Resistant estimation for simultaneous-equations models using weighted instrumental variables. (English) Zbl 0583.62095

A bounded-influence estimator is introduced for the estimation of a single structural equation in a simultaneous equations model. The estimator is a weighted-instrumental-variables estimator which can be resistant to heavy-tailed errors, or aberrant data in either the endogenous or exogenous variables.
An empirical example is provided to demonstrate the ability of this approach to identify influential observations and outliers. This work extends the authors’ previous work on regression models [see J. Am. Stat. Assoc. 77, 595-604 (1982; Zbl 0501.62062)] to the simultaneous equations models.
Reviewer: L.F.Lee


62P20 Applications of statistics to economics
62F10 Point estimation


Zbl 0501.62062
Full Text: DOI