to work under those terms and expect the same effort.
The good news is that these are policy choices written in statute, and Congress writes statutes. The Preamble lists "promote the general Welfare" among the Constitution's purposes, and Article I gives Congress the spending power to do it. A disciplined fix can start with lawmakers reading their own phase-out rules side by side, with no new bureaucracy required. The Atlanta Fed's Leia example shows why: a hypothetical single mother of two in Georgia who trains to become a nurse loses more than $11,000 in her first three years as a nursing assistant. The researchers trace that to fewer work hours while in school and to benefits cliffs. Over her lifetime, she comes out ahead. The law should make that path easier to start.
Congress has already shown it knows how. The child care law's 12-month eligibility period and graduated phase-out are smoothing tools, and they're on the books. Extending that thinking, staggering phase-outs so families exit programs one at a time, and pairing work expectations with rules that let work pay are steps a serious Congress can take. Neither party has to give up its principles to do it.
If you've ever turned down a raise or extra hours because of what it would cost your family, I'd like to hear about it. Which line in the law hit you hardest?
What the Right Says
The conservative and Republican case
The right's case starts with incentives. They shape behavior, and the current structure punishes the very effort we say we want.
On the core economics, CBO's analysts agree. In a 2017 explainer, CBO wrote that increases in marginal tax rates, "on net, decrease the supply of labor by causing people already in the labor force to work less." It added that phasing out a benefit such as SNAP "increases the marginal tax rate and reduces the incentive to work."
The research also shows how steep the climb gets for the families the right worries about most. In ASPE's data, households with children just above the poverty line had a median rate of 51 percent, and it never fell below 42 percent until income approached twice the poverty line. Families on SNAP, the EITC, child tax credits, Medicaid or CHIP, and housing aid together faced a median of 63 percent. For families on child care subsidies earning 100 to 149 percent of poverty, a $5,000 raise came with a median rate of 73 percent. They netted $1,350.
The Atlanta Fed's numbers are even starker for one hypothetical D.C. family: a single parent with a three-year-old, in the city's rehousing program. That family ends up with the same $8,500 in net resources at $11,000 of earnings as at $65,000. Public assistance falls from $68,686 to $22,709 across that range, and taxes rise from $842 to $12,835. The researchers estimated a median rate of 173 percent for participants at 325 to 349 percent of the poverty line, mostly from losing the child care subsidy all at once.
Conservatives also want work and help to move together. In the law signed July 4, 2025 (Public Law 119-21), under a subchapter titled "Increasing Personal Accountability," Congress required states to ask expansion-group Medicaid adults to show community engagement. That means 80 hours a month of work, community service or a work program, half-time school, or equivalent earnings, starting no later than the first quarter after Dec. 31, 2026 (Sec. 71119). The same law modified SNAP's work requirements for able-bodied adults (Sec. 10102). The message is clear: work is the expected path, and the law should be built around it.
Sources & Citations
- Congressional Budget Office, “Effective Marginal Tax Rates for Low- and Moderate-Income Workers in 2016”, November 1, 2015.
- Congressional Budget Office (Shannon Mok), “How Taxes and Transfers Affect the Work Incentives of People With Low and Moderate Income”, March 17, 2017.
- HHS, Office of the Assistant Secretary for Planning and Evaluation (Suzanne Macartney and Nina Chien), “Marginal Tax Rates: A Quick Overview (Brief #1)”, March 1, 2019.
- HHS, Office of the Assistant Secretary for Planning and Evaluation (Nina Chien and Suzanne Macartney), “What Happens When People Increase Their Earnings? Effective Marginal Tax Rates for Low-Income Households (Brief #2)”, March 1, 2019.
- HHS, Office of the Assistant Secretary for Planning and Evaluation (Nina Chien and Suzanne Macartney), “Effective Marginal Tax Rates for Households with Child Care Subsidies (Brief #3)”, March 1, 2019.
- Federal Reserve Bank of Atlanta (Elias Ilin and Alvaro Sanchez), “Mitigating Benefits Cliffs for Low-Income Families: District of Columbia Career Mobility Action Plan as a Case Study”, September 1, 2023.
- Federal Reserve Bank of Atlanta, “What Are Benefits Cliffs?”.
- U.S. Government Publishing Office, “Public Law 119-21 (H.R. 1), Sec. 71119”, July 4, 2025.
Positions above are attributed to the public statements and published reporting cited here. Quotations are taken from these sources; Radio Purple does not publish unattributed or invented quotes. See an error? Request a correction.
