Information on the composition of reserve transactions is needed to balance the global matrix, especially in portfolio investment and banking flows. The Working Party secured this information from a large number of important countries and used it in parts of this report. The accounting for transactions of certain countries with the European Monetary Cooperation Fund gave rise to imbalances—for which the Working Party has made adjustments—in gold and foreign exchange reserve statistics. Several major central banks report claims on domestic banks as reserves, a practice that may cause other asymmetries in balance of payments and banking figures.


As illustrated by Table 3, there is sometimes a large difference between recorded official reserves and liabilities constituting foreign authorities’ reserves. This difference—which averaged –$15 billion in the 1986–89 period—is part of the global capital account discrepancy. Details on reserve transactions are needed to construct a complete picture of both official and private transactions in portfolio securities and other assets acquired for reserve purposes. One of the Working Party’s major efforts, therefore, was to ascertain the composition of transactions in countries’ foreign exchange reserve assets to help explain this part of the global imbalance. As a result of this effort, the Working Party was able to quantify inaccuracies in reported LCFAR figures and to eliminate the reserves/LCFAR imbalance through reclassifications from portfolio investment and “other capital” into LCFAR. To a large extent, these adjustments did not affect the global imbalance, but they reduced offsetting discrepancies in important categories of the capital account.

Changes in official reserves are a highly visible part of international capital movements and, in principle, they should be among the most reliable and best-measured elements of capital flows. Using the Fund’s standard components, Table 41 illustrates reserve transactions made during the period of this study.

Table 41.

Reserve Transactions by Standard Fund Components, 1986–89

(In billions of U.S. dollars; outflows ( – )) 1

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

Outflows equal an increase in reserves.

Formerly designated credit from the Fund and Fund-administered resources.

Three of the reserve components are related to IMF activities: special drawing rights (SDRs), reserve position in the Fund, and use of Fund credit and loans. The last of these is a reserve asset in international organizations’ accounts and the only liability among reserve components in countries’ accounts. In the Balance of Payments Statistics Yearbook, statistics on Fund-related reserves are taken directly from Fund records, rather than from country submissions.111

In principle, transactions in SDRs, use of Fund credit and loans, and monetary gold should balance to zero at the global level, but nonzero totals sometimes are found in the Yearbook accounts. One reason for the imbalance in monetary gold results from accounting practices for gold swaps of countries that belong to the European Monetary System; these swaps will be discussed later in this chapter.

By far the most important component of official reserves is foreign exchange assets (shown in Yearbook Table C-35). “Foreign exchange” usually is understood to mean gross claims—securities, bank deposits, and the like—which are held by reserve authorities of one country and for which other countries have counterpart liabilities.112 Additionally, under the standards of the Balance of Payments Manual, external claims that are held by deposit money banks and that “are readily available to the authorities to meet a balance of payments need” (in other words, that could be requisitioned by the authorities) also may be counted as reserves. Reserve holders seldom reveal the specific composition of their foreign exchange assets.

Instrument Composition of Reserve Transactions

Research into reserve composition usually has dealt with the currency composition of reserve transactions and holdings. For the purposes of the Working Party, however, it was more important to identify the instrument composition of reserve transactions. As noted in Chapter 1, balances in the principal categories of capital flows exclude reserve transactions on both the assets and liabilities sides (although nonreserve “resident official” transactions are included). Discrepancies within specific categories of nonreserve flows provide important clues to measurement flaws in the system as a whole, especially for securities and banking claims. Without instrument information for official holders, however, it is not possible to assess fully the quality of the data that purport to represent only private capital flows.

One approach to this problem is to secure information from debtors about whether counterparties to certain capital flows are reserve authorities. Accordingly, the Fund has requested countries to identify LCFAR in their submissions to the Statistics Department.113 For example, the major reserve center, the United States, reports in its balance of payments accounts transactions in “foreign official assets” for both securities (primarily U.S. Government issues) and banking positions. Overall, however, LCFAR reporting is uneven and, in total, clearly deficient. Only a few other countries report LCFAR transactions in securities. A somewhat larger number report LCFAR in banking. Some reserve centers, like Switzerland, report to the Fund only small amounts of LCFAR transactions; others, like Japan, report none at all.

