Chapter

National Compilation Systems

Author(s):
International Monetary Fund
Published Date:
June 1992
Share
  • ShareShare
Show Summary Details

The Working Party on the Measurement of International Capital Flows asked a number of countries to prepare brief descriptions of their systems for compiling data on international capital accounts, together with an exposition of the problems they have encountered with their systems and the steps being taken to improve the data. In the following section, 11 of these statements are presented, modified to some extent to meet requirements for length and uniformity of presentation. It is hoped that these statements will illustrate the range of statistical systems in place as well as the problems and possible solutions connected with them.

Australia

Main Features of Capital Account Compilation

Data for Australia’s balance of payments capital account are derived from a number of sources, of which the most important is the Survey of Foreign Investment (SFI). Data are collected by the Australian Bureau of Statistics (ABS).

Survey of Foreign Investment

The SFI is a survey conducted under the authority of the Census and Statistics Act. The act provides the Australian Statistician with the power to direct disclosure of the desired information; penalties can be imposed for noncompliance. The act also stipulates that replies are held confidential.

The SFI consists of small monthly telephone surveys and much larger quarterly and annual mail-based collections. The SFI contains a variety of reporting forms designed to fit the activities of all types of international transactors and covers data on capital flows, investment stocks, and related income items. Many details are required, including types of transactions and geographic allocations.

The monthly telephone survey covers 18 enterprise groups.1 It obtains information on the foreign borrowing activity of the state governments and on nonresident holdings of debt securities issued in Australia by the Commonwealth and state governments and by public nonbank enterprises. Respondents include the central borrowing authorities of each state government, the Reserve Bank of Australia, and the large nominee enterprises holding Australian securities on behalf of foreign principals.

The quarterly collection covers just over a thousand enterprise groups that have foreign investment activity in excess of certain thresholds, plus about 80 nominee and securities dealers that act on behalf of foreign principals. Information is collected on all aspects of foreign investment activity of the selected enterprise groups, except on the stocks of foreign portfolio investment in Australian corporate equities and reinvested earnings attributable to direct investors. Nonetheless, quarterly estimates of both types of investment are made as described below.

The annual collection covers the two aspects of foreign investment activity not covered in the quarterly collection and any foreign investment activity of enterprise groups not included in the quarterly collection. The annual collection covers all large transactors and a stratified random sample of other enterprise groups. Approximately 4,200 enterprise groups are surveyed annually from about 6,200 enterprise groups that fall within the scope of the SFI. Many enterprise groups are included in both the quarterly and annual collections but do not need to resubmit data already reported in one or the other survey.

Other Sources

Some elements of the capital account are collected outside the SFI. In particular, data are collected separately from the Commonwealth Government and the Reserve Bank on their holdings of certain foreign financial assets and liabilities, including borrowing in foreign markets and official reserve assets. In addition, the principal Australian export marketing authorities are separately approached for information on accounts receivable.

The country reports were edited by Arie C. Bouter and Neil Patterson, who both served as members of the technical staff.

Although quarterly details on the stock of foreign portfolio investment in Australian corporate equities are not collected in the SFI, estimates of this component are made using both the latest annual SFI stock data, supplemented by capital transactions data collected in the quarterly SFI from brokers and other security dealers, and relevant data from other sources, such as the financial press and the Australian Stock Exchange. These sources are also used to supplement stock information annually reported by companies and nominee enterprises in order to revise estimates of capital transactions.

Estimation Methods

Australia compiles comprehensive capital account statistics (and corresponding statistics on stocks and investment income) on a quarterly and annual basis as described above. However, in compiling preliminary quarterly and annual statistics from the SFI, estimates are included for late respondents. Individual estimates are made for large nonrespondents and global estimates are made for smaller ones. Quarterly published data include a coverage adjustment, based on historical trends for those entities not approached quarterly. When the annual SFI results are available, the coverage adjustment is replaced with an estimate based on the actual annual collection for that year. For quarterly reinvested earnings, estimates are made by extrapolation until survey-based data are available from the annual SFI. Two basic approaches are used to prepare final quarterly estimates from annual SFI data: first, when no better basis is available, the annual data may be allocated evenly to the quarters; second, annual data may be allocated to quarters on the basis of the information reported by quarterly respondents.

