The 1980s saw a marked expansion in the scale of cross-border securities transactions as financial markets became increasingly integrated. During this period, the global data for portfolio transactions exhibited large discrepancies, especially on bond transactions. The Working Party identified numerous factors that contribute to discrepancies in portfolio investment flows and derived adjustments for many of these issues. However, these steps did not correct the overriding problem—the under-recording of capital outflows—in the global data on bond transactions. Unless a concerted effort is made to improve these data, the statistics will deteriorate rapidly.


Changes in international capital markets during the 1980s opened new possibilities for raising capital, investing, and hedging risk, and a wide range of new participants emerged in these markets. The increasing importance of borrowing and lending in the form of securities led to a surge in the issuance of bonds in international markets. Banks and securities firms expanded their operations into foreign markets after the deregulation of securities markets in the United Kingdom and elsewhere. New channels opened through which securities could be bought and sold in foreign markets, and cross-border securities trading increased sharply. Institutional investors such as pension funds, insurance companies, and investment trusts diversified heavily into foreign securities, as restrictions on investment activities were lifted in many countries, notably Japan. Increasingly, large investors dealt in foreign markets or directly with foreign issuers of securities. The trend to diversify portfolios also increased considerably the demand by private individuals for foreign securities, especially in Europe.

Faced with these changes, balance of payments compilers in many countries have found it increasingly difficult to measure portfolio capital flows. At the global level, there were large imbalances in the components of portfolio investment throughout most of the 1980s, partly because the activities of new participants were not sufficiently covered in national compilation systems. Table 16 provides an overview of the global discrepancies on portfolio investment flows for the years 1986 through 1989, as published in the 1990 Balance of Payments Statistics Yearbook. In order to compare global assets with liabilities, recorded “liabilities constituting foreign authorities’ reserves” were excluded from the figures, because the counterpart securities assets were recorded under the category of reserves.

Table 16.

Discrepancies on Global Portfolio Investment Capital Flows, 1986–89

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

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

The overall imbalance in portfolio capital transactions shown in the table was not large (except in 1989), but it concealed more serious problems within the bonds and equities components. Bond transactions showed an excess of capital inflows in most years, with the measured imbalance reaching $59 billion in 1989. Large discrepancies also were recorded for the earlier 1980s. The imbalance on equities showed an excess of measured outflows, especially for 1987 and 1989. The discrepancies for earlier years were not large.

The Working Party identified numerous factors that have contributed to discrepancies in global data on portfolio transactions, and it has derived adjustments for many of these issues. Table 17 provides an overview of the net effect of the adjustments on the global discrepancy on portfolio transactions for the four years under review. In the case of bonds, the adjustments did not reduce the large positive discrepancies in the global accounts, with the exception of a small reduction for 1989. The large negative imbalances in global equities transactions for the years 1987 and 1989 were reduced but remained sizable. It should be stressed that a small discrepancy in any one year cannot be taken as confirmation that the global data are correct, because there may well be offsetting errors on both sides of the accounts.

Table 17.

Global Portfolio Investment Discrepancies and Adjustments, 1986–89

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

article image

Balance of Payments Statistics Yearbook, 1990.

The Working Party employed a number of approaches to identify sources of asymmetry in these data. It used information from various sources, including the data base on international bonds developed by the Bank for International Settlements.56 In addition, a supplementary survey on national practices in compiling cross-border securities transactions was distributed to various industrial countries.


Portfolio investment, as defined in the Balance of Payments Manual, covers transactions in long-term bonds and corporate equities other than those included in direct investment and reserves. Long-term bonds include marketable bonds, debentures, and notes that have an original contractual maturity of more than one year. Long-term negotiable certificates of deposit also are treated as bonds. Equities include company shares, investment fund units, and documents denoting ownership of equity (for example, American Depository Receipts). Securities transactions related to direct investment and official reserves are excluded because financial items in the balance of payments are classified by function rather than by instrument.

Country Practices and Adjustments


The Working Party identified a wide range of country practices that give rise to discrepancies in global data on portfolio capital flows, and it derived a number of adjustments to correct for missing data or deviations from international standards in balance of payments reporting to the Fund. The detailed adjustments to portfolio bond transactions, shown in Table 18, did not, on balance, reduce the large discrepancies in recorded bond transactions, with the exception of a small reduction for 1989. At $53 billion, the imbalance for 1989 remained exceptionally large, and for most years the totals continued to exhibit an excess of capital inflows.

Table 18.

Adjustments to Transactions in Bonds, 1986–89

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

article image

Balance of Payments Statistics Yearbook, 1990.

Liabilities constituting foreign authorities’ reserves.

International Bonds

Because of the importance of the international bond market, the issuance of bonds in this market was identified as a major area of investigation. According to the BIS, the outstanding stock of international bonds increased by $1,000 billion from 1982 to the end of 1989. (Here, the term “international bonds” is used to denote bonds issued originally in foreign and international markets.)

The Working Party found that, in reporting to the Fund, many countries classified transactions associated with the issuance and redemption of international bonds as part of “other long-term capital” rather than as portfolio investment. It was found that national collection systems frequently did not provide compilers with sufficient information to distinguish international bonds from long-term loans. Also, in a few countries, international bonds issued by banks were included as part of deposit money bank transactions, while other bonds were recorded as portfolio investment.

