Income on direct investment comprises the parent company share (after corporate taxes—including withholding taxes—in the host country) in (1) the earnings of foreign subsidiaries and other incorporated affiliates, (2) the earnings of unincorporated branches in foreign countries, and (3) interest received from (or paid to, as a negative item) foreign-incorporated affiliates and branches. It includes neither royalties for the use of patents, trademarks, and the like nor fees for administrative and other services, which are classified as miscellaneous services in the balance of payments accounts.

There are various possible causes of asymmetrical reporting of direct investment income other than simple differences in measurement of the same transaction by the two sides. For one thing, the definition of direct investment is difficult to reduce to a simple quantitative measure of degree of control;5 it is not certain that a particular entity would be considered a direct investment by both countries involved. However, in practice this problem is not quantitatively very important, because the bulk of direct investment is in the form of branches or majority-owned subsidiaries, which are virtually certain to be defined as direct investment by both the investing country and the host country.6 Nevertheless, there may sometimes be a problem, in large and complex economies, in identifying those enterprises that are effectively controlled from abroad or that have foreign affiliates.

Moreover, the proper geographic allocation of direct investment income is often elusive, especially when there are intermediary affiliates in third countries. Exchange rate fluctuations may also affect the way in which direct investment earnings (particularly, as will be noted below, retained earnings) are converted from the host currency to the investor currency, or to a third currency.

In the Fund’s statistics, direct investment income, as defined above, is separated into two parts: reinvested earnings (RE) and other direct investment income (ODI). The former includes the investor’s share of that portion of the net profits of subsidiaries and other affiliates not paid out as dividends7 plus, where identifiable, that portion of the earnings of branches not remitted to the home office. The latter (ODI) then consists of dividends, interest, and branch profits (to the extent they are not included in RE). The extent to which branch earnings are in practice divided between RE and ODI in the statistics is not clear. Among major creditor countries only the United States, the Federal Republic of Germany, and the Netherlands are known to make this accounting division;8 most other countries seem to enter total branch profits as income debits or credits,9 with any unremitted portion being reflected (without separate identification) as part of the capital flow from the parent organization. The separation of reinvested branch profits does not affect the international accounts; only the division between RE and ODI is involved.

The net overall discrepancy on direct investment income in 1983 (the year on which the study is focused) was only minus $1,594 million, but this consists of an excess of credits on reinvested earnings of $9,935 million and an excess of debits on other direct investment income of $11,529 million. As will be explained in more detail below, the large excess of credits on RE is more than accounted for by the failure of most countries, in their balance of payments reports to the Fund, to include undistributed profits in either their current or capital accounts. However, since several investor countries that are major creditors in this respect (the United States, the United Kingdom, Australia, the Federal Republic of Germany, and—in its response to the income questionnaire—the Netherlands) do include them, it is not surprising that reported RE receipts greatly exceed reported payments. Other specific adjustments to RE, which, on balance, added to the excess of reported credits, are shown in Table 10 and are described in detail in the immediately following section. In the section after that, the adjustments to ODI (Table 12) are discussed; some of the factors involved are common to both sets of data, such as the inconsistent treatment of branch profits and the entire omission of certain economies from the Fund’s data.

Table 9


(In billions of U.S. dollars)

article image
Table 10


(In millions of U.S. dollars)

article image

Total capital gains and losses, realized and unrealized.

To treat all branch earnings as remitted.

Amounts received by the United States, the United Kingdom, the Federal Republic of Germany, and the Netherlands from countries not reporting reinvested earnings Includes the excess of credits reported by those countries vis-à-vis the Netherlands Antilles. Panama, and Switzerland over total debits reported by the latter three countries.

Amounts debited by the countries mentioned in footnote 3 against countries not reporting reinvested earnings credits.

Table 11


(In millions of U.S. dollars)

article image
Note: The U.K. data exclude the petroleum industry (See Appendix II.)

Estimated from combined total for the United States and Canada.

Table 12


(In millions of U.S.dollars)

article image

The United States, the United Kingdom, the Netherlands, and the Federal Republic of Germany.

