Infinite variance stable moving averages with long memory

We investigate the notion of long memory for infinite variance moving averages with stable non-Gaussian innovations and regularly varying coefficients. Regularly varying coefficients decay to zero like jPL(j) as j → ∞, where L is a slowly varying function. We study the asymptotic behavior of two measures of dependence, the codifference and the covariation, which are extensions of the covariance.