Least Squares with Error in Variables

by   David A. Hirshberg, et al.

Error-in-variables regression is a common ingredient in treatment effect estimators using panel data. This includes synthetic control estimators, counterfactual time series forecasting estimators, and combinations. We study high-dimensional least squares with correlated error-in-variables with a focus on these uses. We use our results to derive conditions under which the synthetic control estimator is asymptotically unbiased and normal with estimable variance, permitting inference without assuming time-stationarity, unit-exchangeability, or the absence of weak factors. These results hold in an asymptotic regime in which the number of pre-treatment periods goes to infinity and the number of control units can be much larger (p ≫ n).



page 1

page 2

page 3

page 4


Regression-adjusted average treatment effect estimates in stratified and sequentially randomized experiments

Stratified and sequentially randomized experiments are widely used in fi...

No-harm calibration for generalized Oaxaca-Blinder estimators

In randomized experiments, linear regression with baseline features can ...

An alternative to synthetic control for models with many covariates under sparsity

The synthetic control method is a an econometric tool to evaluate causal...

Inference for Batched Bandits

As bandit algorithms are increasingly utilized in scientific studies, th...

Do We Exploit all Information for Counterfactual Analysis? Benefits of Factor Models and Idiosyncratic Correction

The measurement of treatment (intervention) effects on a single (or just...

Distributional Counterfactual Analysis in High-Dimensional Setup

In the context of treatment effect estimation, this paper proposes a new...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.