Private nonbank capital transactions are underreported in many countries. International banking statistics provide alternative information for one category of nonbank transaction—deposits in and borrowings from foreign banks. For 1986–89, banking statistics show changes in nonbank assets that average $67 billion a year higher than counterpart changes in balance of payments statements. In liability comparisons they average about $45 billion higher than balance of payments figures. The banking statistics thus suggest that net capital outflows by private nonbanks were an average $21 billion a year above the amounts included in Table 3.


There is a broadly held perception that private non-bank capital flows are seriously understated in many national balance of payments accounts and that this poses a major problem for the quality of capital account statistics. Capital flight, money laundering, and illegal flows are often named as reasons for this problem, but there are broader sources. Nonbank transactors are a diverse group that includes both individuals and businesses of many kinds. Without exchange control systems, compilers must go to considerable effort to measure the transactions of this group. Many countries are not able to support the statistical systems that would even ensure adequate measures of legal and open capital flows of nonbanks, much less of concealed flows. As a result, some balance of payments statements entirely omit nonbank flows or carry rough estimates based on partial indicators. In some cases compilers turn to external data on the subject, most notably international banking statistics.

International banking statistics give an “other-party” view of part of these flows, namely the transactions that are with foreign banks. This chapter presents comparisons between figures on nonbank flows derived from international banking data and those that appear in national balance of payments statements for private nonbanks. The comparisons support the impression that nonbank capital flows are understated in many countries and that the shortfall with respect to cross-border claims on banks is larger than the under-measurement of borrowing from banks. The under statement of net outflows appears to be at least $20 billion a year over the 1986–89 period. The Working Party’s adjustments to nonbank capital flows that emerge from these comparisons are summarized in Table 31 and shown in greater detail later in the chapter. Nonbank “other capital” flows are also affected by adjustments described in other chapters of this report, and the full effect of all adjustments to this balance of payments category is given in Table 23.

Table 31.

Adjustments to Nonbank “Other Capital” Suggested by International Banking Statistics, 1986–89

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

article image

International Banking Statistics

Banking data used in this chapter are compiled by the Bank for International Settlements and the Fund.88 The BIS and IBS statistics differ in important ways, but they are also closely related. The BIS collects and publishes cross-border positions of banks in 18 industrial countries and 7 offshore financial centers. The figures give a broad view of the international banking activity of the 25 countries: their total cross-border assets and liabilities, positions vis-à-vis nonbanks, currency of positions, and positions vis-à-vis central banks.

Most important for the subject of this chapter, the 25 BIS countries each report their banks’ positions vis-à-vis a long list of partner (or “reported”) countries. Geographic detail provided by most of the reporting countries includes both the total position of their banks against each reported country and also, as a component of that total, the position against nonbanks. When the geographic detail is added up across all BIS reporters, the data can be used to read the asset and liability positions for individual partner countries vis-à-vis banks in all the BIS reporters as a group. These BIS totals are, of course, less than reported countries’ positions with all foreign banks, but a very large part of most countries’ activities with foreign banks is with the BIS group. For reported countries, the BIS totals are “derived” measures of positions with foreign banks, that is, measures that have sources outside national statistical systems. The BIS publishes these derived positions for reported countries, both the totals that include interbank positions and the positions vis-à-vis nonbanks.

The Fund’s IBS presentation is fundamentally different in structure from the BIS compilation in that it includes external banking positions reported by every country, rather than just 25 countries, vis-à-vis foreign banks and nonbanks. Most of the IBS reporters, however, give no geographic distribution of their banks’ assets and liabilities that can be used to derive positions for individual partner countries against the group. Although the IBS data constitute a more complete measure than the BIS figures of world and area totals of cross-border banking activity, the derived positions extracted from the IBS compilation are based mainly on the geographic detail submitted by the 25 BIS reporters, plus a few others that make similar detail available to the Fund.89

Both the BIS and the Fund calculate net flows from the banking position data, with adjustments for breaks in series and valuation effects of exchange rate changes. BIS and IBS methods for calculating flows are based on similar principles, but they differ in details of application that can be seen in comparisons of the two sets of data.90

The adjusted net changes that are calculated in international banking statistics are, for the reported countries, the derived flows that can be compared with their balance of payments measurements of transactions with foreign banks. The comparisons in this chapter are limited, however, to derived transactions of nonbanks with the reporting international banks. While international banking statistics yield derived interbank as well as nonbank flows, almost all countries have good records of the foreign transactions of their banks. The international banking statistics have little to add to balance of payments sources for interbank transactions.

