Because of the inherent uncertainty, promotion of innovation critically depends on screening mechanisms to select projects. This paper studies the relationship between bureaucracy and financial constraints as two such mechanisms. The lack of commitment to hard financial constraints interferes with its ex post screening capability; ex ante bureaucratic screening is optimally chosen as a substitute. However, bureaucracy makes mistakes by rejecting promising projects and delays innovation, and the efficiency loss due to soft financial constraints increases as prior knowledge becomes worse and as research stage investment requirements become lower. In a centralized economy, bureaucracy may reduce the number of parallel projects, particularly for projects with higher uncertainties and less research stage requirements. This theory fits much of the evidence and in particular it explains why the computer industry, but not the nuclear or aerospace industries, has fared so poorly in centralized economies.

We model an organization as a hierarchy of managers erected on top of a technology (here consisting of a collection of plants). In our framework, the role of a manager is to take steps to reduce the adverse consequences of shocks that affect the plants beneath him. We argue that different organizational forms give rise to different information about managers' performance and therefore differ according to how effective incentives can be in encouraging a good performance. In particular, we show that, under certain assumptions, the M-form (multi-divisional form) is likely to provide better incentives than the U-form (unitary form) because it promotes yardstick competition (i.e. relative performance evaluation) more effectively. We conclude by presenting evidence that the assumptions on which this comparison rests are satisfied for Chinese data.