4. Data Availability

Neil Patterson, Marie Montanjees, Colleen Cardillo, and John Motala
Published Date:
September 2004
  • ShareShare
Show Summary Details

FDI Statistics Compiled by International Organizations

4.1 Many countries compile and disseminate in national publications and/or websites information on FDI, typically in the context of balance of payments and IIP statistics. To facilitate analysis, these statistics are often extended to provide geographic information on FDI transactions and positions vis-à-vis individual partner countries or regions. Many countries also compile FDI statistics broken down by industrial sector.

4.2 Various international and regional organizations—such as the IMF, the OECD, UNCTAD, and Eurostat1—have long-established programs for the collection and dissemination of these FDI statistics and also work with countries to improve data compilation practices in this area of statistics.2 The IMF’s statistics on direct investment are disseminated as a component of the balance of payments and IIP statistics, while the other organizations have publications devoted to FDI statistics.

4.3 The primary publications (all annual) of these four organizations that cover FDI statistics are:

  • IMF’s Balance of Payments Statistics Yearbook (BOPSY);

  • OECD’s International Direct Investment Statistics Yearbook;

  • UNCTAD’s World Investment Report; and

  • Eurostat’s European Union Foreign Direct Investment Yearbook.

4.4 Both the IMF and UNCTAD compile global statistics on FDI, whereas the focus of the OECD and Eurostat is on a narrower group of countries. UNCTAD’s World Investment Report contains world tables that aggregate national data on both FDI transactions and stocks and also provides a comprehensive analysis of issues related to FDI, which in recent years have included cross-border M&A (2000 edition) and operations of transnational corporations (2002 edition). The IMF’s BOPSY contains world tables on FDI capital flows and investment income. The OECD and Eurostat disseminate information on the geographic breakdown of FDI transactions and stocks vis-à-vis selected partner countries. The OECD and Eurostat also disseminate data on FDI transactions and stocks by industrial sector. The IMF does not disseminate geographic or industrial sector breakdowns of FDI statistics.

4.5 In late 2002, the OECD updated an earlier joint IMF-OECD survey to explain significant differences that were found to exist in the FDI data that each organization disseminated. Because both institutions adopt common methodological recommendations, the data that countries report to the two institutions should, in principle, be identical or very similar. The study, based on reports from 26 of the 30 OECD countries surveyed, showed that there were differences in the following areas:

  • Methodology and coverage. Twenty-two of the respondents apply the same methodology in reporting to the IMF and the OECD. However, four countries—Denmark, Japan, Korea, and the United States—apply different methodologies.3 In addition, the coverage of the data of three countries—Germany, Iceland, and Switzerland—differs between the two sets of data. Iceland and Switzerland exclude real estate, and Germany excludes both real estate and reverse investments from the data reported to the OECD, while including these items in the data reported to the IMF.

  • Data sources. Japan and the United States use different sources for the data reported to the IMF and those reported to the OECD.

  • Reporting periods. The data that Japan reports to the OECD are compiled on a fiscal year basis, while the data reported to the IMF are on a calendar basis.

  • Time of reporting. There are significant differences in the time of reporting the original data and revised data to the IMF and to the OECD. Only 11 countries send the primary data to both organizations at the same time; only 8 of those 11 countries also send revised data to both organizations at the same time for both flows and stocks.

  • Reporting of revisions. Only 19 countries send revisions of FDI transactions data to both organizations for the same number of years, while only 17 countries cover the same number of years for submitting revised FDI position data.

4.6 A comparison was also undertaken of the IMF and UNCTAD data sets for FDI stocks of individual countries for selected years. The statistics were found to be broadly comparable.4 Differences could largely be explained by the following: different publication release dates, which can involve the availability of revised data at different times; the data reported by member countries to the IMF are a mixture of both book values and, for a number of countries, market values, whereas UNCTAD uses book values to the extent possible; and the UNCTAD data contain estimates for countries that have not reported stock data or that do not compile these data.