Nonbank Transactions with Foreign Banks

Comparisons between banking and balance of payments figures for nonbanks are not close for most countries. Most balance of payments statements do not show enough detail to identify nonbank transactions with foreign banks alone; comparisons usually must be based on some broader measure of nonbank “other capital” transactions with nonresidents. Another important difference is that the banking figures include portfolio positions vis-à-vis nonbanks; securities must be removed to enable comparisons. Because of custodial activities, the banking data also may duplicate or omit nonbanks’ commercial claims and cross-border loan transactions between nonbanks. Finally, definitions of banks vary among countries, resulting in uncertainties about the meaning of both “bank” and “non-bank.”91

International banking data nevertheless allow boundary tests that can be revealing when compared with balance of payments data. A strong indication that national figures are understating nonbank transactions is when the banking statistics show flows of nonbank claims larger than the balance of payments measure of total nonbank “other” claims. For countries that show nonbank asset detail, such as financial versus commercial claims, the comparison can be closer. In such cases, an IBS excess measures more accurately the understatement in the balance of payments version. Finally, some countries show nonbank transactions with foreign banks very specifically in their balance of payments, and for these countries the comparison can be almost exact. The principal allowance is for transactions with banks outside the IBS-reporting group. In any of these cases, the banking statistic is useful only when the derived nonbank asset or liability flow is larger than its balance of payments counterpart. Moreover, the excess must be persistent over a number of years to constitute persuasive evidence of an understatement in the balance of payments.

The numerical difference between banking and balance of payments items usually does not yield an exact adjustment to balance of payments entries because of the differences in coverage. Nevertheless, whenever the banking version is larger, a substitution of the banking figure tends to shift balance of payments accounts toward a more complete estimate of nonbank capital flows. Compilers therefore should look closely at the banking data as a possible source of external information.92

Nonbank Claims on Foreign Banks

Asset comparisons are presented in this section in three tiers. First, a few countries are now directly using BIS or IBS information on nonbank positions in their balance of payments accounts, and relationships for these countries appear in Table 32. Although the two sets of flows are not exactly equal for most of these countries, the differences are small. These differences are excluded from the capital flow adjustments in this chapter. Second, comparisons for another group of 16 countries that have large nonbank positions with foreign banks according to IBS are given in Table 33. These comparisons were made at the most specific balance of payments level of detail available in the country data. Third, for all other countries, the comparisons are between banking figures and the total of nonbank “other capital” asset flows recorded in their balance of payments accounts; comparisons for these countries are shown in Table 34.

Table 32.

Nonbank Claims on Foreign Banks: Comparison of IBS and National Balance of Payments Estimates, Annual Average Flows, 1986–89 Countries That Use International Banking Statistics

(In billions of U.S. dollars; net increase ( + ))

article image

Flows derived from the Fund’s international banking statistics.

Flows published in national balance of payments accounts.

These countries use parts of IBS and in some cases convert stocks to flows by national procedures.

Data consist of IBS flows with an addition of national figures for financial institutions, which are not assumed to be included in IBS nonbank flows.

Table 33.

Nonbank Claims on Foreign Banks: Comparison of IBS and National Balance of Payments Estimates, Annual Average Flows, 1986–89

Major Countries That Do Not Use International Banking Statistics but That Have Balance of Payments Items Comparable to IBS

(In billions of U.S. dollars; net increase ( + ))

article image

Item in national balance of payments accounts that is closest in coverage to nonbank claims on foreign banks. For all countries in the table, the national item is more specific than Balance of Payments Statistics Yearbook total “other capital” asset flows for nonbanks.

Includes special financial institutions.

Table 34.

Nonbank Claims on Foreign Banks: Comparison of IBS and National Balance of Payments Estimates, Annual Average Flows, 1986–89

Countries Not Included in Tables 32 and 331

(In billions of U.S. dollars; net increase ( + ))

article image

Countries for which 1986–89 average IBS net flows are larger than total nonbank “other capital” asset flows.

Balance of Payments Statistics Yearbook: total “other capital” assets of nonbanks.

For both IBS and balance of payments, claims net are of liabilities.

