This paper uses household panel data to provide robust evidence on the effects of
BRAC’s Targeting the Ultra-poor Program in Bangladesh. Our identification strategy
exploits type-1 errors in assignment, comparing households correctly included with those
incorrectly excluded, according to program criteria. Evidence from difference-in-difference
matching and sensitivity analysis shows that participation had significant positive effects on
income, food consumption and security, household durables, and livestock, but no robust
impact on health, ownership of homestead land, housing quality and other productive
assets. Using quantile difference-in-difference, we find that the income gains from program
participation is smaller for the lowest two deciles.