The direct investment data show reported global outflows persistently exceeding reported global inflows since the mid-1980s. The Working Party has been able to explain a large part of this discrepancy and to calculate adjustments to eliminate it. The most significant source of the discrepancy is the failure of many countries to compile data on reinvestment of earnings; however, there are other important inconsistencies and deficiencies in national statistics.


Interest in direct investment, long a significant subject to economic analysts and policymakers, has increased in recent years. One reason is the rapid growth in global direct investment: such flows (investment net of disinvestment) more than doubled between 1986 and 1989 to about $200 billion, and the international stock of direct investment (measured at book value) exceeded $1,000 billion by the end of 1989. Another reason is the changing contributions of major countries and country groups—especially the United States, Japan, and the European Community—to global flows.16 A third reason is the possibility offered by foreign direct investment for channeling resources and technical know-how to developing countries.

At the international level, a number of organizations have a policy interest in direct investment. Organizations such as the Fund, the Organization for Economic Cooperation and Development, the United Nations Center on Transnational Corporations (UNCTC), and the Commission of the European Communities have noted the increased importance of, and interest in, direct investment and several are taking steps to update and expand their data bases.17

Most countries compile statistics on direct investment, and there is widespread recognition of imperfections in the data. The Fund has drawn attention to the imperfections in the data on direct investment income in its Report on the World Current Account Discrepancy. The OECD, the Statistical Office of the European Communities, and the UNCTC have commented on the need for caution in analyzing direct investment statistics because of asymmetries in recording between different countries.18

One symptom of the difficulties with the data is the global statistical discrepancy in the recording of direct investment capital flows. In each of the years 1986 through 1989, the sum of “direct investment abroad” recorded by all countries exceeds the sum of “direct investment in the reporting economy” recorded by all countries. Consequently, there is a “negative discrepancy” in each year.

Table 10 summarizes, for the years 1986 through 1989, the world discrepancy on direct investment and its two major components: “reinvestment of earnings” and “other direct investment (ODI)” capital flows.19 For the four years, the negative discrepancy on overall direct investment flows (an average $16.5 billion per year) consists of an even larger negative discrepancy on the reinvestment of earnings, partly offset by a positive discrepancy on ODI flows.

Table 10.

Discrepancies on Global Direct Investment Capital Flows, 1986–89

(In billions of U.S. dollar; outflows ( – ))

article image
Source: Balance of Payments Statistics Yearbook, 1990.

The Working Party identified many differences in country practices that contribute to these discrepancies.20 Table 11 shows adjustments calculated by the Working Party to improve the data; these adjustments reduce the direct investment discrepancy by an average of 70 percent for the four years.

Table 11.

Global Direct Investment Discrepancy and Adjustments, 1986–89

(In billions of U.S. dollars; outflows ( – ))

article image

Balance of Payments Statistics Yearbook, 1990.


According to the Balance of Payments Manual, “Direct investment refers to an investment that is made to acquire a lasting interest in an enterprise operating in an economy other than that of the investor, the investor’s purpose being to have an effective voice in the management of the enterprise. The foreign entity or group of associated entities that makes the investment is termed the direct investor. The unincorporated or incorporated enterprise—a branch or subsidiary, respectively—in which direct investment is made is referred to as a direct investment enterprise.”21

It is the purpose of the investment, together with its economic consequences, that distinguishes direct investment from other forms of international capital flows. However, defining the border between direct investment and other forms of capital is not simple. Since some degree of equity ownership is almost always associated with an effective voice in the management of an enterprise, it is convenient, for statistical purposes and for international comparability, to use a minimum proportion of cross-border foreign ownership of an enterprise as evidence of direct investment. At present, the international standard—as set out in the OECD’s “Benchmark Definition” and proposed in the draft (fifth) edition of the Fund’s Manual—is a cross-border holding of 10 percent or more of the ordinary shares or voting power in an incorporated enterprise or an equivalent ownership interest in an unincorporated enterprise.22

In practice, a number of important countries—including the United States and Japan—use the 10 percent equity threshold while others—including France, Germany, and the United Kingdom—use a higher threshold. Some—including Belgium-Luxembourg and the Netherlands—use no specific threshold but treat each case on its merits. The result is that certain capital flows regarded by some countries as direct investment will be considered by others as a different type of investment.