Portfolio Security Debt of Reporting Banks

All of these comparisons are potentially affected by the question of debt securities issued by banks, since these securities are nominally in the totals of bank liabilities as found in the IBS. The BIS reporting group of banks had outstanding liabilities of about $260 billion in international bonds and notes at the end of 1989, and banks reporting to the BIS are instructed to include in the banking statistics their security debt in cross-border liabilities. In balance of payments accounts, however, long-term securities are in the category of portfolio investment rather than “other capital.” Bank debt securities are a structural inconsistency between banking and balance of payments data that tends to make the banking figures larger than “other claims” of nonbanks when comparisons are made.

In practice, however, securities issued by banks do not cause much problem in comparisons with national balance of payments figures. Like other cross-border issuers, banks have little knowledge of who holds their securities or whether holders are banks or nonbanks. Most countries reporting in BIS/IBS systems, therefore, relegate the portfolio debt of their banks to the “geographically unallocated” category, while other reporters exclude bank debt security from their reports entirely.93 Consequently, very little of the security debt of reporting banks has been allocated to any holder countries in IBS nonbank asset flows. As a result, BIS and IBS data have almost the same coverage, in practice, as the balance of payments category of “other capital” flows that excludes banks’ portfolio liabilities. In general, bank-issued securities are not an important problem in IBS comparisons with national figures on nonbank “other asset” flows.

Countries Using International Banking Statistics

Five countries now use international banking statistics as a data source for nonbank asset flows in their balance of payments compilations. Table 32 presents comparisons for these five countries between banking versions of the data and corresponding balance of payments items. The table shows balance of payments items at the most specific level of detail at which non-bank claims are published by each country. In each case the item is more specific than the more aggregated asset entries in the Yearbook.

Of the five countries, Ireland and the United Kingdom use derived nonbank figures directly from IBS, although Ireland makes its own adjustments for valuation changes. The other three countries include some part of the international banking data: Canada uses BIS data with adjustments; Germany has used important elements of BIS data in the recent past; and Mexico uses net increments in IBS stocks as stated in dollars. Table 32 shows that, for the four countries other than the United Kingdom, the comparisons are close for the 1986–89 period. The largest differences—which arise mainly from the definition of banks—are for Germany.94

Capital flows as published for the United Kingdom are radically different from IBS, even though U.K. compilers use IBS flows directly and exactly. The difference arises only because the compilers add to IBS nonbank amounts their own estimates of flows for a group of U.K. financial enterprises located mainly in the Channel Islands and the Isle of Man. These institutions are not part of the United Kingdom’s banking sector, but U.K. compilers assume that foreign countries classify them as U.K. banks in their BIS and IBS reports. On the basis of that assumption, U.K. compilers add their own estimates for such businesses to IBS nonbank flows for their own balance of payments statement.

Major Countries Not Using International Banking Statistics

Table 33 gives nonbank asset comparisons for 16 countries that, according to the IBS data, hold large claims on foreign banks. These countries compile non-bank asset flows without using international banking statistics but, for each of them, the balance of payments item in the table is conceptually a fairly close measure of claims on foreign banks.

In total, the IBS measure of nonbank asset flows for these countries averages $40.9 billion more a year over the 1986–89 period than the comparable total of balance of payments flows did. Almost half of the excess originates in the U.S. figures, for which IBS-derived amounts are outstandingly different from balance of payments counterparts. The $21 billion excess for the other 15 countries is widely distributed among them. Only Norway and Venezuela have nonbank outflows in the national data that are larger than IBS versions.95

For three of the countries in Table 33—Austria, Italy, and Spain—the balance of payments statement explicitly excludes entries for private nonbank claims on foreign banks. Nonbanks in these countries generally were not authorized to hold foreign bank deposits during the years under review, and outflows were correspondingly omitted from the accounts. (Circumstances have changed since 1989 in each of these three countries and, in future, such entries will be made.) The Hong Kong balance of payments also omits nonbank outflows to foreign banks. However, the Hong Kong statement is estimated by the Fund rather than by local compilers, and it is used at the Fund in regional and world totals in Part 2 of the Yearbook. For Hong Kong, therefore, the Fund might look more closely at international banking statistics to improve these figures.

Other Countries

The 21 countries in Tables 32 and 33 held about 70 percent of all nonbank claims on foreign banks in IBS totals in the late 1980s. In the comparisons in those tables, national sources were searched to find the most appropriate balance of payments item to match against IBS-derived figures. For the rest of the world—the other 126 countries in Part 2 of the Yearbook—a somewhat coarser comparison with IBS appears in Table 34. This comparison uses national figures for total asset flows of private nonbanks, together with certain non-reserve parts of official asset flows. This balance of payments asset total is typically much broader than claims on banks alone. It includes a variety of receivables, as well as cross-border lending by insurance companies and finance companies. In the 1990 Yearbook, “other capital” also includes commercial paper and other short-term marketable claims, both of which were active forms of credit flow during the 1980s. The balance of payments total is much broader in asset coverage and, if well measured, can be expected to be much larger than IBS counterparts.

