Ghost Beneficiaries in a Cash Transfer Program
A Nigerien NGO, Amani Relief, distributes monthly cash transfers of XOF 20,000 to 12,000 vulnerable households in Tahoua Region through a mobile-money provider. Enrollment is done by 24 community agents using tablets; the same agents also help beneficiaries activate SIM cards. The donor requires a biometric or national ID check at enrollment, but the field coordinator waived the biometric check for 1,800 households in two communes because of network problems. A data review shows that 480 beneficiary phone numbers are duplicated; 300 of those belong to only 12 phone numbers registered to two agents; 62 beneficiaries share the same birth date and similar name spelling; and 41% of transfers to the duplicated numbers are withdrawn within 10 minutes at the same agent outlet. The program manager says the agents are trusted community members and that delays would have caused unrest. The next payment cycle starts in three weeks.
1. What fraud schemes are indicated, and how would you separate agent collusion from program design failure?
Read the worked answer
The pattern indicates ghost beneficiaries and agent collusion, possibly identity fraud. The duplicated phone numbers, shared birth dates, similar names, and withdrawals within 10 minutes at one agent outlet are consistent with agents enrolling people who do not exist or controlling beneficiaries SIMs and cash. But the design failure is also real: the same agents enroll beneficiaries and help activate SIMs, biometric checks were waived for 1,800 households, and no independent verification exists. Separating the two requires testing: if duplicate numbers cluster under specific agents and withdrawals occur at those agents outlets, collusion is likely; if duplicates are spread across agents and reflect data-entry shortcuts, design and validation failure may dominate. The waiver is a control override that created opportunity. A fraud examiner should not accept trusted community members as evidence; trust is not a control. The scheme may be asset misappropriation through fictitious payees, with the mobile-money provider as payment channel.
2. Design an investigation plan and specific data tests to quantify the loss and identify the beneficiaries.
Read the worked answer
First, preserve evidence: obtain raw enrollment and payment logs from the mobile-money provider, agent tablet records, and SIM registration data under a documented chain of custody. Second, reconcile the beneficiary roster to national ID numbers and phones, and flag duplicates, shared device IDs, shared agent IDs, and names with near matches. Third, test timing and location: payments withdrawn within 10 minutes at one outlet suggest agent control; compare withdrawal outlet to the beneficiary commune. Fourth, sample 100 flagged and 100 unflagged households for field visits or calls, using an independent team rather than the enrolling agent. Fifth, interview agents separately, starting with non-suspects, and review their contracts and commissions. Quantify loss as payments to invalid or duplicate beneficiaries plus any agent commissions. Use a conservative threshold before calling a household a ghost; data errors are common. Document exclusions and have the NGO counsel review referral decisions.
3. What preventive controls should be added before the next payment cycle, and which is most urgent?
Read the worked answer
Most urgent: suspend the two agents and independently re-verify the 1,800 households whose biometric check was waived before any new payment. Then add a unique phone per household or documented exception; biometric or scanned national ID at enrollment; segregation between enrollment and SIM distribution; agent rotation and outlet monitoring; real-time exception reports for duplicate phones, shared devices, and rapid withdrawals; beneficiary receipt confirmation by an independent call or SMS; and a grievance hotline. The program manager concern about unrest is legitimate, so the NGO should communicate that verification protects legitimate beneficiaries. The controls should be tested in one commune before scaling. Prevention here is not just policy: the payment data must be monitored continuously because the scheme is data-driven. The examiner should also assess whether donor reporting was misleading, since the waiver may have been known and not disclosed.