In this article we propose a simple method of identifying, at an earlier st
age of analysis, the nested structure among the coefficient matrices in mul
tivariate regression models. When the limiting distribution of the estimato
rs of the coefficient matrices are jointly normal, the Wald type statistics
based on the proposed method is asymptotically a chi-squared random variab
le. A numerical example that arises in cointegration analysis is provided t
o illustrate the method and a small simulation study is provided to illustr
ate its effectiveness.