We consider possible measures of dependence for two symmetric alpha-stable (SαS) random variables. Some results are given which enlighten a few deficiencies of these measures. We propose a new measure which partially solve these problems. The results are illustrated by simulations

This paper contains a Lagrange multiplier test of the hypothesis that the covariance matrix of a multivariate time series model is constant over time. It is further assumed that under the alternative, the error variances are time-varying whereas the correlation remain constant over time. Under...