Targeting South Africa's Social Relief of Distress Grant

South Africa is the most unequal country in the world, and one in six people live in food poverty. During the COVID-19 pandemic, the government introduced the Social Relief of Distress (SRD) grant, South Africa's first welfare program for able-bodied unemployed adults below pension age. The program currently supports eight million beneficiaries with a monthly unconditional cash transfer of about $50 USD (PPP). Our modelling suggests the SRD lifts 1.8 million people out of food poverty each month. Randomized controlled trials show such unconditional cash transfers are effective poverty reduction tools across contexts (Bastagli et al. 2016).

This team has worked with the government on the design of the SRD grant since 2020, contributing to major policy changes including expanding eligibility to caregivers who receive child support benefits and adjusting benefit levels for inflation. This work has been supported by previous J-PAL IGI funding.

In February 2026, the government announced that the SRD will transition from temporary disaster relief into a permanent Livelihoods Support Grant, the first unconditional provision for all unemployed adults in Africa. A key challenge is how to target such grants to maximize poverty reduction under fiscal constraints (Coady et al. 2004; Alatas et al. 2012). Eligibility is currently determined using a "bank means test" based on monthly inflows to applicants' bank accounts, but this criterion is crude. For example, bank inflows from rotating savings groups or transfers between household members can lead to loss of eligibility. A recent court case brought by civil society requires the government to improve targeting and reduce exclusion errors.

The Presidency and the Department of Social Development have asked this team to conduct a study recommending the best feasible targeting approach, which will be scaled nationally beginning in mid-2027. The aim is to continue with the grant application, targeting, and payments being fully digital. Applications are submitted through web and USSD platforms, and targeting will use existing administrative or banking data, or self-reported information collected digitally.

To test alternative approaches, this team will collect a nationally representative household survey measuring income, assets, and consumption to establish a ground-truth poverty benchmark. Survey data will be linked to administrative records to evaluate the accuracy and cost-effectiveness of several targeting approaches, including self-targeting, proxy means testing, administrative data linkages, satellite-based poverty prediction, and combining data with machine learning methods. Follow-up phone surveys will measure poverty dynamics and assess how well each targeting approach captures them.