Assuming a regression equation y=a+bx+u. Where, the point of interest is the coefficient of x "b", which is required to regress on some other variable, let's say "z". In this case, can we generate a series for the beta as "b= (y-a-u)/x ", is it statistically/econometrically appropriate to carry on with this?  or is there any other way to approach this issue. 

Looking forward to your advice. 

More Abdul Rahman Nizamani's questions See All
Similar questions and discussions