These differences in threshold may not be a significant source of global asymmetry because a large proportion of direct investment occurs in branches or in majority-owned subsidiaries, which both the investing and host countries would regard as direct investment enterprises.23 Information obtained from six industrial countries shows that the percentage of the stock of direct investments held by majority-owned affiliates (including branches) was very high.24 With respect to outward direct investment by these countries, ownership by majority-owned affiliates ranged from 83 percent to 97 percent. Only three countries provided data on inward direct investment, and the percentages ranged from 94 to 96 percent. These findings, and additional information from two countries showing that equity holdings in the range of 10 to 20 or 25 percent accounted for only 1 or 2 percent of the stock of direct investment, suggested strongly that the use of different thresholds in the recording of direct investment flows may not contribute significantly to global asymmetries.

Again, according to international standards, the statistics should cover all enterprises in which the investor directly or indirectly has a direct investment interest. As a result, once the 10 percent “across-the-border” direct investment link is achieved with an enterprise, certain other enterprises related “down the line” to the first enterprise will also be regarded as direct investment enterprises. Direct investment measures should cover transactions between the direct investor and the first enterprise and certain of its affiliates and between the affiliates themselves if they are in different countries.

A significant number of countries—including about three-fourths of the industrial countries—take account of indirectly owned enterprises in their statistics. However, the procedures for determining the coverage of indirectly owned enterprises vary considerably from one country to another. This variation in procedure is a likely source of asymmetry, particularly in recent years as company structures have become even more complex, but its impact cannot be readily quantified.25

According to the Manual, direct investment capital flows cover all capital transactions between enterprises in a direct investment relationship.26 These flows include equity investment, reinvestment of earnings—the investor’s share in the undistributed earnings of the affiliate—and “intercompany debt,” such as trade credit and other loans, whether long-term or short-term. The definition also encompasses so-called reverse transactions, that is, investments by affiliates in their direct investors.

The fact that many countries do not include data on all the required transactions in their direct investment statistics is a significant source of global asymmetry.27 For example, about half of the industrial countries exclude reinvestment of earnings from their statistics, and a similar proportion exclude short-term transactions between affiliated enterprises.

Reinvestment of Earnings

Description and Global Imbalances

Reinvested earnings are defined in the Manual as the unremitted earnings of branches and of other unincorporated direct investment enterprises and the direct investor’s portion of earnings of incorporated direct investment enterprises that are not formally distributed. They are conceived of as providing additional capital to the direct investment enterprises. In recording such earnings in the balance of payments, therefore, entries should be made for both direct investment income (in the current account) and direct investment capital. These entries should be equal in magnitude and offsetting. In the capital account, they are described as “reinvestment of earnings.” Also, according to the Manual, earnings should be calculated net of taxes due for payment during the accounting period and net of provision for depreciation of fixed capital, valued at current replacement cost. Capital gains and losses, and any losses on the writing off of bad debts, should be excluded from the calculation. Earnings should be recorded in the periods in which they are made.

Published data for the period 1986 through 1989 show that the outward reinvestment of earnings of all countries exceeded the inward reinvestment of earnings of all countries by an average of $22.2 billion per year (see Table 12).

Table 12.

Global Reinvestment of Earnings Discrepancy and Adjustments, 1986–89

(In billions of U.S. dollars; outflows ( – ))

article image

Balance of Payments Statistics Yearbook, 1990.

A more detailed breakdown of these adjustments is given in Table 69 of Appendix IV.

Country Practices and Adjustments

The Working Party identified a number of important instances in which country practices differ from international standards and from each other. Its efforts to quantify these differences reduced the world discrepancy on reinvestment of earnings by an average of $18.2 billion per year (Table 12), some four-fifths of the published world imbalance in this component, for the period 1986 through 1989. The adjustments reduced the discrepancy in each year to an average of –$4.0 billion.28

Gaps in Reporting

About half of the industrial countries—notably Canada, Japan, Belgium-Luxembourg, France, and Italy—and many developing countries do not report data on the reinvestment of earnings to the Fund. Also, some economies (including some important offshore financial centers) report no data at all to the Fund. However, among the countries that do report these figures are several significant investor countries that have an excess of outward reinvestment of earnings over inward reinvestment of earnings. This is a principal reason for the excess of outflows in the published global balance.

