Graph regularization improves stability of synthetic control estimates
Graph-Regularized Synthetic Control for Stable Counterfactual Estimation under Donor Redundancy
Machine LearningArtificial Intelligence
Summary
Synthetic control methods try to estimate what would have happened to a treated unit if it had not been treated by mixing other similar untreated units called donors. When there are many donors that behave alike, this mixing can be unstable and hard to interpret. The authors introduce Graph-Regularized Synthetic Control, which encourages donors that behave similarly to have similar weights, making the final mix more stable and meaningful. Their method keeps good accuracy while simplifying the interpretation of which donors matter most in the estimate.
What this means in practice
- •For policy analysts: Provide more stable and interpretable estimates of policy effects when multiple similar comparison regions exist.
- •For autonomous vehicle engineers: Improve counterfactual trajectory predictions by weighting similar driving scenarios more coherently.
Authors
Yanlin Zhang, Sungyong Chung, Alireza Talebpour
Abstract
Synthetic control methods estimate counterfactual outcomes by weighting untreated donor units to match a treated unit before treatment. In recent applications, however, the donor pool can be large, highly correlated, and short in pre-treatment history, yielding many near-equivalent solutions with unstable and difficult-to-interpret donor weights. We propose Graph-Regularized Synthetic Control (Graph-SC), a convex estimator that augments the classical pre-treatment matching objective with a graph Laplacian penalty. The graph encodes similarity among donor trajectories, encouraging coherent weights over structurally related donors while preserving the transparency of synthetic controls. We provide theoretical results characterizing the bias-variance tradeoff induced by graph regularization and establish efficient convergence guarantees with an accelerated proximal gradient method. Across a controlled simulation under donor redundancy, the California Proposition 99 benchmark, and observational counterfactual analyses on the Waymo Open Motion Dataset, Graph-SC improves weight stability, reduces sensitivity to redundant donor pools, and maintains competitive pre- and post-treatment accuracy relative to classical and regularized synthetic control baselines. Our results suggest that graph regularization is most valuable not as a universal accuracy booster, but as a principled mechanism for selecting stable and interpretable counterfactuals from near-equivalent synthetic controls with comparable pre-treatment fit.