9.4 Conclusion
From a variety of different perspectives, and utilising range of models with different assumptions and setups, the literature has focused on the trade-offs between dispersion and coordination. In particular, it has identified the fact that decentralisation has the advantage of harnessing information about the identification of projects, and motivating agents by maximising the private benefits they obtain from the ability to choose projects. However, decentralisation has the disadvantage that agents may choose excessively expensive projects (since yardstick competition between different agents to have their projects accepted is more limited), and may veto projects that require coordination to maximum social welfare because those projects are less personally attractive. This is important because, no matter how efficient the implementation, social welfare will not be improved, if the wrong projects are often being implemented under the decentralised structure.
Decentralisation will be least costly when the moral hazard problem in the selection of projects is low, and the path of technical change is uncertain (so that allowing different divisions to pursue their own strategies provides benefits of diversification). It is also advantageous for services that are, or could be, commercially provided, since in that case it may be possible to introduce competition for state-provided services that will promote efficiency in the use of resources. Integration has the advantage of providing a mechanism for controlling moral hazard in the selection of projects, and providing senior managers with the ownership rights to implement some of the good ideas identified by individual workers or teams. Integration also has the advantage that the chief executive of the integrated firm will have less personal empathy with individual divisions or projects, thus more effectively focusing resources on those projects proposed by individual units that have the highest potential to increase value.
In the public sector, this may be translated as suggesting that specialised departments and agencies are better at identifying projects (have more local knowledge), worse at selecting which projects to invest in, and more efficient in implementing the projects that are chosen. If the identification of projects ex ante is not critical (for example, if this is driven by ministers, rather than by the specialised knowledge of civil servants), but decisions about which projects to invest in are critical, then large hierarchical government departments with multiple divisions may be preferred. A model in which different agencies have ownership rights in respect of different projects and issues is likely to be the worst of all possible worlds (providing neither the benefits of coordination nor the benefits of decentralisation).
In general, centralisation of decision-making is preferred when the gains to coordination are large. The more the public sector focuses on policy, as opposed to service delivery and asset ownership, then the more likely it is that coordination will be important. In the New Zealand public sector, coordination of performance and resource allocation by central monitoring agencies has proved difficult, because those monitoring agencies are not residual claimants in the activities that they are charged with monitoring. Thus it is likely that a smaller number of larger public-sector organisations will be preferred, because they may provide individual units with incentives to develop specialised expertise and collect information, while also providing high level coordination of the investment and resource allocation decisions made.