Largely on the basis of information from the BIS data base, the Working Party derived adjustments, where necessary, to reclassify these borrowings to the category of portfolio investment. The adjustments, shown in line 2a of Table 18, added about $16 billion a year on average to recorded bond liabilities, most of which pertained to borrowings by Australia, Italy, Denmark, and New Zealand. Apart from the issue of classification, numerous comparisons of national balance of payments data with BIS figures did not uncover convincing evidence that borrowing in the international bond market was being underreported.57

Japanese Eurobonds Purchased by Residents

BIS statistics show that, in recent years, Japanese firms have been the largest borrowers in the international bond market, with net borrowings approaching $200 billion for the 1986–89 period. Some three-quarters of these borrowings involved equity-related bonds, mostly denominated in U.S. dollars and Swiss francs. A large portion of these bonds reportedly was purchased by Japanese residents.

The reporting forms used by the Japanese compilers to record purchases of foreign currency bonds do not distinguish between foreign securities (assets) and domestic securities issued in international markets (liabilities). While this reporting practice has no effect on the global portfolio discrepancy, it gives rise to an overstatement of both cross-border assets and liabilities whenever Japanese residents purchase these foreign currency obligations in international markets.

Adjustments were derived to reclassify, from the asset to the liability accounts, Japanese purchases of equity-related bonds issued abroad by residents; these estimates are shown in lines 2b and 6a of Table 18. The estimates were based on market sources of information. The adjustments for 1989 were especially large—liability inflows and asset outflows both were lowered by $45 billion.

Offshore Transactions

The channeling of funds through Special Purpose Entities of multinational enterprises is known to be a major source of asymmetry in the recording of world capital flows. Many SPEs, in the form of financing subsidiaries or holding companies, operate in certain industrial countries and in offshore financial centers, and they are active participants in international capital markets.

A number of countries permit special classes of business to carry out international business activities with relative freedom from taxes and regulation. Frequently the scale of cross-border transactions effected by these operations is unrelated to the size of the domestic economy. For this reason, in some industrial countries, recording of the activities of these entities is not articulated according to balance of payments norms, while, in OFCs, cross-border transactions of these entities usually are omitted from balance of payments statements. Also, some OFCs compile no balance of payments statements.58 These activities, commonly referred to as offshore transactions, contribute to asymmetries in the recording of world capital flows whenever there is a transformation in the form of capital channeled through SPEs and other offshore entities.

For example, since the early 1970s, an increasing number of special financial institutions have been established in the Netherlands to raise funds from nonresidents and to channel these funds to other nonresidents. To prevent national statistics from being dominated by financial transactions of institutions that had little impact on the domestic economy, Netherlands compilers have recorded only the net flows of all such transactions under “other long-term capital.” SFIs were large borrowers in the international bond market, and asymmetries arose in global portfolio flows because the Dutch accounts showed no portfolio liability inflows to match the asset outflows recorded by the countries investing in SFI securities. On the basis of information provided by Netherlands compilers, the Working Party made adjustments that added about $10 billion a year net to that country’s recorded bond liabilities over the 1986–89 period (line 2c of Table 18). It appears that these borrowings were channeled largely to parent companies and other affiliates abroad, as discussed in the previous chapter.

Holding companies in Luxembourg also conduct various operations involving nonresident parties. Their function is similar to that of SFIs in the Netherlands: to receive funds from one country and transfer them to others. The net effect of these transactions has been recorded as a one-line entry in the balance of payments of Belgium-Luxembourg, similar to the Netherlands treatment of SFIs. However, unlike the Netherlands case, compilers in Belgium-Luxembourg do not gather enough information to classify holding company capital flows by type, thus no adjustment was attempted. On a gross basis (total of inflows and outflows), these flows exceeded $250 billion in 1988.

In the 1960s and early 1970s many U.S. financing subsidiaries were established in the Netherlands Antilles to issue bonds in the Euromarket and to relend the proceeds to their parents. After the repeal of the U.S. withholding tax on bond interest payments to nonresidents in 1984, U.S. corporations began paying off loans from their Antilles affiliates and raising funds directly in the Euromarket. The transactions of the U.S. financing affiliates were not recorded by the Netherlands Antilles, which considers such entities to be nonresidents. At the same time, countries that held the bonds of these entities would have recorded net inflows of portfolio capital when these securities were redeemed. (U.S. compilers recorded net outflows on direct investment abroad when the loans were repaid.) The Working Party made adjustments based on BIS bond data to account for the declining portfolio liabilities of these U.S. affiliates in the Netherlands Antilles (line 2d of Table 18). The adjustments had the effect of reducing bond liabilities by an average $3.2 billion a year for the four years under review. Small adjustments also were made for bonds issued by other SPEs in the Netherlands Antilles and in other OFCs, such as the Cayman Islands (line 2e of Table 18).

Asymmetries also arise in the global portfolio statistics when SPEs or other offshore entities are used as a conduit to invest in foreign securities. The Working Party identified various instances of such activities. For example, U.S. corporations have captive insurance companies in Bermuda that invest premium income received from abroad in securities and other assets; some investment activities also are conducted by Netherlands SFIs; and Japanese financial institutions use offshore entities to engage in securities transactions in foreign markets. In the last example, Japanese compilers record direct investment outflows, while the investments of their SPEs are reflected in another country’s portfolio inflows.

