Conceptual frameworks and experimental design in simultaneous equations☆
Section snippets
Simultaneous equations and experimental design
The classical linear simultaneous equations model has enjoyed renewed interest of late as a consequence of the problems associated with inference in weakly-identified models; see, for example, the papers discussed by Chesher et al. (2007). Several authors have addressed the specific problem of inference on the coefficient of an endogenous regressor in a structural equation and various suggestions have been made about how to proceed in such circumstances. Using invariance principles and similar
Discussion
Let us state at the outset that we are not suggesting that either ED1 or ED2 is incorrect. Nevertheless, when faced with alternative specifications one is required to make choices. A preference for the specification that is comprised of both (i) a structural equation and (ii) the complete reduced form rather than just a subset of the reduced form, might be justified on two grounds. First, the overall reduced form is an unrestricted regression and so the accompanying conditional distribution of [
References (8)
- et al.
Endogeneity, instruments and identification
Journal of Econometrics
(2007) - et al.
Approximating the distribution of the instrumental variables estimator when the concentration parameter is small
Journal of Econometrics
(2007) - et al.
Estimation of the parameters of a single equation in a complete system of stochastic equations
Annals of Mathematical Statistics
(1949) - et al.
Optimal two-sided invariant similar tests for instrumental variables regression
Econometrica
(2006)
Cited by (4)
On the power of the conditional likelihood ratio and related tests for weak-instrument robust inference
2023, Journal of EconometricsPrediction in linear index models with endogenous regressors
2015, Stata JournalInference in the presence of weak instruments: A selected survey
2012, Foundations and Trends in Econometrics
- ☆
We thank both Adrian Pagan and an anonymous referee for useful feedback on an earlier version of this note, and acknowledge the support of the Australian Research Council Grant DP0771445.