For any country with IBS-derived flows that are persistently larger than balance of payments totals over several years, this comparison gives a strong indication that balance of payments versions of capital outflows are too small.96 The excess of IBS over balance of payments for these countries normally is a minimum measure of the understatement in balance of payments estimates, and totals for them should be viewed as only a first, or base, estimate of flows that may be missing from the world figures. More accurate figures can come only from detailed analysis of national source data.

Table 34 presents the results of this broad comparison for all countries not in Tables 32 and 33. The table includes those countries for which IBS-derived flows were larger than balance of payments net asset outflows of nonbanks during 1986–89. Many other countries had IBS asset flows that were smaller than their balance of payments flows, and such countries are not in the table. The totals in the table come to a $25.8 billion annual average of excess IBS flows over balance of payments figures.

This $25.8 billion, in combination with the $40.9 billion excess that is shown in Table 33, results in a tentative annual average of $67 billion that the Working Party has used to adjust world totals of nonbank asset outflows. This amount is tentative, even speculative, but it is also about 70 percent of the total cross-border flows from nonbanks to banks recorded in IBS. It is by far the largest outflow adjustment to recorded capital flows and greatly reduces the global imbalance. Even if an overstatement, it implies that a very large proportion of asset transactions of private nonbanks with foreign banks is not captured in national balance of payments compilations.

As with the major asset holders, there are reservations about these findings, and the relations should be looked at carefully by compilers to test their validity.97 For many of the countries in Table 34, balance of payments accounts show no entry at all for foreign assets of domestic nonbanks. For countries showing no entries and for others with very small recorded flows, IBS flows undoubtedly add information for compilers. Three offshore centers—the Cayman Islands, Bermuda, and Gibraltar—are not included in the Yearbook figures at all, and the IBS-derived nonbank asset outflows from these three countries, averaging $4.6 billion, are part of the more complete picture for offshore centers discussed in Chapter 9.98

Nonbank Liabilities to Foreign Banks

Nonbank liabilities to foreign banks are derived from the IBS data base by a process parallel to that for assets: amounts reported for bank claims on each partner country’s nonbanks are cumulated across the IBS-reporting countries. The results again do not encompass debt to banks that are not in the BIS/IBS reporting group, but the reporters are a large part of the world banking system. As on the asset side, these derived liability figures can be compared with their balance of payments counterparts. The boundary conditions described earlier in this chapter apply to liability comparisons in the same form as with assets: most of the balance of payments measures of nonbank liabilities cover more than borrowing from banks. In normal circumstances such flows can be expected to be larger than those in IBS. As with the asset comparisons, the following tables give some close comparisons for major countries as well as a broader sweep for others.

Major Debtor Countries

The base for the comparison of nonbank liabilities is 24 countries that are major debtors according to IBS. However, few of these countries had balance of payments entries that could be reasonably interpreted as borrowing from banks. Table 35 gives the comparisons that are possible between IBS-derived liabilities and balance of payments items that are close in coverage to nonbank borrowing from foreign banks. The table lists just eight countries, considerably fewer than in Table 33 on assets. The comparisons in liabilities are also more ambiguous than those in assets.

Table 35.

Nonbank Liabilities to Foreign Banks: Comparisons of IBS and National Balance of Payments Estimates, Annual Average Flows, 1986–89

Major Countries with Comparable Balance of Payments Item

(In billions of U.S. dollars; net increase (+))

article image

National balance of payments accounts item that is closest in coverage to nonbank “other capital” liabilities to foreign banks. For all countries in the table the item is more specific than the Balance of Payments Statistics Yearbook measure of private nonbank total liability flows.

Not included in totals in Tables 36 and 37 and not used as adjustments to capital account flows.

Belgium-Luxembourg, Denmark, Germany, and the United States.

Balance of payments entries for the United Kingdom are IBS flows plus national estimates for financial institutions not included in IBS-derived numbers for nonbanks.