Three sets of information were used by the Working Party to estimate the impact of missing country data in the Yearbook. (These adjustments are combined on line 2a of Table 12.) First, to world totals was added new information obtained from responses to the Special Questionnaire on International Capital Flows and from discussions with national compilers. On average, these adjustments increased the world discrepancy on reinvestment of earnings by $1.7 billion per year. The discrepancy widened because most of the new information was obtained from industrial countries for which outward reinvestment flows exceeded inward reinvestment flows.

Second, use was made of detailed tabulations provided by four major investor countries—the United States, Germany, the Netherlands, and the United Kingdom—of the reinvestment of earnings classified by partner country (geographic breakdowns).29 Hence, data on outward reinvestment of earnings from the four countries were used to estimate inward reinvestment of earnings for those partner countries that do not report such data to the Fund. Overall, these estimates for missing inward reinvestment averaged $14.5 billion per year for the period 1986 through 1989. The addition of these missing inflows significantly reduced the published global discrepancy. Similarly, inward reinvestment of earnings data from the four countries were used to estimate outward reinvestment of earnings for partner countries that do not report these data. These estimates had a marginal effect on this component of the world discrepancy.

Third, additional estimates of reinvestment of earnings for six offshore financial centers were derived by a method described in Chapter 9. These estimates—relating to The Bahamas, Bahrain, the Cayman Islands, the Netherlands Antilles, Panama, and Singapore—reduced the discrepancy by an average $3.9 billion per year.

Time of Recording

Germany records reinvested earnings in the year after they are earned rather than, in accordance with the Manual and other countries’ practices, in the year that they are earned. An adjustment was made to move the reported German data back one year; the average effect was a marginal increase in the world imbalance on reinvestment of earnings (line 2b of Table 12).

Capital Gains and Losses

Notwithstanding international guidelines, many countries include realized capital gains and losses in their reinvested earnings data, presumably because of the difficulties of identifying these gains and losses in the company records used to report these data.30 A smaller number of countries also include unrealized gains and losses. The United States is among the countries that include both types of gains and losses in their data, but it is the only country known to publish separate entries on these gains and losses. An adjustment to deduct the U.S. gains and losses from the published statistics reduced the world discrepancy by an average $1.5 billion per year (line 2c of Table 12). Offsetting entries to the U.S. capital gains and losses are embedded in the statistics of many U.S. partner countries, and the adjustment therefore overcorrected for asymmetrical reporting of flows between the United States and its partners. However, the amount of this overcorrection may not be significant, particularly because some important partners of the United States do not report any reinvested earnings data.31

Bilateral Comparisons

Although a considerable part of the global discrepancy on reinvestment of earnings has been explained, a gap remains. It appears that a part of the residual is the result of inconsistent reporting among countries that do include reinvestment of earnings data in their regular reports to the Fund. This inconsistent reporting is illustrated by bilateral comparisons that have been compiled from the geographic data supplied by the four major investor countries referred to earlier.32

In each year, the total amounts reported by the four countries as outward investments to one another exceeded the total amounts reported by the four host countries as inward investments from each other. The excess was about $1.9 billion in 1986, but it increased sharply to an annual average of $5.9 billion in 1987 and 1988.33 These differences, while large, were obtained after the netting out of many significant positive and negative bilateral discrepancies among pairs of countries. (The results are summarized in Table 13.)

Table 13.

Excess of Bilateral Outward Reinvestment of Earnings over Inward Reinvestment of Earnings Among Four Major Countries, 1986–881

(In billions of U.S. dollars)

article image
Source: National data shown in Table 70 of Appendix IV.

Negative entries indicate that the reported outward investment exceeded the reported inward reinvestment.

