To a considerable degree the existence of large discrepancies in world aggregates of current account transactions and in many national balance of payments statements is the result of the great and growing diversity and complexity of international economic relationships. Essentially a country’s balance of payments account should be a detailed record of all the transactions in a given period between residents of that country and nonresidents. The broad principles and definitions to be employed by all members of the Fund are set forth in the Fund’s Balance of Payments Manual, and the Fund staff regularly collects and publishes national statements and aggregations of national statements based on those principles. (See Appendix I.) Of course, strict conformity with these principles and definitions is not always possible, and interpretations differ, leading to some differences in the way countries report the same transaction. However, practical obstacles and difficulties in compiling the comprehensive information needed to produce the accounts and the lack of adequate reports from some countries on some types of transactions are far more important sources of actual gaps and discrepancies in the accounts—the main focus of the Working Party.

Compilers of the international accounts are not in command of a unified and internally consistent set of accounting data, as is the case for any business concern. Instead, compilers use a variety of unrelated sources: for instance, merchandise trade figures are derived from customs documents, freight charges from reports by shipping organizations, and the resulting changes in international bank accounts from either banks’ balance sheets or from transaction records compiled by banks or others. The main types of data collection systems are discussed below. They rely on reports based on individual exchange transactions supplied by banks or the transactors themselves, on reports from transactors that give results over a period rather than attempting to identify and measure each transaction, or from some combination of those sources. Such methods must be supplemented in nearly all cases by formulating estimates for missing elements. It is clear from this study that there is no one method that is best for all purposes, and even the most comprehensive statistical systems must be supplemented by the imaginative use of estimates and collateral data from many sources.

Some of the adjustments to the reported data shown in detail in other chapters are intended to correct asymmetries. They simply involve transpositions between different components of the current account (e.g., between direct investment income and “other” income) and do not affect the overall world current account discrepancy. It will be seen, however, that the larger part of the significant adjustments, which are aimed at eliminating discrepancies as such, reflect failures of the compilation procedures used. Among these, the most important are the following:

(1) The complete omission by many countries of transactions that are not caught in a system based on currency exchange or payment records, especially income earned abroad on direct investments that is not remitted as dividends to the country of the investor. Virtually the only way of obtaining data on the undistributed profits of direct investments is to get it directly from the firms involved.

(2) Similarly, the exchange or payment record does not capture unremitted income on cross-border bank accounts or securities, and reliance on other types of reports submitted by the asset holders themselves is likely to be ineffective. In this case the compilers’ principal recourse is to make use of all available data on the stocks of such cross-border assets (or liabilities) to create estimates of income credits or debits. There are many other types of transactions between residents and others that are not settled by a payment or an exchange transaction recorded in the country of the compiler. This is certainly the case, for instance, when a country’s ship owners or operators conduct most of their business abroad. In such cases, any exchange records must be supplemented by information supplied by the resident ship operators or by shipping agents.

(3) From the point of view of world aggregations, there is also a gap when some countries or economies either do not prepare or publish balance of payments accounts or compile inadequate accounts. This gap is only partly filled by estimates made by the Fund staff, so that the potential for omissions or errors is considerable.

In the course of conducting this study the Working Party has had to deal with each of these major deficiencies in compilation procedures. It turned out to be possible to adopt methods that could be used either to close the gaps or to point the way for further work that had a reasonable chance of reducing future gaps. These methods can also be adopted by national compilers, but they are not substitutes for improvements in basic statistical programs that are needed where existing statistics are clearly deficient.


The assignment of the Working Party to examine the discrepancies in the global current account should be sharply differentiated from the problem of the residual, or “errors and omissions,” item in national balance of payments statements. Residuals in national statements reflect the compilation difficulties mentioned above; the national compilers are either omitting some transaction by residents or are not obtaining matching data for the two sides of each transaction, as is implicit in the double-entry bookkeeping system. Consequently, when the totals of a country’s international credits and debits do not add to zero, as they should, it is not possible to determine with any certainty whether the error is in the compilation of current or capital transactions, or whether the residual, which is a net result, reflects a set of large or small gross errors or omissions. In practice it is extremely difficult to trace the sources or nature of the residual in national statements simply by analyzing the data as given, though sometimes the surrounding circumstances give some clues as to the most likely type of omission or error.