For tax reasons, large numbers of mutual funds have been established in OFCs, and many of these funds invest in foreign bonds. According to data obtained from Lipper Analytical Services, international bond funds in OFCs in the Caribbean held more than $4 billion in assets at the end of 1989. Asymmetries arise because countries whose residents invest in these mutual funds record an outflow into foreign equities, while the investments by these funds are recorded else-where as an inflow into bonds. There are also many international bond funds and other investment funds in the Channel Islands; most of the international transactions of these funds are not recorded by U.K. compilers. Activities of banks in OFCs also contribute to the problem. Total foreign claims of such banks exceeded $1,000 billion at the end of 1989. These claims include securities, but as in most of the other examples, few data were available on securities transactions carried out by these offshore entities.

While the issuance of international bonds by entities in OFCs can be estimated to a considerable extent by using BIS data, little is known of the various portfolio investments made by these centers. The Working Party relied heavily on bilateral statistics published by several industrial countries, supplemented with other information from national compilers, to derive partial estimates for net transactions in foreign bonds by offshore entities. The adjustments are shown in line 6b of Table 18 and average about $2 billion a year of net bond purchases.

As these instances show, the absence of data for OFCs and entities conducting similar operations in industrial countries poses growing problems in compiling globally consistent capital account statistics. The Working Party urges the authorities in offshore financial centers to produce balance of payments statistics on the international transactions of those sectors of their economy that have a significant counterpart in the international accounts of other countries.

Liabilities Constituting Foreign Authorities’ Reserves

It was indicated in connection with Table 16 that recorded LCFAR was excluded from portfolio liabilities so that assets and liabilities could be compared at the global level. As most countries have difficulties in identifying and measuring LCFAR, the Working Party sought a more complete measure of these transactions from the creditor side (that is, the reserve holders). Table 19 compares the figures obtained on reserve transactions in foreign bonds with reported data on portfolio LCFAR (the latter were reported only by U.S. and U.K. compilers).

Table 19.

LCFAR Compared with Net Reserve Transactions, 1986–89

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

article image

Balance of Payments Statistics Yearbook, 1990.

See Chapter 8 for discussion of the Working Party’s research into reserve composition and LCFAR.

During the four years under study, monetary authorities increased their holdings of foreign bonds by $129 billion—about $30 billion more than was identified as LCFAR by debtor countries. It should not be assumed, however, that the larger transactions reported by reserve holders were not recorded by the debtor countries; it is more likely that they were classified as “other bond liabilities.” The understatement in LCFAR can be taken as a factor contributing to an excess of credits on global portfolio bond transactions. The adjustments to remove the additional LCFAR from portfolio liabilities amounted to about $8 billion a year on average and are shown in line 2f of Table 18.

Repurchase Agreements

Repurchase agreements (repos) are arrangements involving the sale of securities (bonds or short-term instruments) to investors with an agreement to repurchase them at a specified price at a specified date in the future. These transactions are typically viewed as short-term loans secured by top quality assets. According to the Manual, repos backed by long-term obligations are to be recorded as part of portfolio investment.59

Several industrial countries—including the United States, Canada, France, and the Netherlands—record repos as short-term loans rather than as portfolio transactions. In Germany, some repurchase transactions also may be recorded as loans. Japanese compilers record repos (gensaki) as short-term securities in their national publications but report them to the Fund under the Manual convention as portfolio investment.

Japanese financial institutions have been active borrowers in the international repo market in recent years, generating a net capital inflow of more than $13 billion in 1988. In the Yearbook this activity is treated as a sale of securities. However, a large part of Japanese gensaki borrowing was probably from major countries that record repos as loans—both in their national presentations and in the Yearbook. This asymmetry can be removed by adjusting their portfolio assets, so that the repo figures are all given on the “portfolio investment” basis of the present Manual. These adjustments, derived on the basis of Japanese data, are shown in line 6c of Table 18, and raise portfolio bond assets by $3 billion a year on average.

Bonds Issued by Financing Trusts

In 1988 and 1989, a number of military sales loans previously extended to foreign governments by the U.S. Government were refinanced in the U.S. market through the issuance of long-term promissory notes by special financing trusts set up by the foreign governments in the United States. The trusts, in turn, lent the funds to their governments, which repaid the military sales loans. According to the Manual, the trusts or similar entities of a foreign government are to be considered as nonresidents by the economy in which they are physically located.

In the U.S. balance of payments, however, the trusts were treated as resident entities; purchases by U.S. residents of the notes did not appear in the accounts. (The advances made by the trusts to foreign governments and the repayments of the military sales loans both were recorded under loans.) This recording created asymmetries because some of the foreign countries recorded the trusts’ new borrowings as bonds issued by themselves. On the basis of information provided by U.S. authorities, adjustments were made to U.S. data and data for some debtor countries in order to reclassify the trust borrowings to portfolio investment and to identify U.S. residents as the purchasers. These adjustments added about $9 billion to U.S. bond assets over the two-year period (line 6d of Table 18).