A study of Table 13 suggests that, among the four countries, there was either an overrecording of outward reinvestment or an underrecording of inward reinvestment, or some misstatement in each of the series. Furthermore, the sharp increases in the “excess” in 1987 and 1988, in comparison with that in 1986, coincide with the increase in the “adjusted” statistical discrepancy in 1987 and 1988 over the adjusted discrepancy in 1986.

These indications notwithstanding, the evidence is not strong enough to make a further, reliable adjustment to the global figures. First, the Working Party was aware of some specific factors that contributed to the bilateral asymmetries among the major investor countries, but it did not have sufficient data to quantify their impact.34 Second, bilateral comparisons can be misleading; because of differences in the ways that countries allocate transactions by partner country, a particular asymmetry may not generate an imbalance at the world level. A transaction that country A records as being with country B may be recorded by country B as being with country C. Such cases are especially likely when there are intermediate affiliates (such as holding companies). Hence a particular bilateral asymmetry between countries A and B may be compensated elsewhere, as in a comparison between countries B and C.

Concluding Remarks

The global statistical discrepancy on reinvestment of earnings mainly occurs because many countries do not compile these data. National compilers should prepare these data and report them to the Fund, in accordance with the recommendations of the Manual. It is particularly important that major countries—such as Canada, Japan, and France—that do not presently report the data should provide them.

Other Direct Investment Capital Flows

Description and Global Imbalances

According to the Manual, ODI capital flows represent all transactions between enterprises in a direct investment relationship, except reinvestment of earnings.35 Among countries reporting ODI data to the Fund, there are many deviations from this standard.

Published global inward ODI flows exceed global outward ODI flows by an average $5.7 billion per year in the period 1986 through 1989 (see Table 14). The bulk of this positive imbalance is concentrated in 1989; for the other years, the discrepancies are relatively small. However, in view of widespread variations in country practices, it is certain that the small discrepancies in the earlier years conceal offsetting errors or inconsistencies among country reports.

Table 14.

Discrepancy on Global Other Direct Investment Capital Flows and Adjustments, 1986–89

(In billions of U.S. dollars; outflows ( – ))

article image

Balance of Payments Statistics Yearbook, 1990.

A more detailed breakdown of these adjustments is given in Table 71 of Appendix IV.

Country Practices and Adjustments

The Working Party investigated many sources of errors and omissions in the measurement of world ODI flows. It examined variations in country practices from international standards and from practices of other countries, as well as gaps in reporting to the Fund. In some cases, it was able to quantify the impact of these variations and gaps. Its efforts at quantification on average significantly reduced the ODI discrepancy for 1986 through 1989 (line 2f of Table 14).36

Special Purpose Entities

Problems arise in recording direct investment because of the treatment of so-called special purpose entities (SPEs) of multinational enterprises. Set up in one economy with their parent in another, SPEs are enterprises that engage primarily in international transactions and do little or no local business.37 Types of SPEs include “financing subsidiaries,” which are set up abroad to raise and channel funds to their parent enterprises, and “holding companies,” which are set up mainly to hold investments in third countries. These companies are often located in offshore financial centers, such as the Cayman Islands and the Netherlands Antilles.38 Other important cases are the special financial institutions (SFIs) in the Netherlands and holding companies in Luxembourg.

Differences in the methods used for determining the residence of these enterprises, and the classification and recording (or nonrecording) of their transactions, lead to asymmetric recording in the world direct investment accounts. The fourth edition of the Manual does not specifically discuss the treatment of SPEs, but the draft fifth edition states that they should be treated as direct investment enterprises resident in the economies in which they operate. All transactions between SPEs and residents of other economies should be recorded in balance of payments statistics.

In practice, however, some of the economies in which SPEs operate—for example, The Bahamas and the Netherlands Antilles—do not consider them to be residents for balance of payments purposes, and these host economies do not record data on SPE transactions with residents of other countries. Other economies—such as the Netherlands and Luxembourg—do regard the SPEs as residents, but they publish the net sum of all the SPE transactions with nonresidents in “other capital.” Others—for example, the Cayman Islands and Hong Kong—do not report any balance of payments data to the Fund.

