Estimation and Testing in Truncated and Nontruncated Linear Median-Regression Models.
Abstract
A number of important recent advances in econometric theory are related to the methods of truncated regression model - the regression model in which the range of the dependent variable is restricted to some interval of (-infinity, infinity), usually the non-negative half-line, such as the income of an individual. Powell used the L sub 1-norm criterion with some modifications in estimating the regression coefficients in truncated linear models. He proved the consistency and asymptotic normality of his estimates under a set of conditions. On the other hand, Nawata's paper uses the ordinary L sub 2-norm (least square) criterion, along with a grouping and adjustment of the observed data. In his view, his method has the merit of easy computation compared with the method of Powell. This paper borrows the basic idea of Nawata in grouping and adjusting the observed data. But the authors make simplifications in the procedure of grouping, which enables us to make substantial extensions of the results of Nawata's paper under weakened conditions. Keywords: Linear median regression; Truncated regression; Parameters; Linearity.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1986
- Accession Number
- ADA186317
Entities
People
- Paruchuri R. Krishnaiah
- X. R. Chen
Organizations
- University of Pittsburgh