The chain ladder method is one of the most common methods of loss reserving. From the practical point of view, there are two arguments in favour of the chain ladder method: The method is simple, and it exploits all data from the run-oft triangle. These arguments, however, do not settle the question whether or not the chain ladder method is preferable to other methods of loss reserving. The quality of the chain ladder method depends on the stochastic mechanism, or stochastic model, generating the data. In the present paper we consider the model of Schnaus. Under the assumptions of this model, we show that the family of all chain ladder predictors is, in a reasonable sense, superior to many other families of predictors. This result is of interest with regard to asset liability management.