Evaluating policy feedback rules using the joint density function of a stochastic model

The practise of evaluating feedback rules in stochastic models using marginal distributions implies the loss of considerable information. It is possible to use the full joint density function. This is illustrated by analysing three rules in a large stochastic model.