The chain ladder method is one of the most common methods for reservin
g, and various attempts have been made to justify it in a stochastic m
odel. A particularly interesting model was proposed by Mack (1993, 199
4a,b): Under the assumptions of his model, Mack proved that the chain
ladder predictors of non-observable aggregate claims are unbiased, and
Schmidt and Schnaus (1996) proved that the chain ladder predictor for
the first non-observable calendar year is also optimal in the sense t
hat it minimizes expected squared error loss over a wide class of unbi
ased predictors. In the present paper, it is shown that optimality fai
ls for the chain ladder predictor for the second non-observable calend
ar year. (C) 1997 Elsevier Science B.V.