International Organizations

International organizations are important participants in world capital markets, both as borrowers and investors of funds. Revised data were obtained for their portfolio transactions on the basis of information from new reporting forms introduced by the Fund’s Statistics Department in 1991.60 The adjustments are shown in lines 2g and 6e of Table 18. These revisions are in addition to the large portfolio corrections, based partly on the Working Party’s research, that were already incorporated into the 1990 Yearbook. These earlier revisions averaged $10 billion a year from 1983 to 1988 and mostly added to net inflows on bond liabilities.

Other Adjustments to Bond Transactions

The Working Party identified a number of equity transactions reported as bonds in countries’ balance of payments submissions to the Fund. Data reported on the Special Questionnaire and additional information from national compilers were used to make large corrections to the reported data on bond liabilities for a few countries. The amounts involved were especially large for 1988 and 1989 when a total of $27 billion of net inflows were subtracted from bond liabilities and transferred to equity accounts. In other cases, bond transactions were found to be misclassified to other capital account categories. Small adjustments were also made to remove some transactions in short-term instruments and syndicated loans from recorded portfolio flows.

Transactions associated with the recording of bonds acquired under debt-restructuring agreements involving developing countries were investigated. In some instances, it appeared that these securities were not recorded as portfolio capital by creditors. In addition, small adjustments were made to pick up new information reported on the Special Questionnaire, and bilateral statistics compiled by several industrial countries were used to construct partial estimates for countries that had not reported any portfolio transactions. The net effect of these various adjustments, mostly classification corrections, is shown in lines 2h and 6f of Table 18.

Other Sources of Asymmetry

No adjustments could be made for some other factors that contributed to the asymmetries in bond transactions. These factors include the recording of transactions in long-term negotiable certificates of deposit issued by foreign banks as part of “other capital,” rather than as portfolio investment; inclusion of fees and commissions in the reported values of securities transactions, rather than in the current account; and differences in the recording of transactions in zero coupon and other deep-discount securities (recorded on an accrual basis rather than on a due-for-payment basis).

Concluding Remarks

On bond liabilities, the Working Party had access to a considerable amount of external information, such as the BIS data base and comprehensive information from Netherlands compilers on bonds issued by SFIs. On the asset side, there was no comprehensive source of information with suitable geographic detail that could be used to evaluate how well securities transactions were being measured. By contrast, considerable creditor data, used by the Working Party in its research, were available on direct investment (Chapter 3) and banking transactions (Chapter 6).

Research on portfolio bonds revealed that serious problems affect the measurement of these capital flows. On balance, the adjustments derived by the Working Party widened the global imbalance on portfolio bonds. The large positive discrepancies on net bond transactions suggest that there is an underrecording of capital outflows, a reflection of gaps in national reporting systems. The issue of coverage of portfolio transactions is discussed later in the chapter.


Table 20 provides an overview of the Working Party’s adjustments to global equity statistics. While the adjustments reduced the large negative imbalances for 1987 and 1989, they remained sizable.61

Table 20.

Adjustments to Transactions in Equities, 1986–89

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

article image

Balance of Payments Statistics Yearbook, 1990.

Liabilities constituting foreign authorities’ reserves.

Classification Corrections

In a number of instances, countries reported equity transactions under the category of bonds or “other capital” flows in their balance of payments submissions to the Fund. Adjustments to reclassify these transactions to the equity accounts are shown in lines 2a and 6a of Table 20 and were especially large for 1988 and 1989, when a total of $27 billion of inflows was added to equity liabilities. The net effect of classification errors on the asset side was considerably smaller.

Liabilities Constituting Foreign Authorities’ Reserves

No entries were identified in the Fund’s statistics for equity LCFAR, which must be excluded from liabilities if portfolio assets and liabilities are to be compared. Only the U.S. data collection system identified foreign official purchases of equities, and these data were used to adjust equity liabilities on the assumption that the “foreign official” transactions in the U.S. accounts were for reserve accounts. The amounts involved were relatively small and are shown in line 2b of Table 20.

Offshore Transactions

Balance of payments reporting to the Fund by OFCs was incomplete. Bilateral statistics compiled by several industrial countries were used to make partial estimates for net transactions in foreign equities by selected OFCs. These estimates are shown separately in the case of assets (line 6b of Table 20), but, because a broader procedure was used in deriving adjustments to liabilities, adjustments for equity inflows into selected OFCs are embedded in other adjustments.

Other Adjustments to Equities

International investment funds established to invest in the equity markets of developing countries, or so called “emerging markets,” were an important trend in the 1980s. The Working Party sought to determine how well these and other equity liabilities were recorded by developing countries. Excluding Thailand, which showed sizable foreign inflows into domestic equities, and a large net disinvestment recorded by South Africa in 1986, the net inflow of foreign investment into portfolio equities recorded by other developing countries averaged less than $300 million a year during the 1986–89 period.

Two sources of information were used to evaluate these balance of payments figures. The Working Party obtained information from Lipper Analytical Services for closed- and open-end international mutual funds specializing in “emerging markets”; that source was used to estimate transactions of OFC funds that invest in “emerging markets.” The Working Party also used bilateral statistics published by several industrial countries as a basis for estimating other transactions in developing countries.62 The derived estimates suggested an understatement of inflows recorded by developing countries (and other emerging markets), especially for 1989. The adjustments are shown in line 2c of Table 20. Other adjustments to equity assets (line 6c) included estimates for transactions of pension funds of international organizations and for developing countries not reporting any portfolio transactions.

