Heterogeneous Treatment Effects of Place-based Policies: Which Cities Should be Targeted?

In this paper, we empirically assess the treatment effects of a Japanese place-based policy known as the urban revitalization zone (URZ) program. on regional economies. We propose socially desirable assignment rules for the URZ program. The mixed results of previous empirical studies on place-based policies suggest that the treatment effects are regionally heterogeneous. In order to account for such heterogeneity across regions, this study estimates the conditional average treatment effect (CATE) for each region using a marginal treatment effects framework. We then use the estimated CATE parameters to construct empirically welfare-maximizing treatment rules based on the following three types of regional characteristic variables: demographic variables, suburbanization variables, and local production-network variables. Our results indicate that the treatment choice rule based on the suburbanization variables is the most successful. In particular, we find that cities with high numbers of cars per household and low percentages of large-scale stores should be targeted for URZ.