A NEW LOOK AT OPTIMAL-GROWTH UNDER UNCERTAINTY

We consider the discounted optimal growth model under uncertainty, and model the stochastic technology as a transition probability mapping i nputs into random outputs. This formulation allows for weaker assumpti ons of monotonicity and convexity on the technology (associated with s econd-order stochastic dominance and permitting some increasing return s), preserving all the conclusions of this model. Furthermore, in this framework, the differentiability of the (unique) optimal policy is ob tained in an elementary way from the smoothness of the data. Our resul ts provide support for the contention that stochastic modeling in econ omics should be more than a mere extension of the deterministic realm.