Multicollinearity and maximum entropy estimators

Multicollinearity hampers empirical econometrics. The remedies proposed to date suffer from pitfalls of their own. The ridge estimator is not generally accepted as a vital alternative to the ordinary least−squares (OLS) estimator because it depends upon unknown parameters. The generalized maximum entropy estimator depends upon subjective exogenous information. This paper presents a novel maximum entropy estimator that does not depend upon any additional information. Monte Carlo experiments show that it is not affected by any level of multicollinearity and dominates the OLS estimator uniformely. The same experiments provide evidence that it is asymptotically unbiased and its estimates are normally distributed. Quirino Paris is a professor at the University of California, Davis and a member of the Giannini Foundation. I am indebted to Michael R. Caputo and Art Havenner for stimulating discussions that have improved an earlier version of this paper. All errors are mine. Citation: Paris, Quirino, (2001) "Multicollinearity and maximum entropy estimators." Economics Bulletin, Vol. 3, No. 11 pp. 1−9 Submitted: July 29, 2001. Accepted: August 17, 2001. URL: http://www.economicsbulletin.com/2001/volume3/EB−01C20002A.pdf