THE COD MODEL: SIMULATING WORKGROUP PERFORMANCE

Though the question of the determinants of workgroup performance is one of the most central in organization science, precise theoretical frameworks and formal demonstrations are still missing. In order to fill in this gap the COD agent-based simulation model is here presented and used to study the effects of task interdependence and bounded rationality on workgroup performance. The first relevant finding is an algorithmic demonstration of the ordering of interdependencies in terms of complexity, showing that the parallel mode is the most simplex, followed by the sequential and then by the reciprocal. This result is far from being new in organization science, but what is remarkable is that now it has the strength of an algorithmic demonstration instead of being based on the authoritativeness of some scholar or on some episodic empirical finding. The second important result is that the progressive introduction of realistic limits to agents' rationality dramatically reduces workgroup performance and addresses to a rather interesting result: when agents' rationality is severely bounded simple norms work better than complex norms. The third main finding is that when the complexity of interdependence is high, then the appropriate coordination mechanism is agents' direct and active collaboration, which means teamwork.