The short- and long-run comparative statics of uncertainty

If a two-input stochastic reward function is supermodular, then a first-degree stochastic shift increases both inputs, and the more variable factor is more responsive in the long run. These results hold for second-degree dominance under additional curvature conditions