[R] lm.ridge in library(MASS) produces inconsistent parameter estimates as compared to matrix algebra

[Georg D. Blind]

Dear all,

using the hkb estimator obtained from lm.ridge in the below equation
(Formula 6 from this article:
http://www3.interscience.wiley.com/cgi-bin/fulltext/122484280/PDFSTART

beta^hat(k) = ((x'x + kI)^-1)x'y, where
x matrix of independent variables
y vector of dependent variables
k hkb estimator
I identity

then I am getting smaller coefficient estimates than from within 
lm.ridge. The difference is not due to lm.ridge$scales.

Any hints as to this?
Thank you and kind regards.
Georg

__________________________________________________

Georg Blind M.Sc. (Econ.), M.A. (Japanese Studies)
Email: georg.blind at gmx.net