Poverty targeting algorithms expose flawed World Bank cash transfer program in Jordan 

 An automated cash transfer programme in Jordan, commonly known as Takaful, funded by World Bank, is undermined by algorithmic flaws, discriminatory policies, and stereotypes about poverty, Human Rights Watch revealed in a report released on 13 June entitled ‘Automated Neglect: How The World Bank’s Push to Allocate Cash Assistance Using Algorithms Threatens Rights’. 

According to Jordan’s National Aid Fund, households that meet the programme’s essential eligibility criteria are ranked from poorest to richest using an algorithm that estimates their income and wealth using 57 socio-economic indicators. Cash transfers are then distributed to the households deemed to be the most impoverished, depending on available funds. 

Human Rights Watch has found that some indicators must fully capture people’s economic complexity. Because their hardships fall outside an algorithm’s flawed poverty model, many people in Jordan do not receive financial assistance. It also fuels social tensions and widespread perceptions of injustice by distilling people’s hardships into a ranking that turns one household against another for support. Interviewees also spoke of difficulties in applying for benefits online and withdrawing their payments using electronic wallets, as well as in understanding and appealing against decisions regarding eligibility.

The report recommends that the World Bank and the government of Jordan should move away from targeting cash transfers based solely on income and economic status and towards a universal social protection system that ensures that all people receive income support at critical times in life.

In response to this, the World Bank said it is working to refine the targeting algorithm, noting that Takaful has proven to be one of the most cost-effective poverty reduction programmes currently operating in Jordan.