A functional conditional symmetry test for a GARCH-SM model: Power asymptotic properties

Abstract In this paper we consider the empirical process of the errors appearing in a generalized autoregressive conditional heteroskedastic with stochastic mean (GARCH-SM) model. Various functional tests of conditional symmetry can be built on the basis of the limiting distribution of this process. In particular, a Cramér–von Mises-type test is considered. Its theoretical power is studied under fixed and local alternatives. Using the Karhunen–Loève decomposition, the limiting law of the latter is approximated by a chi-square distribution under both null and alternative hypotheses. The local power under a sequence of alternatives is also computed.