A Note on the Distribution Functions of LIML and 2SLS Structural Coefficient Estimators in the Exactly Identified Case
Publication Type
Journal Article
Publication Date
1979
Abstract
Many alternative techniques of estimating coefficients appearing in the structural representation of econometric models have been developed. The techniques of two-stage least squares (2SLS) and limited information maximum likelihood (LIML) yield identical estimates in the exactly identified case. The derivations of the density functions of these estimators for overidentified structural equations do not formally include the exactly identified case. In this note we demonstrate that the previously obtained density functions include results consistent with exact identification as a special case. A correction to the expression for the LIML density function obtained by Mariano and Sawa is also reported.
Discipline
Econometrics
Research Areas
Econometrics
Publication
Journal of the American Statistical Association
Volume
74
Issue
368
First Page
847
Last Page
848
ISSN
0162-1459
Identifier
10.1080/01621459.1979.10481040
Publisher
Taylor and Francis
Citation
Mariano, Roberto S. and McDonald, J. B..
A Note on the Distribution Functions of LIML and 2SLS Structural Coefficient Estimators in the Exactly Identified Case. (1979). Journal of the American Statistical Association. 74, (368), 847-848.
Available at: https://ink.library.smu.edu.sg/soe_research/156
Additional URL
https://doi.org/10.1080/01621459.1979.10481040