On the other hand, parent countries usually regard transactions between parent and SPE as direct investment transactions with countries in which SPEs operate.39 Similarly, other countries in which SPEs raise funds or make investments usually will record transactions with the SPE; of course, these transactions may be classified as either portfolio investment (for example, if SPEs raise funds in capital markets), as direct investment (for example, if SPEs are holding companies investing in affiliates), or as almost any other type of capital.

Several adjustments can be made to global totals to account for the direct investment transactions not recorded by host countries of the SPEs. Netherlands compilers made available to the Working Party data on the transactions of the Netherlands SFIs, which are not included in national direct investment figures. These data can be added to the global direct investment totals. The effect of this adjustment for the four-year period was to reduce the positive global ODI imbalance by an average of $9.5 billion per year (line 2a of Table 14).40

This adjustment was supplemented by estimates of the unreported direct investment capital flows for seven offshore financial centers: The Bahamas, Bahrain, Bermuda, the Cayman Islands, Hong Kong, the Netherlands Antilles, and Panama. These estimates were derived from geographic detail on ODI flows of seven partner countries: the United States, Australia, Japan, France, Germany, the Netherlands, and the United Kingdom.41 Estimates of the offshore centers’ outward direct investment were derived from amounts that partner countries reported as inward direct investment from these centers, while estimates of offshore centers’ inward direct investment were derived from amounts that partner countries reported as outward direct investment to these centers. Including these estimates increased the global ODI discrepancy by an average of $5.9 billion per year (line 2b).42 Further work, described in Chapter 9, suggests that these estimates may understate the unreported net inflows somewhat, although there is no firm basis in the data to make an additional adjustment.

Short-Term Flows

About half of the industrial countries—including Canada, Japan, France, and Germany—and many developing countries advised that, contrary to the international standard, they do not include short-term capital flows between affiliated enterprises in their national definitions of direct investment.43 Several of these countries—including Canada and France—report supplementary data on these flows for inclusion in the Yearbook. However, the omission of short-term flows by many countries creates major inconsistencies in world recording. In order to quantify the omitted flows, a number of important countries that do not report these flows were asked to provide supplementary data about them. However, only Germany was able to provide additional figures. Including short-term German transactions between affiliated enterprises reduced the ODI discrepancy by an average of $1.1 billion per year (line 2c of Table 14).

“Reverse Transactions”

According to the Manual, investment by a direct investment enterprise in its direct investor (for example, net purchases of shares by an affiliate in its parent or loans extended to the parent) is as much direct investment as the more common situation of investment by a parent in its affiliate. In the case of loans, there are many countries—including about one-third of the industrial countries, among which are Japan and Germany—that do not record any such “reverse loans” as direct investment. Several more countries—including Canada and Australia—that normally treat reverse loans as direct investment exclude these loans when they come from a finance affiliate set up abroad solely to raise funds for its parent. Loans from affiliates to parents are known to generate significant capital flows for certain countries. A number of major countries that do not treat these flows as direct investment were unable to provide supplementary data to the Working Party.

In the case of “reverse equity investment,” national treatments also vary. Some countries treat all equity investment by a direct investment enterprise in its parent as direct investment, as recommended by the fourth edition of the Manual. However, others, including Germany and the United States, regard such investments as portfolio investment until the equity threshold is reached, after which they are treated as direct investment.44 The Working Party could not quantify the impact on global accounts of the inconsistent recording of reverse equity transactions.

Real Estate Investment

According to the Manual, investment in land and structures abroad is direct investment. Direct investment should cover “noncommercial” investment, such as that by households in second homes abroad, as well as investment for commercial purposes.

In practice, some countries—for example, Italy and Spain—exclude all cross-border purchases and sales of real estate in reporting direct investment flows to the Fund. Many additional countries exclude purchases and sales of “noncommercial” real estate from these direct investment reports. More than half of the industrial countries—including the United States, Japan, and Germany—and many developing countries exclude these “noncommercial” investments.