In a sense the task of the Working Party is more straightforward. Each payment for goods and services by one country should have a counterpart in the receipts of another country, so that when aggregated over all countries, the individual components of the current account, as well as the total, should equal zero. There is always some possibility that what appears in one country to be a current transaction will appear in the partner country or countries to be a capital account transaction, but the scope for such transpositions is not large relative to the size of the overall current account discrepancy. Thus, the Working Party was able to concentrate on the reasons why particular kinds of current account transactions were out of balance or were recorded in different current account categories. Even when concentrating on the current account, however, it is not possible to escape the problems of obtaining accurate data on international capital flows or stocks of assets. Filling the gaps in the international income accounts depends crucially on being able to establish reasonable levels of cross-border assets and liabilities, and this, in turn, depends in large part on being able to measure resident-nonresident capital transactions. As noted in Chapter I, however, this procedure is less useful for direct investments.

To some extent, therefore, the problems of national and global residuals are intertwined, and further study of the nature and causes of the global current account discrepancy may also help to identify the character of national residuals, at least in broad terms, and vice versa. However, the distinguishing characteristic of the global discrepancies is that they result from inconsistencies in the amounts actually reported by different countries for the same transaction—either party or both parties reporting incorrectly, or one not reporting at all. Consequently, the problem of establishing consistency in reporting across countries is different from, but related to, the problem of compiling consistent accounts for a single country.


Each country has evolved over time a method of collecting and aggregating data on its international transactions that reflects its institutions, capabilities, and interests. In nearly all cases, the basic statistical framework which organizes these data is compatible with the list of standard components recommended by the Fund, though with many variations. However, compilation methods and the quality of the results vary considerably from country to country.

Two main types of data collection methods can be distinguished, on the basis of the sources of the data. One type is based on the reporting of the data by intermediaries; the most important example is the use of the exchange transactions data typically compiled from the records maintained by commercial banks. The second type is based on the reporting of the data by the transactors themselves, in the form of either reports of individual transactions or reports summarizing groups of transactions over time.

In almost all cases, the collection of raw data is only part of the process through which entries are finally made in the international accounts. In many cases estimation procedures must be developed, drawing on statistics collected to permit computation of the balance of payments, statistics collected for other purposes, or statistics compiled in other countries or by international institutions. The results of the compilations and of the estimating procedures have to be verified wherever possible by checking with other information available within the country or, sometimes, with information available in partner countries.

An important consideration in all these systems is the extent to which the office responsible for preparation of the accounts has authority to obtain essential information from residents carrying out transactions with nonresidents. Such authority can range from an absolute requirement for reporting transactions under an exchange control regime to a purely voluntary system of questionnaire surveys. Closely related to the question of authority is the question of the burden and expense of obtaining this information, which involves both the statistical offices using the data and the public supplying the information.

While some countries rely on a single source for basic data on investment income, such as the records derived from an exchange control system, most countries find it necessary to use a variety of methods to cover this increasingly elusive international sector of their economies. Indeed, the hallmark of an effective data gathering system is flexibility and adaptability to the types of transactions involved and to changing circumstances. In this section we are mainly concerned with the procedures used to collect data and to prepare estimates related to international investment income. To a considerable extent these procedures are linked to those used in obtaining information on international capital flows. We do not consider the problems connected with sectors of the international accounts, especially merchandise trade, which were outside the terms of reference of the Working Party. However, problems of the statistics used in the transportation and unrequited transfer accounts are discussed in Chapters VII and VIII.

a. Data Systems Based on Reports from Intermediaries

In these systems the data are provided to the balance of payments compilers by intermediaries rather than by the transactors. The most widely used example is the exchange record system, in which the intermediary is most often the resident banking system, which administers the exchange controls. In some countries other forms of reporting by intermediaries exist: for example, in the United States, there is reporting by banks of foreign assets and liabilities held on behalf of their domestic customers, and reporting by securities firms of transactions in long-term securities with nonresidents which the former undertake on behalf of their domestic and foreign customers. Data systems based on reports by intermediaries utilize the unique position of the intermediary as the possessor of comprehensive information on the relevant international transactions or positions, which under existing conditions could not be obtained or would be very difficult to obtain in any other way—for example, because the transactors are widely dispersed, are very numerous, or would have difficulty in providing the data. Under some conditions the use of intermediary reporting has considerable advantages. The importance and widespread use of exchange record systems make it useful to discuss these systems in some detail.

