M. Mangeas et Jf. Yao, ON LEAST-SQUARES ESTIMATION FOR NONLINEAR AUTOREGRESSIVE PROCESSES, Comptes rendus de l'Academie des sciences. Serie 1, Mathematique, 324(4), 1997, pp. 471-474
Assuming the stability of a nonlinear autoregressive process, we give
simple conditions ensuring strong consistency, asymptotic normality an
d a law of the iterated logarithm for the least squares estimator. The
last result yields an almost sure identification of the the model usi
ng a suitably penalized contrast.