The distribution of World Bank funds to poor families in Jordan using an algorithm has been found to be inaccurate and biased by a non-profit human rights organization. The algorithm evaluated households on 57 socio-economic factors, leading to the exclusion of some people in need of help.
The program, called Takaful, was launched by the National Assistance Fund in Jordan in 2019 and distributed about $1 billion to 220,000 families. It uses an algorithm to track certain indicators to determine which families qualify for assistance. The indicators include the availability of property, the presence of a car and business, the number of family members, and electricity and water consumption.
According to Human Rights Watch (HRW), the algorithm did not consider the economic difficulties faced by those on the poverty line. The HRW team conducted 70 interviews with families and business owners in Jordan and found that while the algorithm took into account some social realities of the country, other factors like ownership of assets and electricity consumption were not suitable measures of poverty.
The study found that some families who were excluded from social support had recently inherited property but could not afford their needs, while applicants who owned a car over five years old or a business worth more than $4300 were automatically excluded by the algorithm, even if they were struggling financially.
While families led by a woman had a better chance of qualifying for assistance due to the small representation of women in the labor market in Jordan, the algorithm’s limitations meant that some people in need were excluded from the list of recipients.
The HRW report highlights the need for greater accountability and transparency in the use of algorithms in distributing social support to vulnerable populations.