AEA Papers and Proceedings
ISSN 2574-0768 (Print) | ISSN 2574-0776 (Online)
Negative Weights Are No Concern in Design-Based Specifications
AEA Papers and Proceedings
vol. 114,
May 2024
(pp. 597–600)
Abstract
Recent work shows that popular partially linear regression specifications can put negative weights on some treatment effects, potentially producing incorrectly signed estimands. We show that this is not an issue in design-based specifications, in which low-dimensional controls span the conditional expectation of the treatment. Specifically, the estimands of such specifications are convex averages of causal effects with ex ante weights that average the potentially negative ex post weights across possible treatment realizations. This result extends to design-based instrumental variable estimands under a first-stage monotonicity condition and applies to “formula” treatments and instruments such as shift-share instruments.Citation
Borusyak, Kirill, and Peter Hull. 2024. "Negative Weights Are No Concern in Design-Based Specifications." AEA Papers and Proceedings, 114: 597–600. DOI: 10.1257/pandp.20241046Additional Materials
JEL Classification
- C21 Single Equation Models; Single Variables: Cross-Sectional Models; Spatial Models; Treatment Effect Models; Quantile Regressions
- C26 Single Equation Models: Single Variables: Instrumental Variables (IV) Estimation