Grants or Food Parcels? Targeting Poverty in Kavango East
Kavango East, Namibia. A constituency of 18,000 people has 45% unemployment, and nationally Namibia's income inequality is among the highest in the world. About 60% of local households contain someone receiving the non-contributory old-age grant. The regional council must choose between two anti-poverty proposals. Proposal A expands the old-age grant and adds a universal child grant of N$300 a month for every child under 18. Proposal B gives a food parcel worth N$600 a month to the 'poorest 20%' of households, selected by a proxy means test using housing materials, livestock and wage income. A pilot of Proposal B found that 30% of the intended poorest households were excluded, while 15% of parcels went to non-poor households. San households living at the edge of the settlement were least likely to be registered, and many work as seasonal farm labourers. The council wants the biggest reduction in poverty for its budget.
1. Distinguish between the poverty headcount ratio and the poverty gap, and explain why a councillor needs both to choose between the proposals.
Read the worked answer
The headcount ratio is the share of the population below the poverty line. The poverty gap measures how far, on average, the poor fall below the line, expressed as a share of the line; it captures depth. A third measure, severity, weights the poorest more heavily.
The councillor needs both because the proposals do different things. A food parcel of N$600 targeted at 20% may reduce the gap substantially for those who receive it, but if 30% of the poorest are excluded, the headcount may barely move. A universal child grant of N$300 may reach many households and reduce the headcount among families with children, but the amount may be too small to lift a household with several children above the line, so the gap remains.
Consider a simple illustration. If the poverty line is N$1,000 and a household earns N$400, the gap is N$600. A N$300 grant closes half the gap but does not change the headcount. A N$600 parcel closes the gap and changes the headcount, but only for those who receive it. Headcount answers 'how many?', gap answers 'how poor?', and the budget should be judged on both - as well as on exclusion, which neither measure captures well.
2. Compare universal and targeted approaches using the pilot evidence, and explain how social exclusion affects the choice.
Read the worked answer
Targeting aims to concentrate the budget on the poorest, giving more help per rand to those most in need. Universal provision gives the same benefit to all, which is expensive but avoids selection errors and stigma.
The pilot evidence is the key. Proposal B excluded 30% of intended households - a large error - and leaked 15% to non-poor households. That means roughly half of the programme's value was misallocated. Proxy means tests using housing, livestock and wage income are poorly suited to this constituency: San households may be mobile, may lack documents, and may not fit the asset profile; seasonal farm labourers have fluctuating incomes that a snapshot survey misses. The San are also socially excluded, so they are less likely to appear on a register or to challenge a rejection. Targeting therefore reinforces existing exclusion.
Universal grants avoid the means test and the stigma. The old-age grant already reaches many households and is administratively proven. A universal child grant would reach San children if registration is mobile and documents are flexible.
But universalism is not automatically inclusive: if registration requires a birth certificate or a fixed address, the same households can still be left out. The comparison therefore depends on delivery design, not only on the principle.
3. Which proposal would you recommend, and what design changes would reduce exclusion? Justify your answer.
Read the worked answer
Recommendation: adopt Proposal A - expand the old-age grant and introduce the child grant - with a deliberate registration strategy for excluded groups, and phase in the child grant over two years.
Reasoning. Proposal B's exclusion rate is too high to justify its theoretical efficiency. A programme that misses 30% of the poorest, and misses the San most often, fails both the poverty-gap and the equity test. Universal grants reduce stigma, have lower administrative costs per rand, and can be delivered through existing payment systems. The child grant also targets a period of life when poverty has long-term effects, so it is likely to improve school retention and nutrition.
The design changes matter. Registration should be mobile, take place at schools, clinics and cattle posts, and accept alternative proof of identity. Community workers from San settlements should be hired to assist registration, and data should be published by settlement so that gaps are visible.
Cost is the main objection. A phased rollout, starting with the poorest constituencies and younger children, keeps the budget manageable. Monitoring should report the exclusion rate, the poverty gap and child school attendance. If exclusion remains high after two years, a top-up for remote settlements should be added.