International Capital Flows
Of all the data on U.S. international economic transactions, capital flow statistics are the most subject to errors and gaps. Although the United States collects as much detailed data on its capital flows as any country in the world, the explosion in direct and portfolio investments across U.S. national boundaries in the 1980s outpaced improvements in the statistical system that monitors them. Specifically, the existing system is designed to collect most of the information on international capital flows from a group of large financial intermediaries and corporations. Yet the integration of world financial markets and innovations in electronic and communications technology have greatly increased the number of financial transactors and facilitated new modes of transactions, bypassing traditional financial intermediaries and channels. This change has made it increasingly difficult and costly to capture comprehensive and accurate capital flows under the current reporting systems. In addition, growing numbers of new financial instruments have outstripped the coverage of existing data, further rendering them incomplete and inaccurate.
Efforts to collect accurate and comprehensive information on the rapidly growing capital flows should yield high payoffs in improving the usefulness of existing data on U.S. international transactions because capital flows are of substantial magnitude and are inadequately measured. In fact, over the past decade, values of capital flows surpassed those of trade flows. Informa-
tion on capital flows is especially needed to examine the extent of internationalization of the U.S. economy; the changing operations of U.S. financial markets; the impact of foreign direct investment on the domestic economy; the incomes on U.S. international investments and net servicing burden on U.S. external indebtedness; and the relationship of flows of foreign capital and U.S. interest and foreign exchange rates. In addition, since net capital flows should match the balance of international transactions in goods and services and transfers (the current account), improved information on capital flows would help illuminate the accuracy of data on the U.S. current account. Improved capital flow data would also contribute to more accurate estimates by the Bureau of Economic Analysis (BEA) of the U.S. international investment position (which measures the value of accumulated stocks of U.S. assets abroad and of foreign assets in the United States).
More important, improvements in data on international capital transactions can improve the preparation of the flow-of-funds accounts and the balance sheets of the U.S. economy (both published by the Federal Reserve). The flow-of-funds accounts detail the sources and uses of savings in the U.S. economy by sector and by type of transaction. Sectors include households; nonfinancial businesses (corporate and noncorporate); state and local governments; the federal government; foreigners; and financial institutions, including banks, savings and loan associations, life insurance companies, and pension funds. The account for each sector shows changes in financial assets and liability items, such as deposits, mortgages, loans, open-market papers, federal government securities, and bonds. Along with the national income and product accounts published by BEA, the flow-of-funds accounts provide an integrated set of financial accounts that are used to analyze economic developments in the United States. U.S. balance-of-payments data are major inputs to the flow-of-funds accounts; that information is modified and supplemented with other data on international transactions from bank regulatory reports and the Treasury International Capital (TIC) reporting system to make it compatible with the sector and instrument classifications used in the flow-of-funds accounts. The foreign data are also used for estimating aggregate levels of outstanding debt by sector and measures of national net worth. At the Federal Reserve, the flow-of-funds accounts have been useful in assessing the impact of monetary policy and the general financial conditions of the various sectors. The domestic nonfinancial debt aggregate measure that is monitored by the Federal Reserve' s Federal Open Market Committee is
derived from the flow-of-funds accounts. The accounts are also widely used for a variety of purposes in the academic and business communities and elsewhere in the government. Flow-of-funds publications include the quarterly Flow of Funds Accounts, the semiannual Balance Sheets for the U.S. Economy, and the annual Financial Assets and Liabilities.1
DATA ON CAPITAL FLOWS: KEY FEATURES
Capital flows refer to transactions in financial assets between U.S. residents and residents of foreign countries. Financial assets include loans, bank deposits, drafts, acceptances, notes, government and private debt and equity securities, and intracompany accounts for the financing of direct investments.
There are two major types of capital flow transactions: official and private. U.S. official capital flows include changes in the reserves of U.S. monetary authorities in monetary gold, foreign exchange, special drawing rights at the International Monetary Fund, and loans and credits to foreigners by U.S. government agencies. Official capital flows are estimated quarterly by BEA on the basis of data provided to it by Treasury, the Federal Reserve System, and the International Monetary Fund (IMF). Changes in foreign official assets in the United States can take place through transactions in U.S. Treasury securities, other U.S. government obligations, and bank deposits, as well as U.S. corporate bonds and stocks. BEA estimates these transactions quarterly on the basis of data provided by Treasury and a number of other U.S. government agencies that possess relevant information (Bureau of Economic Analysis, 1990c).