Table 35 also differs from Table 33 in that the second column shows the portfolio component of IBS nonbank liabilities—the amount, that is, of each country’s non-bank long-term debt securities acquired by foreign banks. While derived nonbank assets in the IBS include small amounts of bank debt securities that cannot be separated out, IBS geographic detail includes much more of the portfolio component of nonbank liabilities. The BIS has recently developed a survey of the portfolio assets of reporting banks that gives some basis for separating holdings of securities from their other assets. The BIS survey data are the basis for the estimated bank acquisitions of securities in the second column of Table 35. The estimates are tenuous and uncertain, and the comparisons in the far right-hand column are coarse as a result.

As this approach to data comparisons makes clear, isolating the securities component of bank asset positions greatly improves the conceptual fit between banking and balance of payments figures and enables valuable comparisons. The Working Party therefore believes that the BIS work on identifying bank holdings of securities is important to the use of IBS as a check against balance of payments and that it should be pursued actively.

Of the countries in Table 35, only the United Kingdom uses IBS directly in its balance of payments accounts. As with nonbank assets, the published U.K. amounts for nonbank liabilities include borrowing by other institutions, located mainly in the Channel Islands and the Isle of Man, that give a result different from IBS flows alone. As in Table 33, the U.K. figures are not included in the totals.99

Except for the U.S. comparisons, Table 35 shows small differences between IBS and balance of payments for the 1986–89 period. For the group other than the United States, the two sources give almost identical totals for nonbank liabilities. This close result contrasts with the asset comparisons and supports the widespread belief that countries can track liabilities considerably better than assets, although the reasons may differ from one country to another. The comparison for Canada is very close. For three countries in the table—Belgium-Luxembourg, Denmark, and Germany—the banking statistics show larger credit flows to nonbanks than do balance of payments data. For Italy and Japan, national figures show almost the same flow as IBS before the portfolio deduction; after portfolio components are removed, IBS are smaller than national figures.

For the United States, the $25 billion excess in average IBS nonbank liability flows is even larger than the differences in asset flows discussed in connection with Table 33. Among major borrowing countries, only the United States shows a persistently large discrepancy between IBS and balance of payments on nonbank liabilities.

Capital flow adjustments to nonbank liabilities resulting from the comparisons for major countries in Table 35 consist of the differences for the United States, together with those for Belgium-Luxembourg, Denmark, and Germany. They average about $31 billion a year of additional inflows.

Other Countries

Table 36 presents for nonbank liabilities the same form of comparison that Table 34 gives for assets. It includes any OFC or other country for which IBS show larger 1986–89 average inflows than the total of non-bank “other” liability flows in the balance of payments statement. As in Table 34, the balance of payments figures include some official flows—in this case, “other capital” liabilities except loans. Table 36 includes no portfolio adjustment because for these countries the amounts are uncertain and probably small.

Table 36.

Nonbank Liabilities to Foreign Banks: Comparisons of IBS and National Balance of Payments Estimates, Annual Average Flows, 1986–89

Countries Not in Table 351

(In billions of U.S. dollars; net increase ( + ))

article image

Countries for which average 1986–89 IBS net flows are larger than balance of payments counterparts in total “other capital” liabilities.

1990 Balance of Payments Statistics Yearbook: total “other capital” liabilities of nonbanks.

Total for countries in which IBS flows, net of portfolio debt to banks, are greater than balance of payments counterparts.

Nonbanks in offshore financial centers display much larger debt flows in the banking statistics than in balance of payments accounts. However, the excess of asset flows ($8.5 billion, including Hong Kong from Table 33) is more than the $5.5 billion in liability flows, and the net balance in both comparisons is a $3 billion outflow by OFC nonbanks. The effect of these figures on world capital flows is integrated into a more complete statement for OFCs in Chapter 9.

The $9 billion total difference for countries other than OFCs in Table 36 probably exaggerates the underreporting of loan flows in balance of payments statistics because of statistical problems in IBS with debt reschedulings and write-offs. Table 36 excludes Argentina, Brazil, and Mexico because of their major reschedulings and debt adjustments, even though differences between IBS and balance of payments are very large for the three countries.100 These countries are treated separately in Chapter 7 of this report.

Reschedulings and write-offs evidently also affect the comparisons for other countries. Balance of payments figures in Table 36 cumulate to a $5 billion net reduction of debt that does not appear in IBS totals. This result suggests that debt reschedulings and debt write-offs are reflected more fully in balance of payments data than in banking statistics. If so, banking statistics may overstate credit inflows to these countries.