FDI Statistics Shown in the IMF’s Balance of Payments Statistics Yearbook

4.7 FDI statistics are reported to the IMF’s Statistics Department as part of national submissions of balance of payments and IIP statistics for publication in BOPSY and International Financial Statistics (IFS). The 2002 BOPSY contained balance of payments statistics for 164 economies, about the same number as in the 1996 issue. Over that six-year period, about two dozen countries, mostly in Africa, fell behind in the reporting of balance of payments statistics, but this decline has been largely offset by an increase in the number of reporters from other parts of the world. Of the 164 countries, 140 reported data for 2000/2001.

4.8 Member countries are now reporting more component detail on FDI, which reflects the availability of data from new collections as well as improvements in classification. Figure 4.1 compares the number of countries that reported information for the components of FDI capital flows published in the 1996 and 2002 issues of BOPSY (using data for 1994 and for 2000 and 2001 from these publications).5 It shows that between the reporting periods 1994 and 2001, the number of countries reporting data on reinvested earnings on outward direct investment nearly doubled to 45, while reporting of reinvested earnings on inward direct investment increased by more than a quarter to 84. The number of countries reporting inward data on equity capital also increased significantly, from 92 in 1994 to 123 in 2001 and from 55 to 66 for the outward data. The increases for the data on other capital were from 76 to 91 for the inward data, and from 31 to 54 for the outward data.

Figure 4.1.Countries Reporting FDI Capital Flows to the IMF Statistics Department—1994, 2000, and 20011

1 In order to allow for late reporters, 21 countries reporting balance of payments data for 2000 were included in the counts.

4.9 There has been an even more significant improvement in reporting of IIP statistics on direct investment for inclusion in BOPSY, where the number of reporting countries more than doubled, from 30 to 70, as measured in a study conducted in January 2003 (see Figure 4.2).6 Within the IIP category of direct investment, many countries reported additional component detail in recent years, especially for the other capital component of FDI (that is, inter-company debt). About three times more countries reported data on other capital in 2001 than in 1994 (increases from 16 to 54 for the inward data and from 16 to 46 for the outward data). There were similarly significant increases in the numbers that reported data on equity capital and reinvested earnings (from 24 to 66 for the inward data and from 23 to 60 for the outward data).

Figure 4.2.Countries Reporting FDI Positions to the IMF Statistics Department—1994, 2000, and 20011

1In order to allow for late reporters, nine countries reporting IIP data for 2000 were included in the counts.

4.10 At the national level, countries sometimes supplement their published FDI statistics with various metadata (that is, information on the data) that explain to users of the statistics how the data were compiled, including information on the concepts and definitions employed. At the international level, surveys have been conducted across a range of countries to provide an understanding of FDI recording practices. Such a survey was conducted in connection with the IMF’s 1992Report on the Measurement of International Capital Flows(the Godeaux Report). More recently, there have been two joint IMF-OECD surveys. The first, which was conducted with respect to compilation and dissemination practices in 1997, was called the Survey of Implementation of Methodological Standards for Direct Investment (SIMSDI), and the second, which reviewed recording practices in 2001, was referred to as the 2001 SIMSDI update. The latter covered 61 of the 114 economies that participated in the 1997 SIMSDI.

4.11 The results of the 2001 SIMSDI update were published by the IMF in October 2003 as Foreign Direct Investment Statistics: How Countries Measure FDI 2001.7 The report provides detailed information on data availability, data sources, deviations from the international recommendations, etc. Key findings and selected FDI recording issues from the 2001 SIMSDI update are discussed in Chapter 6.8

FDI International Recommendations and Their Implementation

4.12 There are a number of methodological materials available to national compilers that provide conceptual and practical guidance in the compilation, analysis, and dissemination of FDI and other external sector statistics. The BPM5 provides guidance to member countries in the compilation of balance of payments and IIP statistics, and the OECD’s Benchmark Definition provides operational guidance on how FDI should be compiled to meet the internationally agreed recommendations.9 The two manuals are consistent. UNCTAD is developing a training manual on direct investment to help developing countries enhance the capacity of government agencies to compile data on FDI and the operations of transnational corporations. Some national compilers also prepare and disseminate material on their methodologies for FDI and other external sector statistics in national publications and on websites, including copies of direct investment and other statistical report forms employed. These, too, can be used as aids by national compilers in other countries in developing their FDI statistics.