Existing Problems as Seen by National Compilers

Measurement of capital account transactions has become increasingly difficult during the 1980s. Since October 1983, there have been a number of government initiatives designed to liberalize the financial system and to introduce diversity and competition into the enterprise sector. These measures have included the freeing-up of controls on participation in the foreign exchange market, suspending virtually all exchange controls, floating the Australian dollar, authorizing a number of foreign banking interests to establish banking operations in Australia, allowing many new merchant banks to be set up (most with significant foreign ownership), permitting substantial foreign ownership in Australian stockbroking businesses, and liberalizing controls on foreign investment in Australia. One outcome of these initiatives has been an “internationalization” of the Australian financial system, which has contributed to a significant increase in the number and types of entities engaged in international transactions. New instruments and methods of financing have also been introduced.

These changes to the structure of the financial system in Australia have challenged international accounts statisticians. For example, the correct conceptual treatment for new, often complicated, sets of transactions has had to be determined, and procedures to ensure that new transactors are covered by the SFI have had to be developed. The latter problem has been compounded by the suspension, from December 1983, of exchange controls. Up to that point, information on entities that had been granted exchange control approvals had provided a prime source for identifying new transactors; since then, the ABS has had to resort to various partial sources. Also, as the government has relaxed its monitoring of foreign investment proposals, with the result that fewer transactions require approval, the Foreign Investment Review Board data are less useful as a source for identifying new transactors.

Two other measurement problems also may have had a significant impact. First, problems have emerged in the measurement of gains and losses on foreign exchange transactions, particularly those involving banks. The nature and extent of these deficiencies are under investigation. Second, the potential for inconsistencies in the time of recording and the valuation of the two sides of a transaction has increased with the rapid growth in the volume of international transactions and the increased frequency of exchange rate adjustments since the floating of the dollar and other major currencies.

Foreign investment in Australian real estate became substantial during the latter half of the 1980s. Until recently, such investments were captured in the statistics only when they had been channelled through an entity operating in Australia. The ABS has investigated and estimated the undercoverage of such investments, incorporating the figures into the published statistics from the third quarter of 1989 onward. This adjustment significantly reduced Australia’s balancing item (net errors and omissions) between the periods 1986-87 and 1988-89. However, some undercoverage may still be associated with this type of investment. In addition to producing undercoverage estimates for past years, the ABS has developed a methodology, within the SFI framework, to improve the regular measurement of investment in Australian real estate.

More recently, many Australian enterprise groups have been liquidating both Australian and foreign assets in order to reduce indebtedness. The coverage of the capital inflows associated with the acquisition of these assets by nonresidents is probably incomplete. A further problem confronting Australian compilers is the lagging receipt of data from an increasing number of companies in receivership or liquidation.

Plans for Improvements

A number of projects to improve coverage in the SFI are currently under way: investigation of new data sources (such as the Australian Securities Commission); development of a survey drawn from an alternative population register to quantify as accurately as possible the extent of undercoverage in the SFI and identify units not in the survey; and reconciliation of unit record data reported in the SFI with data reported in other ABS collections. Methodologies for partial coverage and nonresponse estimation are being reviewed in order to develop more quantitative and automated models.

A review of the SFI’s computer editing system is being conducted with the aim of making the system simpler and more effective, thus freeing computing and clerical resources to undertake other tasks. Also, electronic reporting is being progressively extended in the SFI. Data are edited as they are entered into the electronic form by the respondent, thus minimizing reporting errors. This development is expected to improve the quality of reported data and free clerical resources to check paper returns more thoroughly.

Two methodologies are currently being assessed for improving quarterly estimates of reinvested earnings. The first is an estimation model based on estimates of the total operating surplus generated for inclusion in quarterly national accounts. The second is extending the quarterly SFI collection to obtain data on reinvested earnings.

