Contacts Matter : Local Governance and the Targeting of Social Pensions in Bangladesh

We present evidence on the extent and possible causes of mistargeting of a largescale social-pension program in Bangladesh. The evidence stems from surveys and lab-in-the-field experiments that we ran in eight different unions (municipalities) with three different groups: (i) a random sample of the elderly population (potential bene ciaries), (ii) a random sample of newly selected bene ciaries, and (iii) the local government representatives, who were in charge of the last round of selections. On the one hand, our (pre-registered) analysis suggests that personal relationships are crucial for being selected as a bene ciary, which might indicate corruption. On the other hand, our results strongly suggest that a severe lack of state capacity (e.g., knowledge of the o cial rules…