Data Sources Used by National Compilers

International Monetary Fund
Published Date:
June 1992
  • ShareShare
Show Summary Details
Neil Patterson


Countries have developed systems for collecting and aggregating data on international transactions that reflect their institutional structures and capabilities. In nearly all cases, the basic conceptual framework for organizing the data is broadly compatible with the Fund’s Balance of Payments Manual.

Most countries use a combination of sources to compile their balance of payments statements. Even for a single component of the accounts, such as direct investment, a number of sources may be used. It is possible, however, to focus on three features of data sources. First, data collection may be based on the reporting of individual transactions or on the reporting of aggregates. Second, data may be collected by the statistical agency from an intermediary (say, a dealer that handles security transactions for clients) or directly from a transactor. Third, data may be collected on transactions or on stocks of assets and liabilities.

The Fund’s draft Balance of Payments Compilation Guide sets out a classification scheme for data sources. The collection of data on international capital flows falls mainly into four categories: international transaction reporting systems (ITRS); enterprise surveys; official sources; and foreign sources. These data sources are described and discussed in the next section. The final section discusses how these data sources are used to compile three of the four main capital account categories: direct investment, portfolio investment, and other capital.1

Types of Data Sources

International Transaction Reporting Systems2

ITRS encompass both “simple” foreign exchange reporting systems (FERS) and “modified” versions of FERS. Broadly, an FERS measures purchases and sales of foreign exchange by resident banks, classified by the source of funds in the case of purchases and by the purpose for which the funds are required in the case of sales. These systems have usually developed as a by-product of foreign exchange control administration. Therefore, they usually involve the reporting of residents’ individual transactions to resident banks, which then transmit data to the statistical compiler.

FERS provide data on transactions in foreign currencies. These data are suitable for balance of payments measurement in countries where tight foreign exchange controls exist under which (a) nonbank residents are not permitted to hold foreign currency accounts with resident banks or accounts with nonresident banks, (b) nonresidents cannot hold accounts with resident banks, and (c) residents cannot engage in noncash transactions, such as the granting of trade credit and changes in intercompany accounts.

As foreign exchange regulations have been relaxed or abolished, compilers in many countries have modified their reporting systems. These modified systems still depend on the reporting of individual foreign exchange records via bank intermediaries, but these records have been supplemented in some of the following ways: (a) residents with either foreign currency accounts at resident banks or accounts with nonresident banks report details of transactions through these accounts; (b) residents report details of domestic currency transactions with nonresidents; (c) transactions that take place through nonresident bank accounts with resident banks are monitored; and (d) enterprises report details of noncash transactions.

ITRS vary in their comprehensiveness and internal consistency. Some can be described as closed systems. They reconcile individual transactions with corresponding changes in foreign accounts. For example, each form submitted by a client to a bank is matched with an entry in the bank’s foreign currency account. An open system does not require complete accounting and reconciliation. Often the latter is a partial system, in that certain transactions are excluded from the reporting system because the data are collected through other channels.

A survey of balance of payments compilers, conducted by the Fund’s Statistics Department in 1990 (BOP Compilers Survey), obtained information on the data sources used by 58 countries. The survey found that 49 of these countries used ITRS as a data source for at least some balance of payments transactions, although frequently in conjunction with other sources,3 Of the ITRS countries, 28—of which 1 was an industrial country and the rest were developing countries—covered only foreign currency transactions, and the remaining 21—including 11 industrial countries—measured both foreign currency and domestic currency transactions. The systems of 14 surveyed countries—including 6 industrial countries—were described as being “closed”; 24 systems—including 4 in industrial countries—were described as “open”; and 8 systems were described as “partial.”

Neil Patterson was a member of the Working Party’s technical staff. He is now a Deputy Division Chief in the IMF’s Statistics Department.

Enterprise Surveys

Enterprise surveys also cover a wide range of approaches. In them, the measure is usually the sum of all the specified transactions of a business entity, but data on balance-sheet positions may be collected as well, or even instead of transactions data. The draft Compilation Guide distinguishes between

  • —the collection of data from resident asset holders or obligors;

  • —collections from resident intermediaries (such as securities dealers and fund managers); and

  • —collections from resident banks and financial institutions.

The first and third of these categories relate to “own account” activity, while the second relates to activity undertaken on behalf of customers. The third category covers collections that are often conducted by compilers of domestic financial statistics and that may yield only balance-sheet data.