Other Sources of Asymmetry

Other factors that contributed to the global imbalance on portfolio equity transactions included the recording by some countries of all security transactions carried out on a domestic or foreign stock exchange as portfolio investment, regardless of the size of the share acquired in the capital of the enterprise (that is, including direct investment). In addition, the recording of debt-equity swaps by creditor and debtor countries may not have been uniform as regards classification between direct and portfolio investment transactions. Also, Germany, Japan, and several other countries included fees and commissions in the value of the securities settlement, rather than in the current account.

Concluding Remarks

The preceding discussion of country practices and adjustments to portfolio bonds and equities identified numerous examples of countries that have not reported data to the Fund in conformance with recommendations in the Manual. This lack of conformance was shown to produce large asymmetries in the components of world capital flow statistics. The Working Party therefore strongly recommends that countries make greater efforts to report data on portfolio transactions to the Fund on a basis consistent with the Manual.

While there is often a basis for correcting the classification of securities transactions in reports to the Fund, it is more difficult to find and correct gaps in national data resulting from incomplete coverage of transactions. This issue is discussed later in the chapter.

Bilateral Comparisons

In its research, the Working Party used data compiled by major industrial countries to make numerous bilateral comparisons of portfolio capital flows. While bilateral comparisons hold some potential for identifying statistical errors and gaps in national data, the Working Party recognized at the outset that reconciliations would be difficult to achieve. Apart from incomplete data, there are several factors that impede bilateral comparisons. These include

  • —use of different allocation principles. Compilation systems usually allocate portfolio transactions to partner countries in one of two ways: by following the transactor principle or the debtor-creditor principle. In the former, transactions are to be allocated to the country of the foreign party to the transaction. In the latter, transactions in claims and liabilities are to be posted to the country where the foreign debtor or creditor, respectively, is a resident;63

  • —securities acquired as part of reserves. As shown in Table 19, monetary authorities acquired for reserves more than $100 billion of foreign bonds during the 1986–89 period. Debtor countries classify these transactions under their bilateral detail on portfolio flows, while creditor countries record them under reserves; and

  • —other differences in classification. The discussion on country practices identified numerous deviations from the Manual in ways that countries classify portfolio transactions (for example, exclusion of international bonds, inclusion of short-term instruments, and fees and commissions).

The largest discrepancy in bilateral comparisons of bond transactions was that between Japan and the United States. Japanese data show much larger net purchases of U.S. bonds than were recorded in U.S. statistics. Table 21 shows that during the 1986–89 period the difference between these national figures averaged $19 billion a year. While these discrepancies were large, especially for 1986 and 1987, it was difficult to quantify the extent of gaps or statistical errors in either set of data.

Table 21.

Japanese Net Portfolio Investment in U.S. Bonds, Comparison of Bilateral Data, 1986–89

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

article image

Bank of Japan

U.S. Treasury Bulletin, various issues.

First, the two countries used different criteria to allocate securities transactions geographically. The United States allocated liabilities uniformly by residence of transactor, but Japan allocated transactions in foreign securities according to a number of criteria. Transactions in U.S. Treasury securities were allocated on the debtor-creditor principle, while transactions in other foreign markets generally were allocated on the transactor principle.64 (Some transactions in foreign markets were also allocated to the country in which the security was listed, for example, Luxembourg, in the case of many Eurobonds.)

Second, U.S. portfolio liabilities to Japan included transactions with private investors and with official accounts (that is, reserves); Japanese portfolio assets included only transactions of private investors.65

The effects of these two factors on the bilateral imbalance could not be quantified. However, the U.S. figures may have suffered from some underreporting of portfolio inflows through U.S.-based affiliates of Japanese securities firms, which reported only negligible sales of U.S. Treasury securities to Japan during 1986 and 1987.66 Total net sales of treasury bonds and notes to Japan totaled less than $1 billion in these two years, according to the U.S. statistics.

Other bilateral comparisons for various countries were made for both bond and equity transactions. However, the use of different allocation principles and classification systems, the growing volume of securities acquired for reserve purposes, and the inability of debtors to allocate properly to purchasers the large volume of bearer securities issued in international markets made it difficult to determine the extent of any data problems in the statistics.

The Working Party nonetheless sees value in bilateral data as one tool in investigating gaps in national data. It recommends that countries compile geographic details on portfolio capital flows and outstanding stocks and exchange these data with partner countries. Geographic details are useful for analytical purposes, but only when countries compile data on common standards. Compilers are urged to allocate transactions in foreign securities on the basis of the debtor-creditor principle, as recommended in the Manual.

Coverage of Portfolio Transactions

The Working Party’s adjustments had little net impact on the large discrepancies in recorded portfolio bond transactions and, except for 1988, the data continue to exhibit a large excess of recorded inflows. The imbalance totaled $53 billion for 1989. Although adjustments on equities had greater effect, sizable discrepancies remain for 1987 and 1989.

