To obtain standard errors for αcorr, we calculate (8)Var(αcorr)=CAg^C′, where Ag^≡Varαg^ and (9)C≡P[1ρ2ΣGΣG^,zΣz]−1 Vg. The standard errors are the square root of the diagonal of Var(αcorr). Note that equations (7)–(9) are written in terms of population variance-covariance matrices, model coefficients, and the parameter ρ. To implement this correction, we replace each of these terms with its sample counterpart.