Private capital flows encompass direct investment and portfolio investment undertaken by both U.S. residents abroad and foreigners in the United States. As stated in the International Investment and Trade in Services Survey Act, direct investment is defined as “the ownership or control, directly or indirectly, by one person of 10 per centum or more of the voting securities of an incorporated business enterprise or an equivalent interest in an unincorporated business enterprise,” and portfolio investment is defined to include “any international investment which is not direct investment.” These definitions conform to the general guidelines provided in the IMF's Balance-of-Payments Manual. Direct
We appreciate the information on the flow-of-funds accounts provided by Albert M. Teplin of the staff of the Federal Reserve Board.
investment reflects an investor's interest in and influence over the management of an enterprise (Bureau of Economic Analysis, 1990c:84). All investment transactions between parent organizations and their foreign affiliates are direct investment flows. Portfolio investment primarily refers to ownership of financial securities, broadly defined to include sales and purchases of securities and amounts of outstanding claims and liabilities reported by banks and nonbanking concerns.
In the balance-of-payments accounts, incomes on direct and portfolio investment are reported in the current account; exchanges in financial assets between U.S. and foreign residents are measured in the capital account. The value of accumulated stocks of U.S. assets abroad and of foreign assets in the United States—resulting from capital flows in and out of this country over time—is annually compiled and published by BEA in the statement of U.S. international investment position.
The responsibility of gathering data on private capital flows is divided between BEA and Treasury.2 BEA collects information on direct investment and Treasury, using Federal Reserve banks as agents, compiles data on portfolio investment. The two data systems have evolved together over the years and reflect a high degree of collaboration between the two agencies. Various exemption levels are applied to different types of transactions. Since the 1980s—with the influx of foreign direct investment in U.S. industries covering manufacturing, wholesaling, retailing, as well as banking, securities, finance, and other services, together with the surge in foreign portfolio investment in U.S. securities during the same period—the pressure on both BEA and Treasury to collect adequate information on these burgeoning transactions has increased.
DATA SYSTEMS FOR PRIVATE CAPITAL FLOWS
BEA (and its predecessor, the Office of Business Economics) has been collecting data on U.S. direct investment abroad and foreign direct investment in the United States since the 1920s. It has gradually increased coverage and moved from voluntary reporting by enter-
Foreign income statistics compiled from tax returns can be found in the Statistics of Income data series published by the Internal Revenue Service of the Department of the Treasury.
prises to more comprehensive mandatory reporting. Current legal authority for the collection of direct investment data is provided by the International Investment and Trade in Services Survey Act.
BEA uses a hierarchy of surveys to cover both U.S. direct investment abroad and foreign investment in the United States. These include benchmark surveys (censuses) at 5-year intervals, as well as annual and quarterly surveys. Benchmark surveys are designed to collect a wide range of information on affiliates' transactions, as well as to update the universe of the database from which annual and quarterly surveys of smaller samples are drawn.
BEA defines a direct investor as a person who has a 10 percent or more ownership in a business enterprise located in a foreign country. This definition applies to both outward investment by U.S. residents and foreigners' inward investment in the United States. Separate surveys are conducted for U.S. direct investment abroad and foreign direct investment in the United States. Information collected from direct investment surveys includes not only data required by the U.S. balance-of-payments accounts, but also those on affiliates' operations such as their production, sales, trade, employment, financial statements, technology, and external financing (Bureau of Economic Analysis, 1990c).
Benchmark surveys are comprehensive and cover the universe of filers; exemption levels are minimal. These mandatory surveys cover information on a direct investor's share of income, distributed earnings, and capital gains and losses of its foreign affiliates; interest, royalties, fees, allocated expenses, and other charges; and the changes in a direct investor's equity and debt positions in its foreign affiliates. In addition to these balance-of-payments data, benchmark surveys cover financial structures and operations of parent companies and affiliates, including information on balance sheets and income statements, composition of external financing, production, employment, and intracompany trade, as well as technology, property, plant, and equipment. At the time this report was in print, BEA was tabulating its latest benchmark survey for U.S. direct investment abroad, covering 1989. The most recent available benchmark survey on U.S. direct investment abroad covered the year 1982. For that year, 2,245 U.S. parent corporations filed complete financial and operating data on their own operations, and 18,339 reports on their investment positions and transactions with their foreign affiliates. The most recent benchmark survey of foreign direct investment in the United States was undertaken for 1987. For that year, 8,577 reports were filed by U.S. affiliates of foreign direct investors.