Nevertheless, the total difference for Table 36 countries—$14.6 billion—combined with the $30.8 billion from Table 35 (mainly for the United States) gives a total excess of IBS liability flows over balance of payments of about $45 billion a year. This $45 billion is the average amount for 1986–89 that the banking data imply could be substituted into balance of payments statements for nonbank inflows and that the Working Party has used (in annual form) to adjust the recorded figures. It is about one-half of the total of all credit flows allocated to nonbanks in IBS compilations.

Summary Comparisons

Table 37 pulls together the results of the asset and liability comparisons in this chapter and shows, in the final column, adjustments the Working Party made to recorded capital flows. In this table, increases in asset holdings are shown with negative signs, following balance of payments conventions; the presentation differs in this respect from Tables 32, 33, and 34, where asset increases appear as positive numbers. Totals in Table 37 are only for countries that are included in preceding tables, for which an IBS measure of nonbank transactions is larger than the equivalent balance of payments flows. The United States appears separately because of the exceptional size of the adjustments that emerge from the U.S. figures.

Table 37.

Summary Comparisons for Nonbank “Other Capital” Flows: IBS and National Balance of Payments Estimates, Annual Average Flows, 1986–89

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

article image
Sources: Major countries: Table 33 (asset flows), Table 35 (liability flows); other countries: Table 34 (asset flows), Table 36 (liability flows).

IBS liabilities for major countries exclude portfolio claims held by reporting banks. Table includes countries that have larger nonbank transactions in IBS than in the nearest equivalent balance of payments figures. The United States appears separately because it has large discrepancies and a positive average adjustment (net inflow) that partly offsets the adjustment for other countries.

Actual adjustments to capital flows were made on an annual basis.

The full effect of comparisons in assets and liabilities, including the U.S. figures, is a net nonbank capital outflow that averages $21 billion per year more in the IBS data than in the balance of payments accounts for the countries affected. This $21 billion is about 40 percent of the average world capital flow imbalance for 1986–89, after corrections for discrepancies in direct investment reinvested earnings. For most countries, moreover, the balance of payments item in the comparison has broader coverage, up to and including total nonbank “other” assets and liabilities, than the IBS item, so that the adjustment amount is a kind of lower bound to potentially missing net capital outflows. For some countries the differences with IBS are large relative to entries in their balance of payments accounts, and many are large enough to alter analytic views of recent economic developments.

For the United States, the banking figures show much larger nonbank flows than balance of payments counterparts in both assets and liabilities. Indeed, the IBS net excess of transactions results in an average net inflow adjustment of $5 billion a year for 1986–89. This net adjustment is highly volatile, however, since it is driven by changes in the net balance of large IBS asset and liability figures; the average U.S. adjustment can be very different for other four-year periods.101

For countries other than the United States, IBS imply an average net upward adjustment in capital outflows of $26.5 billion. Asset flows are $47 billion larger than equivalent, or broader, balance of payments flows, while for liabilities the IBS excess is $20 billion. For both assets and liabilities, these differences are larger than the recorded balance of payments flows themselves, and they are large fractions of the IBS flows in the comparisons.

The third column of Table 37 presents the adjustments based on banking statistics that were applied to nonbank “other capital” flows in this report. They were made to the balance of payments accounts of each affected country, and they are included in the capital account adjustments in Table 4 and the adjustments to “other capital” flows that appear in Table 23. Although probably lower bounds to the necessary corrections, such differences should not be used statistically, except as a temporary measure, as this chapter has warned at several points. The comparisons point to possible improvements in balance of payments measurement, but compilers should use them carefully.

Conclusions and Recommendations

International banking statistics indicate that cross-border capital transactions of private nonbanks often are considerably larger than amounts included in balance of payments statements. These larger flows appear for both assets and liabilities. The banking data constitute alternative measurements of such flows that probably can improve the quality of balance of payments statistics for many countries. If the flows derived from banking data are substituted for the national figures of those countries for which banking figures are larger, the effect is to raise world totals of nonbank asset outflows more than liability inflows. The effect of such a substitution is to reduce the positive global capital flow imbalance.102

The Working Party does not argue that international banking figures that are larger than national data are always a better measure of nonbank flows. Indeed, IBS are also subject to compilation problems, such as differences in definitions of banking vis-à-vis balance of payments data and problems related to the inclusion of securities. National compilers must use their knowledge of domestic statistical structure and relationships with foreign banks to judge the quality of banking figures. For many countries, however, the banking statistics point clearly to underreported or omitted flows in national balance of payments statements.

The following summarize the recommendations that have been discussed or developed in this chapter: