Every State Dollar

State governments took in $3.14 trillion of general revenue in fiscal 2024, and $1,072 billion of it, 34%, came from the federal government. That is the largest single source after the states’ own taxes, and in 13 states it is more than two dollars in five. The share runs from 50% in Louisiana to 24% in North Dakota. Maryland, at 34%, is close to the nation; Virginia, at 27%, ranks 47th of fifty: it raises $6,762 of its own revenue per resident against $6,012 nationally, and its Medicaid program costs $2,535 per resident against $2,776.

This study follows that money. It uses the Census Bureau’s survey of state finances for what states receive, the states’ own budget reports for how they spend it, MACPAC for Medicaid and the Department of Agriculture for SNAP. It then sets the figures against what P.L. 119-21, the law of July 4, 2025, moves onto state books, and against what states hold in reserve. The District is a state where the source treats it as one and absent where it does not.

Where the money comes from

Federal money is a larger share of state revenue than it was. In 2012 it was 31% nationally; it peaked at 37% in 2021, and at 34% in 2024 it had settled above where it started. Per resident, the federal government sent state governments $3,159 in 2024: $3,263 to Maryland and $2,517 to Virginia, against $6,963 to Alaska and $5,019 to New York.

The states’ own budget offices count it the same way. NASBO’s expenditure report puts federal funds at 34% of all state spending in fiscal 2025, $1.08 trillion of $3.21 trillion, against general funds of 38%. The two sources differ state by state because they count different things in different years; Louisiana and Alaska lead in both and Hawaii is near the bottom in both, but Arizona is 45% by Census and 24% by NASBO.

What it buys

Most of it is Medicaid. Census’s public welfare category, which Medicaid dominates, was 32% of state general spending in 2024, up from 28% in 2014; the program cost $944 billion in the states and the District in fiscal 2024, of which Washington paid $608 billion, or 64%. The federal share follows each state’s matching rate, from the 50% floor in the highest-income states to 77% in Mississippi in fiscal 2024, with the 90% match for the expansion adults layered on top. Virginia, which expanded in 2019, had a 64% federal share on $22.4 billion; Maryland, which expanded in 2014, 60% on $18.5 billion; the District 73%, because its statutory match is 70%.

The 2025 law

P.L. 119-21 shifts two costs to states that the federal government has carried. The first is SNAP benefits, which Washington has paid in full since the program began. From fiscal 2028 a state pays a share set by its payment error rate: nothing below 6%, 5% from 6% to under 8%, 10% from 8% to under 10% and 15% at 10% or more, with the rate measured three years back, except that for 2028 a state may use its 2025 or 2026 rate. The fiscal 2025 rates published in June 2026 put 19 states and the District in the top band and 9 states in the zero band; the national rate was 10.62%. The second is administration: the federal share of SNAP administrative costs falls from 50% to 25% in fiscal 2027. The Congressional Budget Office put the two at $40.8 billion and $24.7 billion over fiscal 2025 to 2034 in its July 2025 estimate of the enacted law, as tabulated by CRS; the match figure is a third of the House bill’s $128.3 billion because the enacted bands are lower, top out at 15% rather than 25% and exempt states under 6%.

Exhibit 9 is arithmetic on the inputs, not a forecast: at fiscal 2025 benefit levels and fiscal 2025 error rates, the match would move $11.1 billion a year onto state budgets, $1.9 billion of it in California, $266 million in Virginia, $221 million in Maryland and $49 million in the District, which is among the 7 whose start is delayed by the high-error-rate provision. The rolls are already smaller: 36.4 million people received SNAP in June 2026 against 41.7 million a year earlier, a fall of 13% that spans the law’s work requirements and eligibility changes and the October–November 2025 lapse in appropriations, which disrupted November benefits, so the benefit base to which the match will apply is not the one in the table.

Medicaid provisions in the same law require states to set community engagement requirements for certain expansion adults (§71119) and to redetermine expansion adults’ eligibility every six months (§71107), and change the rules for provider taxes (§71115) and state-directed payments (§71116). No federal source allocates their effects by state, and this study does not estimate them.

The cushion

States hold large reserves against this. Rainy day funds totaled $191 billion at the end of fiscal 2025, 14.9% of general fund spending, with a median state at 13.5%; the total is projected at $183 billion in 2026 and $160 billion in 2027, most of the fall in California (NASBO; 26 states expect increases in 2026). Virginia held 13.0%, Maryland 9.2%, New Jersey nothing in its stabilization fund and Illinois 4.3%. Exhibit 12 puts the two measures together: 14 states draw more than the national share of their revenue from Washington and hold less than the median reserve.

What would change it

The arithmetic in Exhibit 9 moves with two numbers the states control in part: the error rate, which falls into a lower band at 10%, 8% and 6%, and the caseload, which fell 13% from June 2025 to June 2026. The fiscal 2026 error rates, due in June 2027, will decide which year each state elects for 2028 and which states qualify for the delay.

The federal share of state revenue in Exhibits 1 and 2 moves mostly with Medicaid, and the next readings of it are Census’s fiscal 2025 survey and the states’ fiscal 2026 budgets; the states’ fiscal 2027 budgets are the first to carry the 2025 law’s administrative cost shift (from federal fiscal 2027, which began October 1, 2026). Those are the two series to watch; this study will be rerun when they are published.