Skip to main content Skip to navigation

Mostly Harmless Simulations? Using Monte Carlo Studies for Estimator Selection

Mostly Harmless Simulations? Using Monte Carlo Studies for Estimator Selection

411/2019 Arun Advani, Toru Kitagawa and Tymon Słoczyński
working papers,behavioural economics and wellbeing
Journal of Applied Econometrics
https://doi.org/10.1002/jae.2724

411/2019 Arun Advani, Toru Kitagawa and Tymon Słoczyński

We consider two recent suggestions for how to perform an empirically motivated Monte Carlo study to help select a treatment effect estimator under unconfoundedness. We show theoretically that neither is likely to be informative except under restrictive conditions that are unlikely to be satisfied in many contexts. To test empirical relevance, we also apply the approaches to a real-world setting where estimator performance is known. Both approaches are worse than random at selecting estimators which minimise absolute bias. They are better when selecting estimators that minimise mean squared error. However, using a simple bootstrap is at least as good and often better. For now researchers would be best advised to use a range of estimators and compare estimates for robustness.

Behavioural Economics and Wellbeing

Journal of Applied Econometrics

https://doi.org/10.1002/jae.2724