In simultaneous equation theory a sample is said to be undersized when it is smaller than the number of exogenous variables. The sample moment matrix of these variables is then singular, but equations can nevertheless be estimated when different moments (hybrid moments based on the...

Canonical correlation theory is extended so that each variate in each of the two observed sets of variates is assigned a unique information-theoretic measure of its contribution to the total amount of information that either set provides about the other.

Root-mean-squared errors and asymptotic standard errors of ML coefficient estimates are compared for equation systems of different size, using normal and discrete bootstrap. It appears that exploiting normality does not yield any gain. Suggestions are made for correcting the downward bias of...

The maximum entropy covariance matrix is positive definite even when the number of variables p exceeds the sample size n. However, the inverse of this matrix can have stability problems when p is close to n, although these problems tend to disappear as p increases beyond n. We analyze such...

This letter concerns the maximum entropy (ME) distribution proposed by Theil and Laitinen (1980); it summarizes some of the results obtained with this distribution and its explores the conditioning of the ME covariance matrix as the number of variables increases.

For allocation models consisting of n = 10, 6 and 4 equations, constrained generalized least squares coefficient estimates are compared with those obtained from the minimum information (MI) criterion in the sense of statistical information theory. The MI estimates are more efficient for n = 10....

A simple measure from statistical information theory is proposed to measure the contribution of each independent variable in a multiple regression.

Simulations based on two sets of data are used to assess the impact of combining equations on the accuracy of parameter estimates and their asymptotic standard errors.