These varying national treatments, together with the difficulties that some countries face in trying to trace real estate flows and the significant size of the transactions, make real estate measurement a likely source of global imbalances. Nonetheless, partial adjustment of the global flows is possible; data on real estate transactions included in other parts of country reports to the Fund can be added to the global direct investment totals. A number of countries—Japan, Denmark, Germany, Italy, and Spain—have published such data separately or provided data in response to a Working Party request. For the years 1986 through 1989, addition of these data marginally increased the average statistical discrepancy on ODI flows (line 2d of Table 14).45 The United States could not be included in this adjustment because it does not compile data on noncommercial real estate transactions, a significant omission from the global picture.

Other Sources of Asymmetry

The Working Party concluded that the four previously mentioned areas covered the most serious sources of inconsistencies in the recording of ODI flows. However, the Working Party investigated many other potential sources of asymmetry and made a number of additional adjustments to published global ODI flows (shown in combination on line 2e of Table 14). For example, compilers in Belgium-Luxembourg and Italy include a number of capital transactions with international organizations in their direct investment reports to the Fund. They should report them under “other capital” in accordance with the Manual. An adjustment to exclude these transactions from direct investment increased the global ODI discrepancy by an average of $0.6 billion.

Two additional problems were identified in the Fund’s balance of payments data on Saudi Arabia. First, during the three years 1986 through 1988, there were large entries for inward direct investment capital flows.46 However, these flows were not described as direct investment, but rather as “oil sector and other capital” in the country’s report to the Fund. Data provided by the United States, Australia, Japan, the Netherlands, and the United Kingdom showed much smaller outward direct investment flows to Saudi Arabia in each year. This comparison suggested that the data for Saudi Arabia are largely misclassified in the Yearbook, and an adjustment to the Fund’s data was made. Second, Saudi Arabia has reported no data on outward direct investment. However, in 1988 and 1989, the United States reported significant inward direct investment flows from Saudi Arabia. These amounts were added to the outward direct investment data in the Yearbook. The two adjustments for Saudi Arabian data increased the global ODI discrepancy by an average $0.4 billion per year.

The Working Party also obtained information on the capital flows of the U.S.S.R.; these are not yet included in the Yearbook. The derivation of adjustments for the U.S.S.R. is described in Chapter 13; on average, they marginally increased the global ODI discrepancy.

Information obtained in the Special Questionnaire on International Capital Flows and in subsequent investigations was used to adjust certain other data provided by countries in their regular reports to the Fund. These adjustments mainly compensated for anomalies in the country reports, but, in one quite significant case, they reflected an industrial country respondent’s recompilation (for 1987 and 1988 only) of its national direct investment data to align more closely with international standards. These adjustments reduced the global ODI discrepancy by an average of $3.1 billion per year.

Bilateral Comparisons47

For ODI capital flows, geographic details are available for more countries than was the case with data on the reinvestment of earnings. To make bilateral comparisons, the Working Party used geographic data for the 11 countries listed in Table 15.48 There are many significant positive and negative bilateral inconsistencies in the figures, but these tend to balance each other out in the total for the 11 countries taken as a group.49 Table 15 shows that, for the average of the three years 1986 through 1988, the total bilateral outflows reported by the 11 countries as investors were $2.6 billion less than the total bilateral inflows reported by the 11 countries as hosts.50

Table 15.

Excess of Bilateral ODI Outflows over Inflows Among 11 Countries1

In billions of U.S. dollars)

article image
Source: National data shown in Appendix IV.

Negative entries indicate that reported outflows exceed inflows.

Not much can be gleaned from the data in Table 15, except that the relatively small average global discrepancy on ODI flows for the period 1986 through 1988 conceals inconsistent, but somewhat offsetting, bilateral reports among countries. The bilateral inconsistencies in ODI flows reflect the many differences among countries in definitions and component coverage, which have been detailed earlier in this chapter. The inconsistencies also reflect differences among countries in the principles and procedures used for the geographic allocation of transactions; these differences can be significant when there are intermediate affiliates in third countries. The fact that national compilers have been able to explain some of the bilateral inconsistencies suggests that the regular exchange of geographic data between countries, and subsequent communication between compilers, can assist in the process of global reconciliation.