4.13 Through training courses, workshops, and technical assistance programs, the international and regional organizations work with countries to implement international statistical methodologies and improve data compilation and dissemination practices.

4.14 Also, several European governments—Austria, Denmark, Sweden, Switzerland, and the United Kingdom—during 2000–02 have funded a series of projects to monitor and analyze private capital flows and stocks and private sector debt covering eight countries.10 These projects were undertaken with technical assistance from Development Finance International11 and regional organizations, such as the Macroeconomic and Financial Management Institute for Eastern and Southern Africa. This involved the development of surveys of private capital flows that helped gather information on FDI. The surveys exposed substantial discrepancies in the published balance of payments data.

4.15 Countries employ a variety of data sources and methods to compile FDI statistics, including enterprise surveys, bank reporting systems and international transactions reporting systems (ITRS),12 administrative data from exchange control or investment control authorities, partner country bilateral data, etc.13 The ability of compilers to implement the recommendations set out in the international statistical manuals may be constrained by the data sources used. For example, bank reports on foreign currency settlements may not be able to correctly distinguish direct investment transactions from portfolio investment transactions.

4.16 Specially designed forms submitted to enterprises are being increasingly used to gather comprehensive information on FDI transactions and stocks. For some components of FDI, such as reinvested earnings data, direct collections from enterprises, rather than bank reporting systems or administrative sources, are the only way that data can be readily obtained, although the reported data may not always conform with the international recommendations because of countries’ accounting or taxation regulations. FDI transactions compiled on the basis of investment intentions or similar information collected by investment control authorities are fraught with difficulties—not least because the planned investment may not take place, or the timing of the investment may not be known. Also, while stock data on FDI compiled by summing capital flow data can be used to fill gaps in the data, using this practice to compile FDI aggregates affects the accuracy of the estimates, since it does not take into account nontransaction changes arising from, for example, exchange rate and price changes, which can be significant.

4.17 Countries may not fully implement the recommendations for FDI statistics in the international statistical manuals for other reasons:

  • Difficulties in gathering the data (for example, resources may not be available to collect data on reinvested earnings or stock data, compilers may not have a legal mandate to collect data, the private sector may be reluctant to complete report forms, etc.);

  • Different accounting and/or administrative procedures (for example, accounting rules may recommend using the all-inclusive concept of income to measure earnings of direct investment enterprises);

  • Legal constraints (for example, investment or other laws may require, say, use of a 20 percent threshold to define a direct investment);

  • Unwillingness to adopt international recommendations that may be viewed as inconsistent with the particular needs and priorities of the compiling economy (for example, certain short-term intercompany debt flows, such as trade credits, may not be considered to represent FDI capital);

  • Confidentiality concerns may preclude disseminating certain datasets because of arrangements with respondents to keep their individually reported data confidential; or

  • Unfamiliarity on the part of compilers with aspects of BPM5 methodology.

4.18 Deviations at the national level from international statistical recommendations on FDI data compilation hamper cross-country comparisons and contribute to bilateral asymmetries in data comparisons with partner countries. They may also contribute to asymmetries with other macroeconomic statistics produced by the country. Similarly, deviations from international recommendations contribute to asymmetries in global FDI aggregations published by the IMF and UNCTAD, which also hamper cross-country data comparisons by users of statistics. More important, unless the statistical methodologies employed by countries are well documented, users may misinterpret the FDI data, and thus surveillance and other policy activities may be hampered.

Discrepancies in Global Balance of Payments Statistics on FDI

4.19 In principle, the global sum of outward flows of FDI recorded by the economies of the direct investor should be matched by the corresponding global sum of FDI inflows in the statistics compiled by the economies of the direct investment enterprises, since the debits of one economy are the credits of another. As is well known, there are discrepancies in the global balance of payments statistics because of gaps in coverage and the use by reporting countries of different definitions and classification systems. The substantial growth in private sector flows during the 1990s and the abolition of exchange controls, without concomitant broadening of statistical reporting, presented new challenges for statistical recording. These included difficulties in identifying the private sector transactors and, sometimes, the poor response rates to enterprise surveys.

