We investigate the possible inefficiency of investment when the manager has better information than the principal. We assume that projects can be grouped into two categories based on their information characteristics of soft and hard information. We consider a case in which the manager faces competition for appointment at the end of term. We consider several cases in which the competitor's ability is known, or proxied by the performance of other managers, or is uncertain. We find that hard information projects are preferred to soft information projects and that the efficiency in project selection critically depends on the competitor's ability. Interestingly, information asymmetry may be socially beneficial, because it can provide the manager with incentives to select high-quality hard-information projects. Welfare improvement can be made by a budget change that is in the opposite direction to the optimum. Several empirical implications of the theory are discussed.