Where does a foreign buyer actually keep the most?

The ranking

Every covered jurisdiction, same deal

On the yield column. The §871(d) election taxes NET rental income, after expenses and depreciation. This engine has no expense input, so the graduated bands are applied to GROSS rent, which overstates the rental tax for an elected landlord. Where several jurisdictions are shown side by side the overstatement is larger in the ones that levy income tax, so it changes the ORDER of the comparison and not only its level.

Buyer profile: non-resident foreign individual, direct ownership, resale property. Withholding at closing (FIRPTA and any state equivalent) is a prepayment credited against the actual tax, so the engine excludes it from cost totals and it is not ranked here; it appears on each jurisdiction's own page for cash-flow planning. Net repatriated compares sale proceeds after exit costs against the capital invested, at the exit price you entered.

Method

How this is computed, and what it is not

Every jurisdiction is run through the same engine on the same inputs, using only human-verified cost rules, each carrying a primary source and an as-of date. Nothing is estimated. A jurisdiction with an unsourced cost line is listed as not ranked rather than scored on partial data, with one exception, stated in full below: three named gap classes still rank, because none of them can flatter a jurisdiction against another. The last column of the table shows how many cost lines resolved for each row, and names the ones that did not.

The three gaps that still rank, and why each one cannot move the order.

• State withholding at closing. State withholding at closing is a prepayment credited against the tax already itemised, so the engine excludes it from every total. A state missing this line still ranks on the same arithmetic as a state that has it.
• Legal / closing costs. Buyer legal and closing costs are not yet sourced for any US state. They are excluded from every row equally, so they shift the level of the whole table and cannot change its order. The day they are sourced for some states and not others, the states still missing them stop being ranked.
• Mansion / luxury transfer tax. A mansion or luxury transfer tax is charged only where a row shows one. This is the weakest of the three: most states carry no rule of any kind here, so a levy we have not sourced would surface as no levy. Sourced negatives for this line are in the review queue.

What this ranking excludes on purpose. This figure counts taxes, government fees and agent commissions only. It excludes operating costs: insurance, HOA or service charge, property management, maintenance and vacancy. Those are building-level commercial terms rather than public record, so they cannot carry a primary source, and this product does not print numbers it cannot cite. The column is therefore yield after tax and before operating costs, and it must not be read as an investor's net yield.

The rental tax is overstated, and it moves the order. The §871(d) election taxes NET rental income, after expenses and depreciation. This engine has no expense input, so the graduated bands are applied to GROSS rent, which overstates the rental tax for an elected landlord. Where several jurisdictions are shown side by side the overstatement is larger in the ones that levy income tax, so it changes the ORDER of the comparison and not only its level. We state it here rather than only on each jurisdiction's own page, because the ranking is the artifact that travels.

The seller's commission is market practice, not law. The 5.4% US seller-side commission is our arithmetic on a Federal Reserve research note reporting a buyer-side average of about 2.7% from CoreLogic MLS data. The rule's own text calls it a documented ceiling-style estimate, not a quote, and records that everything has been negotiable since the NAR settlement took effect on 17 August 2024. It is a single federal constant applied identically to all rows, so it shifts every total by the same dollars and leaves the order untouched, but it is the largest single cost in most rows. An institution disposing of an asset at its own negotiated rate should read that column with that in mind. It is carried in the corpus under an effective-date basis of "statutory", which is wrong for a market average; the correction is queued for the data owner.

How each state is represented. Effective property tax is levied by the county, not the state, so there is no such thing as a state property-tax rate to rank. Each state is therefore run at its median county by effective rate — named in the table above, alongside that county's rate. A median is used rather than a headline county because picking the largest metro would flatter some states and punish others. It follows that these figures describe a typical county in each state, not any specific address: a deal in a high-rate county will cost more than the row shows, and one in a low-rate county less. For a specific address, run True Cost, which resolves that neighborhood's own county.

This is research information, not investment, tax or legal advice, and a ranking is not a recommendation to buy anywhere. Confirm with licensed counsel in the jurisdiction before committing capital.