Skip to content

Asymptotic Properties of MLE

Overview

Under regularity conditions, MLE possesses three key asymptotic properties.

Consistency

\[ \hat{\theta}_{MLE} \xrightarrow{P} \theta_0 \quad \text{as } n \to \infty \]

Asymptotic Normality

\[ \sqrt{n}(\hat{\theta}_{MLE} - \theta_0) \xrightarrow{d} N\left(0, \frac{1}{I(\theta_0)}\right) \]

Asymptotic Efficiency

The MLE achieves the Cramér–Rao lower bound asymptotically, meaning no other consistent estimator has smaller asymptotic variance.

Regularity Conditions

The above results require: (1) identifiability, (2) common support, (3) \(\theta_0\) is an interior point, (4) smoothness of the log-likelihood.