Incomplete reporting alone makes LCFAR an imprecise measure of transactions in reserve assets. Even the recorded entries are potentially misleading because an “official holder” may not always be a reserve authority in the accounts of the claimant. The consequence of this fragmentary reporting is that there is sometimes a large gap between transactions in total foreign exchange reserves as reported by asset holders and those in LCFAR as compiled from debtor accounts. Table 42 gives illustrative figures on this comparison.

Table 42.

Discrepancy Between Reserve Transactions and LCFAR, 1986–89

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

article image

Includes reserve position in the Fund’s accounts.

Results differ from those shown in Tables 3 and 4 because here LCFAR is compared only with foreign exchange reserves plus reserve position in the Fund, rather than with total reserves.

Since reserve holdings invested in portfolio securities appear to be highly concentrated in issues of a few debtor countries that do report LCFAR, it is likely that the residual amounts in Table 42 are mostly unidentified deposit transactions with banks in the Euromarket and elsewhere.114 However, the Working Party needed more explicit information about reserve-investment patterns to resolve this discrepancy.

Special Questionnaire Responses

The Special Questionnaire on International Capital Flows attempted to address the reserves/LCFAR gap by requesting instrument detail on countries’ transactions in foreign exchange reserves. A number of countries complied with the request, but others initially failed to provide the needed breakdowns.115 The Working Party considered this detail to be of sufficient importance that it later repeated its request for this information—with assurances of confidentiality—to 13 countries with large foreign exchange transactions.

Although this request for closely guarded information was unprecedented, virtually all the major reserve holders eventually provided the requested detail. In total, 37 countries eventually provided to the Working Party at least some information about their foreign exchange transactions. Responses to the Working Party’s requests fell into three groups:

  • (1) No detail: 16 countries either made no response or their response included no detail on foreign exchange reserve movements.

  • (2) Partial detail: Questionnaire responses of five countries included information on whether changes in foreign exchange reserve assets were against foreign official or private sectors, but did not provide full breakdowns.

  • (3) Full instrument detail: 32 countries provided a generally complete breakdown of foreign exchange reserve transactions by type of claim.

Countries falling in each of these three groups are shown in Table 43.

Table 43.

Foreign Exchange Reserve Transactions, Detail Provided to Working Party

article image
Source: Special Questionnaire and follow-up letters.

Table 44 summarizes the full-detail data for industrial and developing countries in relation to reported and Yearbook foreign exchange reserve totals. The transactions detail submitted by these countries clarified a very high proportion of total reserve movements. Among industrial countries, which accounted for 70 percent of reserve transactions in the period 1986 through 1989, the coverage by instrument was almost complete. It was much less so for developing countries; for instance, a large gap remained in 1989.116 The “totals reported,” as given in Table 44, differ from Yearbook totals for two reasons. First, questionnaires were not sent to every country and there were some nonrespondents. Second, even among respondents, a certain number submitted figures that differed from their own Yearbook totals. Not all the reasons for this could be clarified. In some cases, respondents appeared to classify, as foreign exchange, assets that are classified as “other reserves” in their regular Fund submissions. Problems with the special calculations may explain Problems with the special calculations may explain other cases.117 Notwithstanding these anomalies, the Working Party’s inquiries yielded an impressive amount of detail about reserve transactions, and this detail was crucial to quantifying certain adjustments to portfolio investment and to the banking statistics in “other capital.”

Table 44.

Foreign Exchange Reserve Transactions, by Detail Provided to Working Party, 1986–89

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

article image
Source: Special Questionnaire and follow-up letters.

Swap Transactions in the European Monetary System

Foreign exchange statistics of countries that are members of the European Monetary System raised certain problems for foreign exchange totals that should be discussed briefly.118 Established in 1979, the EMS works through the European Monetary Cooperation Fund (EMCF). Most relevant to the measurement of capital flows is the EMCF’s “reserve pooling” arrangement, under which participants swap 20 percent of their gold and dollar reserves into the EMCF in exchange for European currency units (ECUs).119 According to Manual standards, the EMCF is an international organization, and participant countries should treat swaps with the EMCF as repeated purchases and sales of dollars and gold against ECUs. These ECUs, in turn, are treated as foreign exchange reserves in the balance of payments accounts of all the EMS members.

