This second publication was drafted by Mr. Simon Quin, Mr. Marco Committeri, and Mr. John Joisce of the IMF Statistics Department. Messrs. Quin, Committeri, and Joisce are also members of the Secretariat to the Task Force on the Coordinated Portfolio Investment Survey.

Final Report of the Working Party on the Measurement of International Capital Flows [Godeaux Report)] (Washington, DC: International Monetary Fund, 1992).

The IMF Committee on Balance of Payments Statistics was established in 1992 (a) to oversee the implementation of recommendations made by two IMF working parties that investigated the principal sources of discrepancies in global balance of payments statistics published by the IMF, (b) to advise the IMF on methodology and compilation issues in the realm of balance of payments and IIP statistics, and (c) to coordinate international data-gathering efforts.

In principle, the combined surpluses and the combined deficits arising from the recording of international financial transactions should offset each other because one country’s credits are another country’s debits. However, sizeable statistical discrepancies are evident in global balance of payments statistics. Since 1994, the discrepancies have been analyzed in the annual reports of the IMF Committee on Balance of Payments Statistics. These reports can be accessed through the Internet at http://www.imf.org.

The terminology of the fifth edition of the Balance of Payments Manual is used throughout Analysis and Plans. Therefore, Godeaux Report references to the Capital Account have been changed to the Financial Account.

Final Report of the Working Party on Statistical Discrepancies in World Current Account Balances [Esteva Report] (Washington, DC: International Monetary Fund, 1987).

Additional sources included data—from the Bank for International Settlements—on international banking statistics and international bond issues, information from the World Bank Debtor Reporting System, balance sheets of international organizations, and published data on financial flows.

These data were collected through the special questionnaire; for many developing countries, assets and liabilities were estimated by cumulating net balance of payments flows at historical values.

Liabilities could be overestimated because of the under-measurement of repatriated securities or because a portion of the existing domestic securities could have been bought by domestic residents through foreign intermediaries or nominee accounts held abroad.

According to the debtor-creditor principle, transactions in a country’s external assets and liabilities should be allocated, respectively, to the country that incurs the liability (the debtor) or to the country that owns the claim (the creditor). According to the transactor principle, financial flows should be allocated to the country of the foreign party to the transaction.

For example, data available to issuers, to custodians acting on behalf of nonresidents, or from security registers may not, for appropriate reference periods, be accurate sources of information on the countries of holders.

Advanced countries and principal geographical regions are defined in World Economic Outlook (Washington, D C: International Monetary Fund, May 1999), 127-136. Countries comprising the emerging market group are those known to have issued bonds internationally. See Appendix A for the classification of countries.

The BPM5 definition of portfolio investment assets excludes securities held as reserve assets. However, the compilation of augmented 1997 CPIS data and the assessment of augmented 1997 CPIS results are based on a broader concept that includes reserve instruments and liabilities constituting foreign authorities reserves (see the BPM5 definition of portfolio investment liabilities) within portfolio holdings.

For Luxembourg, which did not participate in the 1997 CPIS and does not compile either balance of payments or IIP statements, a data model was used to estimate the country’s portfolio assets and liabilities. Balance of payments data available for the Belgium-Luxembourg union were used in the model, and stocks were derived from cumulative flows by using price and exchange-rate adjustments observed for neighboring countries.

To ensure confidentiality, country details were not published or shared with CPIS participants. Monetary authorities’ holdings of foreign securities were included, by some participants, in their CPIS data.

Balance of Payments Compilation Guide (Washington, DC: International Monetary Fund, 1995), 160-61.

In more forma] terms, this procedure can be described thus: K, and F, are, respectively, the stock and flow of portfolio investment assets shown in the IIP and balance of payments statements of a certain country. The estimated value of that country’s end-1997 portfolio investment assets was calculated as the terminal value of the sequence${\hat{K}}_t=K_t$ if IIP data were available; otherwise ${\hat{K}}_t=\max \{0,K_{t-1}+F_t\}$for t = 1970, ..., 1997 . K_l969 was assumed to be nil. This method was applied to all countries except Luxembourg (see footnote 14).

By comparison, the estimate of portfolio investment assets made on the basis of IIP data and cumulative flow estimates (that is, without the use of 1997 CPIS data) indicated that total holdings of portfolio investment assets at end-December 1997 were US$6.6 trillion. However, estimated components accounted for only a small fraction of the total.

Reliance on such arbitrary assumptions reflects the lack of published data for offshore financial centers. However, there are other indications that Bermuda’s cross-border portfolio investment holdings are comparable with those of other offshore financial centers.

