The false economy of means-tested and conditional income support

Battle lines separate the demand for a universal basic income (UBI) from the hold-out view that the support should be rationed and restricted to those who most “deserve” it. Ranged on one side are civil society organizations, trade unions, academic researchers and sections of the ANC; on the other is organised business and the Treasury.

The latter appear to favour an exceedingly narrow, exclusionary and complicated approach that would deprive millions of people of desperately needed income support by adding layers of restrictive terms. The preference is to tightly ration income support by making it conditional (for example, on seeking employment) or “targeting” it at subpopulations (eg the “poorest” households, by testing for income levels). 

Both of those approaches claim they are the most cost-effective way to provide income support (since they only go to the people who are most in need of support). In fact, they are bad at reducing income poverty and near-useless for building social cohesion.

Targeted and conditional schemes rely on detailed information about the targeted beneficiaries (for instance, regularly verifying their incomes) that is often out-of-date, incomplete or plain inaccurate. When tied to job seeking, they force people into the costly theatrics of proving they’re looking for work, even when the odds of finding employment are vanishingly slim, as they are in South Africa. 

It’s been estimated that job searching costs on average around R940 per month (roughly what Statistics South Africa says a month’s supply of basic food and other essential items cost). 

Burdensome and prone to error and delay, targeted programmes routinely miss large proportions of intended beneficiaries. Brazil’s flagship Bolsa Familia programme, for example, missed very large percentages of intended beneficiaries (despite the country’s relatively strong administrative capacity) as did Mexico’s Oportunidades programme. 

None of the 42 targeted social protection schemes examined in a large review had exclusion errors of less than 44%; 12 of them had exclusion errors of more than 70%.

Similarly, the emergency cash payments provided in South Africa during the first wave of the Covid-19 pandemic brought vital relief to millions of people. But they were marred by serious inefficiencies and inaccurate and out-of-date information, which led to huge delays and gaps in coverage. In the grant’s initial cycle, only 6.4 million applicants were approved, even though the eligible population was estimated at 10-12 million people. Almost half of eligible individuals who were not receiving the social relief of distress grant by mid-2020 were in the poorest third of households. 

Means-tested programmes are also highly unfair in places with widespread poverty. They artificially segment populations who, in reality, live in similar desperation – offering one group support, while denying it to the rest. 

Studies of income distribution data from Ethiopia, Malawi and Zambia, for example, have shown that very small actual differences in personal and family circumstances separate people in the bottom 50%-60% of per capita consumption. 

Limiting income support to people earning less than, say, R663 per month (the current “food poverty line”) and denying it to others with, say, R800 entails a basically arbitrary segmentation between people in equally dire circumstances. When eligibility is based on tiny variations which ordinary people don’t see as real differences, it feeds a sense of unfairness, resentment and social tension. 

That is why Thandika Mkandawire argued that means-tested income support does not make sense in places where large proportions of the population are poor. 

Minute changes in circumstance shift people in and out of eligibility for means-tested support. Because the schemes struggle to reliably identify who should receive support at a given point, unmerited interruptions and delays in payment are built-in features. The schemes are typically designed and administered in ways that make it arduous to prove eligibility, remarkably easy to “lose” access, and exceedingly difficult to restore it.

 Tying income support to conditionalities became a staple of social policy during the neoliberal era. This distrustful approach assumes that people with little or no money must be forced to do what’s “good” for them. Defenders say conditionalities can “engineer” certain desirable behaviours (like enrolling and keeping children in school, having them vaccinated or having regular health check-ups). In fact, it’s very often not the conditionalities that lead to the desired behaviours, but the fact that families can afford to keep their children in school, or travel to a health clinic. 

Job-seeking requirements hinge on further fallacies. They imply that living in poverty is the fault of work-shy individuals and they assume that income support encourages idleness. 

The evidence says otherwise. 

Numerous studies, including a World Bank review of cash transfer programmes, show that adults do not work less when they get income support. Income support enables economic activity, increases rates of self-employment, supports job searching, and especially boosts women’s economic independence. In South Africa, with close to half of adults unable to find paid work, a job-seeking conditionality seems especially blind to reality.

Universal is easier, cheaper and fairer