Two EMCF-related complications arise for the measurement of capital flows: (1) there is a reporting asymmetry because the EMCF—an international organization—does not report to the Fund changes in the ECU liabilities (a form of LCFAR) of its members or the counterpart changes in gold and dollar assets, and (2) not all EMS members account for EMCF swaps according to Manual standards. A number of them do not record their gold and dollar swaps as foreign exchange transactions at all, but in the absence of EMCF reporting, no imbalance arises from this divergence. Gold swaps that are properly reported in the Yearbook on the sale or purchase basis, however, do contribute to gold and foreign exchange reserve errors for lack of counterpart liabilities.

Adjustments to foreign exchange figures for EMS members can be made in one of two equivalent ways: either by incorporating the EMCF as an institution into the global statistics and adjusting figures of countries that do not report their swaps as transactions, or by making the EMCF transparent by treating all EMCF swaps as if they were not foreign exchange transactions. By Manual standards, the first approach is conceptually correct, but the required data for the EMCF were not available. The Working Party therefore followed the second approach which entailed adjusting data for countries that follow the Manual norms.

EMS members that follow the recommended accounting fall into two groups. In the first, Spain and Portugal “transacted” in gold insofar as they joined the EMS arrangements during the 1986–89 period and made initial gold swaps that reduced national gold holdings and increased their “foreign exchange” assets.120 With information provided by the two national authorities, the Working Party eliminated the associated variations in the Yearbook’s monetary gold and foreign exchange accounts. In the second group of countries are several EMS members—Germany, Ireland, and Italy—that did not transact in national gold at all during the 1986–89 interval. However, their monetary gold and foreign exchange accounts showed movements resulting from gold revaluation that were part of the EMS swap sequences. The Working Party adjusted the data for these three countries, too, in order to compensate for the reporting asymmetry noted earlier.121

The Working Party’s offsetting adjustments to the gold and foreign exchange accounts of several EMS countries entailed nothing more than a reclassification of reserve components between gold and foreign exchange; they left each country’s total reserve movements unaffected. Thus, these EMS-related adjustments to reserves had no effect on total global capital flows, but they changed the foreign exchange subtotal to be consistent with activity recorded in other kinds of capital flows. As this discussion illustrates, the inconsistency in compilation procedures used by EMS members led to imbalances in gold and foreign exchange statistics. In order to eliminate these asymmetries, and in accordance with Manual norms, the Working Party recommends that EMCF gold swaps be compiled as sales and repurchases in EMS members’ balance of payments accounts. It further recommends that the EMCF transmit data to the Fund on its swap transactions with members.

A 1989 transaction involving Belgium-Luxembourg gave rise to a somewhat broader problem in the statistics. The authorities there sold a large quantity of gold against foreign exchange, and their gold swap with the EMCF therefore decreased. Although the foreign exchange acquired from the gold sale was included in the Belgium-Luxembourg 1989 balance of payments accounts, the increase was partly offset by the drop in ECUs issued against the smaller quantity of swapped gold. The unique aspect of this transaction was that no decrease in monetary gold was recorded that year in the Yearbook accounts. The Working Party’s adjustment to the Belgium-Luxembourg statistics thus consisted of increasing published foreign exchange reserves (to neutralize the drop in ECUs) and decreasing gold holdings by a multiple of the foreign exchange adjustment. This adjustment changed the global total of reserve transactions by several billion dollars.

In addition to adjustments for EMCF swaps and the missing Belgian gold transaction, the Working Party made further small adjustments to total official reserves as a result of its research on the U.S.S.R. (see Chapter 13) and on Hong Kong and Taiwan Province of China. The net adjustments from these various sources are presented later in the chapter.

Instrument Allocation of Foreign Exchange Transactions

Table 45 summarizes the Working Party’s detailed allocations of foreign exchange transactions to net purchases of particular assets. These distributions have been used elsewhere in this report (especially in Chapters 4 and 5) to help resolve imbalances in other kinds of capital flows and the relations of the totals to the reported figures found in the Yearbook. Total foreign exchange flows in Table 45 differ from amounts shown in the 1990 Yearbook for two reasons: (1) the EMCF adjustments described in the preceding paragraphs, and (2) the Working Party’s inclusion of foreign exchange estimates for the U.S.S.R. as mentioned previously.122

Table 45.

Instrument Detail of Transactions in Foreign Exchange Reserves, 1986–89

(In billions of dollars; outflows ( – ))

article image
Source: Data derived from Special Questionnaire and follow-up letters.

Includes deposits with BIS.

