What-if analysis through simulation-optimization hybrids

This paper proposes to improve traditional what-if analysis for policy making by a novel integration of different components. When a simulator is available, a human expert, e.g., a policy maker, might understand the impact of her choices by running a simulator on a set of scenarios of interest. In many cases, when the number of scenarios is exponential in the number of choices, identifying the scenarios of interest might be particularly challenging. We claim that abandoning this {\em generate and test} approach could greatly enhance the decision process and the quality of political actions undertaken. In this paper we propose and experiment with one approach for combining simulation with a combinatorial optimization and decision making component. In addition, we propose two alternative approaches that can reasonably combine decision making with simulation in a coherent way and avoid the generate and test behaviour.