Foreign delegations come to Washington to meet a counterparty that is not there

A government that wants to understand what its money will meet in the United States books its time where the foreign policy is made. The itinerary is federal, the briefings are federal, and the document signed at the end is a memorandum with a federal department or agency on the other signature line. This is entirely rational. It is also, for most of what the delegation actually wants, the wrong address.

The taxes a foreign project will pay, the incentives it might be offered, the permits its buildings will need, the workforce programmes it can use and the utility that will connect it are, with few exceptions, not federal. They belong to one of fifty-one jurisdictions that the federal government does not control and cannot commit. The collections file this publication holds gives a plain view of how far apart those jurisdictions are, and the distance is larger than the language of a bilateral relationship suggests.

My position is simple. The most expensive habit in economic diplomacy with the United States is asking the federal government to guarantee things it does not levy, does not own and cannot deliver, and then treating the resulting warm language as progress.

Fifty-one tax codes and no template

The Census Bureau's survey of state government tax collections for fiscal 2025 names five taxes: general sales and gross receipts, individual income tax, corporation net income tax, property tax and severance tax. Across the 51 jurisdictions, 20 record all five. Twenty-one record four. Eight record three. Two record only two of the five. There is no standard American state tax code, and there is no state that can be treated as representative of the others.

The absences are specific and they are recorded as facts. Six jurisdictions levy no individual income tax: Alaska, Florida, Nevada, South Dakota, Texas and Wyoming. Four levy no corporation net income tax: Nevada, Texas, Washington and Wyoming. Four have no general sales and gross receipts line: Alaska, Delaware, Montana and New Hampshire. Fourteen record no state-level property tax at all: Colorado, Connecticut, Delaware, Florida, Hawaii, Idaho, North Carolina, New York, Ohio, Oklahoma, South Dakota, Tennessee, Texas and Utah. Fifteen record no severance line: the District of Columbia, Delaware, Georgia, Hawaii, Iowa, Massachusetts, Maryland, Maine, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, South Carolina and Vermont.

Texas is the clearest case. Of the five named taxes it carries two: general sales and gross receipts at 56,499,029 thousand dollars and severance at 8,001,451 thousand, against a total of 89,943,632 thousand. Delaware carries the opposite pair, individual income at 2,754,586 thousand and corporation net income at 577,320 thousand, against a total of 6,949,753 thousand. In Delaware's case the two named lines account for less than half of what the state raises, and the same is true in New Hampshire, where the three named lines come to 1,615,827 thousand against a total of 3,372,460 thousand. Even the five-item vocabulary the federal statistical system uses does not describe these states.

A delegation that asks what its investment will pay in tax therefore has no national answer available to it, not because the information is secret but because the question has fifty-one answers and the person who can give any of them works in a state capital.

The label a delegation carries is not what the file says

Shorthand travels well and survives contact with evidence badly. Washington is widely described abroad as a state without an income tax. The file records no corporation net income tax line for Washington and an individual income tax line of 545,712 thousand dollars. The file does not say what that line taxes, and I am not going to guess, but a delegation that arrives with the phrase and not the number has already lost the argument it came to have.

The same applies in reverse. Texas levies no individual income tax, no corporation net income tax and no state-level property tax, and a foreign reader can be forgiven for concluding that the state is indifferent to revenue. It is not. It raised 89,943,632 thousand dollars in fiscal 2025, largely through a sales and gross receipts line and a severance line, both of which are triggered by the kind of activity an industrial project creates. The state is funded. It is funded by different things, and those things sit closer to a project's operations than an income tax does.

Scale is not authority

California collected 275,041,333 thousand dollars in fiscal 2025, roughly 275 billion. Alaska collected 1,686,939 thousand. The instinct that follows from those two numbers, to spend the visit on the biggest, is reasonable for a market and misleading for a negotiation. Total collections describe the size of the economy under a state government. They do not describe what that government can decide, how fast it can decide it, or whether the decision a project needs belongs to it at all.

The property tax is the sharpest illustration. For most industrial projects it is the largest recurring local cost and the most negotiable item on the table. Fourteen jurisdictions record no state-level property tax line whatsoever, and this file, by design, carries nothing raised below the state. The most consequential number in a project's operating model is thus invisible in the federal statistics on state taxation, because it is set and collected by a county, a municipality or a school board that no one in the delegation has heard of and that no federal official can speak for.

The practical hierarchy is close to the inverse of the itinerary. The federal layer sets the conditions of entry. The state layer sets much of the economics of operating. A local layer that appears in none of these files often sets the number that decides whether the model closes.

It is worth being clear about why this catches capable people out. In a unitary state the ministry that signs an agreement is usually also the body that levies the tax, issues the permit and owns the utility. The mental model a delegation brings is not careless; it is accurate at home. The American arrangement breaks the link between the office that can promise and the office that can perform, and it breaks it in a way that is invisible in a meeting, because the federal official is not lying when they express support. They are expressing support for something they do not control. The cost of the misreading is not a refusal. It is eighteen months of courteous progress toward a commitment that was never available, and the foreign side usually discovers this at the point where it has already committed capital.

What cannot be settled in any meeting

There is a further limit, and honesty requires putting it next to the rest. None of these files carries state history. The collections file is a single fiscal year. This publication's state metrics dataset, fetched on 7 September 2026, holds one observation and one prior period for each series. A delegation that wants to know whether a state's revenue base has been stable, or whether an incentive regime has survived a change of administration, cannot learn it from the public files an outsider would naturally reach for. It has to go to the state's own accounts and its own legislative record, which is a different kind of work and a longer visit.

The strongest objection to everything above is practical. Foreign ministries are built to talk to national counterparts. They have one embassy, a handful of consulates, and a minister whose diary is finite. Asking a ministry to run fifty relationships in place of one is not advice; it is a complaint about the constitution.

The objection is right that fifty relationships are impossible. It is wrong that the alternative is one. What a delegation needs is not coverage but correct attribution: for each variable it cares about, knowing which layer owns it. Three questions do most of that work, and none of them requires a new post.

1. Which of the five named taxes does this state actually levy, and what is its total collection.
2. Which of our costs are not in this file at all, and which body sets them.
3. Who signs, on what timetable, and what happens to that signature when the administration changes.

There is a diplomatic consequence that goes beyond any single project. Agreements between national governments increasingly contain commitments that can only be executed sub-nationally: a supply chain located somewhere, a facility built somewhere, jobs in a place. The national government that signs cannot compel the place, and the public statistical record at that level, as these files show, is thin enough that neither party can verify much afterwards. That is not a legal problem. It is an information problem, and it is the reason so many industrial memoranda age badly.

None of this argues against the Washington leg of the trip. It argues against mistaking it for the visit. The meeting that determines whether a foreign project works is usually held in a building the delegation did not plan to enter, with an official whose name does not appear in the communique, about a tax that does not appear in the federal file.