Three economists caught aid theft without an audit, an informant, or a single cooperating witness

the problem

Proving that foreign aid to a corrupt government is being stolen normally requires the cooperation, access, or paper trail that the people suspected of stealing it control and have every reason to withhold

background

what everyone would do

Send in auditors, demand project records, or set up a whistleblower channel inside the recipient government. This fails in exactly the countries the researchers studied: the people who control the records and the people suspected of diverting the money are the same people, so there is no independent paper trail to audit and no safe channel for an insider to use.

what they saw

You do not need to observe the theft to prove it happened — you only need to find the one place the stolen value is structurally forced to reappear. Diverted aid that becomes private wealth has to be stored somewhere, and secrecy-haven bank deposits are one of the few asset classes built specifically to receive money whose owner does not want it traced. That gave the researchers a downstream signal that required no access to the theft itself, only two datasets neither side of the transaction controlled.

the move

Rather than try to trace the theft at its source, Andersen, Johannesen and Rijkers combined World Bank quarterly aid-disbursement records with confidential Bank for International Settlements data on cross-border bank deposits, and looked for where the money would have to surface if it were being skimmed: foreign-owned deposits in secrecy-havens like Switzerland, Luxembourg and the Cayman Islands. Across the world's most aid-dependent, most corrupt countries, aid disbursements equal to 1% of GDP were followed, within the same quarter, by a 3.4% jump in deposits specifically in haven banks — with no matching jump in deposits at comparable non-haven financial centers, and no similar spike around other capital inflows of the same size.

why it works

Aid disbursed to a government creates an opportunity to divert funds; a share of that opportunity, if taken, converts into private wealth that its owner wants both liquid and hidden, which routes it toward haven bank deposits rather than domestic spending or investment. Because haven deposits are a narrow, unusual destination for money with no other reason to go there, a spike in haven deposits timed to the same quarter as an aid disbursement — with no matching spike in ordinary financial centers or around other comparably sized inflows — is difficult to explain by anything other than diversion. The correlation isolates the mechanism precisely because legitimately spent aid has no reason to produce that specific pattern, while stolen aid does.

the payoff

The implied leakage was roughly 7.5% of aid at the sample mean, rising further in the most aid-dependent countries — a number derived entirely from public and confidential financial-flow data, with no site visit, informant or admission from anyone involved. Published in the Journal of Political Economy (vol. 130, no. 2, 2022), the paper also became the center of a public controversy: The Economist and Financial Times reported that the World Bank's own leadership sought to delay the working paper's release in early 2020, a delay that collapsed within hours after co-author Niels Johannesen, protected by tenure outside the Bank, posted the draft to his personal website — forcing the Bank to publish its own working-paper version the same day.

where it breaks

The method only works where a genuinely distinctive downstream destination exists and is separately measurable — it cannot detect diversion into domestic real estate, cash, or local asset purchases that leave no cross-border trace, and it says nothing about which individuals did it, only that the pattern is present at the country level. It also depends on having a clean comparison group (non-haven deposits, non-aid capital flows) to rule out confounds like general capital flight during a crisis; without that contrast the correlation could reflect unrelated economic shocks instead of theft. And the approach proves a population-level pattern, not a prosecutable case against any one official, which is also exactly why it can be published and defended even when the underlying government refuses all cooperation.

what came after

The method has since been cited as a template for a wider line of research using downstream financial-flow proxies to detect aid diversion without on-the-ground access, including Ensminger and Leder-Luis's 2022 Benford's-Law-based method for flagging fabricated project expenditure reports — a complementary technique that looks for the fraud's fingerprint in reported numbers rather than in where the money lands. The 2020 publication dispute is separately cited in discussions of research independence at multilateral institutions that fund the work reviewing their own effectiveness.

references

1. [1]Elite Capture of Foreign Aid: Evidence from Offshore Bank AccountsJournal of Political Economy (University of Chicago Press), 2022nielsjohannesen.net
2. [2]The World Bank's "Papergate": Censorship Is Not the Best Way to Stop Development Aid From Fueling CorruptionProMarket (Stigler Center, University of Chicago Booth School of Business), 2020promarket.org

keep it

same kind of clever

• 1958 · Bank of America (the 'Fresno Drop')No bank knew if ordinary people would actually use unsecured credit — so instead of running a pilot with volunteers, Bank of America just mailed 60,000 live credit cards to an entire city and watched what happened.
• 240 BCE · Eratosthenes, Library of AlexandriaEratosthenes measured the size of the entire planet without ever leaving Alexandria
• 1998 · Innocent DrinksInnocent's founders let a festival crowd's empty cups, not a survey, decide whether they should quit their jobs