Concluding Remarks

National compilers should pay particular attention to international recommendations regarding the treatment of transactions via SPEs, short-term transactions, “reverse transactions,” and real estate transactions. Compilers should provide data on such transactions to the Fund, even when not shown in national presentations (for example, as the Netherlands now provides data on transactions of “special financial institutions”).

The examples and adjustments previously provided demonstrate that geographic data can be used to fill gaps and identify errors and inconsistencies in national reports. Therefore, compilers are urged to prepare and exchange geographic breakdowns of their direct investment data; the method of geographic allocation should conform with recommendations of the Manual.

Data Sources and Methods

Discrepancies in global statistics may also arise from imperfections in the way compilation systems cover transactors or transactions, from errors in the data reported to statistical agencies, or from the relative strengths or weaknesses in checking and compilation procedures.

Countries use a variety of data sources and methods in compiling direct investment statistics, but there are two key features that help to describe the national sources. First, data collection may be based on the reporting of individual transactions or it may be based on the aggregate reporting of transactions. Second, data may be collected by the statistical agency from an intermediary (for example, a bank that handles foreign exchange transactions or an official source that is responsible for exchange control or investment approvals) or directly from transactors.

About two-thirds of the industrial countries and many developing countries base their collections on individual transaction reporting, and the bulk of the information comes from intermediaries. The collection of individual transactions data from financial intermediaries can provide adequate data on balance of payments transactions associated with cash flows. However, this information will be insufficient when an economy is a significant provider or recipient of direct investment flows. Because of the special nature of the relationship between a direct investor and its affiliates, there will be many cases in which goods, services, and technical know-how are provided from the parent to the affiliate without any corresponding cash flow. In such circumstances, there will be merely a book entry that records increased share capital issued to the direct investor or a loan provided by the direct investor. Although these are not cash transactions, they are direct investment transactions that should be recorded in the balance of payments. Similarly, among related enterprises, there may be payments reflecting the net amount of many “gross” transactions, some of which may be capital transactions. Also, reinvestment of earnings represents the undistributed earnings of the direct investment affiliate attributed to the parent; there is no associated cash flow.

Information on noncash capital flows can only be obtained from the accounts of direct investment enterprises or direct investors. Therefore, current gaps in the recording of reinvestment of earnings and some other flows can only be closed by direct inquiries to enterprises.51

A number of countries that once obtained their data wholly from bank reports have supplemented them with enterprise reports.52 In particular, five industrial countries have added annual enterprise surveys that collect, among other things, reinvestment of earnings data and have included these figures in their reports to the Fund. Two of these countries (Finland and Sweden) introduced their new data during 1991. These developments are consistent with a principal recommendation of the Report on the World Current Account Discrepancy. Also, France recently commenced an annual survey of outward direct investment only, but reinvestment of earnings data collected there are not yet included in its national balance of payments statement, nor are they reported to the Fund. Compilers in Belgium-Luxembourg, too, currently are developing a new enterprise survey. On the other hand, there are also countries (the most important is Japan) that have not expanded their systems sufficiently to collect all the required figures. Therefore, the Working Party recommends that compilers periodically survey direct investors and their affiliates to collect data on direct investment capital transactions, including reinvestment of earnings, and related data on investment income and stocks.

About one-third of the industrial countries and many developing countries base their direct investment collections upon aggregate reporting, usually by transactors themselves. That is, these countries obtain their direct instrument data predominantly or entirely from enterprise surveys. These surveys usually yield data on reinvestment of earnings and other noncash transactions, as well as other data on flows and positions.53

When surveys are conducted, whether to obtain specific data or to supplement bank reports, it is important to achieve a satisfactory level of coverage of transactors and their transactions. The general approach among industrial countries that base their data collections on surveys is to conduct annual (or less frequent) benchmarks covering a large number of enterprises and quarterly surveys covering a sample of larger enterprises. However, the U.S. experience shows that a system designed to concentrate on major reporters in interim surveys can be subject to significant error. On the basis of a 1987 benchmark survey and other information, U.S. compilers recently announced that past estimates of inward direct investment had been significantly understated and past estimates of outward direct investment had been significantly overstated. The effect of these U.S. revisions on the global discrepancies was considerable. The announcement of similar revisions throughout the 1980s demonstrated that failure to account for capital flows of nonreporting and late-reporting enterprises created a significant problem for U.S. statistics.

