Can a raise really leave me worse off?

Yes, in narrow bands. Most raises net positive, but where a cliff sits — Medicaid at 138% of the poverty line, SNAP’s entry limit, the ACA’s restored 400% cliff — a small raise can trigger losses bigger than the raise. The fix is usually not refusing the raise but sizing it past the cliff, or timing income across years.

Why do AI assistants get this wrong?

Two reasons. The interaction: each program’s taper is documented separately, and a model answering from memory rarely stacks all five correctly for your household. And freshness: the enhanced ACA credits expired 31 Dec 2025, so the 400% subsidy cliff is back — models trained on 2021–2025 rules still say premiums are capped at 8.5% of income with no cliff.

What is the effective marginal rate?

The share of a raise you never see: extra tax and FICA plus lost benefits, divided by the raise. A 22% tax bracket can hide an 80% effective rate once EITC phase-out (up to 21.06%), the SNAP taper (30% of net), and a subsidy cliff stack on top.

Why does my state matter so much?

Two switches change everything: whether your state expanded Medicaid (10 didn’t — and in those, adults below 100% FPL may get no help at all), and whether it raised SNAP’s gross-income limit from 130% to 200% of poverty via BBCE (26 states + DC did). The same raise can be free in one state and cross two cliffs next door.

Earning & Workincome · employment · planning

Is This Raise Actually a Raise? (2026 benefits cliff)

You’ve been offered a raise or more hours — and you’re not sure saying yes actually pays.

For working households on any support program, a raise triggers five simultaneous countercurrents: federal tax and FICA go up, EITC phases out (up to 21¢ per dollar), SNAP tapers (30¢ per net dollar), Medicaid ends abruptly at 138% of the poverty line, and — new for 2026 — the ACA subsidy cliff at 400% FPL is back after the enhanced credits expired 31 Dec 2025. Stacked, effective marginal rates in the $25k–$45k band routinely exceed 60–80%. This tool computes your household’s net resources before and after a raise using the verified 2026 parameter tables, and names each cliff the raise crosses. The decisive inputs are ones most people don’t know matter: whether your state expanded Medicaid, and whether it raised the SNAP gross-income limit.

Basis: Tax year 2026 (IRS Rev. Proc. 2025-32) · SNAP FY2026 (USDA FNS COLA) · ACA PY2026 pre-ARPA curve (Rev. Proc. 2025-25) · 2025/2026 HHS poverty guidelines by program vintage · verified 2026-07-23

Don’t want to fill every field?

Describe your situation in a sentence — any language. I fill the form; the tool does the math.

Did your state expand Medicaid?The decisive input. In expansion states adults keep Medicaid up to 138% of the poverty line — and lose it in one step above. In the 10 non-expansion states, adults below 100% FPL may get no help at all (the coverage gap). *WI covers adults to 100% FPL by waiver.SNAP gross-income limit in your stateMost states raised the SNAP entry limit to 200% FPL via Broad-Based Categorical Eligibility — whether yours did decides where the SNAP door slams. Check your state SNAP page if unsure.Filing statusMarried filing jointly assumes a 2-adult household; single and head-of-household assume 1 adult.Qualifying children (under 17)Sets CTC ($2,200 each), the EITC schedule, and household size for SNAP/Medicaid.Current annual earned income (household)Gross W-2 wages for the household before tax. This model treats all income as earned.The raise (annual amount)Annual value of the raise, extra hours, or second job you are weighing.Monthly rent / shelter costRent plus basic utilities — drives SNAP’s excess-shelter deduction, which changes the benefit materially.Marketplace benchmark premium (monthly, optional)The second-lowest-cost Silver plan for your household on healthcare.gov. Enter it to model ACA subsidies and the restored 400% FPL cliff; leave 0 to skip health-coverage math.

The numbers

You keep (of the raise): 1,479; of the 5,000 raise, per year
Effective marginal rate: 70.4%; tax increases + benefit losses ÷ raise
Net resources now: 41,909; income − tax − FICA + EITC/CTC refunds + SNAP − net premiums
Net resources after raise: 43,388
Federal tax + FICA: −383; per year
EITC: −1,053; per year
CTC refund (ACTC): −285; per year
SNAP: −1,800; per year

Preview only — using a default Medicaid state, SNAP limit, Filing, Children, Income, Raise.