The exchange record system of compiling balance of payments data is widely used by countries which have, or have had, some form of exchange control. During all or part of the period since the establishment of the Fund, many countries have maintained systems of exchange controls with varying degrees of comprehensiveness and severity. Such systems have been viewed as a ready source of presumably consistent data for the balance of payments accounts and are usually based on sufficiently strong authority to assure a substantial degree of compliance. For some countries and under some conditions the exchange record may be the most effective means of obtaining the necessary data. In principle, such records should contain information on virtually all of a country’s international transactions that are required to be settled through the control mechanism. However, even the more comprehensive exchange control systems do not purport to cover all of the international transactions of a country, and in some cases the information collected is based on permits granted, which may not correspond, either in timing or amount, to actual transactions.

There are substantial difficulties with reliance on bank-reported exchange transactions. The major problem is the amount of international business conducted without the use of exchange conversions through resident banks. For example, multinational corporations can easily retain earnings abroad and can readily finance their operations by borrowing abroad via foreign subsidiaries and moving funds through their network of intercompany accounts. Similarly, individuals who have accumulated investments abroad can retain or spend the earnings abroad thereafter, without any exchange transaction being recorded. The exchange record system puts a heavy reporting burden on the banks, which must record and tabulate a very large volume of foreign exchange transactions. The burden is considerably increased for banks with many resident branches. There is also a substantial difficulty for banks in creating records consistent with balance of payments criteria; for example, the basis of residence for balance of payments purposes may differ from the address records required for the bank’s own purposes, and distinctions required between loans and securities and between short-term and long-term items may require breakdowns in records beyond what is needed for banking purposes. Moreover, where exchange controls are being liberalized or dismantled, it is very likely that the statistics based on these systems will deteriorate in quality and coverage. Finally, wherever there are systems of this kind, incentives may be created for transactors to conceal their operations, and this, too, can lead to large gaps in the statistical record. Concealment, whether deliberate or not, is now widespread because of the use of offshore financial centers and innovative financial instruments.

Data derived from the exchange record are still a part of the balance of payments reporting systems of a number of industrial countries, but some of these countries are adapting their systems to decrease or eliminate their reliance on the exchange record, in line with the elimination of all or part of the exchange controls themselves. The exchange record remains a significant source of data in many developing countries. About 30 percent of the total private portfolio investment income credits and 45 percent of the debits reported on the Working Party’s investment income questionnaire are related to exchange record systems.

b. Data Systems Based on Reports from Transactors

Data collection systems based on reports from the individuals, business enterprises, or other economic entities involved in transactions with nonresidents, rather than on reports from intermediaries, take a variety of forms. In some cases—the Federal Republic of Germany is the most prominent example—the reports cover mainly separate transactions, which are coded and tabulated by the balance of payments compilers. In other cases the transactors report periodically the summary results of their transactions with nonresidents on prescribed statistical forms.

In transactor reporting systems, the reports can cover transactions that do not require bank settlements, such as reinvested earnings and intercompany account transactions of direct investment enterprises, as well as financial or other transactions effected through nonresident intermediaries. These systems can therefore provide much more complete coverage of some types of external transactions than are captured by exchange record systems, but the problem is to locate and obtain reports from the relevant transactors.

Systems based on individual transaction reports typically evolve from exchange record systems, in cases where exchange controls have been eliminated. However, reports from individual transactors may also be used, on a sampling basis, to obtain data (such as average daily expenditures by tourists) that can be used, in conjunction with other data, to produce estimates of some elements in the international accounts.

