1: \begin{abstract}

2: In this paper, we study parametric nonlinear regression under the

3: Harris recurrent Markov chain framework. We first consider the

4: nonlinear least squares estimators of the parameters in the

5: homoskedastic case, and establish asymptotic theory for the proposed

6: estimators. Our results show that the convergence rates for the

7: estimators rely not only on the properties of the nonlinear regression

8: function, but also on the number of regenerations for the~Harris

9: recurrent Markov chain. Furthermore, we discuss the estimation of the

10: parameter vector in a conditional volatility function, and apply our

11: results to the nonlinear regression with $I(1)$ processes and derive an

12: asymptotic distribution theory which is comparable to that obtained by

13: Park and Phillips [\textit{Econometrica} \textbf{69} (2001) 117--161].

14: Some numerical studies including simulation

15: and empirical application are provided to examine the finite sample

16: performance of the proposed approaches and results.

17: \end{abstract}\