Dear Allstaters, I am looking for references, papers that show
(i) how to run linear regressions
(ii) without assuming that errors are normally distributed
(iii) that show how to compute exact (ie correct) significance levels
(so not via asymptotic theory).
My reference list to date is very short:
(1) Dufour & Hallin (JASA 1993): one covariate, errors distributed
symmetrically
(2) Schlag (2008) Universitat Pompeu Fabra working paper 1097: one
covariate, bounded dependent variable
http://www.econ.upf.edu/docs/papers/downloads/1097.pdf
and our newest paper
(3) Gossner & Schlag (2010) Universitat Pompeu Fabra working paper 1212:
many covariates, bounded dependent variable
http://www.econ.upf.edu/docs/papers/downloads/1212.pdf
please help. i will summarize the responses, thanx, karl
--
---------------------------------------------------------------------
Karl Schlag
Professor Tel: +34 93 542 1493
Dept. of Economics and Business Fax: +34 93 542 1746
Universitat Pompeu Fabra email: [log in to unmask]
Ramon Trias Fargas 25-27 NEW: www.econ.upf.edu/~schlag/
Barcelona 08005, Spain room: 20-221 Jaume I
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|