On the Nonparametric Identification of Nonlinear Simultaneous Equations Models: Comment on B. Brown (1983) and Roehrig (1988)
16 Pages Posted: 8 Jun 2006
Date Written: April 19, 2006
Abstract
This note revisits the identification theorems of B. Brown (1983) and Roehrig (1988). We describe an error in the proofs of the main identification theorems in these papers, and provide an important counterexample to the theorems on the identification of the reduced form. Specifically, the reduced form of a nonseparable simultaneous equations model is not identified even under the assumptions of these papers. We provide conditions under which the reduced form is identified and is recoverable using the distribution of the endogenous variables conditional on the exogenous variables. However, these conditions place substantial limitations on the structural model. We conclude the note with a conjecture that it may be possible to use classical exclusion restrictions to recover some of the key implications of the theorems in more general settings.
Suggested Citation: Suggested Citation
Do you have negative results from your research you’d like to share?
Recommended Papers
-
Identification and Estimation of Triangular Simultaneous Equations Models Without Additivity
By Guido W. Imbens and Whitney K. Newey
-
Panel Data Estimators for Nonseparable Models with Endogenous Regressors
By Joseph G. Altonji and Rosa L. Matzkin
-
Nonparametric Censored and Truncated Regression
By Arthur Lewbel and Oliver B. Linton
-
By Jean-pierre Florens, James J. Heckman, ...
-
Estimating Derivatives in Nonseparable Models with Limited Dependent Variables
By Joseph G. Altonji, Taisuke Otsu, ...
-
Estimating Derivatives in Nonseparable Models with Limited Dependent Variables
By Joseph G. Altonji, Hidehiko Ichimura, ...
-
Estimating Derivatives in Nonseparable Models with Limited Dependent Variables
By Joseph G. Altonji, Hidehiko Ichimura, ...
-
Estimating Features of a Distribution from Binomial Data
By Arthur Lewbel, Oliver B. Linton, ...
-
Effective Nonparametric Estimation in the Case of Severely Discretized Data