Advanced Econometrics Takeshi Amemiya
The Asymptotic Properties of the Second-Round Estimator in the Newton-Raphson Method
Ordinarily, iteration (4.4.2) is to be repeated until convergence takes place. However, if 0t is a consistent estimator of 60 such that 'ІТфі — 0O) has a proper
limit distribution, the second-round estimator в2 has the same asymptotic distribution as a consistent root of Eq. (4.1.9). In this case further iteration does not bring any improvement, at least asymptotically. To show this, consider
where в* lies between б, and в0. Inserting (4.4.6) into (4.4.2) yields
m - во)- {і - [Щ-ИаШJ} m - e°> <4A7)
1 PQ I
Тдж6} Vr дв],;
But, because under the condition of Theorem 4.1.3
we have
(4.4.9)
which proves the desired result.