The excess of recorded inflows on bond transactions suggests underrecording of asset acquisitions, but net liabilities inflows may also have been overstated. If, for example, redemptions of domestic bonds held by nonresidents are understated, this would contribute to an overstatement of liabilities. Whether this is a significant problem cannot be judged from available data. Several large industrial countries (for example, the United States and the United Kingdom) do not distinguish in their statistics between bond redemptions and other cross-border bond transactions. Where separate balance of payments data were available on the redemption of international bonds, comparisons were made with records from the BIS data base. Reported data on redemptions generally matched or exceeded BIS estimates. No checks could be made on redemptions of bonds issued in domestic markets or on the ways that creditors reported these flows.67

Another activity that could potentially result in over-stated inflows on portfolio liabilities would be domestic transactions in domestic securities when intermediated by foreign dealers or managers. It is more likely that “inflows” from such transactions would be recorded in reports filed by domestic banks and securities dealers than would be the case for the related “outflows” to the foreign manager, especially for transactions by private individuals.

Several areas where there are known to be gaps in national data and difficulties in accurately recording certain aspects of portfolio flows are discussed below.

Private Individuals

In the data collection systems of many countries, gaps exist in the recording of portfolio investments carried out by private individuals via nonresident banks and brokers.68 In countries that rely on surveys of domestic intermediaries and other corporations, data usually are not collected directly from private individuals on securities bought and sold in foreign markets. In Europe, where many countries use direct reporting systems (reports filed on individual transactions) to measure transactions of residents, coverage has been less than complete because of noncompliance with reporting requirements by individuals.

By way of examples, a change in tax regulations in the Netherlands in 1987 and a proposed withholding tax in Germany in 1988 triggered large capital outflows from these countries. In Germany, sizable amounts appear to have been channeled into bond funds in Luxembourg. More generally, capital flight from developing countries frequently escapes statistical recording; some part of the capital would be invested in foreign securities.69

Banks and other financial institutions in Switzerland have a long tradition of managing asset portfolios for nonresidents. These “custody accounts” contain securities and other assets that are managed on a discretionary or advisory basis.70 Holdings in Swiss custody accounts are said to total $1,000 billion, a large part of which is believed to be held for foreign individuals. Similar fund management services are conducted in other industrial countries and in offshore financial centers.

The impact of such arrangements on global portfolio imbalances is similar to trading by individuals through nonresident banks and brokers. If only a fraction of the reputed capital flows through Swiss custody accounts and similar accounts elsewhere was invested in securities, the impact on the imbalance could be considerable. It is much more likely that the capital inflow into securities will be recorded in some set of balance of payments accounts than would be the case with the corresponding outflow. An added complication is the reinvestment of income on holdings of foreign securities abroad. Asset balances may rise through time because of reinvested income. This more general problem applies to banking as well as to securities flows.71

Financial Intermediaries

Most industrial countries rely heavily on information reported by banks and securities dealers involved as intermediaries, custodians, or agents in the settlement of cross-border security transactions.72 For balance of payments reporting purposes, the ability of market intermediaries to distinguish between domestic and foreign residents is of paramount importance.

On the issue of determining residency, custodial (nominee) relationships may give rise to measurement problems. A major concern in many countries is whether securities held by domestic nominees on behalf of foreign clients are captured in reporting by intermediaries. Investors frequently find it convenient to lodge their securities with nominees when investing in overseas markets, and compilers have no way of knowing whether financial intermediaries correctly capture all transactions involving foreign accounts held in nominees’ names. The issue of residency is likely to be of much greater importance to the measurement of portfolio capital than to other types of capital flows where more of the data are collected from principals than from intermediaries.

The way that intermediaries classify securities may be another pervasive problem for compilers. In some countries, reporting requirements for intermediaries are very exact: they may, for instance, specifically call for the consolidation of transactions of different departments of the firm but require the exclusion of certain types of security transactions captured elsewhere (for example, repurchase agreements). However, verifying the resulting information is difficult.73

After deregulation of its securities market in 1986, the United Kingdom suspected major problems in the reporting of securities transactions by intermediaries. The emergence of a large and growing positive balancing item in the U.K. balance of payments and implausibly high figures in the U.K. financial accounts for domestic bond holdings, especially by the personal sector, supported the view that foreign investment in U.K. bonds might be understated. This was believed to be linked to problems in defining residency. Investigations revealed the following anomalies:

  • —omission of some transactions with overseas parents;

  • —misclassification of some offshore U.K. residents (such as the Channel Islands);

  • —misclassification of U.K. and overseas securities; and

  • —failure to exclude some short-term paper, options, and currency swaps from securities reports.

The identified deficiencies in reporting by intermediaries led to large upward revisions, totaling $18 billion for the period 1984 to 1988, to net foreign investment in U.K. bonds. The current methodology for measuring foreign transactions in domestic bonds is indirect and based on the U.K.’s domestic financial accounts. Any net transactions in U.K. bonds not attributed to the domestic sectors of the United Kingdom are attributed to nonresidents. More recently, large upward revisions were also made to net foreign investment in U.K. equities, on the basis of findings from a 1989 share register survey.

Other Financial Institutions

Large institutional investors such as pension funds, insurance companies, and investment trusts accounted for most of the growth in international portfolio investment in the 1980s. As a result of more sophisticated management of portfolios, large investors shifted part of their assets to foreign managers and often bought and sold securities directly in foreign markets.

