In the recent literature, it is shown that a recursive set up for risk
averse dynamic stochastic programming problems ensures time consis-
tency of the generated optimal policies. However, a lack of suitable eco-
nomic interpretation for this complex objective function is the main reason
why this formulation is not commonly used in practical applications. In
this paper, we develop a clear economic interpretation for this recursive
objective function as the certainty equivalent w.r.t. the time consistent
dynamic utility generated by one period preference functionals. In order
to motivate this modeling choice, we use a CVaR based portfolio selec-
tion problem to show some practical consequences of a time inconsistent
optimal policy and propose a time consistent alternative. We use a nu-
merical example to compare those optimal solutions and to illustrate our
economic interpretation.
|