International Banking Statistics in Capital Flow Measurement
- International Monetary Fund
- Published Date:
- June 1992
- Stephen P. Taylor
|Table 2 (concluded)|
|Part H. Percent of totals, excluding Luxembourg|
|Part I. Japanese bank holdings of international securities|
|Owed by nonbanks||21,560||29,61O||34,350||39,000||…|
|Net changes in nonbank|
|Distributed as in Part H|
Japanese investors, as reported by the Japanese Security Dealers Association. These transactions include sizable amounts of securities bought in Luxembourg that are not relevant to this exercise because Luxembourg plainly is a trading location rather than the residence of the security issuers. Although other countries in the listing also may serve partly as trading locations, the lack of further information means that they are treated as actual security issuers. On that basis, Part H distributes net Japanese purchases geographically, according to each country’s percentage share in the group total, omitting Luxembourg.
Finally, the trading proportions for all Japanese investors are used in Part I of the table to distribute Japanese banks’ net buying of nonbank securities among the group of issuing countries. A major assumption in this step is clearly that Japanese banks trade in foreign securities in the same geographic markets as other Japanese investors. Another important assumption is that nonbank securities are only 30 percent of total portfolio purchases by Japanese banks, rather than the 75 percent used for other BIS banks in Part E.3 The figures in Part I are the Japanese-bank amounts included in Part F, and it is the three-year averages in Part F that are used in Table 35 of the Report on Capital Flows,
Annual Comparisons Between IBS and Balance of Payments Statistics for Nonbank “Other” Capital Flows of Major Countries
The three tables in this section present the annual data used in Chapter 6 of the Report on Capital Flows to make the 1986-89 average comparisons between BOP and IBS flows of “major countries.” The tables here elaborate on Tables 32, 33, and 35 of the Report on Capital Flows. They cover countries with large non-bank positions in IBS and for which balance of payments entries are closely comparable to the nonbank flows derived from IBS. The balance of payment items in Tables 3, Table 4 and Table 5 are those described earlier in Table 1.
|Residence of nonbanks||1986||1987||1988||1989||1986-89|
|Totals, countries above|
|United Kingdom, with additions to IBS1|
U.K. national amount consists of IBS flows plus national figures for financial institutions assumed not to be included in IBS nonbank flows.
|Residence of nonbanks||1986||1987||1988||1989||1986-89 average|
|Totals for above countries|
|Totals including United Slates|
Includes special financial institutions.
|Residence of nonbank||1986||1987||1988||1989||1986-89 average|
|Of which portfolio||−0.2||−||0.2||−0.8||−0.2|
|Of which portfolio||—||0.5||−0.8||0.2||—|
|Of which portfolio||−0.2||0.3||−0.8||—||−0.2|
|Of which portfolio||2.6||4.0||4.1||−0.2||2.6|
|Of which portfolio||1.1||0.6||2.0||0.8||1.1|
|Of which portfolio||2.9||3.6||1.2||4,0||2.9|
|Totals for above countries;|
|Of which portfolio||6.3||9.1||5.9||4.0||6.3|
|Of which portfolio||3.5||3.7||1.8||5.1||3.5|
|Totals including United States|
|Of which portfolio||9.8||12.8||7.7||9.1||9.8|
|Memo: United Kingdom1|
|Of which portfolio||1.3||1,8||1.9||0.3||1.3|
As in Table 3, the U.K. balance of payments flows are those calculated in IBS plus amounts for other financial institutions assumed not to be included in IBS nonbank amounts.
United States: A Closer Look
In the derived nonbank data in Tables 4 and Table 5, the United States shows the largest discrepancies with national balance of payments figures by far, almost as large as those for all the other countries combined. If U.S. compilers used the IBS figures in national balance of payments accounts, the change would reduce the U.S. errors and omissions item but would also increase the global capital account asymmetries and global errors and omissions.