The Working Party’s supplementary portfolio survey requested that each country provide an institutional profile of resident entities directly reporting balance of payments data on portfolio transactions. As countries employed different collection systems, the actual count of reporting transactors was secondary to the question of whether transactions of large institutional investors were properly covered in national collection systems. The following paragraphs discuss certain findings of the supplementary survey that relate to reporting coverage.74

The collection system in the United States required reporting by any entities or persons who engaged directly with nonresidents (those who did not use a domestic intermediary) in transactions of $0.5 million or more per month in long-term securities. Some 300 entities filed monthly reports on cross-border securities transactions. Two-thirds of these were brokers-dealers, banks, and other depository institutions; the remaining 100 were financial institutions such as insurance companies, pension funds, mutual funds, and money managers. Despite the relatively low reporting threshold, the very low number of U.S. direct reporters suggested that U.S. coverage of direct transactions in foreign markets was less than complete. U.S. compilers were aware of this problem, and work has already begun to strengthen the reporting system.

In the United Kingdom, compilers regularly surveyed an established panel of more than 1,000 insurance companies, pension funds, and investment and unit trusts for information on portfolio transactions. The survey data covered all securities bought or sold, whether the transactions occurred via domestic intermediaries or directly in foreign markets.75 Recently, the Central Statistical Office (CSO) found that, because of the age of the sample, the quality of data for private and public sector pension funds had deteriorated. U.K. compilers intend to conduct a full benchmark study in the near future, when a complete register of pension schemes becomes available to the CSO for the first time.

It was found that reporting systems employed in the rest of Europe relied heavily on banks acting as intermediaries or agents in the settlement of securities bought or sold abroad. Accordingly, relatively few reports were filed by other financial institutions.

Direct reporting by institutional investors (and borrowers) may also have been incomplete in other countries. Although direct reporting often was required by law or regulation, compilers have faced at least three difficult hurdles in ensuring accurate results: (1) obtaining reports at all, especially if they did not have full access to lists of transactors and therefore could not identify nonreporters; (2) avoiding double counting (direct transactions with nonresidents must be isolated from those completed through domestic intermediaries that file statistical returns); and (3) verifying the quality of submissions. All three were potentially serious issues and the last, verifying data quality, applied equally to submissions made by direct reporters and intermediaries alike.

In its research, the Working Party found that balance of payments compilers in a few important countries did not have access to the statistical returns filed by domestic reporters, as these were collected and aggregated by other government departments. In some cases compilers did not even have the names of reporters. Under these circumstances, compilers are unable to fully verify the data used to construct their balance of payments estimates, or to identify nonreporters and bring them into the system.

Concluding Remarks

Although the demand by individuals for foreign assets probably has increased and the measuring of such transactions is difficult, the bulk of cross-border transactions in portfolio securities has been associated with institutional transactions. Compilation systems in most countries are mainly oriented toward institutional reporters, usually intermediaries, but difficulties have arisen as wider ranges of institutions transact directly with nonresidents. The Working Party therefore recommends that compilers ensure that portfolio transactions of financial intermediaries and large institutional investors and borrowers be adequately covered by their compilation systems.

Most countries generally believed that accurate information was being collected from financial intermediaries. Nevertheless, the Working Party urges compilers to conduct periodic consultations with large banks and securities dealers in order to stay abreast of new market developments and to assess the impact of these developments on current collection systems. In order that new instruments and other innovations be treated uniformly, the Fund’s Statistics Department periodically should consult with financial experts and national compilers to discuss financial market developments and to agree on the treatment of new instruments in the context of balance of payments statistics.

Compilers require better access to original statistical information in order to research and integrate sources of financial information in the balance of payments without omissions or duplications. The Working Party encourages greater sharing of statistical information on portfolio transactions between national compilers and other government departments that collect the basic data used in balance of payments compilations.

Stock of Cross-Border Bonds

As a further check on its adjustments to the flow statistics and to assess the cumulative effect of past mismeasurements of cross-border bond transactions, the Working Party examined the international investment position statistics for cross-border bonds. Several large industrial countries, including the United States, Japan, Germany, and the United Kingdom, derived some portfolio stocks by cumulating flows. Also, investment income estimates frequently were derived by applying market yields to these position estimates. While stock figures were requested on the Special Questionnaire, the reported data were incomplete in many respects. It was therefore necessary to supplement this information to construct a more complete picture of the global stock of bond assets and liabilities.76 (The methodology used to establish estimates of this stock for the year 1988 is discussed in Appendix VII.)

The estimates derived by the Working Party are summarized in Table 22 and show the global stock of cross-border bond liabilities to be about $1,600 billion at the end of 1988. It is estimated that $200 billion of these securities were held by foreign monetary authorities as part of reserves, with the balance of $1,400 billion being owed to portfolio investors. Measured from the creditor side, portfolio holdings of cross-border bonds totaled only $1,000 billion. This left an imbalance of $400 billion, which reflected errors and gaps in data and differences in methodologies and valuation.

Table 22.

Stock of Cross-Border Bond Assets and Liabilities, Year-end 1988

(In billions of U.S. dollars)

article image
Source: As reported in the Special Questionnaire, plus estimates by the Working Party. See Appendix VII for note on methodology.

Includes equities.

Includes special financial institutions.

Assets exclude bonds held as part of reserves.

Liabilities constituting foreign authorities’ reserves.

