Michael Lechner (St. Gallen/UWashington)

"Machine Learning Estimation of Heterogeneous Causal Effects: Empirical Monte Carlo Evidence"

Nov 02, 2018
from 12:10 PM to 01:30 PM

1131 Social Science and Humanities Gold room


We investigate the finite sample performance of causal machine learning estimators for heterogeneous causal effects at different aggregation levels. We employ an Empirical Monte Carlo Study that relies on arguably realistic data generation processes (DGPs) based on actual data. We consider 24 different DGPs, eleven different causal machine learning estimators, and three aggregation levels of the estimated effects. In the main DGPs, we allow for selection into treatment based on a rich set of observable covariates. We provide evidence that the estimators can be categorized into three groups. The first group performs consistently well across all DGPs and aggregation levels. All estimators in this group take explicitly account of selection into the treatment and the outcome process. The second group shows competitive performance only for particular DGPs. The third group is clearly outperformed by the other estimators.



Seminar is open to the public, space is limited.
Filed under: Econometrics