A Discontinuity Test for Identification in Triangular Nonseparable Models
This is a joint project with Christoph Rothe and Nese Yildiz. We develop a test of the validity of the control function approach in a nonseparable model. This paper was submitted for publication and it is currently R&R at Journal of Econometrics.
A Test of Endogeneity without Instrumental Variables
This project began with my Ph.D. dissertation on testing endogeneity using discontinuities in the unobservables. This test is interesting because it can still be applied no matter whether the functional form of the model is assumed or not, and no matter the estimation technique chosen. The paper is now published at Econometrica.
Identifying Marginal Effects Using Covariates
This is a joint project with Juan Carlos Escanciano. We propose a method to identify the marginal effects of a complex variable (which can be continuous, or a vector) using a single binary instrument. This method can be implemented with simple two-stage least squares regression on Stata.
The effects of smoking on pregnancy and child development outcomes
This project began as a spin-off of the main example I used in the discontinuity test of endogeneity paper. I found strong evidence that the effect of smoking in the baby's weight at birth is strongly confounded. Ken Chay, Umair Khalil, and I are currently delving much deeper into this problem, and we will be ready to report our findings soon. Contact me if you would like more information on this project.