A Model of Comparative Statics for Changes in Stochastic Returns with Dependent Risky Assets.

In this article, we show how the order of Linear Stochastic Dominance proposed by Gollier (1995) can be applied to situations with dependent risky assets. This order was shown to be the least constrained necessary and sufficient condition to guarantee that all risk-averse agents reduce their risky position when an increase in risk is imposed. This was done in a model with only one source of risk, as in the standard portfolio problem with one safe asset and one risky asset. We obtain the necessary and sufficient condition for a change in the joint distribution of returns to yield an unambiguous comparative statics result when the two assets are risky. We show in particular that the concept of Linear Stochastic Dominance is sufficient to generate the desired result. These results are linked to existing sufficient conditions in the one-safe-one-risky-asset model, as the condition of strong increase in risk or the monotone likelihood ratio order. They are also compared to those in models where restrictions are on the set of concave utility functions. Copyright 1996 by Kluwer Academic Publishers