Workforce identity is often the right place for a government to start: the population is bounded, the financial return is measurable within a year, and it builds the institutional muscle a national programme will need.
Reconciliation is not a phase that ends. A programme scoped as "enrol everyone, then finish" produces a clean register that decays from the day it is signed off.
That comparison is where ghost workers surface, and it is a data exercise rather than a biometric one. The biometrics establish that each enrolled person is a distinct human being. The reconciliation establishes which of those human beings the state is actually obliged to pay.
Most programmes budget generously for capture and thinly for reconciliation. The ratio should be closer to the other way round.
Removing names from a payroll has consequences for real people, including people whose records are wrong rather than fraudulent. A teacher enrolled under a married name, a nurse transferred between districts without the file following, a worker whose date of birth was recorded differently in three places — all of them look like ghosts in a naive reconciliation.
A programme without a well-resourced correction and appeal process generates injustice and a backlash that can stop it entirely. There are programmes that recovered substantial sums in year one and were suspended in year two because a union, a regional governor or a newspaper found the cases that were handled badly.
Budget for the appeals process at the same time as the enrolment, and publish its turnaround time.
Verification at the point of payment
Turns a one-off cleanup into an ongoing control. Without it, the ghosts return within two payroll cycles — the same practices that created them are still in place.
Offline verification
Matters wherever payment happens beyond network coverage. If verification requires connectivity that the payment point doesn't have, it will be waived, and a waived control is no control.
Deployment reach
Enrolment must go to the workplace. Asking staff to travel to a provincial capital produces exclusion concentrated among the most remote, who are also the hardest to verify by other means and the most likely to be wrongly classified as ghosts.
- Enrolment completed but never reconciled against payroll, so nothing changes and the programme is judged a success on the wrong metric
- No ongoing verification, so the register decays immediately
- Appeals process under-resourced, so legitimate staff are removed and the programme loses political cover
- Success measured in enrolments rather than in payroll value corrected
- Reconciliation run once, at handover, rather than continuously
Payroll value corrected, net of appeals upheld.
The gross figure — total salaries removed — is the one that gets announced. The net figure, after wrongly removed staff are reinstated, is the one that tells you whether the programme worked. A wide gap between them is not a rounding difference; it is a measure of how many people were harmed on the way to the headline.