tries, arranging joint missions and sharing staff, and making more effective use of the Internet to share databases and internal Websites. Though these are steps in the right direction, they may not go far enough. Assessments suffer from several other problems related to cooperation and coordination, in particular:

The somewhat divergent interests and objectives associated with the development of the instruments reviewed in this study means that they are similar but slightly different, and there is not necessarily agreement on the fiduciary protection provided in a particular country.

Ensuring country ownership of the diagnostic process and open provision and exchange of information is likely to be much harder if a development agency appears to be setting the agenda or collecting information for fiduciary purposes or to set conditions for aid disbursement.

None of the instruments provides an entirely satisfactory measure of fiduciary risk.

The potential conflict between the need for government ownership and the robustness of an assessment is especially relevant to financial accountability assessments.

Against this background, it is recommended that:

A common definition of fiduciary risk be agreed upon between governments and donors.

The relationship between the role of the instruments in evaluating fiduciary risk and contributing to long-term development goals be clarified.

Consideration be given to dividing the fiduciary and development aspects of diagnostic work into separate processes and reports and to creating a more independent process-with some element of joint ownership by donors and external quality control and validation-for carrying out assessments. Doing so would allow donors to develop stronger partnerships with countries, reduce the possibility of conflicts of interest, and strengthen the focus on development issues and capacity building.

In addition, important issues arise in the coordination of activities between the European Commission and the World Bank and IMF-particularly between EC audits and CFAAs. The evaluation work related to EC budget support has moved away from traditional ex post audits, focused

on establishing the eligibility of expenditures financed by aid, to an approach based on learning from other institutions diagnostic work, filling gaps where necessary, and conducting compliance tests and annual updates (using audit techniques) to establish whether public expenditure management systems are operating efficiently. Linked to this approach is an EC proposal to develop indicators for monitoring improvements in public financial management.

In principle there is close complementarity between EC compliance tests and CFAAs, because the CFAA is not an audit and the Bank has eschewed audit techniques. But more needs to be done to increase collaboration. For example, the European Commission and the Bank need more experience working together in the field (some joint work has already been done, especially in Africa), with case histories and lessons recorded from these experiences, and should work together with governments. Moreover, as noted, the methodology of EC compliance tests has not been fully defined and documented. The Commission should fill this gap, drawing on the experience and expertise of the consulting companies that conduct most EC audits and compliance tests-and consider how information deriving from this work can be linked with the development of performance indicators (PEFA 2003).

In addition, the Bank and IMF could seek input from the two or three largest partner development agencies in each recipient country, allowing them to influence the design of assessment work (for example, through consultations on initiating concept memorandums and terms of reference), consulting with them during assessments, and inviting them to participate in country dialogue (for example, through government and donor workshops) before assessments are finalized. Doing so would make partner agencies more inclined to rely on Bank and IMF instruments instead of conducting their own analyses, while enabling them to shape and share assessment results. As noted, the agencies involved would not relinquish their rights and responsibilities for independent judgments. But such judgments would be exercised through careful scrutiny and synthesis of other agencies analyses-without necessarily carrying out multiple analyses or duplicating efforts.

Often, though, what aid agencies view as better coordination is seen by recipient governments as ganging up. This difference in viewpoint will persist as long as assessments are top-down exercises carried out by aid agencies with only superficial participation by recipient governments. For coordination to provide the benefits of scale economies, lower transaction costs, and better advice without the threat of heavier interference and coordinated leverage, assessments must be at least partly owned by recip-

ient governments. Similarly, dialogue on public expenditure management reform is more likely to achieve its desired outcomes if governments can be reassured that assessment results are generally accepted (if not formally agreed) by the donor community.

Two final points are worth mentioning. First, donor coordination issues also occur during efforts to implement assessment recommendations. Again, high transaction costs may arise for recipient countries if the supervision efforts of multilateral and bilateral donors are poorly coordinated. Two country examples show how this outcome can be avoided. In Mozambique, bilateral agencies conduct an annual joint review as part of joint budget support to implement the actions identified in the Poverty Reduction Strategy Paper. One of the reviews main goals is to evaluate progress in improving public financial management.

Similarly, in Tanzania bilateral donors have combined their budget support under the Poverty Reduction Budget Support program. The donors have developed a matrix to assess progress in implementing the Poverty Reduction Strategy Paper, and public financial management is among the key areas assessed. Moreover, these donors are engaged in discussions with the World Bank to develop a common matrix for Poverty Reduction Budget Support and Poverty Reduction Strategy Papers, and to sponsor joint missions for evaluating progress.

Second, the OECD Development Assistance Committees Task Force on the Harmonization of Donor Practices recently defined good practices for measuring performance in public expenditure management (DAC 2003; see also annex 2). These practices include planning and conducting diagnostic reviews, ensuring quality, sharing reports with stakeholders, engaging in follow-up activities, updating reviews, and monitoring performance. Donors should consider this guidance when conducting reviews and coordinating activities.

Coordination within the World Bank

Internal coordination problems affect every large organization. But when it comes to assessment instruments, the people interviewed for this study and a review of existing guidelines indicate that the main ambiguities occur between the World Banks Financial Management network, Poverty Reduction and Economic Management (PREM) network, and Procurement network. The first is responsible for CFAAs; the second for PERs and for leadership on adjustment lending in countries where such lending is