Estimation of weak ARMA models with regime changes (Q1984643): Difference between revisions

The authors study the asymptotic properties of the least squares estimator (LSE) of autoregressive moving-average (ARMA) models with regime changes under the assumption that the errors are uncorrelated. The model considered is \(X_t-\sum_{i=1}^p a_i^0(\Delta t)X_{t-i} =\epsilon_t-\sum_{j=1}^q b_j^0(\Delta t)\epsilon_{t-j}\) where the innovation process \(\epsilon_t\) is a weak white noise, i. e. a stationary sequence for which \(E\epsilon_t=0\) and \(E\epsilon_t\epsilon_t=\sigma^21_{t=t'}\) and \(\Delta_t\) is a stationary ergodic observed process with values in a finite set \(S\). Conditions are given for the consistency and asymptotic normality of the LSE. The theoretical results are illustrated by means of Monte Carlo experiments.