Title:Convolution and convolution-root properties of long-tailed distributions

Abstract:We obtain a number of new general properties related to long-tailed distributions that are of interest themselves and may be applied in many models that deal with 'plus' and/or 'max' operations on heavy-tailed random variables.
The classes of subexponential and long-tailed distributions are among the most important classes of heavy-tailed distributions. It is known that an infinitely divisible distribution (IDD) is subexponential if and only if its Levy measure is subexponential, see [13] . It is natural to conjecture that the same 'if and only if' result may hold for long-tailed distributions.
Recently Leipus and Siaulys [24] proved this conjecture in the 'forward' direction: an IDD is long-tailed if its Levy measure is long-tailed. In this paper, we show that the converse does not hold, in general, and support our conclusion by examples. Namely, we introduce two classes of heavy-tailed distributions that are not long-tailed and study their properties. These examples help to provide further insights and, in particular, to show that the properties to be both long-tailed and so called 'generalised subexponential' are not preserved under the convolution roots. This leads to the negative answer to a conjecture of Embrechts and Goldie [10, 12] in the class of long-tailed and generalised subexponential distributions.