提高国际货币基金组织COFER数据的分析有用性（英）2025

N O T E S&M A N U A L SN O T E S&M A N U A L S Improving the AnalyticalUsefulness of the IMF’sCOFER Data Glen Kwende and Erin Nephew Improving the AnalyticalUsefulness of the IMF’sCOFER Data Glen Kwende and Erin NephewAuthorized for distribution by Bert Kroese Cataloging-in-Publication DataIMF Library Names: Kwende, Glen, author. | Nephew, Erin, author. | International Monetary Fund, publisher.Title: Improving the analytical usefulness of the IMF’s COFER data / Glen Kwende and Erin Nephew.Other titles: Technical notes and manuals.Description: Washington, DC : International Monetary Fund, 2025. |TNM 2025-14. | Nov. 2025. | Includesbibliographical references.Identifiers: ISBN:9798229004855 (paper)9798229028370 (ePub)9798229028219 (web PDF)Subjects: LCSH: International Monetary Fund—Statistics. | Foreign exchange reserves—Statistical methods.Classification: LCC HG3881.5.I58 K8 2025 DISCLAIMER: This Technical Guidance Note should not be reported as representing the views of the IMF. The viewsexpressed in this paper are those of the authors and do not necessarily represent the views of theIMF, its Executive Board, or IMF management. Recommended citation:Kwende, Glen and Erin Nephew. 2025. “Improving the Analytical Usefulness ofthe IMF’s COFER Data.” IMF Technical Notes and Manuals 2025/14,International Monetary Fund, Washington, DC. Please send orders to:International Monetary Fund, Publication ServicesPO Box 92780, Washington, DC 20090, USATel: (202) 623-7430 | Fax: (202) 623-7201publications@IMF.orgeLibrary.IMF.orgbookstore.IMF.org Contents Executive Summary1 Introduction Methodology: Imputing the Unallocated Share of COFER5Stratified Mean Imputation6Carry-Forward and Mixed Method Imputation8 ResultsChanges to Total ReservesChanges to Allocated ReservesChanges to Unallocated ReservesChanges, by Currency Improving Analytical Usefulness and Strengthening Confidentiality18 Next Steps Annex I. Technical Background and History of the COFER Dataset Annex II. Analytical Usefulness of the COFER Dataset Executive Summary This technical note outlines a new methodology for the IMF’s Currency Composition of Foreign ExchangeReserves (COFER) data, aimed at improving its analytical value and confidentiality. The IMF’s StatisticsDepartment conducts the COFER survey quarterly and publishes aggregated data at the end of eachquarter.1Participation is voluntary except for Special Data Dissemination Standard Plus adherents, who arerequired to complete the survey and disclose their participation. COFER tracks the currency breakdown of the world’s foreign exchange reserves. At present, 149 entitiesreport data, consisting of IMF member countries, nonmember countries or economies, and other foreignexchange reserve holding entities. The data are kept strictly confidential. The IMF does not publish country-level data or breakdowns below the global total. The names of COFER reporters also remain confidentialunless the relevant authorities have granted the IMF explicit consent to disclose them. COFER is widely used by central banks, other official institutions, and the private sector. The data plays avital role in the IMF’s work, enabling deeper analysis of trends in international financial markets. In today’sfast-evolving global landscape, the accurate assessment of reserve data has become more crucial thanever—prompting growing analytical interest in the COFER dataset. Under the new methodology, staff will impute the current “unallocated” portion of COFER to provide acomplete currency composition—expressed in both dollars and shares—that accounts for 100 percent ofthe world’s foreign exchange reserves. Until now, the IMF has compiled data from COFER reporters andpublished aggregate totals that included an “unallocated” category to account for gaps caused by nonre-porting, incomplete submissions, or discrepancies with the IMF’s International Liquidity database. While thisapproach helps preserve internal consistency, it can also produce irregularities or distortions that reducethe data’s overall meaning and reliability. To address this shortcoming, staff use a multipronged statisticalapproach, leveraging the underlying currency composition in COFER to estimate each country’s total foreignexchange reserves. As a result, the new COFER dataset will no longer include an “unallocated” component. This new methodology aims to enhance both the analytical value of the COFER data and the confidentialityof reporters.By imputing missing values across the time series, underlying trends in the currency composi-tion of foreign exchange reserves become clearer. Eliminating the “unallocated” portion removes the needfor users to make assumptions—explicit or implicit—about these reserves, such as presuming that they reflectthe average composition of the “allocated” portion. It also reduces distortions in aggregate trends thatpreviously resulted from the mechanical inclusion or exclusion of countries over time. With missing datanow e