Balancing Weights For Causal Effects With Panel Data: Some Recent Extensions To The Synthetic...
summary
Presents concrete extensions to synthetic control: change the optimization to trade off separate versus pooled balancing with a tuning parameter nu, and augment weighting with a simple outcome model (intercept or ridge) to produce an augmented synthetic control. Implements these in the R package augsynth, applies to 33 state adoptions (example: Ohio, New York) and to teachers’ collective bargaining, and finds no robust 12% spending jump claimed earlier — the methods improve pre-treatment balance and reduce overfitting.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 2:25Panel Data
- 5:00The Synthetic Control Method
- 10:55Mandatory Collective Bargaining Laws
- 14:40Agenda
- 15:25Ohio
- 16:35Synthetic Control
- 21:00Balancing Averages
- 24:55Optimization Problem
- 25:55Results
- 26:35Outcome Model
- 28:05Synthetic Controls
- 28:35Crosssectional Data
- 29:30Two Approaches
- 31:35Wrapping Up
related talks
39mCausal Effect Estimation in Practice: Lessons Learned from E-commerce & Banking
23mUber's Synthetic Control
13mAlgorithmic Advances For The Design And Analysis Of Randomized Control Trials
31mA Gentle Introduction To Causal Inference
11mCausal Inference For Socio-Economic & Engineering Systems
34mAlmost Perfect: A Benchmark on Algorithms for Quasi-Experiments
41mKeynote: Personalization, Bandits, and Causal inference
36mWhat-if- Causal reasoning meets Bayesian Inference