I. External Debt Guide

3. The 2000 Capital Flows Conference showed widespread support for continuing to develop methodological frameworks that all economies could use. These statistical standards lay the ground for comparability of data across countries and consistency of approach between related data series. The recently completed External Debt Statistics: Guide for Compilers and Users (External Debt Guide) is, I believe, a landmark in this field.¹

Box 1. Conference on Capital Flows and Debt Statistics—Can We Get Better Data Faster? February 23–24, 2000 The IMF, in cooperation with the Financial Stability Forum's Working Party on Capital Flows, brought together some 120 senior-level data users, policymakers, and data compilers to consider data on capital flows and debt. The following views emerged: Overview: Policymakers, data users, and a few compilers stressed the need for more comprehensive and timely capital flow and external debt statistics. Compilers from industrial countries and offshore centers generally viewed disseminating such data as a lower priority, reflecting national statistical priorities, concerns about resources, compilation difficulties, and respondent burden. Some compilers from industrial countries noted that service industry statistics might have higher national priority. National (debtor-source) measures of external debt: Views differed about whether all countries, even those in net external creditor positions, needed to disseminate such data. Methodological frameworks: There was wide support from users and compilers for continuing to develop international methodological frameworks. International cooperative efforts: Efforts such as the International Banking Statistics, the Coordinated Portfolio Investment Survey, and the BIS-IMF-OECD-World Bank Joint External Debt Table were seen as very useful. Creditor-source measures of external debt: Efforts to broaden the coverage of creditor-source data and to reconcile debtor- and creditor-source data were viewed as very important. Data on the nonbank private sector: Compilers pointed out that respondents to national statistical surveys expressed concerns about increased reporting burden. On the other hand, the undercoverage of the nonbank private sector was viewed as a key issue to address. Data quality: Many participants raised the issue of data quality. Specific data requirements: (1) Some noted that data needs depend on the questions at issue. For example, to assess liquidity positions, debt valued in nominal terms plus information on liquid assets is useful, while market valuation is important for balance sheet studies. (2) Disaggregation by sector was considered extremely important, because economy-wide totals could mask offsetting imbalances among sectors. (3) Many saw information on the foreign currency component of assets and liabilities as important; others thought the importance of such data depended on the exchange-rate regime. Financial derivatives: Participants saw the clear need for better information and analysis, but compilation systems that measure financial derivative activity were still relatively few. Special Data Dissemination Standard (SDDS) and General Data Dissemination System (GDDS): Compilers and data users commended the beneficial role of the SDDS and GDDS, but expressed less unanimity on proposals to strengthen the external components of both. The background paper for the Conference and the Summary of Proceedings are on the IMF Website: http://www.imf.org (go to "About the IMF" and then to "Statistical Topics").

4. The Inter-Agency Task Force on Finance Statistics, which the IMF chairs, undertook to prepare the External Debt Guide in close consultation with national compilers of external debt and balance of payments statistics. Work began in 1998, and in the end involved representatives from nine organizations.² The Guide is intended to be of use to both compilers and users of external debt statistics. It provides guidance on (1) the concepts, definitions, and classifications of external debt data; (2) the sources and techniques for compiling these data; and (3) the analytical use of these data.

5. The previous international guidance on external debt statistics, commonly referred to as the Grey Book, was published in 1988. Since then there have been new international statistical guidelines for national accounts and balance of payments statistics—the System of National Accounts, 1993 (1993 SNA) and the fifth edition of the IMF's Balance of Payments Manual (BPM5). As well, there has been tremendous growth in private sector financial flows, especially to private sector borrowers, and growing sophistication of financial markets with increased use of debt securities, derivatives, and other instruments to better manage risks. Against this background, the External Debt Guide provides a comprehensive conceptual framework, derived from the 1993 SNA and BPM5, for measuring gross external debt of the public and private sectors. It draws on many of the concepts introduced in the Grey Book and provides clear guidance that can be applied consistently across the different sectors of an economy and the different debt instruments used for borrowing.

6. The External Debt Guide has a practical side as well. It provides a scheme for classifying external debt by instruments and sectors that is developed into a presentation table for the gross external debt position. Data disseminated using this presentation table, and employing the concepts outlined in the Guide, are essential for providing a comprehensive and informed picture of the gross external debt position for the whole economy. For countries with a particular interest in public sector debt, the sector information can be rearranged to give focus to public and publicly guaranteed debt, consistent with the approach used by the World Bank's Debtor Reporting System.

7. The Guide recommends that debt instruments be valued at the reference date at nominal value, and, for traded debt instruments, at market value as well. The nominal value of a debt instrument is the relevant measure of value from the viewpoint of the debtor because at any moment in time it is the amount the debtor actually owes. This value is typically established by reference to the terms of the contract between the debtor and the creditor, and is often used to construct debt ratios. The market value is determined by prevailing market price, which, as the best indication of the value that economic agents currently attribute to specific financial claims, provides a measure of the opportunity cost to both the debtor and the creditor.

8. The External Debt Guide reflects evidence from the international financial crises of the 1990s that suggested that additional data series could help identify potential vulnerability to solvency and liquidity problems arising from the gross external debt position. It provides conceptual guidance, and presentation tables, for data series such as the debt service schedule (especially relevant for liquidity analysis), the currency composition of debt, and other series known from experience of many countries to be of analytical use. The Guide also explains net external debt—that is, a comparison of the stock of external debt with holdings of external financial assets of similar instrument type—and incorporates financial derivative positions into external debt analysis.

9. In conjunction with work on the External Debt Guide, the IMF Statistics Department, with help of other agencies represented on the Inter-Agency Task Force, organized seven regional seminars. These seminars, beginning in 2000, selected officials who might have the responsibility of setting up or improving their country's external debt statistics. Because the end of the transition period for the newly introduced external debt category in the Special Data Dissemination Standard (SDDS) was approaching (see below), the focus was on SDDS subscribers. Over 200 representatives from more than 50 economies took part. About six to nine months after each seminar we followed up to determine whether further training or technical assistance would be useful. More recently, we have begun a series of regional training seminars designed for mid-level compilers of debt statistics. These seminars will cover more comprehensively the methodology and practice of compiling debt statistics. As well, the IMF will take part in the training seminars when resources allow and at the invitation of the members of the Inter-Agency Task Force that maintain debt compilation systems (namely, Commonwealth Secretariat and UNCTAD).

II. Joint Table on External Debt

10. The Capital Flows Conference also expressed the view that international cooperative efforts were very useful, and mentioned two of these in which the IMF is involved as examples. The examples were the BIS-IMF-OECD-World Bank Statistics on External Debt (Joint Table) and the Coordinated Portfolio Investment Survey. I will treat the Joint Table only briefly, because the IMF's share in that enterprise is not a major one.

11. The Joint Table, which appears on the websites of the four cooperating agencies, was designed to promote frequent and timely access to the data they compile and publish. The table presents creditor and market debt data and international reserve assets for emerging market and developing countries. It provides 14 series—stocks and flows where available—along with metadata.

12. Since the Capital Flows Conference, the agencies enhanced the table in several ways. The table's availability was speeded up and it now becomes available with a lag of about one quarter. This table reflects the improvements that BIS is making in its international banking statistics, including extending the coverage to other reporting countries. In May 2002, the agencies agreed to consider further enhancements, including the addition to the table of debt data for industrial countries.

III. Coordinated Portfolio Investment Survey

13. The Coordinated Portfolio Investment Survey (CPIS) had its origins in a major study, the Report on the Measurement of International Capital Flows, published by the IMF in 1992.³ The Report highlighted the increased importance of portfolio investment across international borders, but also the increased measurement difficulties that were being reflected in the imbalances between recorded assets and liabilities at the world level. The imbalances, with higher flows usually recorded for liabilities than for assets, have generally increased over the period since. One recommendation of the Report was to conduct an internationally coordinated benchmark survey of portfolio investment.

14. This idea of an internationally coordinated survey was picked up by the IMF Committee on Balance of Payments Statistics to facilitate cross-country comparison, permit data exchanges, and encourage standardization and best practice. The Committee hoped that a survey of holdings would put many countries, for the first time, in a position to make reasonable estimates of the outstanding balances, at market prices, of the level of portfolio investment their residents held. The IMF organized the first CPIS, using a common set of definitions, with the common reference date of the end of 1997; 29 countries took part.

15. The data from the 1997 CPIS were published by the IMF in 1999.⁴ They were disaggregated by type of instrument—equity and debt securities—and showed full geographical detail for each instrument according to the country of issuer. The results identified additional assets that increased the estimate of global portfolio investment assets from $6.9 trillion to $7.7 trillion. (Additional liabilities were also identified, so the reduction in the global discrepancy was limited to $0.3 trillion.)

16. After completing the 1997 CPIS, the national compilers of the participating countries set up a Task Force to review the results. The Task Force concluded that the 1997 CPIS had been successful in identifying and spreading best practices among participating countries. The resulting global discrepancy of $1.7 trillion between measured outstanding portfolio investment assets and liabilities showed, however, the need for more work. Some of the problems related to gaps in coverage were as follows: Some important investing countries and some countries with offshore financial centers did not participate. Also, participating countries had difficulties in capturing cross-border portfolio investment by households that do not use the services of resident custodians. Given the potential for improvements, the Task Force recommended that the CPIS be repeated with a reference date of end 2001 and regularly after that, arguing that these data were important to capture better the increasing globalization of financial markets (see Box 2).

17. The Task Force's recommendation was accepted. The 2001 CPIS was undertaken, with wider international participation (almost 70 countries and jurisdictions) and broader coverage of holdings. One feature deserves special note: the data are collected by counterpart country, so the data can be used to check estimates and fill gaps. Results are expected to be available later this year.

18. IMF staff made a special effort to include the small economies with international financial centers. While small in population, these economies have a major role in global portfolio investment. In the 1997 CPIS, the only small economy with an international financial center to take part was Bermuda, which alone had more portfolio investment assets than Canada. The result of the effort is encouraging; all the major offshore centers will take part (including the Bahamas, Cayman Islands, Guernsey, Isle of Man, Jersey, and the Netherlands Antilles). These jurisdictions have had special difficulties that result from large financial sectors, few statistical staff, and a limited tradition in statistical compilation. Progress has been achieved as a result of increasing willingness by the jurisdictions to participate in international statistical collections of data on cross-border financial stocks and flows. The IMF provided extensive support in the form of training, visits, seminars, and backup.

19. The CPIS will be conducted again for end-2002 as part of a continuing annual survey. Annual data will help analysts to identify trends. In addition, the patterns in global and bilateral discrepancies will assist in understanding the links between financial markets. Annual data collection is also likely to bring about improvements in data quality as both respondents and statistical compilers become more familiar with the survey.

IV. Research areas: On to BPM6

20. As already noted, the 2000 Capital Flows Conference showed widespread support for methodological frameworks. One of the oldest—if not the oldest—of these is the IMF's Balance of Payments Manual. The first edition came out in January 1948, and the most recent, the fifth edition, came out in 1993. At the time of the Conference, the idea of a sixth edition was still subliminal and hazy. By fall 2000, the IMF Balance of Payments Statistics Committee considered various aspects of the framework that might be reviewed, extended, or updated for a BPM6. A compendium of issues was presented in fall 2001.⁵ The list of issues was wide-ranging:

• Refinement of the concept of residence, especially considering cases where the center of economic interest is ambiguous, such as production with little or no physical presence, entities that undertake no production but hold assets, highly mobile individuals and entities, joint entities, and areas of joint sovereignty



• The statistical treatment of trusts



• Nonperforming loans



• Accrual of interest on debt securities



• Financial intermediation services indirectly measured



• Increased emphasis on services



• Employee stock options.

21. Committee members suggested some other items and noted that emphasis should be on links between the BPM6 and the IMF's data dissemination standards, between micro- and macro-data, and between international statistical and accounting standards. They also stressed the importance of including a chapter on the purposes and uses of balance of payments, and asked that increased emphasis be placed on the international investment position (IIP) and external debt statistics.

22.It was agreed that the process of updating BPM5 should be as public and transparent as possible. All proposed changes to the framework, as well as alternative treatments that have an impact on other parts of the suite of macroeconomic statistics, should be routed through the interagency group for the national accounts. A target date of 2007 was agreed for the release of the new manual.

V. Data Quality

23. At the 2000 Capital Flows Conference, many participants raised the issue of the quality of data on capital flows and external debt. Data quality more generally was already high on the agenda of the IMF Statistics Department (see Box 3). The data quality work is an integral part of the ongoing efforts to strengthen the international financial architecture. It was seen as important in its own right and an important complement to the data dissemination standards (see the next section).

24. To make a long story short in order to stick to my assigned topic, I will say only that, in an interactive and consultative way, we prepared a Data Quality Assessment Framework (DQAF). As many of you know, following the 1997-98 financial crises the IMF was asked to prepare reports on the extent to which countries follow internationally recognized standards and codes in its areas of expertise. These assessments appear as Reports on the Observance of Standards and Codes (ROSCs).⁶ Within the ROSC program, data modules—for which the IMF is responsible—contain an assessment of data dissemination practices complemented with an assessment of data quality using the DQAF. The DQAF provides a common language and structure for assessing data quality. It brings together best practices and internationally accepted concepts and definitions in statistics. These are grouped in several dimensions of data quality: integrity, methodological soundness, accuracy and reliability, serviceability, and accessibility, as well as the related institutional prerequisites (see Box 3). So far, nine data modules using this framework have been published.

25. The Appendix shows an example of the summary table contained in each of these ROSC data modules. As this summary table shows, so far the DQAF includes dataset-specific frameworks for national accounts, prices, government finance statistics, monetary and financial statistics, and balance of payments statistics. The balance of payments dataset-specific framework sheds light as well on the international investment position. Recently we began work toward a framework for external debt statistics. The Inter-Agency Task Force on Finance Statistics, at its May 2002 meeting, confirmed that the quality of external debt data was an important issue, and members agreed to provide comments on the IMF draft of the framework. We hope to have field-tested the dataset-specific framework by October 2002 and will make it available for self-assessment as well as use it in our own work.

VI. Dissemination Standards

26. At the time of the 2000 Capital Flows Conference, the Special Data Dissemination Standard (SDDS) had been in place for four years, and the General Dissemination System (GDDS) for three.⁷ Both compilers and data users commended the beneficial role played by the dissemination standards. We continue to receive such comments. For example, at a Seminar on Governance of National Statistical Systems that we co-organized in May 2002 with the United Nations Statistical Division and the Singapore Department of Statistics, participants noted that taking part in the SDDS or GDDS has helped to increase the credibility and legitimacy of statistics and statistical agencies.

27. Table 1 focuses on the SDDS and tries to put together a more quantitative picture of the roles it has played in improving statistics and statistical practices. It shows that over a three-year period, from 2000 to 2002, the number of subscribers has increased and that an increasing number of countries are known to be working with IMF staff in preparing for subscription. The increase in the number of countries in observance reflects that countries worked off their transition plans—in other words, improved their statistics—to meet SDDS requirements for coverage, periodicity, timeliness, and provision of advance release calendars. The only subscriber not now in observance has one transition plan—the dissemination of a producer price index—still outstanding. All but one subscriber maintains a National Summary Data Page that is hyperlinked to the IMF's Dissemination Standards Bulletin Board (DSBB); this link makes it possible for a DSBB user to go directly to the country's actual SDDS data. Impressive progress can be noted in terms of the number of summary methodologies—a key to gauging data quality—publicly available through the DSBB.

28. The lower panel of Table 1 presents information about on-time release of data. Beginning in May 2000, the IMF has monitored the extent to which countries make data available on the schedule described in their advance release calendars. The percentage of on-time release increased in all categories over the last year. Over 95 percent of releases for consumer prices, the reserves template, and national accounts are on schedule. Several categories—especially central government operations—have room for further improvement.

Table 1. SDDS Indicators and Results of Monitoring Release Calendars
Activity Indicators (number of countries unless noted)	Q1 2000	Q1 2001	Q1 2002
Subscribers	47	48	50
Countries in observance	13	40	49
Countries working with IMF staff for possible subscription	7	7	9
Countries with National Statistical Data Page websites linked to the IMF's DSBB	44	49	49
Countries with summary methodologies posted	43	47	47
Summary methodologies posted (number of methodologies)	453	662	701
On-time Data Release (percentage of releases on schedule)	Q1 2000	Q1 2001	Q1 2002
Monthly data Of which:	n.a.	85.8	92.5
Consumer prices	n.a.	90.4	95.2
Central government operations	n.a.	76.3	79.2
Analytical accounts of central bank	n.a.	82.6	93.2
Reserve template	n.a.	91.4	97.3
Quarterly data Of which:	n.a.	84.9	94.5
National accounts	n.a.	90.6	99.0
Central government debt	n.a.	70.7	87.5
Balance of payments	n.a.	87.6	93.9

Source: Quarterly Update on the Special Data Dissemination Standard-First Quarter 2002, www.imf.org/external/pubs/ft/sdds/q/2002/eng/01/index.htm.
n.a. = not applicable.

29. At the time of the 2000 Capital Flows Conference, the IMF Executive Board was considering steps to strengthen the SDDS and GDDS. The data template on international reserves and foreign currency liquidity had been introduced in 1999, and further changes related to external debt were to be dealt with (see Box 4).

In March 2000 external debt statistics were introduced as a category. Prospective debt service schedules was made an encouraged category under the SDDS. A domestic-foreign currency breakdown of external debt was also encouraged. As they now stand, the SDDS specifications for external debt and the other categories in the external sector are shown in Box 4.⁸

30. The transition period for the reserves template ran through end-April 2000, with the first set of template data for end-May 2000 due no later than end-June 2000.⁹ As of mid-2002, 50 SDDS subscribers and one non-subscribing country have been disseminating the template data on their national websites regularly. Forty-three economies have been reporting their template data on a voluntary basis to the Fund for redissemination on the IMF's website http://www.imf.org/external/np/sta/ir/index.htm (see the next section).

31. From time to time there is discussion about whether the SDDS should be strengthened further. The most often heard suggestion is that data on reserves should be on a more frequent and timely basis. The Institute for International Finance (IIF), for example, urges that gross international reserves be disseminated weekly with a one-week lag and international reserves and foreign currency liquidity be disseminated weekly with a one-week lag. We discussed this question with SDDS subscribers, and while many stressed the importance of frequent and timely dissemination of reserves data, most considered that increasing the frequency and timeliness for the dissemination of reserves template data under the SDDS is not necessary at this time. They noted that moving to weekly frequency and timeliness would raise technical and resources constraints. Some subscribers expressed concern about the potentially destabilizing market reactions to volatile weekly data. We have presented subscribers' views to the Executive Board in June along with the consideration that these additional requirements might be a barrier to new subscribers. Most Directors considered that increasing the frequency and timeliness for the dissemination of the reserves template under the SDDS from monthly to weekly frequency and reporting is not necessary at this time. Directors stressed that priority should be given to ensuring that the maximum number of member countries can subscribe to the SDDS or participate in the GDDS, as appropriate.¹⁰

VII. Dissemination of Data on Capital Flows and External Debt

32. The 2000 Capital Flows Conference did not surface specific views on dissemination of data. It was understood that easy and timely access to data was crucial to its widespread use. The IMF continues to promote the dissemination of data on capital flows and external debt as part of its mission to serve as a center for the exchange of economic and financial data. Many users of such data are quite familiar with the traditional products the IMF makes available—our monthly and annual print and CD-ROM publications.

33. Earlier I mentioned that, as of mid-2002, 43 economies report their template data on a voluntary basis to the Fund for redissemination on the IMF's website.¹¹ The IMF's website presents countries' template data in a common format and in a common currency (U.S. dollars) to facilitate users' access to the data and to enhance comparability of data across countries. In addition to current data, the IMF's database offers historical information and selected cross-country time series. To facilitate users' viewing, printing, and downloading the data for analysis, the IMF database is presented in html (hypertext markup language), in pdf (portable document format), and in csv (comma separated values) files.

34. The joint table on external debt, including its prospective enhancements, has also already been mentioned. There is ongoing discussion of whether, as SDDS subscribers come on stream with more external debt data, a database with some of the same features of easy access to comparable data shown by the reserve database should be put in place.

35. As part of an ongoing effort to improve our statistical products, the IMF has recently launched a complete version of the International Financial Statistics (IFS) database on the Internet. This new product is a major enhancement of the existing version, which contains only a limited selection of time series from the IFS database. I encourage everyone to take advantage of our "60-day free trial offer" and subscribe to this new IFS Online service at the following Internet address http://ifs.apdi.net/imf/logon.aspx. Beyond the trial period, IFS Online will be available by subscription at rates discounted on a sliding scale based largely upon ability to pay. For example, commercial enterprises in industrial countries will pay the regular price but individual, non-commercial users in developing countries will receive a full 100 percent discount. I am hopeful that, by providing IFS Online to users in developing countries free-of-charge, or at a steep discount for commercial users, we will witness a rapid growth in both the number of subscribers and the quality of statistical analysis in those countries. We are planning to offer other products via the Internet in the future and are working toward increasing our frequency of dissemination.

VIII. Conclusions

36. In the past two and half years, I believe that significant progress has been achieved in developing internationally agreed statistical guidelines related to capital flows and external debt, in strengthening relevant collections of data, and in making the data available more rapidly and readily. While I have referred to the IMF's work, certainly these achievements and others in the field reflect the collaborative international efforts; many organizations and many statistical agencies have taken up the cause. The work on international capital data should also be put into a larger context. Work is under way in other areas that are complementary in terms of the objectives of financial stability, sustainable growth, and poverty reduction. As just one example, I would mention the work to develop a compilation guide on indicators of financial stability (previously referred to as macroprudential indicators). Also, if one examines the experiences of the financial crises since the 2000 Capital Flows Conferences to see if they point to additional data needs, one might venture to say that comprehensive fiscal data will increasingly be seen as a high priority. The IMF is moving in that direction with a stepped up emphasis on implementation of the new Government Finance Statistics Manual.

37. I said at the outset that I hope this session would be the occasion for further dialogue on needs and priorities for capital flows and external debt data. In our discussions, we can consider the progress made—looking for, among other things, possible gaps and ways to strengthen the approaches used—and whether recent developments shed any new light on priorities for statistics on capital flows and debt.

Summary Table from an IMF Report on the Observance of Standards and Codes (ROSC) data module
Table 2. Data Quality Assessment Framework: Summary Presentation of Results

Element	National Accounts	Consumer Price Index	Producer Price Index	Government Finance Statistics	Monetary Statistics	Balance of Payments Statistics	Comments*
0. Prerequisites of quality
0.1 Legal and institutional environment 0.2 Resources 0.3 Quality awareness	O O O	O LO O	O O O	LNO O O	O O O	O O O	See paragraph 19 See paragraph 18
1. Integrity
1.1 Professionalism 1.2 Transparency 1.3 Ethical standards	O O O	O O O	O O O	O LO O	O O O	O O O	See paragraph 22
2.Methodological soundness
2.1 Concepts and definitions 2.2 Scope 2.3 Classification/ sectorization 2.4 Basis for recording	LO LO O O	O O O O	LO O LO O	LO LNO O O	O O LO LO	O O LO LO	See paragraphs 24, 26, and 27 See paragraphs 24 and 27 See paragraphs 26, 28, and 29 See paragraphs 28 and 29
3. Accuracy and Reliability
3.1 Source data 3.2 Statistical techniques 3.3 Assessment and validation of source data 3.4. Assessment and validation of intermediate data and statistical outputs 3.5 Revision studies	LO LO O LO N.A.	LO O O O N.A.	LO O O O N.A.	LO O O O N.A.	O O O O O	O O O O O	See paragraphs 31, 33, 34, 35, and 37 See paragraph 32
4. Serviceability
4.1 Relevance 4.2 Timeliness and periodicity 4.3 Consistency 4.4 Revision policy and practice	O O O LO	O O O O	O O O O	LO O LO LO	O LO LO O	O O LO O	See paragraph 40 See paragraph 41 See paragraphs 40, 41, and 42 See paragraphs 40 and 42
5. Accessibility
5.1 Data accessibility 5.2 Metadata accessibility 5.3 Assistance to users	O O O	O O O	O O O	LO LNO O	O O O	O O O	See paragraph 45 See paragraph 45
*Refers to the text of the ROSC. Key to symbols: NA = Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed For country reports see: http://www.imf.org/external/np/rosc/rosc.asp?sort=topic#DataDissemination.

¹¹ See: http://www.imf.org/external/np/sta/ir/index.htm.