4.20 The size of the statistical discrepancies in the global balance of payments statistics has been a cause of concern to the IMF in connection with the analytical implications for the IMF’s surveillance of its members’ economies and its biannual report on this activity, the World Economic Outlook.14 The IMF has undertaken two studies of the imbalances in statistics on global current and capital transactions.15 Both of these highlighted various country recording practices that have contributed to asymmetries in global (and bilateral) balance of payments statistics, including in the area of direct investment, and both made recommendations to address them. Many of the problem areas identified in these studies continue to exist, notwithstanding important statistical improvements made by countries in the past decade.

FDI capital flows

4.21Table 4.1 shows the global discrepancies underlying the statistics on FDI capital flows published in BOPSY. The data show a considerable widening of the discrepancies between recorded inflows and outflows of FDI in 2000 and 2001, to more than $100 billion.16 The data for current periods typically represent preliminary estimates and are thus subject to larger revisions than earlier periods.

Table 4.1.Discrepancies in Global FDI Capital Flows(Billions of U.S. dollars)
In reporting economy197.7327.9372.9461.4690.41076.61489.8729.2
Equity capital1
In reporting economy230.0246.5274.9439.9750.6975.7442.5
Reinvested earnings
In reporting economy1.238.443.565.161.283.8119.762.3
Other capital1
In reporting economy59.582.9121.4189.3242.2394.4224.4
Source: IMF, BOPSY, various issues.

4.22 Looking at the components of FDI capital—(i) equity capital, (ii) reinvested earnings,17 and (iii) other capital (intercompany debt)—the widening of the discrepancies in recent years relates to the recording of FDI transactions involving equity capital and intercompany debt. The discrepancy on inter-company debt transactions was especially large in 2000, when recorded FDI inflows of $394 billion were double the level of recorded FDI outflows.18

4.23 The statistical discrepancy in the recording of data on reinvested earnings in the financial account was in the opposite direction, showing an excess of recorded debits. This has been a long-standing trend. The discrepancy averaged $77 billion during the 1999–2001 period, 40 percent higher than the average of the previous three years. A significant contributor to the discrepancies in this item is likely to be the use of different accounting practices in defining the operating earnings of direct investment enterprises (discussed in Chapter 5).

FDI investment income

4.24 In the global current account, the mirror image of the above discrepancy on reinvested earnings reflects an excess of recorded credits (Table 4.2). Overall, the discrepancies on other FDI income (that is, dividends and distributed branch profits, and interest on intercompany debt from/to direct investors/direct investment enterprises) are smaller than those recorded for reinvested earnings and mostly show the opposite sign—an excess of recorded debits over credits. The reported data on FDI interest flows show a discrepancy of $23 billion in 2001 (22 percent of reported FDI interest receipts and payments). The imbalances on FDI interest flows—excess interest payments—are consistent with the positive imbalances (excess inflows) seen in the data for intercompany debt flows in Table 4.1. For dividends and distributed branch profits, the largest discrepancy was $22 billion in 1999 (8 percent of reported dividend and profit receipts and payments), but it has declined significantly since that time.

Table 4.2.Discrepancies in Global FDI Investment Income Flows(Billions of U.S. dollars)
Dividends and distributed
branch profits1
Reinvested earnings
Source: IMF, BOPSY, various issues.

Discrepancies in Bilateral FDI Stock Data

4.25 The imbalances in the global balance of payments statistics are also manifested in the bilateral balance of payments and IIP statistics disseminated by countries. Table 4.3 compares published bilateral data on inward and outward FDI stocks (on a book value basis) for six of the eight largest investing economies as at end-2001 (2001 bilateral data for France and Germany were not available when the table was prepared). Several of the comparisons reveal very large differences, some of which can be explained by different statistical practices employed by countries. Two comparisons are highlighted below.

