Experimental Design
The sample comprises 89 rural communes selected to receive the land registration program. Within each commune, two to five villages were sampled, yielding 295 villages overall. The 89 communes (and their 295 sampled villages) were randomly assigned to treatment (45 communes, 151 villages) or control (44 communes, 144 villages), with randomization stratified by department. Treatment communes were paired with treated land surveyors, and control communes with control surveyors.
Within each village, a household listing was used to identify eligible households—those that (i) owned at least one plot of land and (ii) had a daughter aged 13–18 living in the household—and eligible households were then randomly sampled. In total, 2,435 households were selected from villages in treatment communes and 1,487 from those in control communes. The 2,435 households in treatment communes were further randomly assigned, stratified by village, into two groups: 1,435 rewards households, whose titles counted toward surveyors’ performance evaluation, and 1,000 non-rewards households.