118. In today’s interconnected world, where local policies and crises can have almost instantaneous global spillovers, it is critical that the IMF has access to the high-quality and timely data it needs to fulfill its mandate. In fixing inherited data problems, and trying to get ahead of the coming ones, it will be important to take into account the interaction among the Fund’s various data-related activities to sustain the needed transformation.
119. That is why, first and foremost, this evaluation recommends the design and implementation of a long-term overarching data strategy for the Fund, one that goes well beyond a data management strategy (see paragraph 110). Indeed, given that data is integral to all of the Fund’s core operations, all members of the Management team would have a role to play in advancing the Fund’s data strategy.
120. In designing such a strategy, consideration could be given to, among other elements, a redefinition—and regular update—of the Fund’s data needs; a discussion of the Fund’s role in regard to member countries’ statistical systems; and measures to ensure that the Fund keeps abreast of new developments in the statistical area. Building on the progress already made in the area of data dissemination (e.g., increasing transparency and data free of charge), the strategy could also include a road map toward the adoption of open data. In preparing such a road map, the Fund would need to ensure that it does not compromise its trusted advisor role; the confidentiality of sensitive data shared by members must be preserved and interference with operations avoided.
121. The evaluation then puts forward four recommendations aimed at addressing the most salient problems. These recommendations concern important elements of the overarching strategy, but their implementation could begin in parallel with the design of the strategy. Some of these will have budgetary implications (a precise estimate of which is beyond the purview and capacity of the IEO). Their costs should, however, be compared with the cost of maintaining the current modus operandi.
122. The current conjuncture provides a window of opportunity for change. The broader awareness of data-related problems in the aftermath of the global crisis and the much greater data challenges arising from the Fund’s reorientation toward multilateral and financial surveillance provide clear rationale for improving IMF data and statistics. At the same time, the progress made under the Fund’s new internal data management structure and the associated initiatives, together with the renewed impetus in STA toward increased cooperation with the rest of the IMF and greater internal-service orientation, offer a solid institutional foundation for transformation. Finally, technological advances provide a strong basis for sustained progress toward a strengthened statistical and data architecture.
Recommendation I: Develop a long-term strategy for data and statistics at the Fund.
This should be based on an overarching vision of how data can best support the IMF’s core operational needs, going beyond just a data management strategy. While such a strategy would likely incorporate new elements (e.g., those listed in paragraph 110 above), one of its key purposes would be to align and articulate all the initiatives already underway and provide them with a common institutional objective.
The implementation of the strategy would need strong and consistent leadership, making the business case for Fund-wide value-added of data, and should incorporate a stronger top-down component than previous efforts. A starting point would be to integrate Management oversight of STA and of the new data management structure to provide high-level strategic guidance and coordination and draw on the synergies with the ongoing work on knowledge management.
Recommendation 2: Define and prioritize the Fund’s data needs and support data provision by member countries accordingly.
The Fund should adopt a proactive approach to identify existing and emerging data issues most relevant from a global stability perspective. Then in close consultation between economists, statisticians, and member country authorities, the Fund’s data requirements should be prioritized carefully, weighing the benefits and costs—for the Fund and for member countries—of any additional data requests. The minimum data necessary for surveillance should also be kept under more frequent Board review, while the Fund’s confidentiality protocols could be clarified to the membership in order to reassure countries in their voluntary provision of data. At the same time, demands for new data could be rationalized if the Fund were to make full use of the data already available, including through more training for staff in how to effectively use new or underused analytical approaches.
The Fund should continue supporting data provision by members, including by (i) providing its well-respected capacity building—aligned with the Fund’s overarching data strategy—particularly to the more resource-constrained low-income countries; (ii) encouraging the adoption of international standards and reporting templates for all data provision to the Fund; (iii) considering a less costly alternative to the now-suspended data ROSCs; and (iv) pushing forward with the work of the Inter-Agency Group to reduce the overall burden of data reporting.
Recommendation 3: Reconsider the role and mandate of the Statistics Department.
The work of STA could be refocused toward what is needed to support the Fund’s core operations, making the provision of services to the Fund the nucleus of the department’s activity. This would entail a change in the department’s culture and organization—including increased attention to the timeliness and operational relevance of the data it manages, reallocation of resources toward activities that more directly support the Fund’s main mandate, and inclusion of more staff with Fund operational experience.
Once STA has undergone the necessary reforms, the Fund’s recently introduced data management structure could then be integrated into STA, and the (new) department would assume the role of central provider of data services to the Fund. This role could include running an integrated database, with homogenized access to all data used at the Fund, and with a full set of embedded validation checks. In the absence of a clear role and mandate, STA would be increasingly marginalized—with its efforts just focused on capacity-building and standard-setting activities—while the new governance structure would effectively undertake the provision of the statistical services the Fund needs.
Recommendation 4: Reexamine the staff’s structure of incentives in the area of data management.
Improving data management practices will require strengthening staff incentives and accountability, both personal and departmental. Among other possible measures, this could entail: ensuring that periodic, third-party assessments of compliance with guidelines are carried out; holding division/mission chiefs responsible for adherence to these guidelines; and clearly recognizing data management skills among development needs in staff’s annual performance reviews. Incentives for staff to candidly assess and discuss data issues in Article IV and FSAP reports also need to be realigned and reviewed. Rather than as a supplement that is largely ignored by country authorities and the Board, the Statistical Issues Appendix should be more fully integrated into Article IV reports. Given its limited effectiveness, the current practice of shoehorning country data into rigid categories of adequacy for surveillance could be rethought or replaced.
Recommendation 5: Make clear the limits of IMF responsibility regarding the quality of disseminated data, and clarify the distinction between “IMF data” and “official data.”
To reduce reputational risk, the IMF should make clear that it does not “endorse” the data that appear in its publications, databases, or the Dissemination Standards Bulletin Board (DSBB) and that there are limits to what it can do about quality. The distinction between “IMF data” and “official data” could be clarified, in part, by providing easy access to metadata for all IMF databases, including full annotation of data in the tables in Article IV consultation reports. While clear, easily accessible metadata would help address problems with data comparability in doing cross-country research, even more important would be the ability to present all “IMF data” in line with international standards, including those in Article IV reports; to achieve this would entail encouraging member countries to move toward adopting common reporting templates for all data they provide to the Fund. The IMF could also more closely review the accuracy of DSBB metadata, together with a willingness to remove violators. Finally, as a first step in moving toward more open data, consideration could be given to facilitating electronic access to the data and metadata included in Article IV consultation reports and IMF Working Papers.