The MD estimator (see Section 6.7) uses the restriction g ( ) = 0

The MD estimator (see Section 6.7) uses the restriction $\pi -\mathbf{g}(\mathit{\theta })=\mathbf{0}$. Suppose more generally that the restriction is $\mathbf{h}(\theta ,\pi )=\mathbf{0}$ and we estimate using the generalized MD estimator that minimizes $Q_N(\theta )=\mathbf{h}(\theta ,\hat{\pi }{)}^{\mathrm{\prime }}{\mathbf{W}}_N\mathbf{h}(\theta ,\hat{\pi })$. Adapt (6.68)(6.70) to show that (6.67) holds with ${\mathbf{G}}_0=\mathrm{\partial }\mathbf{h}(\mathit{\theta },\mathit{\pi })/{\mathrm{\partial }\mathit{\theta }|}_{{\mathit{\theta }}_{0,},{\pi }_0}$ and $\mathrm{V}[\hat{\mathit{\pi }}]$ replaced by ${\mathbf{H}}_0^{\mathrm{\prime }}\mathrm{V}[\hat{\mathit{\pi }}]{\mathbf{H}}_0$, where ${\mathbf{H}}_0=\mathrm{\partial }\mathbf{h}(\mathit{\theta },\mathit{\pi })/{\mathrm{\partial }\mathit{\pi }|}_{{\theta }_0,{\pi }_0}$.