The source of the U.S. problem is well known: private nonbank cross-border flows in “other” claims are measured through a U.S. Treasury reporting form that is returned by only a few hundred large corporations, and many transactions by individuals and businesses escape the reporting system entirely. Compilers want to broaden the nonbank reporting system, but that process can take several years. In the meantime, it is clear that nonbank transactions, except those handled by banks as custodians, are almost entirely missing from the U.S. balance of payments statistics.4
Private Nonbank Capital Flows
Table 6 presents the relation between IBS and U.S. balance of payments figures for stocks of nonbank capital at year-end 1989. The liability balances found in IBS are roughly eight times those recorded in the U.S. investment position figures; the IBS figures likewise show a large asset balance that is not present in the U.S. national statistics.
|Private nonbank financial assets||18.11||258.8|
|Nonbank liabilities to foreign banks|
|Private financial liabilities||17.4||…|
|U.S. government securities held by|
|Net nonbank assets||−8.6||43.0|
The outside sources added by U.S. compilers in 1992 increase this total to about Í107 billion. See Survey of Current Business. June 1992,
The positive net asset balance of U.S. nonbanks recorded in the IBS statistics implies that net investment income of several billion dollars per year is missing from the U.S. BOP accounts; such an income adjustment is included in the IMF’s 1987 Report on the World Current Account Discrepancy.5 The comparison of U.S. national figures with IBS is more complicated, however, because the U.S. current account was in deep deficit during the 1980s, and the United States was a net borrower from the rest of the world. The IBS-derived flow estimates for U.S. nonbanks reflect those capital account developments more completely than the balance of payments; they show inflows of credit from foreign banks that are well above outflows. For 1986-89, the flow comparison appears in Table 7.6
|U.S. balance of payments data1|
|Private nonbank sectors|
|U.S. government securities purchased by foreign banks||0.2||−0.8||−1.5||1.8||−0.1|
|Net nonbank flow2||−5.7||4.5||−0.4||10.9||2.3|
|IBS-derived net flows for nonbanks|
|Claims on foreign banks||−26.2||−22.7||−18.9||−6.4||−18.6|
|Liabilities to foreign banks3||21.2||19.6||34.8||29.1||26.2|
|Net nonbank flow||−5.0||−3.1||15.9||22.7||7.6|
|Excess of IBS net flow over BOP net flow||0.7||−7.6||16.3||11.8||5.3|
|Effect on U.S. errors and omissions|
|Published errors and omissions1||15.8||−6.7||−9.3||18.4||4.6|
|Less: IBS excess of capital flows||0.7||−7.6||16.3||11.8||5.3|
|Less: Unrecorded investment income||6.9||7.0||6.3||4.6||6.2|
|Equals: adjusted to IBS basis||8.2||−6.1||−31.8||2.0||−6.9|
Survey of Current Business. June ¡991, pp. 62-64.
Net sum of private nonbank financial flows and foreign-bank net purchases of U.S. government securities. Maturity of the securities is not available but they are assumed to be short term in this table.
Excluding portfolio liabilities to international banks.
Table 7 indicates the effect that IBS-based estimates might have on U.S. errors and omissions if they were used in the U.S. balance of payments accounts. Substituting IBS figures generally increases net capital inflows (excess of IBS net flows over BOP net flows in Table 7). The IBS figures would also affect the current account, however, because the larger IBS net asset positions for U.S. nonbanks imply a larger net investment income for the United States than is presently captured in the balance of payments. A rough estimate of the income effect is reflected under “unrecorded investment income.” When the two effects—capital flow and income—are combined, U.S. errors and omissions take the form of the last line of the table. The 1986 and 1989 imbalances are much reduced, and the 1988 residual is very deep.7 Going beyond the table, for 1990, the U.S. discrepancy suggests large unrecorded inflows, and international banking statistics could reduce them by about one-half.
The validity of these IBS figures can be tested by comparing derived interbank components of IBS figures with U.S. counterparts. The Report on Capital Flows did not discuss derived interbank positions and flows, but they are part of both the BIS and the IBS systems. Most national data on cross-border activities of banks are reasonably complete, and differences between derived and direct measures of interbank activity usually arise from technical differences such as the definition of banks. The U.S. nonbank differences are so large, however, that a comparison of interbank differences for the United States might suggest whether the nonbank findings are misleading—for example, whether BIS reporters are attributing to the United States large amounts of other countries’ positions or unallocated residuals.
The IMF’s published IBS data do not include derived interbank stocks and flows. The BIS publication, International Banking and Financial Market Developments, shows them implicitly, however. The positions and flows of the 24 BIS reporting countries (other than the United States) Vis-à-vis banks in the United States can be gleaned from these published numbers. On the U.S. side, the Treasury Bulletin publishes positions reported by U.S. banks Vis-à-vis other countries, including BIS reporting countries. For recent years, it is possible to add up these U.S.-reported positions Vis-à-vis BIS reporters and compare them directly with the BIS- derived amounts for U.S. banks Vis-à-vis other BIS countries.
