Laboratory experiments on the efficiency of decomposition principles in decentralized planning

We conducted laboratory experiments on the efficiency of three different coordination procedures with 64 students of finance at the University of Kiel in February 1982. In this paper we focus on the contrast between CP1 - the coordination within a conference - and CP 3 - the coordination by means of a decomposition algorithm. Basically, CP3 performs better than CP1 as regards the quantitative results. It has, however, shortcomings on the motivational side, which is one dimension in our concept for measuring the efficiency of alternate coordination procedures. The results discussed above suggest that further effort should be devoted to integrating people into the decomposition-based coordination as much as possible. This promises to pay off not only in an improved motivation but also in an improvement in the "technical results", i.e. the benefit-cost relation. The results with CP2^rthe coordination by man and machine (decomposition algorithm) strongly indicate: By integrating the managers, near-optimum solutions can be achieved in fewer iterations. Additionally, the level of decision-making autonomy has been perceived to come closer to that in the coordination within a conference. As regards methodology, laboratory experiments on the combination of man and machine in decision-making have proved to be a useful device of research in empirical decision theory.