What is the asymptotic distribution of the Wald statistic $W = (\hat{\theta} - \theta_0)^T [I(\hat{\theta})](\hat{\theta} - \theta_0)$ as $n \to \infty$ under the null hypothesis $H_0: \theta = \theta_0$, where $I(\theta)$ is the observed Fisher Information matrix?