Collaborative Research: Transparency and Misspecification in Structural Estimation

Project Outcomes Statement

The project has produced the following academic articles:
- A Model of Scientific Communication (Andrews, Shapiro). An article using economic theory and statistical decision theory to propose and develop a model of scientific communication and contrast its implications with those of a classical model of statistics.
- Transparency in Structural Estimation (Andrews, Gentzkow, Shapiro). An article defining a working notion of transparency for structural estimation and discussing approaches to achieving it.
- Causal Interpretation of Structural IV Estimands (Andrews, Barahona, Gentzkow, Rambachan, Shapiro). An article studying the effect of instrument and estimator choice on the sensitivity of causal estimates to model misspecification.

The project has advanced knowledge in the following respects:
- Providing a rigorous foundation for concerns about transparency grounded in a model of scientific communication.
- Proposing a working definition of transparency in structural estimation.
- Showing the connection between concerns for transparency and concerns about misspecification.
- Showing conditions on estimators that reduce sensitivity of causal estimates to model misspecification.

The project will benefit society by contributing to the achievement of the following outcomes:
- Improved transparency and clarity of communication of policy-relevant conclusions from empirical research in economics.
- Improved robustness to misspecification of policy-relevant conclusions from empirical research in economics.