On an annual basis, BEA collects key information on the operations of a sample of parent companies and their affiliates. (excluding banks). The annual surveys cover the operational aspects of the affiliates (for example, sales and employment) to supplement the data that enter the balance-of-payments accounts. The results of these surveys are published annually by BEA in detail. For 1988 the survey on U.S. direct investment abroad covered the operations of 9,500 U.S. companies and their affiliates; the one on foreign direct investment in the United States surveyed the operations of 5,500 U.S. affiliates of foreign corporations. Mandatory quarterly surveys cover information required for balance-of-payments purposes. In 1988, 1,300 U.S. parent companies filed 9,100 quarterly reports covering their foreign affiliates, and 3,400 U.S. affiliates of foreign companies filed quarterly reports.
The universe of direct investors is heavily skewed toward large companies and is also heavily skewed toward investors with ownership percentages of 50 percent or more. For instance, the 10 largest U.S. parent companies, in terms of the assets of foreign affiliates, owned 32 percent of such assets; the 100 largest owned 72.4 percent. Smaller foreign affiliates ($15 million of assets or less) account for 47 percent of the number of affiliates, but only 5 percent of assets. Similarly, in the United States in 1989, majority-owned U.S. affiliates of foreign companies accounted for 72 percent of the total assets of U.S. affiliates of foreign firms (Bezirganian, 1991:80).
All this survey information is confidential. Data are published only in aggregate form. When one or two filers account for the dominant share of the information in a data cell, the cell is suppressed. “Secondary suppressions” also are imposed to prevent derivation of such cells by subtraction. To augment the content of the information on foreign direct investment in the United States, the Foreign Direct Investment and International Financial Data Improvements Act of 1990 provides that the Census Bureau exchange information with BEA on data collected on business enterprises operating in the United States. This legislative action is necessary because BEA collects information on foreign direct investment in the United States at the enterprise or consolidated level: that is, if a manufacturing company also owns a retail operation, the manufacturing company would file information with BEA, consolidating the retail operations. The Census Bureau, in contrast, collects information (pursuant to its authority under Title 13, U.S. Code) at the establishment level: that is, information on individual plants. BEA, through its access to information in the
Census Bureau's Standard Statistical Establishment List (SSEL), will be able to produce more disaggregated data on foreign direct investment in the United States, facilitating the analysis of the relative share of such investment in the domestic economy at more detailed industry-by-industry establishment levels.
In June 1991, using two new measures, market-value and current-cost estimates, BEA revaluated U.S. foreign direct investment abroad and foreign direct investment in the United States for 1982 through 1989. These new estimates of inward and outward direct investment represent BEA's attempt to put U.S. assets abroad and foreign assets in the United States on a comparable basis. The comparability problem arises because U.S. direct investment abroad is, on average, older than foreign direct investment in the United States. Together with the revaluation of U.S. gold reserves using the market price of gold as a basis, these new estimates resulted in a change in the value of the 1989 U.S. net international investment position from the previously reported −$663.7 billion to −$281.0 billion under the market-value basis, and to −$464.0 billion under the current-cost basis (Bureau of Economic Analysis, 1991a).
BEA publishes and discusses data on direct investment periodically in its monthly Survey of Current Business and issues more extensive annual and periodic specialized census data in separate publications.
PORTFOLIO INVESTMENT DATA
The Treasury International Capital (TIC) reporting system, with Federal Reserve banks serving as agents for the Treasury Department, provides monthly and quarterly data on holdings of, and on transactions in, portfolio investment: investment in the forms of debt instruments between unaffiliated parties, equity positions of less than 10 percent, and other claims and liabilities (Bureau of Economic Analysis, 1990c:14). Treasury uses a number of TIC forms to capture sales and purchases of securities and amounts of outstanding claims and liabilities reported by banks and nonbanking concerns: the TIC S forms cover securities transactions; TIC B forms, banks' own claims and liabilities and custodial transactions; and TIC C forms, nonbanks' assets and liabilities in relation to unaffiliated foreigners, such as foreign trade credit by corporations.
Filing of TIC S forms is required on sales and purchases of long-term U.S. and foreign securities (original maturities longer than 1 year). Filers include banks, banking institutions, brokers, dealers,
and other persons in the United States who engage in these transactions, as well as corporations that issue Eurobonds directly from their U.S. offices. Monthly reports of sales and purchases are required. Reports must be filed if the total of purchases or sales amounts to $500,000 or more during a given month.
TIC B form filings by U.S. banks, depository institutions, international banking facilities, bank holding companies, brokers, and dealers in the United States report information on loans, advances, and overdrafts; placements of funds; acceptance financing and depositing; and borrowing through repurchase and resale agreements. Also included on the form are operating transactions between U.S. banks and their foreign branches, agencies, and subsidiaries, as well as those between U.S. branches, agencies, and subsidiaries of foreign banks and their parent corporations (excluding their permanent direct investment positions). Depending on the particular institution and form involved, monthly or quarterly reports are required. Banking data are monthly positions as of ends of the month. Reported transactions include both those undertaken for the banks' own accounts and those undertaken for the accounts of their domestic and foreign customers (custodial positions). Reports are required if total claims on, or liabilities to, foreigners are $15 million or more for any month-end closing balance.
Nonbanking businesses are required to file TIC C forms on a quarterly basis. Such businesses include exporters, importers, industrial and commercial firms, some nonbanking financial institutions, and U.S. affiliates of foreign business enterprises. Transactions include financial claims or liabilities (such as deposits in foreign banks or direct borrowings from foreign investors) and commercial claims or liabilities related to the sale of goods and services in normal business operations (for example, trade credits). Direct investment flows are excluded. Reports are required for financial or commercial claims on, or liabilities to, unaffiliated foreigners totalling $10 million or more at the end of the quarter.
According to the Treasury Department, in 1990 approximately 350 institutions filed regular monthly S forms; about 975 filed regular monthly and quarterly B forms; and 475 filed quarterly C forms. TIC forms are filed with district Federal Reserve banks by banks and banking institutions and with the Federal Reserve Bank of New York by all brokers, dealers, and corporations. The costs to the Federal Reserve banks of collecting the TIC data are borne by their own budgets. The Federal Reserve Bank of New York consolidates the data and makes them available to Treasury for publication in the quarterly Treasury Bulletin. The data remain
on the Federal Reserve Bank of New York's computer, loaded to databases to which Treasury has access. Treasury, in turn, transmits aggregate TIC data to BEA, which uses the TIC data to estimate portfolio investment income and compile capital flow statistics for the balance-of-payments accounts.
LIMITATIONS OF THE DATA
Under the current reporting systems, data on direct investment appear to provide a reasonably accurate picture of these transactions, although problems remain. Data on portfolio investment, however, are questionable. The latter are particularly affected by the growing numbers of financial transactors and products, as well as intermediaries. In compiling capital flow data, both BEA and Treasury have also encountered growing technical and operational problems. The increasingly complex nature of the transactions and the less than satisfactory compliance by filers have affected the accuracy and timeliness of the data. Despite BEA and Treasury efforts to broaden the data coverage and update filer lists, major data gaps remain.
U.S. DIRECT INVESTMENT ABROAD
Although it is generally acknowledged that data on U.S. direct investment abroad are of high quality, especially in view of the size of the universe to be covered and the complexity of the financial structures of U.S. multinational companies, sizable revisions of quarterly data do occur at times when tardy reports are received and errors detected. According to BEA, it generally does not attempt to expand, or blow up, reported capital flow data to account for late responses because capital flow data tend to be volatile and are subject to sign reversals. However, BEA plans to study the feasibility of including in direct investment capital flows an allowance for late reports, particularly when systematic patterns of late-reported data can be discovered.
Another shortcoming relates to the fact that the wide range of economic and financial data gathered on operations between parent corporations and their affiliates, beyond the balance-of-payments requirements, are not well synthesized and summarized, making them difficult to use and to interpret. In addition, BEA undertakes few in-depth analyses of these data. As a result, the wealth of information collected is not well understood and utilized by the users who could benefit from greater information on
changes in U.S. direct investment abroad. In effect, the information BEA collects and publishes on direct investment is not sufficiently well known to either the public or specialists.
FOREIGN DIRECT INVESTMENT IN THE UNITED STATES
Except for transactions undertaken by small investors who may not be familiar with the reporting requirements (especially in real estate transactions), data on foreign direct investment in the United States probably accurately represent the overall trend, since most large acquisitions or new investments in the United States by foreigners are well publicized and are monitored by BEA. In addition, BEA uses information collected by industry groups to identify incoming direct investment and to broaden mailing lists of filers. On a limited basis, BEA is currently adjusting its data to account for capital flows related to establishment or acquisition of new affiliates by foreign direct investors. Nonetheless, problems persist.
One problem is that compliance with reporting requirements is not as good as it could be. U.S. affiliates of foreign parent companies often fail to respond or to respond fully and often refer inquiries to their foreign parent companies. Late reporting is also a problem. Increased funding provided to BEA to track new inward investments and to pursue cases of noncompliance and inadequate reporting, together with heightened penalties for noncompliance and greater enforcement efforts, should ameliorate the situation. In the case of foreign investment in U.S. real estate, it is suspected that much of it is not captured in the U.S. balance of payments. This is because many of these transactions are in the form of limited partnerships, which are also not captured under the TIC system unless they are listed on the exchanges (Stekler, 1991b).
Another problem relates to the adequacy of data on income on (payments to) foreign direct investment in the United States. While such foreign direct investment rose by $342 billion between 1980 and 1990 (from $124 billion to $466 billion, valued at current costs), income payments related to it did not show an upward trend; they fluctuated between $3 billion and S11 billion during the period. And in 1990 such payments were reported to be less than $2 billion. The trend of these income payments contrasts with that of income receipts from U.S. direct investment abroad. During 1980-1990, U.S. direct investment abroad rose from $385 billion to $598 billion (valued at current costs), or an increase of
only $213 billion. Yet income receipts from such investment rose from $37 billion to almost $55 billion during the same period (Bureau of Economic Analysis, 1991d). Although variations between incomes on foreign direct investment in the United States and those on U.S. direct investment abroad can result from differing cyclical conditions in the United States and abroad, as well as from variations in foreign exchange rates, the drastic differences in their magnitudes and trends raise concerns about the underlying causes for the relatively small income payments on foreign direct investment in the United States. Deliberate understatements of incomes by filers (often through transfer-pricing practices) can result in losses of tax revenues to the Treasury Department.
There appear to be significant gaps in data on securities transactions. Such lacunae also exist in securities reporting in other countries. The main difficulty with both U.S. and foreign data is that the information collected comes primarily from resident financial institutions. In some countries, that process would be adequate, since residents would almost always effect their international portfolio purchases or sales through a resident bank, broker, or dealer, but it is no longer always the case for many countries. Modern communications facilities and the withering away of exchange controls or other official restrictions have made it not only possible, but often quicker and cheaper, for residents of one country to deal directly with markets in another country. In addition, money managers and large institutional investors (such as pension or mutual funds) increasingly maintain their own facilities for conducting business, without dealing with a financial institution that reports on the TIC S forms. In these cases, unless the Federal Reserve Bank of New York contacts the transactors, these transactions are not captured by the existing system. In principle, the S form reporting requirement extends to all U.S. residents whose transactions are above the exemption level in any month, but there is no system for identifying all such residents. Although reports filed by large financial institutions cover many transactions and transactors, given the changing international financial trading environment, the number of filers of monthly reports—about 350—casts doubt that the data fairly represent the cross-border business done by residents, including those removed from the traditional channels of resident financial institutions. According to the Federal Reserve Bank of New York, there is also
no assurance that the reported transactions are complete and accurate.