Many of the gaps or asymmetries, especially on ODI, do not affect the overall balance on current account or the total balance on account of investment income, since offsetting adjustments are required in “other investment income.” (See Chapter V.) Nevertheless, for present and future analytical purposes, it is important to identify inconsistencies in individual types of transactions, even though they do not affect the investment income accounts or the current account as a whole.

For the convenience of the reader, the various adjustments to the data, shown in detail in Tables 10 and 12, are summarized in Table 9.

As can be seen, a significant part of the initial discrepancies can be explained by identifiable inconsistencies or omissions in the reported data, but a considerable gap is left to be explained. The remaining ODI discrepancy may well include large but offsetting inconsistencies in the reported data. However, the Working Party has identified several countries (Table 16) that reported larger debits on ODI than are accounted for by the major creditor countries that provide geographic breakdowns of their credits on this account. Consequently, the Working Party believes that further progress in reducing the excess of reported debits can be made on the basis of such comparisons. Effecting the necessary corrections will require collaboration among national compilers, assisted by the Fund, to locate the reporting discrepancies and to agree on the appropriate changes necessary to produce consistent statistics.

Table 13


(In millions of U.S. dollars)

article image
Table 14


(In millions of U.S. dollars)

article image

The overall discrepancy in 1979 was only $77 million The United States did not report separately on Saudi Arabia in that year, but it appears that the amount included in the U.S. totals was close to the debit reflected in the Saudi Arabian data.

Table 15


(In millions of U.S. dollars)

article image
Note: Totals exclude pairs where data from only one side are available The U.K. data exclude the petroleum industry (See Appendix II.) The abbreviation nss denotes figures not shown separately.

Excludes that portion of branch earnings classified as “reinvested” in U.S. data

For an analysis of the difference between the U.S. figure and that of the Netherlands, see Marius van Nieuwkerk and Robert P. Sparling, The Netherlands International Direct Investment Position (Dordrecht: Nijhoff. 1985), Appendix C, pp 101 ff.

Table 16


(In millions of U.S. dollars)

article image

The United States, the United Kingdom, and the Federal Republic of Germany.

The United States and the Federal Republic of Germany only. The United Kingdom figure for Algeria and Libya combined is not in excess of $59 million.

Country Notes:Ireland. No apparent reason for the discrepancy It seems logical to suppose that the United States and the United Kingdom would be major investors. Ireland does not report RE; perhaps some of what others record as RE is included in ODI by Ireland.Algeria. Almost 90 percent of the total reported by Algeria is attributed to the construction industry. It seems likely that the figure includes more than just direct investment income—that is it may reflect payments for construction services performed in Algeria.South Africa. The difference here might well reflect payments to other creditors or a different classification of portfolio vs. direct investment.Indonesia. The arrangements for operating the Indonesian petroleum industry are complicated. It may be that some payments classified as income by the Indonesians are reported under some other category by the countries receiving the funds However the Indonesian questionnaire reports $2,944 million as dividends, and these should be easy to measure accurately. It might be that investments of overseas Chinese, especially those residing in Hong Kong, are important The absence of information on Japanese income from Indonesia might be an important factor.Malaysia. The absence of creditor data from Japan; Hong Kong; the Netherlands; and probably, Chinese investors residing elsewhere than in Hong Kong could be part of the explanation.Libyan Arab Jamahiriya. French, Italian, or Dutch investments may be part of the explanation. Or the Libyan Arab Jamahiriya might have a different measure of income than the foreign investor companies employ.


The balance of payments data for 1983 reported to the Fund’s Bureau of Statistics show an excess of credits of $9,935 million on RE account. (See Table 10, line 1.) However, various adjustments to the reported figures are necessary to pick up some revisions by the compilers and to eliminate certain large inconsistencies among reporters.

First, the responses to the Working Party’s questionnaire included certain information, some of which was supplied on a confidential basis, which was not reflected in the Fund’s data. The (net) additions were somewhat larger for debits than for credits (Table 10, line 2). Presumably, some or all of these revisions and additions will be reflected in subsequent editions of the Balance of Payments Statistics Yearbook.

