Block Grants or Categorical Grants, and Why the Strings Matter Most

Here is a claim most Americans would never think to test: two programs can spend the same number of dollars and still be run by different governments.

Suppose Washington sends a state one billion dollars for a purpose. In the first case, the check arrives with a four-hundred-page rulebook, a required list of approved activities, a mandate to match each federal dollar with a state dollar, and a federal auditor who must sign off on every category. In the second case, the check arrives with a single sentence describing the goal. The billion is the same. The power is not. The first program is run from Washington with a state office as the branch plant. The second is run by the state with Washington as a funder.

That is why the type of grant matters more than the amount, and why anyone who wants to understand federalism has to understand the strings. Let me define the terms, put them in a table, and then say which design fits which goal.

Why grants exist at all

The federal government collects revenue nationally and spends some of it through the states. The constitutional basis is the Spending Clause, Article I, Section 8, Clause 1, which gives Congress the power "to lay and collect Taxes, Duties, Imposts and Excises, to pay the Debts and provide for the common Defence and general Welfare of the United States." Congress has long used that power to offer money to the states with conditions attached.

The Supreme Court has allowed those conditions within limits. In South Dakota v. Dole (1987), the Court upheld a law that withheld five percent of certain federal highway money from states that did not raise their drinking age to 21. The Court said conditions must be unambiguous, related to the federal interest in the spending, and not coercive. Twenty-five years later, in NFIB v. Sebelius (2012), the Court found the limit: a federal threat to withhold all of a state's existing Medicaid money unless it accepted the expansion was "a gun to the head," as the Court's opinions put it, and amounted to coercion. So the Constitution gives Congress wide room to attach strings and a narrow limit on how tight they may be.

The Tenth Amendment frames the stakes. The powers not delegated to the United States "are reserved to the States respectively, or to the people." Grants are the mechanism by which Washington gets influence over what the Tenth Amendment reserves to the states. The more tightly the grant is drawn, the less is reserved.

The five types

Here is the table. The design column describes how much freedom the recipient has, and the last column gives the leading example.

A few words on each, because the table hides the interesting parts.

Categorical grants are the oldest and the most tightly controlled. A state cannot move the money from one category to another, and it must follow the federal program's rules. The administrative cost is high, because every dollar has to be tracked to its category.

Project grants pit applicants against each other. They reward whoever writes the best application, which means they reward grant-writing staff as much as need.

Formula grants are the quiet workhorses. The formula is a political document: a change in how poverty or population is measured can move millions of dollars between states, and the argument over the formula is the real debate.

Block grants are what conservatives usually have in mind when they say "send it back to the states." The sum is fixed, the purpose is broad and the state chooses the details.

Open-ended matching is the most dangerous design for a taxpayer, because the federal bill is set by someone else's decisions. A state that spends a dollar gets back a dollar of federal money, or fifty cents, or whatever the matching rate is. The incentive is to spend more, because the state pays only a portion. That is not a conspiracy; it is arithmetic. A state that pays 40 cents of every Medicaid dollar will treat a $100 service as costing $40.

The history, with dates

The history of block grants is the history of conservative hope and conservative disappointment.

1972: General revenue sharing. The State and Local Fiscal Assistance Act of 1972, signed by President Nixon, sent a share of federal income tax revenue to states and localities with almost no strings. It was the purest form of the idea: the federal government collected, and the states spent. Congress cut the states out of the program in 1980, and ended it entirely in 1986, and the local governments that depended on it lost it. The lesson was bitter: money with no strings has no constituency in Washington, because members of Congress get no credit for it. A program that nobody can take credit for is a program that is easy to cancel.

1974: The Community Development Block Grant. The Housing and Community Development Act of 1974 consolidated several categorical urban programs into a block grant for cities and counties, to be spent on community development in the way local officials judged best. It is the longest-lived and most durable example, still running. But the strings did not stay away. Over the years, Congress and the executive branch attached requirements: environmental review, wage rules for construction, reporting and national objectives that a local government must show it is meeting. The block grant became less of a block. It remains more flexible than a categorical program, but it is far from a no-strings check.

1996: TANF. The Personal Responsibility and Work Opportunity Reconciliation Act of 1996, signed by President Clinton, replaced Aid to Families with Dependent Children, which was an open-ended matching program, with Temporary Assistance for Needy Families, a block grant. The federal commitment was fixed at about $16.5 billion a year for the states' family assistance grants, and the states got broad discretion over how to spend it, subject to work participation requirements and a five-year lifetime limit on federally funded cash assistance for an individual. It was widely considered a conservative victory, and the early results in caseload and employment were what supporters had hoped for.

But look at what happened to the fixed sum. The block grant was set at a nominal amount and has not been adjusted for inflation. Every year that Congress leaves it untouched, it is worth less. States, with broad discretion, have also used TANF funds for a variety of purposes only loosely connected to cash assistance for poor families, and some have used them to fill holes in their own budgets. The federal work requirements, intended as the central guardrail, have proved easy to meet through accounting adjustments. So a block grant delivered flexibility, and flexibility delivered both innovation and leakage. Both are real, and an honest account has to say so.