A second difficulty relates to U.S. borrowers' use of Euromarkets. Eurobonds account for most U.S. corporate bonds sold to foreigners. When a U.S. corporation issues securities directly in Euromarkets (usually through underwriters in London), there may be no U.S. financial institutions directly involved (with the possible exception of an affiliated office in Europe), so that reporting of the inflow and subsequent redemption would occur only as a result of direct reporting by the borrower. Although the Federal Reserve Bank of New York or BEA can track these issues when they are well publicized, comprehensive coverage is difficult to obtain. Similarly, domestic borrowers in domestic markets may not know when the resident lender has resold a loan to a foreigner. Although there is a provision for recovering such resales of loans—on the BL-2 form through custodial reporting when the originating lender acts as the loan-servicing agent and a foreign bank is the purchaser—such transactions involving nonbank entities may not be adequately captured.
In addition to these examples of capital inflows that may not be captured, unreported capital outflows pose even greater problems, since it is more difficult to monitor financial activities undertaken abroad by residents. Global studies on capital flows conclude that most countries, including the United States, are better positioned to capture inflows of capital in their statistical frameworks than outflows by residents. (In some countries, not especially the United States, there is also an incentive to hide outflows of funds illegally or to avoid taxes or other restrictions.) The last Treasury Department benchmark survey of U.S. portfolio investment abroad was undertaken during World War II. This raises the concerns about the adequacy of the coverage of U.S. holdings of foreign securities.
A third difficulty with securities data is the proliferation of short-term marketable instruments and derivative instruments such as warrants, options, puts, and calls. These are not ordinarily covered by the TIC S forms, which deal primarily with long-term securities. The S forms cover warrants and options only when the underlying security is a stock or long-term bond. But since data on purchases and sales of these instruments are not disaggregated with purchases and sales of U.S. or foreign bonds, an evaluation of the adequacy of coverage of these transactions is currently not possible. And all other options or warrants are not covered. Although BEA includes margin accounts and profits and losses on
futures trading by foreigners on U.S. futures exchanges in its estimates of U.S. nonbank liabilities to unaffiliated foreigners, U.S. residents ' trading in foreign futures exchanges are not captured by the TIC system (Stekler, 1991b). In addition, as noted above, limited partnerships are not adequately covered.
A fourth difficulty—which generally applies to data on U.S. international transactions and which particularly complicates any attempt to identify gaps in the reporting of securities—is the confusion of geographical allocations, especially when trading takes place through offshore centers or foreign financial centers. The latter often act as intermediaries on behalf of residents of many countries. In the U.S. data, the geographical identity of the foreign transactor is simply the address from which, or to which, orders are placed or transmitted. As a result, when such transactions are reported, they are often reported as transactions between the United States and countries with well-developed financial markets, such as the United Kingdom and Switzerland, instead of the country that is the ultimate source of foreign funds or the ultimate destination of U.S. funds (Bureau of Economic Analysis, 1990c). Consequently, the U.S. data show the bulk of recorded transactions for major foreign financial centers, regardless of whether the ultimate buyer or seller actually resides in those locations. In those foreign financial centers, the statistical offices tend to net out their foreign-to-foreign transactions on the ground they have no relation to the economies of those centers, so that it is not possible, in most cases, to rely on data collected by partner countries as a check on the data available to the home country.
International comparisons are further made difficult by the varying data definitions used by different countries. For example, there are sizable discrepancies between U.S. and Japanese data on purchases of U.S. Treasury securities, because the Japanese data include all Japanese purchases of U.S. Treasury securities worldwide while the U.S. data include only U.S. Treasury securities sold directly to Japan (Stekler, 1991b).
A fifth difficulty with securities data concerns changes taking place in the institutional and functional arrangements of financial organizations. One change is the increased complexity of individual organizations: as they develop new techniques and instruments and spread around the world, the responsibility for reporting on U.S. statistical forms becomes decentralized, the actual locus of a transaction becomes arbitrary, and participation with institutions headquartered in other countries further obscures the responsibility for statistical reporting.
There are numerous difficulties related to the use of the TIC B forms for banking that can distort the data. A basic problem is that the B form data are end-month positions, which must be converted into flows for use in the balance-of-payments accounts. This means that write-offs or renegotiated balances must be detected and adjusted from the flows, and balances in foreign currencies must be adjusted to reflect changes in the foreign exchange value of the dollar. In practice, however, this is not done in the U.S. accounts because the currency composition is not known. In addition, apparent flows may occur because an address changes from domestic to foreign, or vice versa. All these adjustments can result in large swings in the statistical discrepancies in the balance-of-payments accounts.
Furthermore, although it appears that U.S. banks generally do a conscientious job of reporting under the TIC system, problems arise in dealing with some U.S. affiliates of foreign banks. The problems concern whether the instructions for the forms are always fully understood and acted on properly and whether accounts owned by foreigners in the United States can always be identified as such by reporting banks. It is comparatively simple for a non-resident to acquire a U.S. bank account in the name of a resident or to establish a temporary residence in the United States.
Interbanking business between U.S. and foreign banks seems to be accurately reported, and transactions can be verified with the banking data collected by the Bank for International Settlements from partner banks. However, cross-border banking is increasingly done without the intervention of international banking facilities. This is a gap in the statistics for which banks are not responsible, but it means that reliance on reporting by banks no longer gives the assurance of nearly complete coverage that was assumed in earlier years.
Another change of considerable importance is the growth of direct transactions between nonfinancial lenders and borrowers, with financial intermediaries increasingly acting as agents or arrangers of deals for a fee or commission. This process is partly driven by the tightening of minimum capital standards for internationally active banks by the Basle Committee on Banking Supervision, which indirectly encourages banks to do business in ways that do not appear directly on their balance sheets. Under these circumstances, it is unlikely these transactions are included in the TIC reporting system.
Many attempts have been made to broaden the coverage of the TIC C forms and to review the accuracy of reporting for transactions by nonbanking businesses, but the results of efforts to include more respondents or detect significant errors in reporting have been only marginally successful. Current filers are believed to include only a fraction of those that should report. One reason to doubt the effectiveness of the C forms comes from comparing U.S. data on cross-border positions of nonbanks with the data reported by partner-country banks in the Bank for International Settlements (BIS) and IMF accounting frameworks. In September 1990, for example, TIC data on U.S. nonbank financial assets abroad (including bank custodial accounts for nonbanks) amounted to about $65 billion; the data derived from foreign banks as reported by the BIS and IMF show liabilities to U.S. nonbanks of about $250 billion. On the other side of the accounts, the TIC reports that the liabilities of U.S. nonbanks to foreign banks were about $80 billion; the BIS and IMF report that foreign banks' claims on U.S. nonbanks amounted to $235 billion. Although there are differences between the TIC and the BIS and IMF reporting systems in definition and, in particular, in coverage, such sizable discrepancies suggest the need for an examination of the possibility of underreporting in the U.S. data.
Currently, increases in foreign holdings of U.S. currency are not included in the U.S. balance of payments. Yet these holdings represent noninterest bearing obligations of the U.S. government to foreigners. Unreported currency shipments can be related to drug trade, to money brought in and out of the country by tourists, and to money mailed by U.S. residents to relatives abroad. Although Treasury requires persons taking more than $10,000 in currency into or out of the United States to file a Currency and Monetary Investment Report (CMIR) with the Customs Service, compliance and enforcement are far from satisfactory.
On the basis of various surveys of household and business currency holdings, IRS estimates of the magnitude of drug trade, estimates of currency lost or destroyed, and other sources, it has been estimated that as much as one- to two-thirds of U.S. currency outstanding may be held by foreigners, or $85-170 billion. Increases in foreign holdings of U.S. currency were estimated at
more than $15 billion in 1990 (Stekler, 1991b). An accurate account of foreign assets in the United States would require that data gaps on the Treasury CMIR, as well as currency flows related to drug trade, tourists, and other transfers, be closed.
INCOMES ON PORTFOLIO INVESTMENT
Unlike data on incomes on direct investment, which are reported by filers, data on incomes on portfolio investment are largely estimated by BEA, on the basis of accumulating stocks derived from Treasury 's TIC reports and rates of return on these investments assumed by BEA. Both inadequate data on portfolio investment as filed in the TIC reports and inappropriate rates of return applied by BEA can result in misleading estimates on portfolio investment incomes.