The G-7 countries, together with a few other industrial countries, accounted for most of these cross-border positions. On the asset side, the largest holders of foreign bonds were Japan, Switzerland, the United Kingdom, the United States, and Germany; the combined holdings of these countries exceeded $800 billion, or 80 percent of the total. Cross-border bond liabilities were less concentrated. The five largest debtors were the United States, Japan, Germany, Canada, and the Netherlands; their combined liabilities amounted to some $950 billion, or about 60 percent of the total stock of cross-border bond liabilities.

It was estimated that bonds issued in foreign and international markets (international bonds) accounted for 60 percent of the total stock of cross-border bond liabilities. The balance of some $650 billion represented bonds that were originally issued in domestic markets, mainly in the United States and Germany.

These estimates are subject to the caveats discussed in Appendix VII, but the $400 billion imbalance between assets and liabilities was consistent with the apparent underreporting of assets in portfolio flow accounts. In view of the overwhelming importance of the major industrial countries in the figures above, a large part of the error in either the asset or the liability figures is likely concentrated in these countries.

Uncertainty about the amount of cross-border bonds and the imbalance in the stock of assets and liabilities has implications for global current account estimates. Most notably, an understatement of assets causes reported investment income receipts to be less than reported investment income payments in the global current account. The Fund has for some years made adjustments to investment income to compensate for such problems.77

Determining stocks of assets and liabilities or the geographic distributions of portfolio securities will be difficult to achieve on the basis of flow measurement alone. The most feasible approach to obtaining such information would appear to be through direct readings of the positions themselves, via periodic benchmark surveys conducted by the major creditor and debtor countries.

The United States proposes to conduct in the near future a benchmark on portfolio assets, the first since 1943.78 The Working Party welcomes this initiative. France and the United Kingdom recently conducted portfolio benchmark surveys that produced valuable information. For example, on the basis of a 1989 survey of share registers, U.K. compilers found that their estimates of the stock of foreign holdings of U.K. shares (as constructed from flow accumulations) were considerably understated. These results led to a doubling of the recorded estimates of net inflows into U.K. equities during the 1985–89 period. U.K. compilers now intend to conduct similar surveys annually.

The Working Party strongly recommends that countries undertake periodic benchmark surveys of cross-border portfolio claims and liabilities. The benefits would be greatly enhanced if such surveys could be coordinated in timing and design among major countries. Coordinated surveys potentially could help compilers identify errors in cross-border positions that are presently based on flow cumulations and also could provide better instrument and geographic detail than is currently available. Such surveys could facilitate comparisons among countries and provide useful information to fill gaps in areas where it is difficult for compilers to obtain full survey coverage (such as asset holdings by private individuals). Also, because of the large volume of bearer bonds issued in the Euromarket, data exchanges would afford insight about holders of these securities. The Working Party recommends that the Fund’s Statistics Department assist in the coordination of the surveys among countries and the exchange of information.

In its research, the Working Party made extensive use of information from the BIS data base on international bonds. Such information could also assist compilers in verifying coverage of borrowings in international bond markets. The Working Party recommends that the statistics on international securities, especially those on the classification of transactions and positions by domicile of borrower, be refined further.

Conclusions and Recommendations

The Working Party’s research into portfolio investment revealed serious problems in the measurement of transactions flows and in the associated stock and investment income estimates. While considerable detail was available on international bond liabilities, fewer data were available on the creditor side. The discussion on bilateral comparisons showed the considerable difficulties in interpreting bilateral portfolio flow statistics.

The Working Party found the measurement of portfolio investment to be one of the most difficult statistical problems facing balance of payments compilers. Some of the issues concern matters of identification and classification of securities that, in principle, could be corrected by compilers without great difficulty. Other problems are less tractable and may require international collaboration. Among the principal problems of both types were

  • —incorrect identification of nonresident accounts;

  • —incorrect classification of types of securities;

  • —incomplete coverage of transactions conducted by domestic residents directly in foreign markets; and

  • —lack of data on offshore transactions.

As a result of a general paucity of data, the role played by transactions in repurchase agreements in the portfolio imbalance could not be fully specified, but was probably large. The recording of repos also must be addressed by compilation systems.

Within the limits of data availability, adjustments were made to global statistics for many of these items and other factors contributing to asymmetries. However, many of the adjustments were reclassifications among capital account categories that resulted from identifiable deviations from international standards. Many of the pure measurement problems, such as those caused by incomplete coverage of transactions through foreign markets and in offshore financial centers, could not be quantified fully. Therefore the global imbalances on portfolio transactions remained large, especially for 1989, when the adjusted data on portfolio bonds showed an excess of credits in the amount of $53 billion.

In several important countries, existing data on stocks of portfolio investment are derived from flow data rather than from position surveys. To the extent that the flow data were accurate, countries employing the debtor-creditor principle for attributing geography had reasonable estimates of the geographical distribution of their foreign claims. However, because of intermediaries and nominee relationships, there is little chance that countries relying only on domestic reporters and flow statistics have developed reliable information on who holds their securities.

Imperfections in the measurement of portfolio flows have implications for other balance of payments accounts. Several large industrial countries, including the United States and the United Kingdom, estimate income flows in their current accounts through application of estimated market yields to position figures derived by cumulating capital flows. Thus, the accuracy of investment income estimates depends on the cumulative accuracy of a long sequence of flow estimates.

Against this background, the Working Party makes the following recommendations: