Do Numerical Simulation and Optimization Results Improve Management? Experimental Evidence

Models of different forms are the core tool of economic analysis. The advises econo-mists give decision makers may represent general insights derived from stylized models. Alternatively, economists may construct complex large scale models to simu-late the consequences of different policy option, often without any definite conclusion as to what option is best. Decision makers are likely to blend the results from numerous analyses, focusing on different facets of the real world. We asked the question: do decision tools improve practical decision making? This ques-tion differs markedly from the more usual academic question: what is the quality of a model? A laboratory experiment was used to answer the question. The case was quota setting in a two-species fishery in the Barents Sea. A simplistic optimization model was chosen to reflect economic literature on two-species management under uncertainty. A set of two one-species simulation models was used to represent biological single-species models used to make forecasts. In total 64 students were asked to manage a virtual fishery with or without access to the tools. We find that numerical advises from both the optimization and the simulation tool improve management. When applied together, the tools strengthen each other, i.e. they are complements rather than substitutes. One reason for this is that the two tools had their strengths on different aspects of the management problem. Subjects did not follow advises closes, however, failed to adjust fully for weaknesses of the tools. Without a tool to suggest targets for management (the optimization tool), we found that historical observations tend to serve as targets. The participants came up with severely biased estimates of the value of both tools. The project was funded by the Research Council of Norway.