I have a model y=n1p1+n2p2+n3*p3+noise. I know that p1,p2,p3~normal(mu,var-cov). I observe only y,n1,n2,n3. How do I estimate the var-cov matrix of p? I know how to estimate mu. These are only the estimates from OLS of y on n1,n2,n3.
I derived the first order conditions from the MLE. But to solve them I need to use the Newton-Raphson method. It is easier said than done, because the Jacobian is very very complicated.