This study empirically assesses the treatment effects of a place-based policy in Japan and proposes treatment assignment rules that are socially desirable and practically feasible for the program. To account for the heterogeneity of treatment effects, we estimate the conditional average treatment effect for each region using a marginal treatment effect framework and construct treatment rules based on regional characteristics such as demography, degree of suburbanization, and local production network. We find that the program has positive effects on retail floor space, retail sales per inhabitable area, and retail employment per inhabitable area. However, the treatment effects are significantly heterogeneous and the total welfare gains could have been larger under the proposed treatment rules, which target the cities facing a “modest” decline of city centers. We also demonstrate that the regional characteristics for which the treatment rule achieves the largest welfare gains are different in each outcome. This emphasizes the difficulty of pursuing multiple objectives with a single rule. Although the program was effective for its main target industry, we did not observe any significant effects on general economic outcomes such as land price and income tax revenue.