Economic policies often involve dynamic interventions, where individuals receive repeated interventions over multiple periods. This dynamics makes past responses informative to predict future responses and ultimate outcomes depend on the history of interventions. Despite these phenomena, existing economic studies typically focus on static targeting, possibly overlooking key information from dynamic interventions. We develop a framework for designing optimal dynamic targeting that maximizes social welfare gains from dynamic policy intervention. Our framework can be applied to experimental or quasi-experimental data with sequential randomization. We demonstrate that dynamic targeting can outperform static targeting through several key mechanisms: learning, habit formation, and screening effects. We then propose methods to empirically identify these effects. By applying this method to a randomized controlled trial on a residential energy rebate program, we show that dynamic targeting significantly outperforms conventional static targeting, leading to improved social welfare gains. We observe significant heterogeneity in the learning, habit formation, and screening effects, and illustrate how our approach leverages this heterogeneity to design optimal dynamic targeting.
Authors
Takanori Ida, Takunori Ishihara, Koichiro Ito, Daido Kido, Toru Kitagawa, Shosei Sakaguchi, Shusaku Sasaki
Related Organizations
- Acknowledgements & Disclosure
- We would like to thank Linnea Holy for her exceptional research assistance and Christopher Costello, Michael Greenstone, Ryan Kellogg, Anna Russo and seminar participants at the Coase Project Conference for their helpful comments. We thank the Japanese Ministry of Environment for their collaboration in realizing this study. Kitagawa and Sakaguchi gratefully acknowledge financial support from the ERC grant (number 715940) and the ESRC Centre for Microdata Methods and Practice (CeMMAP) (grant number RES-589-28-0001). The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.
- DOI
- https://doi.org/10.3386/w32561
- Published in
- United States of America
Table of Contents
- Introduction 3
- Conceptual Framework 7
- Dynamic Targeting 7
- Potential Outcomes 7
- Information available to the planner 8
- Optimal Dynamic Targeting 9
- Static Targeting 10
- Welfare Gains from Dynamic Targeting 11
- Learning 11
- Habit Formation and Other Effects 11
- Screening 12
- Decomposition of Welfare Gains 13
- Field Experiment and Data 14
- Field Experiment 15
- Data and Summary Statistics 16
- Optimal Assignment Policy and Welfare Gains 17
- Construction of the Social Welfare Criterion 17
- Estimation: Dynamic and Static Empirical Welfare Maximization 19
- Results of the Optimal Policy Assignment 22
- Mechanism Behind the Dynamic Targeting 24
- Identification 25
- Estimation Results 28
- Conclusion 29
- Online Appendix 38
- Proof of Proposition 2.1 38
- Linear Regression Analysis of Average Treatment Effects 39
- Artificial Test Data 42
- Proof of Proposition 5.1 43
- Learning and Habit Formation Effects 44
- Additional Tables 46
