In a linear model \( {y_i} = {x'_i}\beta + {u_i} \) with E(ux) = 0, where ß is a k × 1 parameter vector of interest, u is the error term, x is a k × 1 regressor vector, and \( \left( {{{x'}_i},{y_i}} \right) \) are iid, the least squares estimator (LSE) for ß is obtained by minimizing
$$ \left( {1/N} \right){\sum\limits_i {\left( {{y_i} - {{x'}_i}b} \right)} ^2} $$
with respect to (wrt) b. LSE can also be viewed as the solution of the first-order (moment) condition of the minimization
$$ \left( {1/N} \right)\sum\limits_i {{x_i}\left( {{y_i} - {{x'}_i}b} \right)} = 0. $$


Asymptotic Property Moment Condition Little Square Estimator Seemingly Unrelated Regression Linear Hypothesis 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer Science+Business Media New York 1996

Authors and Affiliations

  • Myoung-jae Lee
    • 1
  1. 1.Department of EconometricsTilburg UniversityTilburgThe Netherlands

Personalised recommendations