Medicaid: the road not taken. Medicaid, created in 1965, is the largest example of the open-ended matching grant. The federal share depends on each state's income level and ranges widely. In 1995, the Republican Congress passed a budget that would have converted Medicaid into a block grant, and President Clinton vetoed it. Proposals to do the same have returned in nearly every serious budget fight since, and none has become law. Conservatives who want to understand why the largest program stays open-ended should ask who benefits: governors who get federal dollars at a discount, health providers who are paid, and members of Congress who do not have to choose.

What the record teaches

Four conclusions follow.

1. Strings matter more than dollars. A smaller grant with few conditions can give a state more real authority than a larger one with thick rules. When a conservative says he wants federalism, the grant design is the test.
2. Flexibility without a constituency erodes. General revenue sharing died because no one in Washington owned it. A block grant needs defenders, or it is cut or has strings added.
3. Fixed sums protect taxpayers but wear away. The great virtue of a block grant is that Washington's bill is capped. Its great weakness is that a cap not indexed to anything is a cut. A fair design states the indexing.
4. Open-ended matching creates the incentive to spend. The more the federal government pays of each marginal dollar, the less the state has to care about the price. Any serious plan to control cost has to change the match.

Which design suits which goal

The question "which is best" has no answer without a goal. Here is a plain guide.

• If the goal is national uniformity, such as a minimum standard that must apply everywhere, a categorical grant with firm rules is the right tool. The price is the loss of local discretion. Conservatives should use that tool sparingly and admit its cost.
• If the goal is to find out what works, use block grants, so that fifty states try fifty approaches. That is the logic of Federalism Lets Fifty States Experiment, and the Constitution Planned It. The 1996 welfare overhaul is the best example: states tried different work and training approaches, and the evidence about what worked came from the variation.
• If the goal is to buy a single national good, such as an interstate highway or a research breakthrough, use project or formula grants with clear outcomes and an end date.
• If the goal is to control cost, do not use open-ended matching. A capped block grant, with a transparent indexing rule, gives the state both the freedom and the responsibility.
• If the goal is to protect the Tenth Amendment, keep the strings short and the sums predictable, and avoid the coercion that NFIB v. Sebelius condemned.

Where conservatives have been disappointed

I mentioned conservative disappointment, and here it is stated plainly. We were promised that block grants would shrink government and return authority to the states. In practice, three things have happened.

The strings returned. Congress cannot resist adding conditions to a block grant, because members want credit for how the money is used. The CDBG shows it.

The cap eroded. Without indexing, TANF's real value fell, which saved the federal government money but also meant that the program did less than its sponsors intended.

States spent the freedom on themselves. Governors treat flexibility as a budget tool. A conservative who believes in federalism has to accept that states will sometimes use freedom in ways he dislikes. The alternative is the permanent supervision of fifty states from one capital, which no one who believes in the Tenth Amendment can defend.

Those disappointments are not reasons to abandon the idea. They are reasons to design it better: put the indexing in the statute, say what counts as a permitted use, and publish how each state spent the money. For the broader argument about whether Washington should be sending money to the states at all, see The Federal Workforce Is Smaller Than You Think, and That Is the Problem, which describes how much of federal power already works through states and contractors, and Departments, Commissions, Corporations and Who Really Controls Each One on who controls what.

The steelman for categorical grants

I owe the other side its best case. A defender of tight federal strings would say: some national goals are too important to leave to fifty variations. When states were left alone in the past, some denied rights, and the federal government used conditioned funding as a lever to end it. Accountability to the taxpayer demands rules, because a block grant with no oversight invites waste, and the voter who funds the program in Ohio has a right to know the money is not spent in Texas on something else. And without detailed rules, a block grant becomes a slush fund.

That is a reasonable case, and part of it is correct. The answer is that the strings should follow the Constitution. Where the Constitution gives Congress a power, such as protecting the civil rights guaranteed by the Fourteenth Amendment, Congress may legislate directly, and does not need the leverage of a grant. Where it does not, using money to command what Congress could not directly order is exactly the kind of end-run that NFIB v. Sebelius policed. And the accountability argument cuts the other way: a voter can hold a governor accountable for a program the governor runs. He cannot hold anyone accountable for a program that is run by a federal rulebook and blamed on the state.

You can see how your own state's federal funds are divided in the budget documents on its treasurer's or comptroller's website, and the federal side in the national debt picture.

The case in brief

1. How Washington sends money to the states decides who runs the program, and the grant's strings matter more than its size.
2. The five types, categorical, project, formula, block and open-ended matching, sit on a spectrum from tight federal control to broad state discretion, with open-ended matching creating the strongest incentive to spend.
3. General revenue sharing (1972 to 1986), the Community Development Block Grant (1974) and TANF (1996) show that block grants tend to acquire strings, lose value without indexing, and lose their protection when no one in Washington owns them.
4. Each design suits a different goal: categorical for national uniformity, block grants for experiment and cost control, project and formula grants for single national goods.
5. Conservatives should write indexing, permitted uses and public reporting into any block grant, and keep strings tied to powers the Constitution actually grants.

What you can do: find one federal grant program your own state or county receives, read its eligibility and reporting conditions on the agency's website, and ask your state legislator whether the state would be better off with a block grant version.