This Note studies the properties of the random coefficient autoregressive m
odels and the Markov plus i.i.d. noise models. We provide tractable suffici
ent conditions that simultaneously imply strict stationarity, finiteness of
higher-order moments, and beta-mixing with geometric decay rates. These re
sults are applied to GARCH(p, q) and EGARCH(1, 1) models. (C) 2000 Academie
des sciences/Editions scientifiques et medicales Elsevier SAS.