The figures under line 2 of Table 45 show relatively large investments of foreign exchange reserves in long-term (portfolio) securities in the past few years. The scale of these investments certainly modifies the traditional impression that reserve assets are invested predominantly in short-term instruments. The largest part of this investment in securities has been in government bonds. Line 3 of the table, “other capital,” captures all other foreign exchange transactions, details of which are in lines a to d. Short-term official bills, deposits with central banks, and deposits with commercial banks account for almost all remaining reserve changes. As might be expected, movements in these short-term forms of reserves show high year-to-year variability.

Adjustments to the Reserves/LCFAR Discrepancy

Table 46 provides a detailed look at the Working Party’s total adjustments to official reserves and LCFAR. (These adjustments are presented in consolidated form in Table 4 of this report.) Lines 1 and 2 of the table reproduce reserves and LCFAR as they appear in the 1990 Yearbook. Line 3 is the discrepancy between these quantities as shown, for instance, in Table 3. Adjustments to the reserves/LCFAR discrepancy consist of two parts. The “adjustments to reserves” are amounts mentioned in the previous section in connection with the Belgian gold transaction and amendments to data for Hong Kong and a few other countries. The “adjustments to LCFAR” require more explanation. As noted earlier, LCFAR are often a poorly measured proxy for foreign exchange reserves. On the other hand, the Working Party obtained considerable specific information on the composition of reserve transactions. This information filled in the gap between LCFAR and reserve totals, and it also provided an instrument breakdown that permitted adjustments to specific categories of capital flows. The adjustments to LCFAR, therefore, consisted of substituting information provided by reserve holders themselves for the published LCFAR figures in the portfolio investment and “other capital” categories. The adjustment amount was the difference between the calculated and published flows. The “adjustments to LCFAR,” therefore, are the total difference between adjusted foreign exchange reserves, by category (from Table 45), and recorded LCFAR flows (Table 42), except for the reserve position in the Fund component. In effect, the Working Party has corrected flawed LCFAR estimates by substituting for them the directly reported transactions in reserves, and the substitution almost obliterates the reserves/LCFAR imbalance that occurs in the Yearbook. The average adjustment to the reserves/ LCFAR balance was upward (higher inflows), because reported LCFAR have tended to understate purchases of reserve assets.

Table 46.

Reserves, LCFAR, and Adjustments, 1986–89

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

article image

Balance of Payments Statistics Yearbook, 1990.

See Table 3.

See Table 4.

See Tables 18 and 20.

Included in “other capital” adjustments in Tables 4 and 23.

Line 4 of Table 46 shows the reserves/LCFAR discrepancy that results from the combined adjustments previously discussed. (This discrepancy is also found in Table 4 in Chapter 1.) The lower part of the table (lines 5 and 6) then shows the effects in specific capital flow categories of the LCFAR adjustment in line 3. These effects have the opposite sign of the adjustment because LCFAR are subtracted from elements of capital account transactions to produce the balances on private transactions that define discrepancies in this report. The LCFAR adjustment is, thus, a form of reclassification from the reserves/LCFAR discrepancy into specific discrepancies elsewhere in the accounts. Lines 5 and 6 of the table, which are the adjustments to portfolio investment and to “other capital” (and its components), are parts of the total adjustments (shown in Table 4) to these capital flows.

The LCFAR adjustments discussed here represent total amounts by category and do not require knowledge about specific sources of error in LCFAR as published in the Yearbook. However, the Working Party identified two problems connected with the “other capital” category. The first concerns LCFAR not reported by Japan, and the second concerns transactions with domestic commercial banks that some central banks report as foreign exchange reserves. Those parts of the LCFAR adjustments that the Working Party was able to trace to specific sources are shown in Table 47.

Table 47.

Identified Portions of LCFAR Adjustments to “Other Capital,” 1986–89

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

article image

As noted earlier, Japan reports no LCFAR in its balance of payments submissions to the Fund. However, geographic details in statistics published by the United States specifically show changes in its foreign exchange reserve claims on Japan (and on other countries).123 These are presumed to be claims on the Bank of Japan or in short-term official debt. As the first line of Table 47 shows, there is an indicated adjustment in the global capital account (to the resident official sector) for the LCFAR that are missing in the Japanese accounts.