Table 4.3.Bilateral Comparisons of FDI Stocks, Selected Countries, End-2001(Billions of U.S. dollars)
Reported by Investor CountryReported by Host CountryDiscrepancy
United States investment in:
Hong Kong SAR29.420.09.4
United Kingdom249.2188.460.8
Canadian investment in:
United States118.5108.69.9
United Kingdom24.913.111.8
Japanese investment in:
United States140.3159.0–18.7
Hong Kong SAR5.512.4–6.9
United Kingdom33.115.817.3
Hong Kong SAR investment in:
United States2.81.61.2
United Kingdom2.45.1–2.7
United Kingdom investment in:
United States231.7217.714.0
Hong Kong SAR9.44.15.3
Netherlands investment in:
United States87.2158.0–70.8
Hong Kong SAR1.314.5–13.2
United Kingdom26.490.0–63.6
Sources: United States: Survey of Current Business, July, 2002, Tables 2.2 and 3.2.Canada: Foreign Direct Investment, Daily, March 26,2003( Data by country are available in Table 376-0051 (CANSIM).Japan: Balance of Payments Monthly, April 2002, Tables 15.1 and 15.2.Hong Kong SAR: External Direct Investment Statistics of Hong Kong 2001. Appendix III, Tables IIIA and IIIC.United Kingdom Office for National Statistics:

4.26 In the case of the Netherlands, the data of the partner countries are consistently larger than the positions reflected in the Dutch data, both on outward investment to the Netherlands and inward investment from the Netherlands. For example, the United Kingdom shows FDI in the Netherlands of $247 billion at end-2001, which was more than $200 billion above the corresponding liability recorded by the Netherlands ($44 billion). In the other direction, the stock of Dutch FDI in the United Kingdom was $26 billion, compared with the U.K. estimate of $90 billion.

4.27 The major contributor to these discrepancies is the exclusion from the Dutch bilateral data of investment in/by Special Financial Institutions in the Netherlands, which are entities that are directly or indirectly owned by nonresidents and are used mainly to channel funds received from nonresidents to other nonresidents. In effect, the Dutch data suggest that most of the large U.K. investment in the Netherlands is ultimately employed in third countries, with the Netherlands acting as a conduit for tax or other reasons. The BPM5 recommends that the activities of special purpose entities (SPEs), such as the Dutch Special Financial Institutions, be included as part of direct investment. The statistical treatment of SPEs is discussed in Chapter 5.

4.28 Countries generally allocate stock positions bilaterally to the first, or immediate, country of ownership, which is in conformity with the debtor/creditor principle recommended in the BPM5. In Canadian statistics on outward FDI, when the first foreign direct investment enterprise is a foreign holding company, an attempt is made to attribute the investment to the “ultimate country of destination,” to improve the analytical usefulness of the FDI data. This may explain why outward Canadian FDI positions are higher than the inward positions recorded by the other countries in Table 4.3.

4.29 Bilateral comparisons can throw light on the accuracy of direct investment data, but care must be exercised in making use of them. In some instances, the bilateral discrepancies in the stock data discussed above would have no impact on the compiled national aggregates, since the differences may reflect different recording conventions for allocating positions to individual countries rather than indicators of potential gaps in the data. In other cases, bilateral discrepancies may flag gaps or recording errors in the national data. However, investigations into bilateral discrepancies are very resource-intensive and, because of confidentiality constraints, are difficult to resolve. This is perhaps especially so in the area of direct investment, where data on individual transactors or a small number of transactors can have a major impact on the statistics. An OECD direct investment workshop in 2001 decided not to attempt a detailed reconciliation of bilateral FDI data because the exercise was considered to be too difficult until such time as the methodologies that countries use to compile the data are more in line with the international statistical recommendations.

Other regional organizations, such as ASEAN and ECLAC, also disseminate direct investment statistics relating to their members and conduct various activities in support of the development and harmonization of FDI statistics in the regions. The World Bank also provides technical assistance in FDI statistics.

The IMF and OECD also have responsibility for the work on developing the methodology for FDI statistics.

Japan’s data reported to the OECD are currently based on investment notifications (a new system will be implemented in 2005); Korea’s data to the OECD are believed to be based on approval data; and the United States uses historical cost information for stock data reported to the OECD and market valuations in reporting IIP data to the IMF. In addition, FDI income reported to the OECD are net of withholding taxes, while those reported to the IMF are gross of these taxes. Denmark did not specify the differences in the methodology used for the two sets of data.

