Asymptotic properties of a rank estimate in linear regression with symmetric non-identically distributed errors
2013 (English)In: Statistics (Berlin), ISSN 0233-1888, Vol. 47, no 6, 1160-1183 p.Article in journal (Refereed) Published
In this article, a simple linear regression model with independent and symmetric but non-identically distributed errors is considered. Asymptotic properties of the rank regression estimate defined in Jaeckel [Estimating regression coefficients by minimizing the dispersion of the residuals, Ann. Math. Statist. 43 (1972), pp. 1449-1458] are studied. We show that the studied estimator is consistent and asymptotically normally distributed. The cases of bounded and unbounded score functions are examined separately. The regularity conditions of the article are exemplified for finite mixture distributions.
Place, publisher, year, edition, pages
2013. Vol. 47, no 6, 1160-1183 p.
rank regression, symmetric heteroscedastic errors, linear rank statistics, consistency, asymptotic normality, bounded score functions, unbounded score functions
IdentifiersURN: urn:nbn:se:uu:diva-210168DOI: 10.1080/02331888.2012.688206ISI: 000324863100002OAI: oai:DiVA.org:uu-210168DiVA: diva2:661659