Property claim services (PCS) provides indices for losses resulting from ca
tastrophic events in the US. In this paper, we study these indices and take
a closer look at distributions underlying insurance claims. Surprisingly,
the lognormal distribution seems to give a better fit than the Paretian one
. Moreover, lagged autocorrelation study reveals a mean-reverting structure
of indices returns. (C) 2000 Elsevier Science B.V. All rights reserved.