L-1 geometric ergodicity of a multivariate nonlinear AR model with an ARCH term
(2001) Statistics & Probability Letters — Vol. 51, n° 2, p. 121-130 (2001)
Files
pdfdocument.pdf
Restricted Access - Adobe PDF
- 117.59 KB
Details
- Authors
- Abstract
- In this note, the condition to ensure the L-1 geometric ergodicity of a multivariate nonlinear AR model mixed with an ARCH term (also called conditional heteroscedastic autoregressive nonlinear model) is investigated. Under some mild conditions on the white noise process with first absolute moment, a sufficient condition much weaker than that by Ango Nze (C.R. Acad. Sci. Paris 315 ser. 1 (1992) 1301-1304) is derived. As an application, the L-1 geometric ergodicity of an additive AR model mixed with a multiplicative ARCH term is studied. Our condition expands the application of the result in Ango Nze (C.R. Acad. Sci. Paris 315 ser. 1 (1992) 1301-1304) and is interesting for robust modeling when the white noise is fat-tailed with infinite variance. Some additional remarks are also made. (C) 2001 Elsevier Science B.V. All rights reserved.
- Affiliations
Beijing Institute of Information Technology Division of Mathematics Academia Sinica Institute of Systems Science
Citations
Lu, Z., & Jiang, Z. (2001). L-1 geometric ergodicity of a multivariate nonlinear AR model with an ARCH term. Statistics & Probability Letters, 51(2), 121-130. https://doi.org/10.1016/S0167-7152(00)00138-3 (Original work published 2001)