Every remittance dollar received was sent by someone — yet in 2024 the world reported receiving $237 billion more than it reported sending

Personal remittances appear twice in the balance-of-payments framework the IMF and World Bank compile: once as money received by the destination economy (indicator BX.TRF.PWKR.CD.DT) and once as money paid out by the sending economy (indicator BM.TRF.PWKR.CD.DT). The two are opposite sides of the same cross-border transfers — one country's outflow is another's inflow. For 2024, the World Bank's world total records $856.6bn received and $619.3bn paid — a gap of $237.3bn, with received exceeding paid by a factor of 1.383. Summing the two series across the individual economies that report them reproduces those world totals exactly: 160 economies reported a received figure and 162 reported a paid figure for 2024, and their sums are $856.6bn and $619.3bn. The gap is therefore internal to the same universe of reporters, not an artifact of aggregating different country sets. The mismatch is not confined to one year. In every year from 2018 through 2024, reported received exceeded reported paid — by ratios ranging from 1.345 (2023) to 1.537 (2022). The single-year gap ranged from $176.2bn (2018) to $267.7bn (2022); summed over the seven years it totals roughly $1.5tn. Both series carry the same source attribution — "IMF balance of payments data; Staff estimates, World Bank" — and the same database vintage, last updated 2026-07-13. The largest reported receivers in 2024 were India ($137.7bn), Mexico ($67.6bn), the Philippines ($40.3bn), France ($38.8bn) and Pakistan ($34.9bn). The largest reported payers were the United States ($103.2bn), the United Arab Emirates ($58.5bn), Saudi Arabia ($46.6bn), Switzerland ($40.1bn) and Germany ($24.7bn).

The following is analysis, not fact. A remittance is a closed transaction: money that leaves one country's accounts arrives in another's. In a perfectly measured world the two series would nearly reconcile — and to the extent they differed, the paid side should sit slightly *higher*, because intermediaries skim a fee in transit, so the amount debited at origin exceeds the amount credited at destination ^mmodeled. The World Bank's own remittance-cost tracking puts the global average fee near 6% of the amount sent. Fees, in other words, predict paid > received. The data show the reverse, and by a wide margin: received exceeds paid by 38%. Read as an accounting identity, the $237bn is a missing sender — inflows that arrive on 160 countries' books with no matching outflow recorded anywhere on the paying side ^mmodeled. The direction is the tell. Under-counting is easier on the outflow side than the inflow side, because the incentives are asymmetric: a receiving economy has every reason to measure hard-currency inflows carefully, while the compensation of employees and personal transfers leaving a host economy pass through channels — informal operators, cash carried across borders, correspondent-bank routing that obscures the ultimate origin — that its statisticians capture less completely ^mmodeled. The World Bank has for years flagged that remittance *outflows* are recorded less reliably than inflows; the $237bn is the size of that asymmetry in 2024. What makes the pattern legible only in aggregate is that no single country's books look wrong. The United States, the UAE and Saudi Arabia between them report $208bn in outflows — substantial, not obviously deficient. The shortfall emerges only when all 162 payers are summed against all 160 receivers and checked against the identity that the two totals must, in principle, meet. That the ratio has stayed between 1.35 and 1.54 for seven consecutive years — never once inverting — is what distinguishes a structural measurement gap from year-to-year noise.

The strongest counter is that this is a known statistical artifact, not a mystery, and "missing sender" overstates it. Balance-of-payments compilers document several mundane sources of asymmetry: economies differ in how they split "personal transfers" from "compensation of employees," in valuation and timing of when a transfer is booked, and in whether they compile the debit side at all — some report inflows but never estimate outflows, mechanically pushing the world received total above the world paid total. On this reading the gap measures the uneven quality of national statistics, not money that vanished; every dollar was in fact sent, it simply was not *recorded* as sent. That counter is largely correct, and this piece asserts measurement asymmetry, not disappearance. But it does not dissolve the finding — it names its mechanism. Two economies reported a paid figure with no received figure, so missing receivers cannot explain a gap that runs the other way. And the World Bank's figures are themselves staff estimates layered on IMF returns precisely to patch national gaps; that the patched series still leaves a persistent, one-directional $200bn-a-year hole is the point, not a rebuttal to it ^mmodeled. The arithmetic and its persistence are the facts; whether to call the residual "unrecorded outflows" or "statistical noise" is the contested part.

From a receiving central bank, there is no gap at all. The dollars land, they are counted, they clear — Bangladesh's or the Philippines' inflow line is among the more carefully audited numbers in its accounts, because the foreign exchange matters. The mismatch is visible only from the vantage of whoever tries to close the world's books: a statistician reconciling 160 destinations against 162 origins, for whom $237bn arrives in 2024 with no departure stamp. The same transaction is a well-measured fact on one side of the border and an estimation problem on the other, and which one you see depends on which ledger you are keeping.

A correspondent covers remittances one corridor at a time: what India took in, what the Gulf sent home, what a fee cut would save a Filipino nurse. Each of those numbers is individually defensible, which is exactly why the discrepancy is invisible at country scale. It appears only when all 160 receiving economies are summed against all 162 paying ones and held to the identity that binds them — that money received had to be money sent. Do that, and 2024 shows $237bn arriving that no country reports sending, a gap that has pointed the same direction, against the pull of transfer fees, for seven straight years and $1.5tn. The story is not any one corridor; it is that the world's most-cited figure for how much migrants send home is measured far better on the receiving end than on the sending end — and the size of that blind spot is a quarter-trillion dollars a year.