I'm planning to conduct an impact assessment using a cross-sectional data. So, from Endogenous Switching Regression Model (ESRM) and Propensity Score Matching (PSM), which is the most preferable model? Does Endogenous Switching Regression Model has merits over Propensity Score Matching?
The two models give different results depending on the selectivity.
@Kiprotich Kiptum, thank you for the answer. If you have used both models, which one do you recommend me ?
Consult this article in RG
Maximum Likelihood Estimation of Endogenous Switching Regression Models
- August 2004
- Stata Journal 4(3):282-289
- DOI:
- 10.1177/1536867X0400400306
- Source
- RePEc
- 📷Michael Lokshin
- 📷Zurab Sajaia
In your case, PSM could be a preferable model. Good luck
both methods measure ATT, ATU and ATE but the ESR model can correct the selection of unobservables while the PSM cannot. ESR better minimize heterogeneity and endogeneity biases. But for the sake of robustness, if you choose one, estimate the other as well.
Both models can be used for impact assessment. But the ESR model can correct the selection of unobservables, whereas the PSM cannot. It will be ideal to run both models to check for robustness. However, if you opt for PSM, the Rosenbaum bounds procedure can be employed to reduce bias and check for robustness.
Kiprotich Kiptum thank you for the answer. If you have used both models, which one do you recommend me ?
ESR will deal with the 3 types of sources of bias, hence should be superior to PSM (account for selection bias) or IPWAR (implicit endogeneity and misspecifications); both leaving out omitted heterogeneity bias.
ESR has the advantage to minimize endogenous biais and heterogeneous biais better than PSM.
Yes, because, the ESRM help us to address the endogenity and selection bias problem which the PSC can not address.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size