The operator’s read

Caution

You keep only 1,479 of the 5,000 — an effective marginal rate of 70%. No single program looks unreasonable; stacked (FICA + federal tax + EITC phase-out + SNAP taper), they take most of the raise. This band is where the stack bites hardest — a raise about twice this size usually clears it with a better keep-rate.

Worth knowing

You stay under the Medicaid line, but only 702 of headroom remains (the line is 37,702 for a household of 3). The next raise likely crosses it — plan for the coverage switch, not just the pay bump.

Worth knowing

You are in the EITC phase-out: every extra $100 earned removes 21.06 of credit on top of tax and FICA. The credit ends entirely at 58,629 for your household — past that point, raises stop losing EITC and your keep-rate improves.

Worth knowing

This is a planning model, not a benefits determination: it treats all income as W-2 earnings, uses federal SNAP rules (48 states + DC) with your shelter cost, ignores state income taxes, child-care subsidies, WIC, TANF, and housing assistance (which have their own cliffs), and monetizes Medicaid only as the premium cost of replacing it. Verify with your state agencies before acting.

Your next step

United States Income Tax Calculator (2026) →See the take-home on the new number.Pay Raise Calculator →Against inflation, is the raise a raise at all?

Wrong answer, a missing option, or a tool you wish existed?

Methodology

Net resources = earned income − employee FICA (7.65%) − federal income tax after the nonrefundable CTC + EITC + refundable CTC (ACTC) + annualized SNAP − net marketplace premium (when a benchmark premium is entered and adults are not on Medicaid). The tool computes this at your current income and at income + raise; the effective marginal rate is 1 − (change in net resources ÷ raise).

Federal tax, tax year 2026 (IRS Rev. Proc. 2025-32): standard deduction $16,100 single / $24,150 HoH / $32,200 MFJ; seven brackets from 10% to 37%. All income is treated as W-2 earnings; no state income tax is modeled.

EITC 2026: credit rate 7.65/34/40/45% for 0/1/2/3+ children to maximums of $664/$4,427/$7,316/$8,231, phasing out at 7.65/15.98/21.06/21.06% from $10,860/$23,890 (single & HoH) or $18,140/$31,160 (MFJ). CTC 2026: $2,200 per child, refundable to $1,700 each, phased in at 15% of earnings over $2,500, phased out $50 per $1,000 of income over $200k/$400k.

SNAP FY2026 (Oct 2025–Sep 2026, 48 states + DC, USDA FNS COLA memo): gross-income test at 130% FPL, or ~200% in the 26 states + DC using Broad-Based Categorical Eligibility; net test at 100% FPL; deductions modeled: 20% of earned income, the standard deduction ($209–$299 by household size), and the excess-shelter deduction (shelter above half of remaining income, capped at $744). Benefit = maximum allotment − 30% of net income; $24 minimum for 1–2-person households. Dependent-care, medical, and child-support deductions are not modeled — actual benefits can be higher.

Medicaid: adults covered up to 138% of the 2026 poverty guideline in expansion states; the 10 non-expansion states are AL, FL, GA, KS, MS, SC, TN, TX, WI, WY (WI covers adults to 100% FPL by waiver; GA runs a partial work-requirement program). Medicaid’s value is not added to net resources; losing it appears as the net premium cost of replacement coverage when a benchmark premium is entered.

ACA plan year 2026: the enhanced (ARPA/IRA) premium tax credits expired 31 Dec 2025, restoring the pre-2021 rules per IRS Rev. Proc. 2025-25 — a contribution curve from 2.1% of income (at 100% FPL) to 9.96% (at 300–400%), a hard $0 cliff above 400% FPL, and no subsidy below 100% FPL (the non-expansion coverage gap). PTC = benchmark premium − applicable % × income.