Second, the data reported by the United States (credits $9,602 million; debits $96 million) require two adjustments to make them comparable with the amounts likely to have been reported by the partner countries. In the first place, in calculating reinvested earnings, the U.S. compilers deducted from foreign subsidiary and branch earnings an amount of $6,542 million for net capital losses. This amount reflects both realized and unrealized capital gains and losses, the latter resulting primarily from converting the financial assets and liabilities of the foreign enterprises from host country or other foreign currencies to U.S. dollars at a lower rate (for most foreign currencies) at the end of 1983 than that which prevailed at the beginning of that year. No such adjustment would be relevant to the host countries, which generally maintain the accounts in their home currencies. (Any conversions of reported balance of payments data—to, say, U.S. dollars or SDRs—are usually made at the average exchange rate for the period.) An adjustment to eliminate this anomaly is shown in line 4 of Table 10.

Realized gains and losses constitute a small part of the total of $6,542 million, according to information supplied by the U.S. authorities. There is a possibility that at least part of the realized losses may also have been recognized by the partner countries and therefore would not have caused any discrepancy. But the individual country data supplied by the U.S. authorities, which were used in calculating the adjustment shown in line 4, were exclusive of all capital gains and losses.

The comparable entry on the debit side probably does not include any significant amount for exchange fluctuations but does include changes in market values of investment portfolios, particularly in the insurance industry. The adjustment is minus $429 million (line 4, second column)—that is to say, the recorded figure included capital gains of that amount. Again, it is possible, but not likely, that such capital gains would be reflected in partner country statistics.

A second adjustment is made in the U.S. data to transfer “reinvested” branch profits from RE to ODI in order to make the U.S. data more comparable with those of most other countries, which, as stated above, seem generally to include total branch profits in ODI. This adjustment is shown in Table 10, line 5, on both the credit and debit sides of the account. (Reinvested earnings of U.S. branches of foreign enterprises were negative in 1983.)

In addition to these adjustments in reported figures, there is another adjustment required (line 6) to recognize a large gap in the reporting of reinvested earnings. As noted above, credits are reported by four major investing countries10 (the United States, the United Kingdom, the Federal Republic of Germany, and the Netherlands) as having been received from countries or territories not reporting RE, or that are not included in the Fund’s data (Bermuda, the Cayman Islands, and Hong Kong). In addition there is an excess of identifiable receipts reported by these countries from two major offshore areas (Panama and the Netherlands Antilles) and Switzerland over the total amounts of debits shown by those three reporters.

The first of these adjustments was obtained essentially by subtraction. The receipts of the creditor countries from all countries reporting RE debits were calculated. Some estimating was required, notably in the cases of countries included in somewhat broader groupings in the creditor countries’ reports or where RE and ODI were not shown separately. The receipts thus calculated were deducted from total RE credits reported by the four investing countries and the difference, amounting to $10,167 million, is part of line 6 of Table 10.

As to the adjustment for Panama, the Netherlands Antilles, and Switzerland, it seems clear that these countries (among, possibly, several others) do not consider as residents for balance of payments purposes certain local subsidiaries and branches of foreign enterprises which engage only in international transactions and do little or no business with other local residents. The excess of identifiable RE receipts by the four creditors over total RE debits (to all countries) by these three host countries was $2,622 million, the remaining part of the total in line 6.

An analogous adjustment (line 7) was made for RE debits reported by the same four countries against countries not reporting RE credits. Again, the amount, minus $463 million, was estimated by deduction—that is, the amounts reported by these countries as RE debits against each other11 were deducted from their total RE debits. The remainder was assumed to represent debits against (1) countries included in the Fund’s reports but not reporting RE credits, and (2) areas not included in the Fund’s data.

A geographic allocation of the net adjustments to RE is shown in Chapter IX. About half of the total adjustment of $10.4 billion was applicable to industrial countries.

After making the foregoing adjustments there remains a negligible statistical discrepancy (excess of debits) of $443 million on this account. But it may be useful to observe at this point that the failure of debtor (host) countries to include RE in their balances of payments tends to understate both (1) the income accruing to nonresident investors, thus understating their current account deficits, and (2) the corresponding net inflow of foreign capital.