It should be emphasized that these countries were singled out only because they provided data. Classifying portfolio liabilities by country of debt holder is a difficult task, and the issues discussed herein are certain to apply to many other countries.

Only Japan could be assumed to have issued a significant amount of debt securities that were held as reserve assets by Japan’s partner countries. Attempts made to include, on the asset side, estimates of monetary authorities’ holdings broadly confirmed the pattern described previously.

Bilateral comparisons also indicated that the liabilities of some countries were understated because of their failure to identify positions with some major investing countries (for example, the USA, UK, and Japan). These results have been taken into account in Table 5.

It is even more likely that data on financial derivative transactions are reported on a net basis as information may not readily be available to distinguish transactions in assets from transactions in liabilities.

Countries implementing the BPM5 would previously have reported, for the Balance of Payments Statistics Yearbook, financial derivative positions as part of portfolio investment. However, few countries reported financial derivative transactions and positions as separate components of portfolio investment.

Even if position data for financial derivatives were available for countries, a geographic attribution of holdings might not add much to an assessment of country risk because financial asset and liability positions can change so rapidly.

Both External Debt: Definition, Statistical Coverage and Methodology, 1988, and Debt Stocks, Debt Flows, and the Balance of Payments, 1994, were jointly published by the BIS, IMF, OECD and World Bank. The IMF-chaired Inter-agency Task Force on Finance Statistics draws on representatives from these four agencies; the Commonwealth Secretariat, the European Central Bank (ECB), Eurostat, the Paris Club Secretariat, and the United Nations Center for Trade and Development also participate in the work of the task force.

BIS international banking statistics can be useful for the measurement of a country’s external debt liabilities. For example, both the United States and Canada use BIS statistics to supplement data from national data collection systems. Without the additional information, external debt liabilities would otherwise be under-reported.

Information about economic and financial data disseminated by member countries adhering to the SDDS is available at http://dsbb.imf.org.

At the IMF Executive Board’s third (March 2000) review of the SDDS, executive directors agreed to change the time limit for dissemination of annual IIP data from six to nine months if countries were disseminating quarterly external debt data within three months after the reference dates of the relevant IIP statements.

Annual IIP statements prepared according to BPM5 methodology are published in the Balance of Payments Statistics Yearbook published annually by the IMF.

The BIS classifies as offshore banking centers the following 14 countries and territories: Aruba, the Bahamas, Bahrain, Barbados, Bermuda, the Cayman Islands, Hong Kong SAR, Lebanon, Liberia, the Netherlands Antilles, Panama, Singapore (Asian Currency Units—ACUs), Vanuatu, and the West Indies (UK). See Errico and Musalem, “Offshore Banking, an Analysis of Micro- and Macro-prudential Issues,” IMF Working Paper WP/99/5, January 1999, for a list of an additional 56 countries and territories in which OFCs are located. These are Andorra, Anguilla, Antigua, Australia, Austria, Belize, the British Virgin Islands, Campione d’ltalia, the Cook Islands, Costa Rica, Cyprus, Djibouti, Dominica, Dubai, Gibraltar, Grenada, Guam, Guernsey, Hungary, Ireland (Dublin), Israel, Japan (Japanese Offshore Market—JOM), Jersey, Kuwait, Liechtenstein. Luxembourg, Macao, Madeira, Malaysia (Labuan), Malta, the Marianas Islands, the Marshall Islands, Mauritius, Micronesia, Monaco, Montserrat, Nauru, the Netherlands, Niue, Oman, the Philippines, Puerto Rico, Russia, Sark & Isle of Man, Seychelles, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, Switzerland, Tangier, Thailand [Bangkok International Banking Facilities (BIBFs)], the Turks and Caicos Islands, the United Kingdom (London), the United States [US International Banking Facilities (IBFs)], Uruguay, and Western Samoa.

OFCs can also be located in large and medium-size economies. This is the case, for instance, of the US International Banking Facilities, the Japanese Offshore Market, the Bangkok International Banking Facilities in Thailand, Offshore Banking Units in Philippines, and the Labuan International Offshore Center in Malaysia.

Another useful reading is Richard Roberts, ed.. Offshore Financial Centers (Cambridge: University Press, 1994).

According to Cassard, London is the major OFC of this type; by contrast, New York and Tokyo are major domestic financial centers with offshore facilities. Regional OFCs are Hong Kong SAR., Singapore, Luxembourg, Bahrain, Panama, Lebanon, Malta, Labuan, Gibraltar, Cyprus, Dublin, and Madeira. Most regional OFCs lack financial autonomy because of they are located in small local economies; they act as hosts to foreign financial institutions with headquarters located in large countries or dynamic regions (for example, Singapore with South East Asia, Hong Kong SAR with Japan and China, Luxembourg with Germany and France, and Bahrain with Saudi Arabia).

