We compare commune-level poverty rankings in Cambodia based on three different
methods: small-area estimation, principal component analysis using aggregate data, and interviews
with local leaders. While they provide reasonably consistent rankings, the choice of the ranking
method matters. In order to assess the potential losses from moving away from census-based poverty
mapping, we used the concentration curve. Our calculation shows that about three-quarters of
the potential gains from geographic targeting may be lost by using aggregate data. The usefulness
of aggregate data in general would depend on the cost of data collection.