Moreover, it is virtually certain that the small remaining discrepancy conceals inconsistent reporting among those countries that include RE in their statements. As a partial test of this possibility, the bilateral data for four countries were compared for consistency; the results are shown in Table 11. Among the four, reported credits exceeded reported debits in 8 out of 12 cases; the total (net) excess was $1.8 billion. However, the omission of data on the petroleum industry from the United Kingdom data, and the presence of known inconsistencies between countries in their methods of allocating the data geographically, make it inappropriate to use these results to further adjust the world totals.


Other direct investment income (ODI) includes dividends, that portion of branch profits considered to be remitted, and interest—all net of withholding taxes. Interest is measured net—that is, any interest paid by a parent company or by a domestic affiliate thereof to a foreign affiliate is treated as negative ODI receipts and not as income payments by the investor country.12

The asymmetry in this account for 1983, according to Fund data, was minus $11,529 million. (See Table 12, line 1.) Revisions reported in the questionnaires and other data submitted to the Working Party, but not yet incorporated in the Fund’s data, add $2,039 million to credits and $2,540 million to debits.

Again, the U.S. data must be adjusted to make them compatible with the data reported by partner countries. First, the U.S. balance of payments data reflect $4,246 million of interest paid to financial affiliates incorporated in the Netherlands Antilles as negative ODI receipts. These companies issued bonds in the Euromarkets and, in the main, loaned the proceeds to their U.S. parent companies. The interest paid by the latter is used by the subsidiaries to pay interest on their outstanding bonds, to cover non-interest expenses, and to generate net earnings; the latter are included in the U.S. data as earnings on direct investment in the Netherlands Antilles. The balance of payments reports of the Netherlands Antilles apparently omit these transactions entirely. Presumably, the interest received by the bondholders is recorded, if at all, as portfolio investment income received from the United States by the countries where the bonds are owned or held. Information supplied by the U.S. Department of Commerce indicates that $2,760 million of the interest paid to the Netherlands Antilles subsidiaries in 1983 was in effect used to pay interest to nonresidents of the United States, presumably bondholders. Shifting this amount from (negative) direct investment income credits to portfolio investment income debits (Chapter V) increases U.S. and world credits on direct investment income by the same amount. (See Table 12, line 4.)

Second, we must add back to ODI receipts and payments for the United States the amounts of branch profits considered as reinvested, but which were eliminated from reinvested earnings. The amounts for 1983 were $2,762 million in credits, and $320 million in debits. (See Table 12, line 5.)

There are quite a few countries, some of them major, that do not report direct investment income separately but include it with other private income, although one cannot be absolutely certain that it is always so included.13 Some, but not all, of these countries are shown separately in the data for the United States, the United Kingdom, the Netherlands, and the Federal Republic of Germany.14 These four countries combined reported receipts of $2,643 million from, and payments of $501 million to, countries that have not yet made this breakdown. It is appropriate to adjust the world data by these amounts, that is, by adding the receipts reported by the four investor countries to world debits and the payments reported by the same four countries to world credits. This is done in Table 12, line 6. Again, offsetting adjustments will be required in portfolio income.

There are three economies with which the United States and the United Kingdom report direct investment transactions (Bermuda, the Cayman Islands, and Hong Kong) that are not national states and do not provide balance of payments statements for inclusion in the Fund tabulation.15 In addition, the United States reports income under the area category “international” (mainly from the operation of merchant ships and oil drilling rigs) that may not be reported by any partner countries. The United States and the United Kingdom together reported $1,903 million of ODI receipts from, and $30 million of payments to, these areas. Adjustment for this factor is made in Table 12, line 7.

Another identifiable discrepancy involves the balance of payments accounts of Saudi Arabia, which show $4.3 billion of “oil sector investment income,” classified as an “other direct investment income” debit in the Fund’s balance of payments statistics. However, very little is reported as direct investment income receipts from Saudi Arabia in the geographic data supplied by countries that were the major investors. The United States, the United Kingdom, and the Federal Republic of Germany report specifically on Saudi Arabia, noting total receipts of $117 million. In the absence of any evidence that the income is being reported somewhere else in the world as an income credit, it seems appropriate to adjust ODI payments downward by, say, $4.0 billion (Table 12, line 8).

The foregoing adjustments reduce the aggregate asymmetry on this item by about half, but many questions remain. Bilateral data between several major pairs of countries (Table 15), as well as other apparent inconsistencies (Table 16), provide evidence that the remaining discrepancy on direct investment income account may well reflect significant bilateral differences. These bilateral discrepancies will require further study and, as suggested in the last section of this chapter, the data could undoubtedly be improved through collaboration among the compilers in the countries concerned.


For reinvested earnings, the same situation prevailed in 1979 as in 1983; the United States and the United Kingdom reported receipts (credits) for reinvested earnings from countries not reporting this item (or not included in the Fund totals) of $13,401 million. This amount was added to total debits, as was done for 1983 in Table 10, line 6; the results are shown in Table 13.

While the remaining discrepancy is $1.6 billion, it must be noted that, as was true for the 1983 data, the adjustment for omitted debits can only be approximated. Similar calculations were not made for the intervening years; nevertheless, it seems reasonable to assume that the bulk of the asymmetry on RE in every year results from this factor.

The capital gains/losses included in the United States data were as follows: 1980, loss, $1,624 million; 1981, gain, $426 million; 1982, loss, $2,146 million. Adjustment for this factor would enlarge the excess of credits in 1980 and 1982 and reduce it in 1981.

With respect to other direct investment income, it appears that inconsistent treatment of income on direct investments in Saudi Arabia was the major factor in producing the excess of debits in most recent years, as shown in Table 14.


Within the time and resources available to the Working Party and the Technical Staff, it was not possible to reduce the remaining ODI discrepancy of minus $6.2 billion in 1983 by quantifiable measures.

However, with the extensive bilateral data available from four major countries, two further avenues of partial investigation were possible. The first was to compare the bilateral data among the four countries themselves (and, in part, with Canada); the second was to compare ODI receipts of the four countries from a number of host countries with total ODI debits reported by these host countries. The results are presented in Tables 15 and 16, and analyzed in the remainder of this section.

Except for the United States, the figures in Table 15 are those furnished in reply to the Working Party’s income questionnaire. For the United States, they are taken from articles in the Commerce Department’s Survey of Current Business and from tabulations received from the Department, with capital gains and losses removed and after the other adjustments delineated above.

Not much can be gleaned from the data in Table 15; the overall discrepancy is small, but that is probably coincidental. There are several possible explanations for bilateral differences, including the following:

  • (1) The U.S. definition of direct investment is broader than that generally employed elsewhere, in that it includes cases where as little as 10 percent of the equity is owned by the investing company. This is probably a minor factor; in 1982, 94 percent of all direct investment income (except of banks) was received from majority-owned affiliates, including branches.

  • (2) The U.K. geographic data do not reflect oil industry transactions. Such transactions (as revised herein) accounted for 34 percent of total direct investment income credits (including reinvested earnings) and 55 percent of debits. If these same ratios were applied to U.K. transactions with the four other countries in Table 15, total U.K. credits would be $1,773 million (instead of $1,172 million) and debits $2,854 million (instead of $1,290 million). For all four countries total credits would be $9,768 million and total debits $10,478 million. The discrepancy would be a $710 million debit instead of the present $253 million credit.

  • (3) But the major differences are likely to lie in the geographic allocations. That is, partner countries may assign transactions to different areas, or even to different transaction categories. This is especially likely when there are intermediary subsidiaries in third countries, and even more so if such intermediary subsidiaries are jointly owned.

Thus, it is by no means certain that these apparent discrepancies are real. For instance, it is quite possible that some of the transactions recorded by the United States vis-à-vis the United Kingdom are reported by the latter as having occurred vis-à-vis a third country, in whole or in part. In such a case, there would not necessarily be a world discrepancy, even if the third country had omitted the transaction on both sides of its accounts.

Nevertheless, taken by themselves, transactions with the United States as host country reflect an excess of debits of $2.3 billion. This discrepancy may well be due to (1) the omission of the petroleum industry from the United Kingdom data or (2) different country classifications, especially for investments of large European multinationals. But it is possible that, for some reason, at least part of the U.S. debits are not being reflected in credit entries elsewhere.

The second avenue of investigation, illustrated in Table 16, can only be imprecise, in part because detailed geographic breakdowns are unavailable for several countries known to have significant direct investments abroad: France, Belgium, and Japan, among others. Even so, the debits reported by the six developing countries included in the table significantly exceed the credits reported by three major investing countries, suggesting an overstatement of debits under this heading, though there may be some creditors not reporting country detail that could account for part of the remainder. In some cases, debits under “other investment income” seem small compared with debt outstanding, which may indicate some misclassification—that is, perhaps part of the income debits reported under direct investment should have been considered as income on portfolio investments. However, the evidence is not clear enough to justify any adjustment on our part. Some possible explanations for the differences are noted below, but in each case a more thorough investigation would be needed to establish whether, in fact, any of these countries have overstated their debits (and, if so, by how much), whether the creditor totals have been understated, or whether the difference has some other explanation. It seems reasonable to conclude that, unless the credits are being reported by some countries other than the three major investing countries, the bulk of the remaining discrepancy shown in Table 12 can be accounted for by the differences in the reports of these pairs of countries.16


As we have shown, most of the world discrepancy on direct investment income account is attributable to omissions: the omission of reinvested earnings by a large number of countries; the omission of certain “international” companies’ activities, especially in offshore financial centers; and the omission of certain countries and areas from the Fund’s Balance of Payments Statistics Yearbook tabulations. Another major factor in the discrepancy is the treatment of branch earnings and capital gains and losses by the U.S. compilers. Finally, there may be significant, though unmeasurable differences in recording the same transactions—differences in amount, in classification, and in geographic allocation. Our recommendations for dealing with these issues are both immediate and long run in nature.


(1) In the immediate future, the Fund could considerably reduce the reported discrepancies by applying the same adjustment procedures used in this chapter. The Working Party realizes that many of the adjustments could not appropriately be applied to present tables in the Yearbook itself, either in Part 1 or Part 2. But in other publications, such as the World Economic Outlook, most, if not all, of the adjustments we have made seem feasible, and the Yearbook could contain adjusted summary tables showing revised totals. In this connection,

(i) With respect to reinvested earnings, the Working Party of course encourages the Fund to continue working with countries that do not compile data on these earnings, or whose data do not agree with those of partner countries. However, pending significant results from this action, the Working Party recommends that global summations of the current account and the investment income account omit reinvested earnings entirely. (Of course, the data as reported would be available in other tables.)

(ii) World aggregations of balance of payments data on direct investment income could include an “adjustment” column for those transactions only one side of which was presumably included in the reported data. This could include all, or virtually all, of the adjustments made in this chapter, although some countries, notably the United States, would need to be called upon for unpublished, though not confidential, details. Conceivably, this “adjustment” column could also include an allowance for apparent discrepancies in bilateral reporting not reflected in other adjustments.

(2) Bilateral comparisons. Data on direct investment income must, in general, be built on reports by individual companies. Because reporters could be requested, at least by the host countries, to produce country-by-country figures, such data are particularly susceptible to analysis by comparing partner country compilations. Four creditor countries which account for 76 percent of total ODI receipts reported to the IMF (the United States, the United Kingdom, the Federal Republic of Germany, and the Netherlands) have virtually complete country breakdowns in their data or at least extensive country detail. Thus extensive bilateral comparisons would be possible if these countries would make available the data they have in the greatest detail that confidentiality restrictions would permit, either in published form or to principal partner countries, or to the Fund, which could coordinate such comparisons—either through correspondence or, if the discrepancies remained large, through face-to-face discussions. As part of this process other creditor countries might obtain some geographic detail from resident multinationals, or could arrange to do so.

(3) As noted above, discrepancies can arise from the inclusion in balance of payments income data of unrealized exchange gains or losses resulting from the conversion of financial assets and liabilities into the currency of another country. Removing these potential discrepancies would enable closer bilateral comparisons to be made and would be consistent with the principles of the national accounts. Therefore, the Working Party recommends that such gains or losses not be included in balance of payments data. However, the total amounts of capital gains or losses could be given in a footnote or as a reconciliation to position tables.

(4) In order to present direct investment income in a consistent manner across countries, and in accordance with the Fund’s Balance of Payments Manual, we recommend that direct investment income be divided into the following components:

  • Total direct investment income

  • Branch profits (Countries that wanted to divide these between remitted and reinvested could do so.)

  • Total direct investment income

  • Income of incorporated affiliates

    • Dividends

    • Undistributed profits of incorporated affiliates

  • Interest

(5) The Working Party recommends that the U.S. compilers change their treatment of interest paid to Netherlands Antilles affiliates (and to other areas where similar transactions may occur) in the manner done in this chapter; reconciliations with other countries would be much simpler and closer to the real economic significance of the transactions if the United States would change its method of accounting to reflect the fact that the bonds essentially represent borrowing by the parent companies and would be viewed as such by the investors.17 The same issue exists, on a smaller scale, for other countries.

(6) The Working Party takes note of, and commends the action by, the authorities in the United Kingdom to include income related to the petroleum industry under direct investment rather than other investment income and to separate retained from distributed earnings. We also note that there may be some inconsistencies between the treatment of certain jointly owned petroleum enterprises in the U.K. statistics and in the statistics of the Netherlands and of the other countries in which these enterprises have investments. We recommend that efforts already begun, bilateral and multilateral, to arrive at consistent treatment of these and other similar situations, wherever they may exist, be expanded and intensified.


International insurance transactions, apart from those related to merchandise trade, usually take one of three forms:

(1) Direct writing. The insured deals directly, or more usually through insurance brokers, with an insurance company located in another country.

(2) Branches and subsidiaries. A foreign insurance company is licensed to do a domestic insurance business, either directly (as a branch) or through a domestically incorporated subsidiary. In principle, such activities are direct investments, and they are so treated in the Report.

(3) Reinsurance. By definition, such transactions occur between insurance and reinsurance companies.

So-called captive insurance companies may be considered a special example of the second case just mentioned. These companies originated in the desire of large manufacturing companies to minimize taxes (since in the United States premiums on self-insurance are not deductible) and to reduce other insurance costs wherever possible through the ownership of insurance firms favorably located. Even though some of the initial advantages have been reduced as tax and other regulations have changed, many of the companies have developed into viable entities in their own right and have become especially useful in the light of the recent difficulties of obtaining liability coverage in the normal way. In part, the exposure of the captive companies is covered by reinsurance with old established insurance organizations.

Certain features of the operations of captive companies have special implications for balance of payments reporting:

(a) They are not, typically, subsidiaries of insurance companies, but of industrial companies or even of groups of individuals or companies. There are instances, however, where the ownership is not sufficiently concentrated to qualify as a direct investment.

(b) Their transactions, both with their parents and with their clients, are not usually reflected in the Fund’s balance of payments data. This is because they may be operating in countries that do not report to the Fund (e.g., Bermuda); or they may be omitted from reported balance of payments data (e.g., The Bahamas) because they deal only with nonresidents.

(c) On the other hand, transactions with the captives are likely to be reflected in the balance of payments reports of the partner countries and therefore, at least to some extent, are reflected in the adjustments made in Chapters IV and V of the Report. This applies, in particular, to the dividends and profits accruing to the parent companies (Tables 10 and 12) and the earnings of the captives on their investments in international financial markets (Chapter V).

But clearly the premium receipts of, and the claims payments by, the captives are not included in either the Fund’s data on insurance or in the adjustments proposed in this report. The possibility of asymmetric reporting exists to the extent that the partner countries do report these insurance transactions as such or because the flows of funds appear as capital flows to the financial markets where reserves are held. Total assets of Bermuda insurance companies alone (not all of which are captives) were reported to amount to $22 billion in 1984.