Víšek, J. Á. : White's estimator of covariance matrix for instrumental weighted variables
| Author(s): | prof. RNDr. Jan Ámos Víšek CSc., |
|---|---|
| Type: | Article in collection |
| Year: | 2008 |
| Number: | 0 |
| ISSN / ISBN: | |
| Published in: | COMPSTAT 2008 Proccedings, 355 - 362. |
| Publishing place: | Porto, Portugal |
| Keywords: | Robustness, heteroscedasticity, Instrumental Weighted Variables, White estimator |
| JEL codes: | C51, C21, C23 |
| Suggested Citation: | |
| Grants: | 402/06/0408 Robustification of generalized moment method |
| Abstract: | Under heteroscedasticity of disturbances the significances of explanatory variables in a linear regression model have to be established employing the White estimator of covariance matrix of the (Ordinary) Least Squares estimator of regression coefficients. When the orthogonality condition is broken the Instrumental Variables (in econometrics, sociology, etc.) or the Total Least Squares (in natural sciences) are used to preserve unbiasedness of estimation. If moreover, data are contaminated a robust version of instrumental variables called the Instrumental Weighted Variables is to be used to cope both with the break of orthogonality condition as well as with contamination. Significance of explanatory variables (and of instruments) is to be examined by a robust version of White estimator of covariance matrix. |
| Downloadable: |
Paper in pdf |