The Cayman Islands, the Bahamas, the Netherlands Antilles, the British Virgin Islands, Aruba, Bermuda, Barbados, the Channel Islands, Seychelles, Turks and Caicos, the Marshall Islands, and Vanuatu are considered by Cassard to be offshore centers.

Because of the low “sunk costs” related to establishing a presence in booking centers, foreign institutions can leave these jurisdictions on very short notice. Booking centers offer highly attractive legislation for offshore activities.

According to the Basel Core Principles, consolidated supervision requires that a parent financial institution and parent supervisory authority must be able to monitor the risk exposure (including reputation risk) for the total business, wherever conducted, of the banks or banking groups for which the parent institution and supervisory authority are responsible. OFCs may contribute to the difficulty of implementing effective consolidated supervision because these centers can restrict foreign supervisors’ access to information on subsidiaries located therein and increase the complexity of relevant corporate structures and jurisdictions.

In a majority of cases (especially those of OFCs located in large or medium-size economies), regulatory responsibilities and supervision of offshore banking and financial activities are normally assigned to local monetary authorities, which may also act in cooperation with governmental bodies. Supervision of OFCs located in small economies is usually performed by government bodies called Financial Services Commissions. Eighteen OFCs are members of an association called the Offshore Group of Bank Supervisors (OGBS). These countries are Aruba, the Bahamas, Bahrain, Barbados, Bermuda, the Cayman Islands, Cyprus, Gibraltar, Guernsey, Hong Kong SAR, the Isle of Man, Jersey, Malta, Mauritius, the Netherlands Antilles, Panama, Singapore, and Vanuatu.

Aruba, the Bahamas, Bermuda, and Barbados do not cover such transactions because of the difficulty of obtaining data on offshore operations. See the 1999 IMF Balance of Payments Statistics Yearbook, Part 5, 116, 123-124, 128.

Notable exceptions within these groups are Luxembourg and Hong Kong SAR. Luxembourg and Belgium compile a consolidated balance of payments. External transactions of offshore entities located in Luxembourg are, in principle, covered in the Belgium-Luxembourg statement—with the exception of those vis-à-vis Belgian residents. Hong Kong SAR released balance of payments data for the first time in 1998; since no details on the financial account were made available, such data could not be included in IMF publications. In general, only a few regional centers (Mauritius, Bahrain, and Ireland) compile IIP data. Offshore entities are treated as residents in the latter two countries’ balance of payments and IIP statements.

Supplementary data from the 1997 CPIS on bank holdings of foreign debt securities were available only for Hong Kong SAR and Singapore.

The United Kingdom and Ireland are not included in this list because they participated in the 1997 CPIS and offshore units were covered in their surveys. Luxembourg is also excluded because balance of payments statistics are compiled by that country, jointly with Belgium, which cover offshore entities. Data for the Channel Islands (Guernsey and Jersey) and the Isle of Man have only recently become available.

The large amount (US$130 billion) of portfolio investment assets reported by Bermuda is still an underestimate, according to Bermuda’s compilers. Available data from regulatory and market sources suggest that the order of magnitude for Bermuda could well apply to other OFCs. If so, a large part of the global discrepancy of US$1.7 trillion could be attributable to OFCs, especially if assets under management by trust companies are included.

About 20 of the countries that participated in the 1997 CPIS voluntarily provided geographic attributions for short-term portfolio investment assets. These countries indicated that amounts of cross-border investment in short-term portfolio investment instruments were comparatively small. However, it is probable that countries significantly underestimated cross-border activity. It is generally easier to collect data on position than on large numbers of transactions. Therefore, data on transactions may be estimated from changes in positions.

Even when incomplete and not performed on a debtor/creditor basis, geographic attributions of portfolio investment liabilities voluntarily reported in the 1997 CPIS were useful for gap-filling purposes. Voluntary reporting, in the 2001 CPIS, of disaggregated third party holdings is expected to be equally useful.

Such information is also used by countries of the European Monetary Union (EMU) to estimate area-wide IIP data on portfolio investment assets and liabilities. There is no other way to obtain net estimates that exclude intra-EMU claims.

This group mainly comprises countries and territories in which offshore booking centers are located. See Section V.

Some countries that participated in the 1997 CPIS did not, for reasons of confidentiality, disclose positions with certain of their partner countries. Confidential data were reported separately to the IMF and recorded as such.

Individual country statistics are not published in the World Economic Outlook; however, WEO files contain information for individual countries.