In one common type of system the business enterprises or other transactors report periodically the results of their transactions with nonresidents on statistical forms prescribed by the balance of payments compilers. These forms can include (as in the U.S. system) annual or quarterly surveys of direct investment enterprises, providing balance sheet and income statement information; monthly or quarterly statements of bank and nonbank asset and liability positions vis-à-vis nonresidents; and monthly statements (by financial intermediaries) of transactions with nonresidents in portfolio securities. In this type of system the enterprises conducting the transactions are asked to summarize and tabulate the results of individual transactions with nonresidents, so that the balance of payments compilers deal only with the summary statistical reports. In general, the statistical reports are designed to conform with the normal accounting statements of business enterprises, or with the records normally maintained by banks for their own purposes. However, there often are major difficulties in conforming to balance of payments definitions and in obtaining consistency between the reports of different types of enterprises. The accounts of business enterprises may use different definitions than the balance of payments accounts, and business accounting years may differ from the calendar years preferred for the balance of payments. Similarly, some types of enterprise may account for their transactions in unique ways—for example, the insurance industry.

As noted in the following section, questionnaire surveys on a regular schedule are employed by many industrial, and by some developing, countries to obtain information from multinational corporations on their direct investments. In addition, they are widely used to obtain data on other international service transactions, such as shipping and transportation transactions, fees and royalties, institutional remittances, engineering and construction fees, etc. An advantage of surveys is that they can be limited to samples, or to the relatively small number of major participants in a particular type of international transaction, and can be tailored to specific types of transactions. To supplement sample surveys there might be occasional more complete surveys, or recourse to other sources, to develop universe estimates.

While the use of periodic questionnaires to collect data on international investment income seems to be superior for covering direct investment income, it is not likely to be effective for other interest or dividend credits or debits. This is partly because the potential reporters are very numerous and difficult to locate and partly because, even if they were all identified, it would be very difficult to obtain accurate and timely reporting. Under the best of circumstances the collection of data on such income is unlikely to be fully comprehensive, and the collection systems usually have to be supplemented with an estimating technique of some kind.

National compilers also draw on administrative records, such as tax records, the accounts of official agencies, or records obtained by regulatory bodies for other purposes. The difficulty often encountered is that the important definitional concepts of the balance of payments accounts, especially the question of residence, do not fit easily into many of these administrative records.

c. Estimation Procedures

As noted above, many, or perhaps all, countries find it necessary to estimate some of their balance of payments figures, as well as to collect basic data through various forms of reporting arrangements. Estimation procedures have a place in virtually all segments of the balance of payments accounts. They are employed whenever the results of data collection systems are considered to fall seriously short of providing reasonably complete figures for a given category of data, or where no feasible basis exists for direct reporting of the data required. Often, the data collected may be from a sample which provides the basis for estimates of the universe being measured; the travel accounts are typically based on such an estimating process. Another example would be the situation, covered in Chapter V, where data on asset stocks may be available but no comparable data for income is available except that produced by employing an estimation process using information on rates of return. As a practical matter, many countries which have advanced data collection systems rely on estimation for some parts of their balance of payments data; and some countries, of necessity, make very substantial use of estimating techniques.

Estimates are used for a wide range of items in both the current and capital accounts, including travel and tourist expenditure and portfolio investment income. It is more difficult to rely on estimates to derive capital flows, since they are much more volatile, but where figures on asset and liability stocks are available, as in the case of banking positions, they can be used to derive both income and capital flows. Valid estimating techniques are often based on data generated by a country’s own data collection systems and other internal information. However, where such domestic information is unavailable or inadequate, data from external sources, such as data provided by partner countries in direct investment situations and data generated by international organizations can sometimes be very helpful. The international banking data published by the BIS and the Fund and the external debt data published by the OECD are important examples of the latter.

It is important that balance of payments compilers exercise ingenuity in their use of estimating techniques and that they utilize external sources of relevant information, as well as information produced by their own data collection systems. Often reference to external sources can serve as a check on national data.


The systems actually used by countries in the compilation of their balance of payments statistics have been shaped by each country’s particular circumstances. It is beyond the scope of this report to discuss entire balance of payments systems, but a number of systems are described in detail in documents published by the countries, and the systems of several European countries are described in publications of Eurostat. However, a rather comprehensive picture of data systems used for the compilation of the investment income accounts can be presented on the basis of the Special Questionnaire on International Investment Income Accounts employed by the Working Party for the present study. (A copy of the questionnaire appears in Appendix II.)