The IMF does not compile world tables on FDI stocks, since only 70 countries reported IIP statistics on FDI to the Statistics Department of the IMF. Reported IIP data appear in the individual country tables in Part 1 of BOPSY and in the IMF’s International Financial Statistics. To compile the world aggregates of FDI stocks published in the World Investment Report, UNCTAD accumulates data on reported FDI transactions for the large number of nonreporting countries, without making adjustments for nontransaction changes, such as price changes. UNCTAD made these estimates for about 120 countries in the case of the inward stock data for the year 2000, published in Annex Table B.3 of the 2002 edition of the World Investment Report.

In order to allow for late reporters, 21 countries reporting balance of payments data for 2000 were included in the counts, with supplemental information from the February 2003 IFS.

Of the 85 countries with IIP statements published in the October 2003 issue of IFS, at the time of the January 2003 study 8 did not report data for FDI stocks: Bangladesh, Costa Rica, Jordan, Lesotho, former Yugoslav Republic of Macedonia, Maldives, Mauritius, and Uganda. A further four countries did not have IIP data for the 2000–01 periods: Botswana, Senegal, Vanuatu, and Yemen. Three new countries have also begun to report IIP data since the study: Cyprus, Korea, and Ireland.

The IMF prepared summary metadata for each of the 61 countries that participated in the 2001 SIMSDI update, in consultation with the authorities in each country. The summary metadata indicate whether a country’s practices are in accordance with the international recommendations set out in the IMF’s BPM5 and the OECD’s Benchmark Definition. Cross-country comparison tables, organized by issue, were also prepared. The metadata for the 56 countries that agreed to make their information available to the general public, and the cross-country comparison tables, were posted on the IMF’s external website in October 2002 (see

Other methodological materials published by the IMF that provide information on direct investment include the Balance of Payments Compilation Guide (1995), Balance of Payments Textbook (1996), Financial Derivatives: A Supplement to the Fifth Edition of the Balance of Payments Manual (2000), International Investment Position—A Guide to Data Sources (2002), Recommended Treatment of Selected Direct Investment Transactions (2002), and “Classification of Financial Derivatives Involving Affiliated Enterprises in the Balance of Payments Statistics and the International Investment Position (IIP) Statement” (2002). All these documents are available in electronic format on the IMF’s website at

The Gambia, Ghana, Guyana, Malawi, Tanzania, Trinidad and Tobago, Uganda, and Zambia.

Development Finance International is a nonprofit economic and policy advisory, research, training, and capacity-building group based in London.

An ITRS measures individual balance of payments cash transactions that pass through domestic banks and may also measure (i) individual cash transactions through enterprise accounts abroad; (ii) noncash transactions; and (iii) stock positions. Statistics are compiled from forms submitted to domestic banks and may also be compiled from forms submitted by enterprises to the compiler.

The IMF’s Balance of Payments Compilation Guide reviews the advantages and disadvantages of the main sources of information on direct investment (Table 16.1, page 153).

See, for example, the essay on external imbalances in Box 2.1 in the September 2002 issue of the IMF’s World Economic Outlook.

Report on the World Current Account Discrepancy (September 1987) and Report on the Measurement of International Capital Flows— the Godeaux Report (IMF, September 1992).

In the context of global cross-border capital flows, the discrepancies on portfolio investment and other investment (for example, trade credits, deposits, loans) transactions have, until recently, been much larger than those seen on FDI transactions. For example, in 1997 and 1998, the discrepancies on recorded portfolio investment transactions exceeded $200 billion a year.

That is, the direct investor’s share (in proportion to equity held) of earnings not distributed as dividends by subsidiaries and associated enterprises and earnings of branches not remitted to the direct investor during the reporting period.

Like other components of the balance of payments, the discrepancies do not provide a complete indication of the underlying data problems, since there are offsetting errors—for example, when both parties to a transaction fail to report.

    Other Resources Citing This Publication