Work is also under way to align more closely the measurement of debt securities with the market-price principle of valuation and to improve the measurement of discount and premium income. Various inconsistencies and other deficiencies have been identified in these data but are expected to be overcome by modified SFI reporting requirements that were introduced during the first half of 1992.

Work on an overall quality framework for the SFI is in progress. The aim is to provide regular measures of data quality against predetermined targets so that problems can be identified and addressed quickly.

Belgian-Luxembourg Economic Union

Main Features of Capital Account Compilation

The balance of payments of the Belgian-Luxembourg Economic Union (BLEU) is compiled by the National Bank of Belgium (NBB) broadly according to the principles recommended by the IMF. The NBB obtains the necessary data from the Belgian-Luxembourg Foreign Exchange Institute (Institut Belgo-Luxembourgeois du Change, or IBLC), which collects data on international payments.2 Until 1990, the prevailing exchange control regulations established a two-tier exchange market, consisting of a regulated and a free market. On March 5, 1990, the two-tier exchange market was abolished, but the data collection was continued.

The IBLC collects data from resident banks and other companies. Resident banks report their own international transactions and the transactions of their customers. All settlements are coded by the banks by type of transaction, by country of the foreign transactor, and by currency; the list of codes is provided by the IBLC, The amount of the settlement is reported in the currency used and is converted into francs by the IBLC at the corresponding monthly average exchange rate. AH settlements must be reported; there is no minimum threshold. On a daily basis, the total of the reported settlements must equal the change in the net currency position of the bank; thus, the reporting system is the “closed” type.

Resident companies report transactions that are settled through foreign financial intermediaries, by way of bilateral or multilateral clearing of mutual debts, or to some extent in bank notes. Their reports also provide details using the IBLC list of codes.

For balance of payments purposes, data received from the IBLC are supplemented by the NBB with information provided by other sources, mostly concerning current account transactions. Furthermore, some transactions are reported to the IBLC on a global basis only—broken down by currency and country of the foreign party, but not by type of transaction. They concern both settlements of the European Communities in Belgian francs and in European currency units with residents of the BLEU and transactions in bank notes. The former are classified on the basis of information provided by the European Community; the breakdown of the latter is estimated on the basis of a set of economic and financial indicators. The large fluctuations in the bank note transactions of the BLEU probably reflect capital transactions, given the well-developed financial center in Luxembourg and the absence of a withholding tax. Capital flows involving trade credit are computed as the difference between data on merchandise transactions by date of payment and that by date of invoice. The latter are not derived from customs statistics but from the same source as the payments data.

Existing Problems as Seen by National Compilers

The closed reporting system of the BLEU banks and other companies with nonresidents is complete in principle. Only “free market” transactions settled without the intervention of a domestic bank were missing before March 5, 1990, and it does not appear that this gap was very important. However, since the banks and other companies are responsible for coding the transactions, the coding list cannot be too long or too complicated, and it cannot be changed frequently. Consequently, the available data are not very detailed and the system is inflexible in response to new demands.

In addition, it appears that the transactions of holding companies operating in Luxembourg, those of some “coordination centers” operating in Belgium, and those of other foreign-owned direct investment enterprises are causing problems for the compilers in the BLEU. Although foreign-owned holding companies operating in Luxembourg handle huge flows of capital, details on these flows are insufficient to permit comparisons with the figures published by other countries on direct investment in the BLEU.

Until recently, some of the resident coordination centers of multinational enterprises, which can carry out only limited activities for their affiliated enterprises, were considered by the IBLC to be nonresident companies in order to reduce the administrative burden of exchange control. Coordination centers could benefit from that status only if they could prove that in more than 80 percent of their activities they acted as an intermediary between two nonresidents. Consequently, none of their settlements with nonresidents was recorded in the balance of payments of the BLEU, but their payments to Belgian or Luxembourg companies or individuals were considered to be international transactions. Since this status was given to both foreign- and BLEU-owned entities, the creation of a coordination center in Belgium by a Belgian company was recorded as direct foreign investment by the resident company. As a result, BLEU statistics on direct investment differed from those published by BLEU partners. In September 1991, these centers were reclassified as residents.