The investment income questionnaire contained a checklist of the types of data sources used by the respondent countries for their data on direct investment income and private portfolio investment income in 1983 and also asked for a detailed description of the origins of each of the income items reported on the questionnaire. The range of sources used is indicated in Table 7, which is based on the responses to the questionnaire. The breakdowns provided cannot be precise, however, because many systems are composites. Most countries still depend heavily on reporting by resident banks, though it is not clear in some cases whether the banks report only for their own accounts or for both themselves and their customers. It is also shown that data on direct investment income is often obtained by questionnaires, but that this method is used less widely to obtain information on portfolio investment income.

Table 7


article image
Source: Working Party’s Questionnaire on International Investment Income Accounts.

Debits only.

Credits only.

Six countries which responded to the questionnaire did not provide information on data sources and reported negligible amounts of investment income for 1983.

The relative quantitative importance of the sources of the data reported by the countries which provided data on the questionnaire is shown in Table 8. Data for the 17 countries which indicated more than one source for data reported on the investment income questionnaire are shown in the first four columns of the table as derived from “multiple sources.” The last four columns contain an estimated distribution of the “multiple source” totals among the basic sources of the data for 9 of the 17 countries, based on additional information provided by several of the countries and on staff estimates for others where a basis for estimation could be found.

Table 8


(In billions of U.S. dollars)

article image
Source: Working Party’s Questionnaire on International Investment Income Accounts and estimates provided by selected participating countries.

Incorporates estimated distribution to individual data sources of amounts initially reported by nine countries as derived from multiple sources.

Includes unspecified credits of less than $50 million.

It should be noted that the investment income questionnaire was directed toward countries which provide the overwhelming bulk of the balance of payments data reported to the Fund. The coverage of that questionnaire is shown in Appendix II. The total amount of all forms of 1983 investment income, official and private, reported by questionnaire respondents constituted 90 percent of the credits and 91 percent of the debits for all countries in the Fund’s Balance of Payments Statistics Yearbook for 1985.

The checklist on data systems used in the questionnaire was not structured in terms of the distinction between exchange record systems and transactions systems; the information provided is based on the means of calculating the investment income accounts rather than on the basis of the principles of the data collection. In Table 8, the category “Reports from [resident] banks” probably comes closest to representing the systems based on exchange transaction types of recording, but there are cases where the bank is merely providing information on its own operations. As shown in the table, countries relying on reporting by resident banks accounted for only a very small part of direct investment income (4 percent of credits and 7 percent of debits, including an estimate for the share of bank reporting in composite systems). However, reporting via resident banks was the most important source for data on other private investment income. Including an estimate for the share of bank reporting in composite systems, information obtained via resident banks accounted for over half of both credits and debits.

Information on direct investment income was obtained primarily through the use of questionnaires—86 percent of credits and 60 percent of debits—including an estimate for questionnaire reporting in composite systems.

The technique of using data on asset stocks and yields as the basis for estimates is unique to the investment income accounts and was quite important but somewhat one-sided, covering about 37 percent of the credits but only 22 percent of the debits (including an estimate for this data source in composite data systems). Of course, these figures reflect only the amounts reported on the investment income questionnaire—in Chapter V it is made clear that a major deficiency of balance of payments compilation practices is the failure to account adequately for income earned by resident nonbanks on cross-border deposits.


The Fund has a central role in the compilation and analysis of world balance of payments statistics. Article VIII of the Articles of Agreement requires Fund members to furnish balance of payments information to the Fund, and requires the Fund, in requesting data, to take into account the varying ability of members to provide the data. Article VIII, Section 5(c) also requires the Fund to act as a center for the collection and exchange of information on monetary and financial problems.2 In agreement with the United Nations, the Fund assumed responsibility for carrying forward the compilation of balance of payments statistics in standardized form (continuing the earlier work in this field by the League of Nations). In formulating its requests to members for balance of payments data, the Fund takes into account the needs of the United Nations and the international organizations associated with it.3 The Fund therefore has unique responsibilities in the development of adequate and consistent balance of payments statistics for the world community.

a. Conceptual Framework

In its work in this field since 1948, the Fund has made substantial progress in developing a set of balance of payments concepts and in securing their adoption to a large degree by its member countries. A considerable effort has been made to train balance of payments compilers in the application of the concepts to the situations of their countries and, to a lesser extent, to aid in applying certain techniques of data collection. New problems in these areas continue to emerge, however, and continuous attention to changes in the economic environment is necessary for adequate implementation of the concepts and for the production of accurate balance of payments data consistent across countries.