In some countries, enterprise surveys involve large-scale collections covering thousands of enterprises. In others, collections from a few large companies may be all that is required, particularly if these are the only companies that engage in significant international transactions or if the collection is used to supplement other data sources. In enterprise surveys, particular attention must be paid by compilers to ensuring satisfactory coverage of transactors.

Of the 58 respondents in the BOP Compilers Survey, 34 countries—including 11 industrial country respondents—said that they conducted surveys of asset holders and obligors; 17—including 10 industrial country respondents—said that they conducted collections from intermediaries; and 39—including 12 industrial country respondents—said that they conducted banking collections. Since there were more data sources than respondents, it was evident that many countries used more than one source.

Official Sources

“Official sources” cover a wide range of other data sources. They may be government financial accounts, records of the particular agency responsible for debt management, foreign investment approvals, or applications to acquire foreign exchange. In some instances, these sources may yield individual transactions data. In others, such as foreign investment approvals, they may provide data based on anticipated rather than actual flows, or on initial investments but not on incremental investments or disinvestments.

Official sources may often provide valuable data at relatively small cost to the balance of payments compiler. However, the statistician may have limited influence over these collections, since they have usually been designed for administrative or revenue purposes rather than to meet statistical needs. In this respect, these sources differ from ITRS and enterprise surveys, which have usually been designed for balance of payments purposes.

Foreign Sources

Data from foreign sources, namely partner countries or international organizations, are occasionally used in balance of payments compilation. For example, the Fund and the BIS collect data on the external claims and liabilities of banks vis-à-vis foreign banks and non-banks. Individual countries’ banking claims on and liabilities to the rest of the world can be derived from these figures. Some countries use these statistics to derive certain capital flow items in the absence of complete national data.

Data Sources Used in Capital Account Compilation


Table 1 is derived from responses to the BOP Compilers Survey and summarizes the use of various data sources in the compilation of capital account items.4 The table shows the number of data sources used for the three main types of capital flows (with direct investment split into reinvested earnings and other flows). Many countries used more than one source to construct a capital account item.

Table 1.Data Sources Used by 58 Countries(Number of sources)
Reinvested earningsOther direct investmentPortfolio investmentOther capital
Data sourcesIndustrial countriesDeveloping countriesTotalIndustrial countriesDeveloping countriesTotalIndustrial countriesDeveloping countriesTotalIndustrial countriesDeveloping countriesTotal
Enterprise surveys
Data supplied by

resident asset

holders and obligors
Data supplied by


Data supplied by

Data from official

Data from foreign

Item not measured or

Source: BOP Compilers Survey, conducted by the Fund’s Statistics Department in 1990.

International transaction reporting systems.

Source: BOP Compilers Survey, conducted by the Fund’s Statistics Department in 1990.

International transaction reporting systems.

Direct Investment

According to the survey, the most frequently used sources of direct investment data were enterprise surveys and ITRS. A selection of responses to the BOP Compilers Survey was examined in further detail for the direct investment category. The replies of 54 compilers were classified by the principal data source used (see Table 2).

Table 2.Direct Investment: Principal Data Sources in 54 Countries(Number of countries)
Data sourcescountriescountriesTotal
ITRS, supplemented by
enterprise surveys515
Enterprise surveys51621
Data From official sources110311
Items not measured33
Source: BOP Compilers Survey

These measure both foreign currency and domestic currency transactions.

These measure foreign currency transactions only. In a few cases, they were supplemented by data from official sources.

In a few cases, these were supplemented by enterprise surveys.

Source: BOP Compilers Survey

These measure both foreign currency and domestic currency transactions.

These measure foreign currency transactions only. In a few cases, they were supplemented by data from official sources.

In a few cases, these were supplemented by enterprise surveys.

Industrial Countries

Of the 15 industrial countries covered in Table 2,9 national compilers reported using ITRS to compile direct investment data, 5 reported using only enterprise surveys, and 1 reported using official sources. Of the group of nine, all the collection systems were based on compulsory reporting of individual transactions—from the banking system, directly from enterprises, or from a combination of both.5

All nine compilers reported that their systems extended beyond FERS to cover domestic currency transactions as well as foreign currency transactions. Nonresidents generally were permitted to hold accounts with domestic banks, and residents were allowed to hold foreign currency accounts with either domestic or foreign banks. Most compilers indicated that their systems were designed to record transactions occurring through these accounts.

Of the nine compilers, five—including Germany and the Netherlands—conducted supplementary surveys to collect additional direct investment data. Two of the five, Finland and Sweden, have introduced surveys recently. All five compilers collected and published data on reinvested earnings; the others did not. The same five reported that noncash transactions between a parent and its affiliate abroad were recorded in the direct investment statistics; the others did not. A further compiler, Belgium, was developing a supplementary survey.

Of the five compilers that used only enterprise surveys, all conducted annual (or less frequent) benchmark surveys that covered a very large number of enterprises and quarterly surveys that covered a sample of these enterprises, including the more important ones,6 The five compilers generally reported that few enterprises (those with minimal investments) were excluded from their benchmark surveys. The coverage in their quarterly surveys was quite high (80 percent or more, by total value of the direct investment activity of all enterprises). Of the five compilers, four recorded reinvested earnings in their balance of payments statistics.7

The fifteenth compiler reported using official records and data supplied by the authorities responsible for foreign investment approval.

Developing Countries

The responses of 39 developing countries are covered in Table 2. Broadly, 10 compilers reported using ITRS to compile direct investment data, 16 reported using enterprise surveys, 10 reported using official sources, and 3 reported that direct investment transactions were not measured.

Of the group of 10 compilers that used ITRS, none had extended their systems to cover domestic currency as well as foreign currency transactions. A few of these systems were extensions of the basic FERS model; several compilers reported that their systems captured resident holdings of foreign currency accounts. Several countries supplemented their FERS using data from official sources and measured reinvested earnings.

Of the group of 16 countries that used enterprise surveys, 15 reported that they collected reinvested earnings data (in some cases not separately identified in their statistics) and 4 reported that they supplemented their survey data using FERS. The surveys conducted by this group of compilers ranged from comprehensive inquiries to limited inquiries covering only a small number of firms.

The 10 compilers that used official data most frequently described their information source as foreign investment approvals or applications to acquire foreign exchange. In several instances, these two types of data were combined, and in one case they were supplemented by tax records. In some instances, the official data were supplemented by surveys to obtain reinvested earnings data. Reinvested earnings were compiled by six of the ten economies, although not always from the survey source.

Portfolio Investment

Table 1 shows that the 58 respondents in the BOP Compilers Survey reported 104 types of data sources for the measurement of portfolio investment; hence many countries used several sources. The data were obtained in most countries from financial intermediaries, either as the product of an ITRS or as a statistical report, but quite a few countries collected data directly from issuers and holders of securities or from official sources. For example, the United States and Germany relied heavily on reports from financial intermediaries but also used data supplied by issuers and holders.

Investigations by the Working Party on the Measurement of International Capital Flows determined that, in most countries, portfolio investment data were based on the collection of transactions information. Only a few countries—one was Switzerland—derived portfolio data from changes in stock positions. Financial intermediaries played a dominant role in the provision of portfolio data, especially for secondary-market transactions. Across countries, practices varied considerably regarding how much securities detail was collected from intermediaries. France and Canada collected a large amount of detail on individual securities bought and sold to nonresidents, but in many other countries the transactions data were collected in highly aggregated form.

Other Capital

In view of the diversity of transactions covered in “other capital,” it is not surprising that Table 1 shows that the 58 respondents in the BOP Compilers Survey reported 118 different types of data sources. Data were obtained most frequently from financial institutions as well as from resident enterprises and official sources. In the absence of complete domestic data, a small number of countries—including Canada, Germany, and the United Kingdom—compiled certain items of their non-bank sector from the international banking statistics of the Fund and BIS.

The frequent use of banking reports is partly because the transactions of resident banks form an important part of “other capital.” Information on the sources of banking flows for 38 countries was extracted from the Working Party’s Special Questionnaire on international Capital Flows. This information is shown in Table 3.

In the case of banking flows, nearly all countries relied on data reported by resident banks and a majority—including Canada, France, Germany, India, Italy, Japan, the United Kingdom, and the United States—based their flow data on changes in the positions reported by banks, sometimes in combination with data reported on a flow basis.

Table 3.Banking Flows: Data Sources in 38 Countries(Number of countries)
Flow data reported by resident banks56
Position data reported by resident banks1617
Combination of flow and position data
reported by resident banks119
Data from foreign sources21
Other, including combinations of listed
Source: Special Questionnaire on International Capital Flows.
Source: Special Questionnaire on International Capital Flows.

    Other Resources Citing This Publication