The least weighted squares I. The asymptotic linearity of normal equations

0410954 - UTIA-B 20020168 RIV CZ eng J - Journal Article
Víšek, Jan Ámos
The least weighted squares I. The asymptotic linearity of normal equations.
Bulletin of the Czech Econometric Society. Roč. 9, č. 15 (2002), s. 31-58. ISSN 1212-074X
R&D Projects: GA AV ČR KSK1019101
Grant - others:GA UK(CZ) 255/2002/A EK /FSV
Institutional research plan: CEZ:AV0Z1075907
Keywords : the least weighted squares * robust regression * asymptotic normality and representation
Subject RIV: BA - General Mathematics

For the least weighted squares we show that normal equations may be asymptotically approximated by a form which is linear in regression parameters. The proof employs Skorohod embedding into Wiener process. The asymptotic linearity of the normal equations allows (it is done in the part II) to derive asymptotic representation of the estimator and its asymptotic normality.
Permanent Link: http://hdl.handle.net/11104/0131041