AID"s experience with decentralization and the delegation of project review and approval

promoted decentralization and the redelegation of authority to its field offices as a means of increasing efficiency. This study examines how well A.I.D."s Washington-based Bureaus have succeeded in delegating authority and describes their project review and approval systems. The study found that the Bureaus have developed slightly different systems for project and nonproject review, depending on the management conditions in their region and the nature of the assistance. All three geographic Bureaus use the new project descriptions contained in action plans to decide whether to delegate PID and/or Project Paper approval to the Missions. The Bureau for Latin America and the Caribbean delegates many more PID"s to the field than do the Africa or Asia/Near East Bureaus, and compensates for this by requiring more detailed action plans and project implementation summaries. All three Bureaus selectively choose to approve Project Papers in A.I.D./Washington when they have concerns about the novelty of the project concept or the difficulty or sensitivity of the policy issues involved. The study"s major conclusions are as follows. (1) Several informants indicated that A.I.D./W reviews have deteriorated in quality, partly because lower level staff are leading the issue and approval meetings and often failing to maintain the control needed for constructive dialogue. (2) A.I.D./W is generally not concerned with improving the technical design of a project; rather it is concern for policy that makes A.I.D./W unlikely to redelegate its remaining approval authority. (3) There is no significant difference in quality between design documents approved in Washington and in the field. (4) Except in cases of new, particularly complex, or controversial policy issues, the field is in a better position than A.I.D./W to design projects responsive to local conditions. Finally, in a broader context, the study findings suggest that A.I.D."s programming process, with its historical emphasis on design documentation, fails to provide the kind of information and incentives needed to effectively manage program performance.