***multicolinearity

use longley_data.dta, clear
reg employment military popsizeyear ntpri cesgnp essm if popsizeyear>=1947 & popsizeyear<=1961
reg employment military popsizeyear ntpri cesgnp essm if popsizeyear>=1947 & popsizeyear<=1962

***in the second regression you are just including one additional observation

***notice how the coefficients chang between one regression and the other; also the high standard
***errors of some variables, regardless of a high R-squared

***this are clearly symptoms that assumption 2 is violated, i.e. X'X doesn't have full rank k