It is recognized that, for many countries, truly comprehensive surveys of direct investment are only feasible at intervals longer than required for immediate balance of payments purposes. At the same time, the U.S. experience suggests that special care must be taken to ensure that accurate aggregate measures are obtained between benchmark surveys. Accurate measures may be achieved by including a sufficient sample of transactors in interim surveys or by developing suitable procedures to adjust interim survey aggregates to be compatible with benchmark coverage.

Measurement of real estate transactions is a particular problem for certain countries that rely upon enterprise surveys. Numerous, relatively small transactions are common, but they can sum to quite significant amounts. For example, in the United States, no direct investment reports are required on residential real estate for personal use or real estate investment through limited partnerships. Gaps such as these—between benchmark surveys and omission of certain transactions—give rise to important imbalances in the direct investment accounts. National authorities must ensure that their systems provide comprehensive coverage of all transactions on a continuing basis to meet the requirements of balance of payments accounting.

Measurement of direct investment flows has become more difficult in recent years as a result of the growing complexity of multinational company structures and transactions within these structures. Traditional forms of direct investment increasingly have been supplemented by cross-border mergers and acquisitions and international joint ventures, especially in the United States and the European Community. Noncash transactions, share exchanges, and cross-participation also became major features of direct investment in the late 1980s. These factors have made it even more difficult for countries to identify all transactors in a chain of related enterprises (and to identify which ones are direct investors and which are affiliates) and all the transactions between them. National compilers require an adequate level of resources to analyze these complicated accounts. Also, to help ensure consistent treatment of transactions of multinational enterprises, compilers should develop regular contact with their counterparts in other countries. The increasing complexity of the subject, the expected further growth in direct investment activity, and the sharp increase in the published global discrepancy in 1990 make more urgent the need to prevent further deterioration in the statistics.54

Conclusions and Recommendations

The Working Party’s research into direct investment measurement has found that, notwithstanding international standards, there are widespread variations among countries in the definition, classification, and coverage of data used in the measurement of direct investment capital flows. There are also differences in data sources and methods of collection. These factors undermine the quality of national statistics and hamper efforts to reconcile data at the global level.

The most significant sources of global imbalances that the Working Party identified were

  • — the failure of many countries to record reinvestment of earnings;

  • — inconsistencies in the recording of transactions of the special purpose entities of multinational enterprises;

  • — the exclusion from many countries’ direct investment data of “short-term” capital transactions among related enterprises;

  • — inconsistencies in the treatment of investment by affiliates in their parents (that is, “reverse investment”);

  • — inconsistencies and poor coverage in the recording of real estate transactions; and

  • — the failure of collection systems to provide comprehensive coverage of all direct investment transactions.

Despite the many variations in national practice, the detection and correction of gaps and discrepancies are more readily achievable for direct investment than for many kinds of international capital flows. The Working Party has shown that using supplementary detailed information already in the hands of compilers to make adjustments to reported data can eliminate a significant part of the world direct investment discrepancy.

Two types of national data are particularly useful in detecting and quantifying gaps and discrepancies:

  • — supplementary data provided by countries to “bridge,” when necessary, the national and Yearbook presentations; and

  • — national data providing geographic details of capital flows vis-à-vis partner countries.55

The measurement of direct investment transactions has become more difficult in recent years as a result of the increasing range and volume of international transactions and the numbers and types of transactors as well as the increasing complexity of company structures. Reflecting these difficulties in some countries, including the United States, collection systems have not provided comprehensive coverage of transactions on a continuing basis. In others, including Japan, collection systems have not yet been adapted to capture data on international transactions that do not pass through the domestic banking system. All these factors make it more urgent that action be taken to prevent further deterioration in the statistics.

In summary, the Working Party’s recommendations are