The second—and sometimes large—indicated adjustment to LCFAR shown in the table is to the commercial bank sector of “other capital.” It ranges between –$11.7 billion and $24.9 billion a year. This adjustment is indicated because central banks of several industrial countries disclosed to the Working Party that they include, as part of their official reserve transactions, foreign currency deposit claims on domestic banks.124 Treating these assets as part of foreign exchange reserves may be in accordance with the Manual if the banks have equivalent external claims deemed to be under the effective control of (and readily available to) domestic monetary authorities, but whether this is the situation could not be determined. In any case, LCFAR reported by deposit money banks should be adjusted to include domestic “reserve” deposits as well as genuine cross-border flows. Banks in one country with liabilities to deposit money banks in another cannot be expected to know, or report to the Fund, that these liabilities are treated as reserve assets in the holder country. Even a country that routinely reports LCFAR would be unable to do so in this circumstance.

Taken together, the two identified adjustments in Table 47 to LCFAR of “other capital” are substantial and fluctuate between –$10.6 billion and $19.6 billion; they average $4 billion a year, about one-fourth of the total adjustment to LCFAR made on the basis of the reserves data. Although isolating specific problems with LCFAR reporting is not necessary to effect the adjustments shown in Table 46, these instances do serve to highlight certain particular sources of the reserves/LCFAR discrepancy, which could be reduced if LCFAR were better recorded or aligned with reserve transactions.

Several other factors may account for the remaining gap between foreign exchange and LCFAR transactions reflected in the sizable adjustments in lines 5 and 6 of Table 46. For example, one of the important countries that reported to the Working Party that it treats claims on domestic banks as reserves did not provide specific figures that could be used to make further adjustments to the imbalance. There may be still other countries—unknown to the Working Party—that account similarly for reserves. Faulty reporting or non-reporting of LCFAR by commercial banks in various countries (and offshore centers) is another likely problem area. Reserve acquisitions of securities, especially international bonds, may also be badly measured. Line 5 of Table 46 shows a persistent imbalance—averaging almost $8 billion a year—in this part of the accounts. Finally, as mentioned earlier, some reserve centers do not report LCFAR in any form.

Notwithstanding the large corrections the Working Party was able to make to LCFAR, not all the sources of the excess of reserve transactions over reported LCFAR could be clarified. Eliminating this gap is a necessary step to reconciling capital flows. In future, the gap can only be bridged by improved LCFAR reporting by debtor countries—including offshore financial centers—and by greater transparency on the part of countries about their reserve transactions. Of those two options, the LCFAR route has various pitfalls and does not appear to be a promising one. The Working Party believes that better knowledge, based on direct information, about the asset composition of foreign exchange transactions is a more accurate answer to this particular problem with the global accounts.

The treatment of domestic deposits as reserves entails another danger to balance of payments statements besides misstatements of LCFAR: potential double counting. Individual deposit money banks may not be aware that certain of their foreign investments are regarded as transactions in reserves, and they will file balance of payments reports showing private activity. If domestically held deposits or certain foreign assets of the banks are recorded as reserve flows, compilers must be able to make adjustments to banks’ balance of payments reports in order to exclude such transactions from private capital flows and to reclassify them as reserves. Compilers must have sufficient information to make such an adjustment.125

Even eliminating this double counting does not resolve other problems. Certain other inconsistencies in financial data may arise from the treatment described in the preceding paragraphs. Notably, some disjunctions may occur between international banking statistics (see Chapter 6) and balance of payments data for deposit money banks. In some countries, the BIS/IBS data are submitted by offices not connected with balance of payments compilers, and the figures may not be adjusted. Even if national banking data are adjusted, IBS reporters in other countries will (correctly) report that their banks have liabilities to deposit money banks rather than to monetary authorities. In view of the various problems that can result, the Working Party believes that aspects of reserve treatment and repercussions on other statistics should be reviewed whenever such practices are followed. National authorities are encouraged to publish enough detail about foreign exchange reserves in their balance of payments accounts to make compilation practices transparent.

Conclusions and Recommendations

A large number of countries cooperated with the Working Party in providing extensive detail on their transactions in foreign exchange reserve assets. These details helped the Working Party resolve capital flow imbalances in the portfolio investment and “other capital” parts of the global capital accounts. At the same time, this information revealed several problems with published gold and foreign exchange reserve statistics. Some of the problems arise from differences in accounting for gold swaps among EMS member countries. Also, at least a few important countries include as reserve assets in their balance of payments statements claims on domestic institutions. The Working Party made further adjustments to reserve statistics on the basis of its research into the U.S.S.R. and certain other socialist countries. The following recommendations emerged from this investigation of reserve transactions and LCFAR.