Such a comparison appears in Table 8 in the form of year-end stocks of assets and liabilities for the 1986-90 period. The table gives derived totals for U.S. interbank assets and liabilities Vis-à-vis other BIS countries as percentages of the U.S. Treasury’s measure of U.S. bank positions against banks in those same BIS countries. This is perhaps as specific a comparison as is possible between derived and national versions of international banking numbers. For 1986 and 1987, the results are very close for liabilities. (A 1986 Treasury version of U.S. interbank assets was not available.) In 1988, BIS coverage dropped severely, perhaps because of data processing problems of BIS reporters in offshore financial centers. The reduced coverage affected U.S. liabilities more than it did assets, but both sides show some instability in the relationship from 1987 to 1989. For assets, however, the two sources seem much closer over the entire 1987-90 period. The Treasury figures include bank custody assets, following the common assumption that BIS reporters include all custody claims on them as interbank positions. The close fit seems to support that assumption,8
|BIS/U.S. (percent ratio)||99.0||99.5||90.6||96.3||95.6|
|BIS/U.S. (percent ratio)||…||100.4||98.5||102.3||100.6|
|Net interbank liability|
This comparison of the interbank figures is only an indirect test of differences in the data for U.S. nonbank capital flows, which according to the IBS are on the scale of $25 billion a year. The interbank numbers in Table 8 give no indication, however, that BIS reporters are exaggerating U.S. nonbank positions by allocating irrelevant assets or liabilities to the United States. There is also no evidence that BIS reporters are attributing large positions to U.S. nonbanks that would be included in the banking totals of the U.S. accounts. Thus, consistency in the interbank figures should lend credibility to the IBS-derived figures for U.S. nonbanks.
The question remains why IBS stocks of U.S. nonbank assets and liabilities are so much larger than the U.S. counterpart data. Although not a new question, it has not yet been clearly answered. The asset difference may result, in part, from unrecorded accrued income earned on assets that were established abroad before 1980. In the IBS numbers, the stock of U.S. nonbank assets grew at an annual rate of 9 percent between 1981 and 1989, which is close to the rate that could be expected from reinvestment of income. The stock of U.S. nonbank liabilities grew at a 20 percent rate over this period, however, rising from 37 percent of assets in 1981 to 87 percent in 1989. Such borrowings are almost entirely missing from the U.S. balance of payments, and they are much larger than interest income can explain.
Data Comparisons for Other Countries
Table 9 and Table 10 summarize the results of annual comparisons of balance of payments and banking statistics for all countries not among the 24 surveyed in the earlier sections. These tables compare total asset and liability flows in the balance of payments data with the IBS-derived flows. The two tables include all countries for which the average IBS flow over the 1986-89 period exceeded the corresponding average balance of payments flow. The banking figures are not necessarily larger for each of the four years separately.
|Offshore financial centers|
|Other developing countries|
Countries other than those in Tables 3 and 4 for which the 1986-89 average IBS flow for nonbanks is larger algebraically than the nonbank total asset flow in the 1990 Balance of Payments Statistics Yearbook.
For both IBS and balance of payments data, asset flows less liability flows.
|United Arab Emirates|
|China. People’s Republic|
Annual Data for Offshore Financial Activity
A final group of tables presents annual data on offshore financial activity to support the summary tables that appear in Chapter 9 of the Report on Capital Flows. Table 11-Table 15 cover six offshore centers. Table 15 concludes with annual capital account adjustments and the implications of these adjustments for current account balances and for errors and omissions. The table thus corresponds to Table 52 in the Report on Capital Flows.
