A causal framework for evaluating drivers of policy effect heterogeneity using difference-in-differences.

Area of Science:

• Economics
• Public Health
• Policy Evaluation

Background:

• Traditional policy evaluation methods, such as difference-in-differences (DiD), often focus on average treatment effects.
• Assessing the sources of heterogeneity in policy impacts typically lacks a robust causal framework.
• Confounding factors and neighborhood dynamics complicate the accurate evaluation of policy drivers.

Purpose of the Study:

• To present a novel framework for evaluating the drivers of policy effect heterogeneity.
• To apply this framework to understand the specific drivers of the Philadelphia sweetened beverage tax policy.
• To provide tools for assessing policy effect heterogeneity while addressing confounding and neighborhood dynamics.

Main Methods:

• Developing a framework to represent questions of effect heterogeneity under hypothetical interventions.
• Extending recent advancements in estimating causal effect curves within difference-in-differences (DiD) designs.
• Integrating methods to address practical challenges like confounding and neighborhood effects in policy evaluation.

Main Results:

• The proposed framework enables a more rigorous causal assessment of factors contributing to policy effect heterogeneity.
• The application to the Philadelphia sweetened beverage tax demonstrates the framework's utility in identifying key drivers.
• The study provides actionable insights into the complex pathways influencing policy outcomes.

Conclusions:

• The developed framework offers a significant advancement in causal inference for policy evaluation.
• It allows for a deeper understanding of why and how policies differentially impact populations.
• This approach enhances the ability of policymakers to design and implement more effective interventions.