Estimation by nonlinear regression of the parameters for the stationary and invertible autoregressive moving average (ARMA) model with mixing or martingale difference errors is considered. Simple proofs of consistency and asymptotic normality for the nonlinear least squares estimator are given by exploiting results from nonlinear estimation theory and mixing and mixingale theory.
|Number of pages||13|
|Journal||Communications in Statistics - Theory and Methods|
|State||Published - 1 Jan 1993|
- Large sample properties
- Missingale theory
- Nonlinear estimation