Perfect and High Multicollinearity

Gauss-Markov theorem’s second condition

A second condition of the Gauss-Markov theorem is that the independent variables can’t be perfectly correlated. For illustration, we provide an example. Suppose we create a new variable, open4, which simply equals four times the open variable, making the two variables perfectly correlated. For illustration, we reestimate the original model twice—first with open entered before open4 and then in reverse order—using the lm() function. The R code and output are as follows: