Conference Agenda

Background Paper



Summary of Proceedings
Conference on Capital Flow and Debt Statistics:
Can We Get Better Data Faster?

Conference Sponsored by
The Statistics Department and the Policy and Development Review Department of the International Monetary Fund in Cooperation with the Financial Stability Forum Working Group on Capital Flows

Washington, D.C., February 23–24, 2000

1.  An international meeting entitled "Conference on Capital Flow and Debt Statistics: Can We Get Better Data Faster?" was hosted by the International Monetary Fund on February 23-24, 2000 in cooperation with the Financial Stability Forum (FSF) Working Group on Capital Flows. The Working Group had earlier proposed such a conference as a way to generate a dialogue between data users and compilers, discuss recent initiatives to improve data on capital flow and debt statistics, and identify priorities for further work in this area. The focus of the conference was on what actions and resources would be required to provide better and more timely data on capital flows and debt and what should be given priority.

2.  Approximately 120 senior-level data users, policymakers, and data compilers participated in the conference. They represented a wide range of institutions, including finance ministries, central banks, national statistical agencies, private commercial and investment banks, and international organizations.

3.  The conference took place in the context of many international initiatives that have been launched since the financial crises of 1997 and 1998 to enhance disclosure practices and transparency, in order to improve the functioning of markets as well as the bases for policy making. The financial crises underscored again the importance of timely and reliable economic and financial data to assess risks of sharp swings in capital flows.

4.  The conference was timed to fit into the complementary work programs of the FSF and the Fund. The Working Group on Capital Flows was in the process of finalizing its report, which, inter alia, was intended to identify gaps in available data on cross-border capital flows and external positions and the analytical bases for risk and liquidity management. The Fund's Executive Board was soon to discuss proposals to strengthen the IMF's Special Data Dissemination Standard (SDDS) and General Data Dissemination System (GDDS) with respect to external debt statistics, and its work program for the coming months included several other related issues.

5.  The conference began on the evening of February 23 with a working dinner addressed by Mr. Stanley Fischer, Acting Managing Director of the IMF, and Mr. Mario Draghi, Director General of the Italian Ministry of Treasury and Chairman of the Working Group on Capital Flows. Both speakers stressed the critical importance of timely and comprehensive statistics for informed policymaking.

6.  Data users and data compilers brought a variety of perspectives to the discussion of the need for certain types of external-sector data. Policymakers, data users, and a few compilers stressed the need for more comprehensive and timely capital flow and external debt statistics. Data compilers from industrial countries and offshore centers generally viewed the dissemination of such data as a lower priority. This reflected national statistical priorities as well as concerns regarding resources, compilation difficulties, and respondent burden.

7.  Data users stressed the importance of comprehensive, high quality, and timely capital flow and external debt data and noted that the absence of good data has a potential economic cost far greater than the cost of providing data. Indeed, one speaker questioned whether an economy could afford not to have comprehensive and timely data in an era of global capital markets. Other participants noted that, in the recent past, some poor international investment decisions could be attributed to inadequate data on country risk. Users felt that it was in an economy's own interest to make available good data, not only to attract international investment but also to manage risk.

8.  For their part, data compilers warmly welcomed the opportunity to discuss data issues with users, noting how rare an event this was at the international level. Most data compilers felt pressured to produce more timely and comprehensive data, but many lacked the necessary additional resources to undertake this work. It was noted that, while policymakers and users have often called for more resources for statistical work at international conferences, these resources generally were not forthcoming in the domestic budget processes.

9.  A number of compilers from industrial countries mentioned the growing policy interest in the measurement of service industry activities, which accounted for a growing proportion of economic activity. Accordingly, resources that might otherwise be committed to improving capital flow and debt data, which was seen as an international priority, might be focussed instead on improving services statistics or meeting other domestically-oriented statistical priorities. However, in an integrated world, the lack of good data on capital flows and debt in one economy may have consequences for regional neighbors and perhaps beyond, justifying, for some participants, international involvement in priority setting for statistical work.

10.  The SDDS and the GDDS provided an indication of international statistical priorities. The SDDS had identified a framework of core data that were recognized as being essential in any comprehensive statistical system. On balance, both compilers and users had a positive view of the usefulness of the SDDS. The system had broad applicability, had been well accepted, and, as one compiler emphasized, had been beneficial to the international statistical community. However, the proposed strengthening of the SDDS into the area of external debt data had raised the question of whether some SDDS categories might be designed too specifically to address the problems of a subset of countries. There were differing views on whether national (debtor-source) measures of external debt needed to be disseminated on a timely and frequent basis for countries that were in a net external creditor position. Some compilers felt that it would be next to impossible to request budgetary resources for data for which they could see little or no user interest, especially within their own economies. Other participants noted that even countries in net creditor positions may have external debt vulnerabilities. They cited examples such as a net creditor country being a net debtor in one crucial sector or having a large proportion of short-term liabilities together with a small proportion of liquid assets, and also recalled that countries may move from a net creditor to a net debtor position over time. It was suggested that the usefulness of any specific reporting criterion to a particular country might not be dependent on whether the country is a net creditor or not but on the strength of its markets and institutions. Nonetheless, the concern of some compilers was that for a net creditor country the SDDS prescription on external debt data would be in such marked conflict with national priorities that withdrawal from the SDDS might need to be considered. In short, for items such as external debt, these compilers believed that "one size did not fit all."

11.  Nevertheless, there was widespread support from both users and compilers for continuing to develop at the international level methodological statistical frameworks that could be employed by all economies. These standards allowed for comparability of data across countries and consistency of approach between related data series. Also, international cooperative efforts, such as the long-standing International Banking Statistics of the Bank for International Settlements (BIS), together with the more recent IMF Coordinated Portfolio Investment Survey and the Joint BIS-IMF-OECD-World Bank Statistics on External Debt (Joint Statistics), were felt to be very useful. One speaker suggested that more such coordinated efforts should be considered. All participants supported efforts to develop and improve these creditor-based data, which were not only valuable in their own right but could be used to cross-check, and perhaps fill gaps in, debtor data.

12.  Thus, efforts to broaden the coverage of creditor sources of data and to reconcile these data with information from national sources were viewed as important to improving the quality of information for the assessment of external vulnerability. The BIS staff offered to bring creditor and debtor countries together to facilitate the task of reconciling some of the differences between its international banking statistics and those from national sources. It was noted that full reconciliation of these data was not a realistic objective, particularly as data from different sources were compiled to answer different types of questions. For example, the international banking statistics were aimed at identifying the country exposure of banks in the major creditor countries, not the external liabilities or foreign currency risk exposures of individual debtor counties. Nonetheless, the differences between the two data sets could be narrowed. The international banking statistics could be made more suitable for certain other analytical purposes through changes such as the incorporation of maturity information in the locational banking statistics and more detailed reporting of claims in the form of debt securities by offshore centers, while information from sources other than the international banking statistics on foreign holdings of domestically issued debt securities would fill a major gap in the creditor-based Joint Statistics.

13.  Compilers pointed out that respondents to national statistical surveys had expressed concerns about increased demands. The supply of information was a cost to business, which was understandably wary of statistical initiatives. Indeed, in some countries, the official policy was to reduce the paperwork burden upon businesses. On the other hand, a number of participants pointed out that a key issue to address was the undercoverage of the activity of the nonbank private sector. In a world of increasing capital flows, the need to cover nonbank private sector activities had become more important, but this sector was often reluctant to provide data. Some participants felt that a very high priority would be to encourage nonbank private institutions to disclose data for market use; this would improve transparency and support efforts to improve statistical reporting. Others encouraged compilers to consider using new technologies both to obtain data faster and to keep costs under control both for statistical agencies and respondents. Compilers were also encouraged to make better use of existing data sources such as private-sector databases on security market issues.

14.  The issue of data quality was raised by many speakers. Compilers stressed their concern that requiring more data without additional resources could lead to increased use of estimation and consequently a breakdown in user confidence in the data. One user felt that quality of data has deteriorated over the last 25 years, although others were not so pessimistic. Users constantly stressed that the "perfect should not be the enemy of the good." It was suggested, for instance, that, if only data for the largest five businesses could be collected, it would be better for compilers to publish these data and to clearly describe their coverage than to disseminate nothing. If data were not available, the market would make its own estimates and these might well misrepresent the actual situation. This was an argument for openness and transparency. Participants also addressed this issue in a broader context, including initiatives underway in conjunction with the SDDS and with the Fund's surveillance work to promote and assess data quality; some of these initiatives were already reflected on the Fund's website.

15.  The participants discussed more specific data requirements in the field of capital flows and external debt. Policymakers and users generally stressed the need for good quality and timely data on external debt, measured from both debtor and creditor sources. However, some speakers noted that the data required depended upon the questions at issue. For instance, if assessing an economy's liquidity situation was important, then debt valued in nominal terms, together with information on liquid assets, would be particularly useful information. For this purpose, debt data on forward amortization schedules or position data classified by residual maturity would also be needed. On the other hand, for assessing the balance sheet of a country's external assets and liabilities, market valuation of both assets and liabilities would be important.

16.  For most purposes, disaggregation by sectors was considered extremely important, since economy-wide totals could mask offsetting imbalances among sectors that could be sources of systemic vulnerability. Country circumstances were also relevant. For instance, many participants saw information on the foreign currency component of external assets and liabilities as being important, while others thought that the importance of such data depended on the exchange-rate regime of the economy--with foreign currency data being more important in a fixed exchange-rate regime. Also, one speaker noted that a survey of main borrowers had revealed that virtually all of New Zealand's large foreign currency debt had been hedged either through the use of financial derivatives or naturally by foreign currency earnings. Thus, the analysis of foreign-currency denominated positions was not straightforward.

17.   There was discussion of how to take account of financial derivatives in the assessment of vulnerability. There was clearly a need for better information and analysis in this area, but compilation systems that measure financial derivative activity were still relatively few. Some users noted that, while the recent crises had not involved derivatives to a significant extent, the existence of financial derivatives in many cases broke the link between market values and exposures. It was suggested that stress testing be used more extensively to estimate the impact on an entity's balance sheet of possible changes in market variables such as the exchange rate or interest rates. It was felt that entities in the public and banking sectors were probably most advanced in making use of stress tests, although more progress was needed--perhaps by means of experimentation--to aggregate the results of stress testing by individual entities.

18.  The participants also discussed information, outside the context of national or international statistical systems, that may be needed at higher frequencies and with shorter reporting lags. Some countries had established high-frequency reporting systems to monitor the external liabilities of domestic financial institutions, while a few others used such systems to monitor developments in the foreign exchange market more broadly. While there emerged differences of view on the usefulness of compiling high-frequency data, all participants felt that some form of frequent monitoring of foreign exchange market activities and views was important. While some such systems were formal, requiring market institutions to report data regularly, others were informal and relied upon market contacts. The discussion underscored that no single approach appeared likely to fit the circumstances of all countries. A workshop devoted specifically to this topic was held the following day.

19.  In summing up at the end of the conference, Mrs. Carol Carson, Director of the Fund's Statistics Department, noted that a consensus between data users and data compilers on the priorities for producing better data faster had not emerged, but there had been a flow of ideas that would be essential in working towards achieving such a consensus. Mr. Jack Boorman, Director of the Fund's Policy, Development, and Review Department, was impressed by the different perspectives brought to the meeting. He underscored that the international community would need to keep seeking the right fora to develop a consensus on statistical issues that have cross-border implications.