Stability results for nonlinear vector autoregressions with an application to a nonlinear error correction model

This paper improves previous sufficient conditions for stationarity obtained in the context of a general nonlinear vector autoregressive model with nonlinear autoregressive conditional heteroskedasticity. The results are proved by using the stability theory developed for Markov chains. Stationarity, existence of second moments of the stationary distribution, and useful mixing results are obtained by establishing appropriate versions of geometric ergodicity. The results are applied to a nonlinear error correction model to obtain an analog of Granger's representation theorem.