We recently dealt with our worst fraud case to date, with 2% of a round of transfers in Uganda diverted. This incident brought to light vulnerabilities which we have since addressed, and we don’t expect changes in performance going forward. We are making it a point to write openly about this case, however, because nonprofits typically do not discuss fraud risk and management and we want to set a precedent for greater transparency within the sector.

Funds were diverted by staff at a local mobile money agency, in collusion with members of our own team. This group took advantage of the fact that in Uganda (unlike Kenya), we disburse transfers on a scheduled day each month, on which the mobile money agent locates close to recipients. These “cash-out” days create an opportunity to quickly spread misinformation among a large group of recipients. In this case, the agent and our staff told recipients that their transfers would be reduced by 50,000 shillings (20 USD) to cover the costs of mobile phones that had been given to them (and for which they had already paid). Some recipients believed this and others did not, but those who did not had limited recourse, as traveling to a different agent would have been costly and time-consuming. Compounding this problem, our own call center operator was participating in the scam and told recipients who called with questions that the deductions were legitimate.

We learned about this incident in the course of routine back-checks with recipients. We then suspended the staff in question while conducting a systematic investigation, talking with each potentially affected recipient. The testimony we heard from recipients uniformly corroborated the account above. We then dismissed the staff members involved from our side and requested a police investigation, leaving it to recipients to decide on an individual basis whether to participate and seek to recover funds. Given the state of law enforcement in Uganda, we suspect this is, sadly, not a high-return strategy.

Moving forward, we’ve made a series of changes to address the vulnerabilities this episode uncovered. We’ve done so working on our usual assumption that the integrity of the process has to rest on a system of independent checks and balances, not on faith in the goodness of any particular person. We’ve adapted the Uganda model as follows. First, we are requiring more frequent and less predictable presence by senior staff at pay-days. This diverts management time from other important functions (e.g. planning, training) but increases accountability. Second, we now enlist community-nominated monitors to observe the entire payments process and call us to report any irregularities in real time. This gives us one extra set of eyes on the crucial cash-out process. Third, we have increased the separation of responsibilities between call center and field staff. This reduces our operational flexibility to move the most talented staff across functions, but limits the influence any one person has over communication to recipients. Fourth, we have tasked our call center to call recipients in real time during “cash-out” days to create an additional, real-time channel through which information reaches us from recipients. In the longer term, we may also consider moving the payments function in-house entirely in some settings to give us more control than we had in this case over the staff responsible for disbursing cash.

Fraud and corruption are big business in development work, particularly when we deliver resources (like cash) that are actually valuable to the poor. At GiveDirectly we aim to be both the best and the most honest in the anti-fraud business. We’ll continue to take credit for successes and responsibility for failures, and as always welcome your questions or comments.

Back to List