The Fund’s Balance of Payments Manual has been an important means of developing and promoting uniform balance of payments concepts. The first and second editions of the Manual—published in 1948 and 1950, respectively—were essentially basic guides for the reporting of balance of payments data to the Fund, presenting a set of tables for the classification and accumulation of the data. The third edition, published in 1961, went beyond definitions and descriptions of the categories to explain their rationales and described the basic concepts and accounting principles underlying the balance of payments accounts; this edition became “as much an introduction to the principles of balance of payments accounting as a guide to reporting.’4 The fourth edition of the Manual, published in 1977, continued the evolution and development of the document; it presented a more comprehensive discussion of balance of payments concepts and included for the first time a chapter on “analytic presentation” of the balance of payments and a recommended list of standard components. However, this version tended to de-emphasize the need for a uniform structure of reporting essential data, while enlarging on issues of definitional consistency with the United Nations System of National Accounts (SNA).

To implement the requirements of the Manual, the Fund’s Bureau of Statistics carries on technical training activities in balance of payments statistics through courses given in the IMF Institute, sends technical assistance missions to member countries, and conducts in-house training of country personnel temporarily assigned to the Bureau. Bureau staff may also participate in the annual consultations with, or in other missions to, Fund member countries, and may meet with country representatives during the Annual Meeting of the Fund to discuss important general statistical questions.

b. Publication and Analysis of Data

The Fund’s work in developing the basis for consistent recording of international transactions has several important applications. First of all, each country can produce a set of accounts that is internally consistent and also tends to be consistent with the accounts maintained by other countries, at least in principle. This provides an important part of the macroeconomic information needed by each country to appraise the performance of its economy and to conduct a meaningful dialogue with partner countries. Within the Fund itself, the international accounts of a country are an important tool in carrying out the round of country consultations that is an important part of the Fund’s responsibilities.

A second major application of the Fund’s work is to provide statistics on the balance of payments that can be aggregated across countries to provide a global and regional view of trends in the international economy. For this purpose, the uniformity of classification and procedure presented by the Manual is essential. The results of such aggregations are published regularly in detail by the Fund. (See Appendix I for details.) Other organizations prepare aggregations of international transactions that depend on the same basic accounting framework.

A third application, derived from the second, is the undertaking by the Fund’s Research Department of studies on the world economy, leading to the publication by the Fund of regular analytical reports on the state of the world economy and its prospects—including an analysis of developments in world current account positions and in financial flows among countries. It was in this application—especially in the World Economic Outlook—that the emergence of a world discrepancy on current account became an important issue.

c. Problems of Implementation

Although the Fund’s efforts over the years to achieve comprehensiveness and accuracy in the production of data on international transactions have achieved considerable success, some of the problems remain unsolved. Without doubt the Fund has made great progress in designing and promoting uniform balance of payments concepts to be adopted by its members, despite some continuing important differences in country practices. The Manual and the educational efforts of the Fund have been important tools in producing this result. Nevertheless, the emergence and persistence of large discrepancies in the world accounts show that it has not always been possible to adjust to changing circumstances and that there is still a great deal of work to be done in improving the accuracy and consistency of the underlying compilation processes employed by the member countries.

The recommendations in the concluding chapter of the Report reflect the urgency attached by the Working Party to the actions that must be initiated by the Fund if the compilation of balance of payments statistics is to be improved. These range from a review of the present version of the Fund’s Balance of Payments Manual and Balance of Payments Statistics Yearbook, to the production of statistics now omitted from the global aggregations, and, most importantly, to an intensification of efforts to strengthen the statistical base at the country level. Over the longer run, the Fund must take the lead in keeping up with changes in the ways international trade and finance are conducted and in modifying definitions and statistical practices accordingly. Moreover, this activity should continue to be consistent with developments in the System of National Accounts.

The Working Party believes that a practical plan can be devised to deal with these problems comprising a mix of actions on the part of national authorities and the Fund.