Coordination incentives, performance measurement, and resource allocation in public sector organizations

Why are coordination problems common when public sector organizations share responsibilities, and what can be done to mitigate such problems? This paper uses a multi-task principal-agent model to examine two related reasons: the incentives to coordinate resource allocation and the difficulties of measuring performance. The analysis shows that when targets are set individually for each organization, the resulting incentives normally induce inefficient resource allocations. If the principal impose shared targets, this may improve the incentives to coordinate but the success of this instrument depends in general on the imprecision and distortion of performance measures, as well as agent motivation. Besides decreasing available resources, imprecise performance measures also affect agents' possibility to learn the function that determines value. Simulations with a least squares learning rule show that the one-shot model is a good approximation when the imprecision of performance measures is low to moderate and one parameter is initially unknown. However, substantial and lengthy deviations from equilibrium values are frequent when three parameters have to be learned.