Assessment of welfare gains from a targeted social program can be seriously biased unless the endogeneity of program participation is addressed. Bias comes from two sources of placement endogeneity: the purposive targeting of geographic areas, and the targeting of individual recipients within selected areas. Partial decentralization of program placement decisions is common, but the center often retains control of broad geographic targeting. This paper argues that partial decentralization of program placement decisions creates control and instrumental variables useful for identifying program benefits. The center's allocation to a local level of government is presumably based on observable indicators. The central allocation will also influence allocation to individuals but is unlikely to determine individual outcomes conditional on individual program participation. With controls for the welfare-relevant geographic characteristics determining program placement decisions, the center's allocation across areas can be used as an instrumental variable for individual participation. This approach is illustrated with Bangladesh's Food for Education program, which provides a food stipend to poor, primarily rural families in exchange for their children's school attendance. A single post-intervention cross-sectional household survey identified program impact on attendance, using village-level placement as an instrument for individual placement. A detailed community survey, coordinated with the household survey, controlled for sources of heterogeneity in geographic influences on attendance. The program had a sizeable impact on school attendance and was greatly underestimated by a regression estimator that ignored the purposive program placement. (SV)