Key Insights

1. Exclusion error is the failure mode nobody audits

Every targeting system produces two kinds of mistake. Inclusion error puts an ineligible household on the list, and it is visible, auditable, and politically explosive because it looks like fraud. Exclusion error leaves an eligible household off, and it is invisible by construction, because the excluded household appears in no dataset the system generates. Audit regimes and political incentives both push administrators to minimise inclusion error, and the rational response is to tighten criteria, which increases exclusion. A registry serving half the country’s households needs deliberate measurement of who it misses, or it will optimise for the error that embarrasses rather than the error that harms.

2. Registry data ages, and poverty moves faster than enumeration

Household circumstances change with harvests, illness, drought, and family structure, while registry data is collected in enumeration waves years apart. A household that qualified when surveyed may not qualify now, and a household that did not may urgently qualify today. Static registries systematically miss the newly poor, which is the group most responsive to timely support. The design question is not whether data ages but how quickly the system can accept new information, through on-demand registration, community updating, or integration with other administrative data.

3. Interoperability determines whether the registry helps or hardens exclusion

The registry’s value multiplies where it connects to civil registration, the national identity system, and payment infrastructure, because those connections allow verification without repeat enumeration. But the same connections transmit exclusion. A household without national identity documents, disproportionately likely to be poor, marginalised, or living in border and pastoral counties, can be excluded from the registry by a failure in a different system entirely. Integration is a benefit only where the systems being integrated are themselves inclusive.

4. Categorical and poverty targeting solve different problems, and mixing them requires clarity

The Older Persons Cash Transfer is categorical: age determines eligibility, and administration is simple, transparent, and cheap to verify. Poverty targeting is more precise in theory and considerably more contested in practice, requiring data, judgement, and defensible criteria. Consolidating both under one registry is efficient, but the two logics should not blur. A system that applies poverty screening to categorical programmes, or categorical simplicity to poverty programmes, will produce results it cannot defend under either standard.

5. Grievance mechanisms are the registry’s accuracy layer, not its complaints desk

The multi-channel grievance and redress function is often treated as customer service. It is better understood as the primary correction mechanism for a dataset that cannot be perfect. Excluded households appealing successfully are the system telling itself where it is wrong. That only works where the mechanism is genuinely accessible to people without smartphones, transport money, or literacy, which means the design of the appeals route determines how much accuracy the registry can recover.

6. Expansion to 2.5 million households changes the political economy

Scaling from 1.2 million to 2.5 million households, and building registry coverage toward half the country, moves social protection from a programme for the visibly destitute to a system touching a substantial share of the electorate. That raises the stakes on every targeting decision, invites political pressure on the criteria, and makes the technical governance of the registry, who can access it, who can amend it, and under what authority, a question of real consequence.