|Deposit money banks, total net claims|
|Total, six OFCs||−2,019||−592||−1,074||−1,265||−1,238|
|Nonbank “other” net claims|
|The Bahamas||2||− 17||5||20||3|
|Total, six OFCs||356||−1.531||−1,249||−760||−7%|
|Total “other” net claims|
|Total, six OFCs||−1,663||−2,123||−2,323||−2,025||−2,034|
|Total, six OFCs||−27,616||9,181||−15,324||8,846||−6,228|
|Bank flows vis-à-vis foreign nonbanks|
|Total, six OFCs||19,614||−15,278||6,354||−18,562||− 1,968|
|Total bank flows|
|Total, six OFCs||−8,002||−6,097||−8,970||−9,716||−8,196|
|Nonbank flows vis-à-vis foreign banks|
|Total, six OFCs||−8,621||−3,801||554||−644||−3,128|
|Total bank and nonbank flows|
|Cayman Islands||− 12,582||2,231||−10,978||−8,695||−7,506|
|Total, six OFCs||−16,623||−9,898||−8,416||−10,360||−11,324|
|Deposit money banks, total net flow|
|Total, six OFCs||−5,983||−5,505||−7,8%||−8,451||−6,959|
|Nonbank net “other” claims|
|Total, six OFCs||−8,977||−2,270||1,803||116||−2,332|
|Total bank and nonbank flows|
|Total, six OFCs||− 14,960||−7,775||−6,093||−8,335||—9,291|
|Reinvested earnings reported by partner countries1|
|Total, six OFCs||1.204||1,344||¡57||1,509||1,054|
|Other direct investment1|
|Total, six OFCs||4,165||5,670||107||2,130||3,018|
|Total, six OFCs||−5,662||−5,023||−2,967||19||−3,408|
|Netherlands Antilles||—||—||359||− 109||63|
|Total three OFCs||−64||−231||287||777||192|
|Total, six OFCs||−15,317||−6,015||−8,509||−3,900||−8,435|
|Reinvested earnings not reported by partner countries3||1.796||2.656||4.843||6.491||3.947|
|Errors and omissions4||13,521||3,359||3,666||−2,591||4,489|
From estimates underlying Table 3 of Part 2 of the 1990 Balance of Payments Statistics Yearbook. Earnings for OFC banks are based on cross-border asset and liability positions reported by OFC authorities and the market interest rates appropriate to each type of claim. Earnings for OFC nonbanks are based only on their claims on foreign banks as derived from IBS and used in Chapter 6 of the Report on Capital Flows. The estimate excludes earnings on nonbank positions Vis-à-vis foreign nonbanks.
Total income less the amounts from reinvested earnings in Table 14.
The net sum of combined adjustments and unreported retained earnings, with the sign reversed.
Table 16 and Table 17 show the annual adjustments that were developed for Hong Kong and for special financial institutions (SFIs) in the Netherlands. The tables include detail on types of capital flows that have been adjusted.
|Type of capital flow||1986||1987||1988||1989||1986-89|
|Other direct investment||46||−700||−464||−690||−452|
|Deposit money banks||3,929||−90,516||−43,119||−53,338||−45,761|
|Total asset adjustments||7,101||−83,474||−48,750||−59,330||−46,113|
|Other direct investment||539||−387||531||442||281|
|Tola1 liability adjustments||−590||83,878||42,906||54,602||45,199|
|Other direct investment||585||−1,087||67||−248||− 171|
|Total adjustments, net||6,511||404||−5,844||—4,728||−914|
Liabilities constituting foreign authorities’ reserves.
|Type of capital flow||1986||1987||1988||1989||1986-89|
|Other direct investment||−10,416||−24,379||−30,309||−33,248||−24,588|
|Other capital, private nonbank||−2,161||−1,524||−937||−3,930||−2,138|
|Total adjustments to asset flows||12,330||−26,764||−32,129||−38,777||−27,500|
|Other direct investment||6,514||16.386||19,036||18,397||15,083|
|Other capital, private nonbank||283||1.448||−907||5.717||1,635|
|Total adjustments to liability flows||11,940||25,737||31,436||36,133||26,312|
|Other direct investment||−3,902||−7,993||−11,273||−14,851||−9,505|
|Other capital, private nonbank||−1,878||−76||−1,844||1,787||−503|
|Total adjustments, net||−390||−1,027||−693||−2,644||−1,189|
In the Balance of Payments Statistics Yearbook statement for the Netherlands, all cross-border transactions of SF is are combined in a net form as part of a single item in the capital account: “other capital” flows of private nonbanks. The Report on Capital Flows includes, in several chapters, adjustments to redistribute SF1 capital account transactions into the balance of payments categories from which they were removed for the Yearbook. This table presents the SFI adjustments to the Netherlands statement that are included in adjustment totals for industrial countries.
The redistribution of capital account flows in the table do not add to zero. The net amount in the bottom line is the current account balance (with opposite sign) of SFls Vis-à-vis the rest of the world. The SFI current account balance is also included in the Yearbook single capital account entry, but in the report it is transferred to the current account. It is included in the current account adjustments reflected in Table 6 of the Report on Capital Flows.