Testing for changes heavy index under infinite variance observations

Abstract

This paper studies how to detect structural change characterized by a change in heavy index of the infinite variance observations. In particular, we are interested in a process where the heavy index changing from a bigger to a less or vice versa. A ratio type test is adopted and its asymptotic distribution under null hypothesis of have no index break approximates a ratio of a stable process. A Monte Carlo study shows that our test has reasonably good size and power properties in finite samples.