A 1986 study on statistical discrepancies in the world's current account identified portfolio investment incomes as one of the most serious problem areas (International Monetary Fund, 1986a). Reported portfolio investment payments worldwide far exceeded reported portfolio investment receipts. The U.S. transactions account for a large proportion of the world current account transactions. More recently, Federal Reserve and IMF data indicate that the TIC reports underestimated U.S. residents' bank deposits abroad by $80-180 billion at the end of 1988. Using the 3-month Eurodollar interest rates, this implies that U.S. interest earnings were underestimated by $6-14 billion in that year (Stekler, 1990). Furthermore, in estimating income on equities, BEA assumes that income in the current period is the same as in the previous period plus dividends on additions to securities holdings. This methodology cannot yield reliable estimates because it assumes a stock acquired in 1960 is earning the same dividend today as it did three decades ago. Also, the value of holdings on which dividends are calculated is not current because the last benchmark survey of U.S. portfolio investment abroad was undertaken during World War II (Stekler, 1990).
According to the Treasury Department, it is considering conducting a new benchmark survey on U.S. holdings of foreign securities and undertaking a number of research projects to estimate more closely the magnitude and nature of other reporting defects. These projects include an examination of pension funds' offshore purchases of securities, the size of the reporting panel for the TIC C forms and the quality of the nonbanking reports, the volume and nature of currency flows, and effects of synchronization of
payments and delivery. These efforts should yield insights into ways of enhancing the accuracy of existing U.S. portfolio investment data and lead to more accurate estimates of incomes from such investments.
Our recommendations for improving U.S. international capital flow data are listed in order of their relative importance, with the most important one listed first.
Recommendation 6-1 Because the existing system for collecting data on international portfolio transactions was developed before the advent of modern electronic and telecommunications technology currently used by growing numbers of transactors, research is needed to explore alternative methods of data collection. Over the long term, the clearing systems for securities, the payments systems for banking transactions, and other clearing channels for nonbanking transactions should be explored as alternative sources of information. Increased automation in trading and clearing systems would facilitate the compilation of data from such sources.
Recommendation 6-2 Research should be performed to develop ways to broaden coverage and to enforce compliance of reporting by filers. For securities transactions, improvements in coverage of U.S. outward investment in foreign securities and transactions by nonfinancial entities are of particular importance. For reporting by banks, the principal research effort should focus on enhancing the timely and accurate reporting of U.S. affiliates of foreign banks. For transactions on nonbanking concerns, improvements in the coverage of activities of the U.S. nonfinancial population, both corporate and personal, are needed. A broadening of the filing of TIC C forms, perhaps through lowering exemption levels for reporting, should also be considered.
Recommendation 6-3 The adequacy of data on official U.S. government international capital transactions should be examined. Official U.S. government international capital transactions are of growing importance, but the myriad forms of transactions in U.S. official reserve assets and other government assets and the involvement of numerous government
agencies make them susceptible to inadequate reporting. There have also been few evaluations of government agencies' compliance with reporting requirements.
Recommendation 6-4 The methodology of estimating portfolio investment incomes should be improved, including new methods to measure current values of U.S. securities holdings abroad, as well as other forms of U.S. international financial assets and liabilities. Appropriate rates of return can be developed through close consultations with financial institutions at home and abroad.
Recommendation 6-5 In view of the difficulties of obtaining data on residents' direct purchases or sales of foreign securities, research should be undertaken to explore the feasibility of exchanging data with partner countries. This would require, among other things, the development of universal codes for identifying securities and uniform standards for reporting, as well as common definitions of the residence of the immediate transactor and of the ultimate owner.
Recommendation 6-6 To gauge the effectiveness of the existing reporting systems, “experiments” should be conducted in which capital transactions of various types are earmarked and purposefully injected into the economy and then methodically followed. Follow-up actions could then be taken to determine which transactions are captured by the existing systems.
Recommendation 6-7 The Bureau of Economic Analysis, in cooperation with the Internal Revenue Service of the Treasury Department, should examine the causes of the relatively low income payments on foreign direct investment in the United States.
Recommendation 6-8 Additional efforts are needed to exploit more fully the analytic potential of existing data on U.S. inward and outward direct investment and to disseminate them more effectively to users. Existing data should be reviewed to determine if additions, deletions, or modifications are necessary. Research should also be conducted to examine how data can be better synthesized and tabulated in more usable formats to enhance their analytic usefulness.