Testing Phone-Based Targeting for Urban Social Protection

Low-income countries spend an average of 1.5 percent of GDP on social protection programs, but significant gaps in coverage result in only 18 percent of the poorest households receiving the program benefits for which they are eligible.

GiveDirectly aims to address the inefficiency and high cost of identifying and enrolling recipients with MobileAid, its eligibility assessment model that leverages machine learning and cell phone data to identify the extreme poor and then deliver benefits via mobile money transfers.

With funding from DIV, GiveDirectly, in partnership with researchers from the University of Berkeley and the Government of Malawi, will evaluate the comparative accuracy and cost-effectiveness of different targeting methods for identifying households eligible for cash transfers.

GiveDirectlys approach could significantly improve the speed, efficiency, and accuracy of urban poverty targeting in Malawi and serve as a model for other countries.