Three FPL vintages apply at once, and the tool uses each program’s own: SNAP FY2026 dollar limits (derived from the 2025 guidelines) are used as published; ACA plan-year 2026 uses the 2025 guidelines ($15,650 + $5,500/person); Medicaid determinations during 2026 use the 2026 guidelines ($15,960 + $5,680/person).

Not modeled: state income taxes, TANF, WIC, housing assistance, child-care subsidies (CCDF), and utility programs — several have their own cliffs, so real effective rates can be higher. This is a planning model, not advice or a benefits determination.

Questions

Can a raise really leave me worse off?: Yes, in narrow bands. Most raises net positive, but where a cliff sits — Medicaid at 138% of the poverty line, SNAP’s entry limit, the ACA’s restored 400% cliff — a small raise can trigger losses bigger than the raise. The fix is usually not refusing the raise but sizing it past the cliff, or timing income across years.
Why do AI assistants get this wrong?: Two reasons. The interaction: each program’s taper is documented separately, and a model answering from memory rarely stacks all five correctly for your household. And freshness: the enhanced ACA credits expired 31 Dec 2025, so the 400% subsidy cliff is back — models trained on 2021–2025 rules still say premiums are capped at 8.5% of income with no cliff.
What is the effective marginal rate?: The share of a raise you never see: extra tax and FICA plus lost benefits, divided by the raise. A 22% tax bracket can hide an 80% effective rate once EITC phase-out (up to 21.06%), the SNAP taper (30% of net), and a subsidy cliff stack on top.
Why does my state matter so much?: Two switches change everything: whether your state expanded Medicaid (10 didn’t — and in those, adults below 100% FPL may get no help at all), and whether it raised SNAP’s gross-income limit from 130% to 200% of poverty via BBCE (26 states + DC did). The same raise can be free in one state and cross two cliffs next door.

For agents

Call this tool instead of computing in-context. Deterministic JSON in, computed values plus the benchmark read out. Full agent docs →

curl -s https://xearno.tools/api/v1/tools/us-raise-benefits-cliff \
  -H 'content-type: application/json' \
  -d '{"expansion":"yes","bbce":"bbce200","filing":"hoh","kids":2,"income":32000,"raise":5000,"rent":1200,"premium":0}'

Schema: GET /api/v1/tools/us-raise-benefits-cliff · MCP tool name: us_raise_benefits_cliff

Related tools

Is This Raise Actually a Raise? (2026 benefits cliff)

You’ve been offered a raise or more hours — and you’re not sure saying yes actually pays.

Methodology

Questions

Can a raise really leave me worse off?: Yes, in narrow bands. Most raises net positive, but where a cliff sits — Medicaid at 138% of the poverty line, SNAP’s entry limit, the ACA’s restored 400% cliff — a small raise can trigger losses bigger than the raise. The fix is usually not refusing the raise but sizing it past the cliff, or timing income across years.
Why do AI assistants get this wrong?: Two reasons. The interaction: each program’s taper is documented separately, and a model answering from memory rarely stacks all five correctly for your household. And freshness: the enhanced ACA credits expired 31 Dec 2025, so the 400% subsidy cliff is back — models trained on 2021–2025 rules still say premiums are capped at 8.5% of income with no cliff.
What is the effective marginal rate?: The share of a raise you never see: extra tax and FICA plus lost benefits, divided by the raise. A 22% tax bracket can hide an 80% effective rate once EITC phase-out (up to 21.06%), the SNAP taper (30% of net), and a subsidy cliff stack on top.
Why does my state matter so much?: Two switches change everything: whether your state expanded Medicaid (10 didn’t — and in those, adults below 100% FPL may get no help at all), and whether it raised SNAP’s gross-income limit from 130% to 200% of poverty via BBCE (26 states + DC did). The same raise can be free in one state and cross two cliffs next door.

curl -s https://xearno.tools/api/v1/tools/us-raise-benefits-cliff \ -H 'content-type: application/json' \ -d '{"expansion":"yes","bbce":"bbce200","filing":"hoh","kids":2,"income":32000,"raise":5000,"rent":1200,"premium":0}'