Capital flows between affiliated enterprises are classified as direct investment, and no minimum ownership threshold is specified. There are no data on reinvested earnings.

Plans for Improvements

Some changes in methodology are planned for 1992.

  • —A new regulation will introduce a threshold of BF 250,000 for reporting the nature of a transaction and also a new coding list that conforms more closely to Fund classifications.

  • —The stock of direct foreign investment will be surveyed, giving the BLEU an opportunity to verify the flows and to estimate the reinvested earnings on direct investment.

  • —Inquiries to get faster and more complete information on changes in trade credits will be introduced.

  • —More detailed information on portfolio investment will become available as a result of the introduction of new transaction categories in the IBLC’s coding list. In particular, it will become possible to distinguish short-term financial instruments, such as commercial paper and treasury certificates, from long-term bonds.

Canada

Main Features of Capital Account Compilation

Canada’s balance of payments statistics are prepared by Statistics Canada. Two main data sources are used to produce the record of financial transactions with nonresidents.

  • —Statistics Canada collects information on most of the capital account flows from surveys conducted under the authority of the Statistics Act. Information is treated in strict confidence and is specifically exempt from being released under the Access to Information Act.

  • —Administrative data are used to compile external banking flows and claims of the Government of Canada.

About 400 large Canadian enterprises are surveyed on a quarterly basis for direct investment and other international transactions. These preliminary estimates are superseded by information from an annual survey covering the international transactions of some 4,000 enterprises. Estimates of Capital flows for smaller enterprises are derived from information reported on annual international investment position questionnaires sent to 6,200 enterprises for inward direct investment and long-term debt and on those sent to 1,800 enterprises for outward investment.

The coverage of direct investment transactors is reviewed against two major administrative sources. For inward direct investment, the reporting universe is periodically checked against information filed under the Corporation and Labour Unions Returns Act, which requires the reporting of ownership information (in Canadian dollars) from Canadian corporations with assets over $10 million, or gross revenues over $15 million, or long-term debt or equity held by nonresidents exceeding $200,000. For outward direct investment, a new tax form requiring Canadian companies to report transactions with foreign affiliates is being used to update the reporting population.

Monthly estimates on portfolio transactions are obtained from three main sources:

  • —monthly survey of Canadian intermediaries, pension funds, mutual funds, and other institutions that transact directly with nonresidents;

  • —monthly information from Canadian governments and corporations that borrow in international capital and money markets. The financial press is also closely monitored for information on borrowings abroad; and

  • —redemptions of Canadian bonds held by nonresidents, derived from an automated data base that contains particulars on some 10,000 individual bonds held in total or in part by nonresidents.

For transactions in Canadian bonds (and money market instruments), details are collected for individual instruments—both international bonds and secondary- market trading in domestically issued debt instruments. Data on international bonds are checked against annual international investment position surveys covering approximately 7,000 enterprises, in which information is reported on each long-term debt obligation owing to nonresidents.

Data on transactions in Canadian bonds are entered into a data base that contains detailed information on more than 14,000 long-term debt instruments held by nonresidents, 10,000 of which are Canadian bonds. Each issuer is identified separately and coded accord-ing to the country whose residents control the issuer, its province of incorporation, its sector (the various levels of government or private corporations), and its industry. In addition, each issue is identified by type of debt, date of issuance and maturity, currency of issue, rate of interest, frequency of interest payment, geographical detail, and the particulars of any interest rate or currency swap. Since 1991, as available, numbers relating to uniform securities identification have also been used. In addition to information on new issues and redemptions, the system generates estimates of accrued interest payable to nonresidents and offsetting short-term payables as well as estimates of the outstanding stock of debt valued at prevailing exchange rates. In order to alleviate respondents’ burden, the system generates for each issuer a detailed listing of debt, which is preprinted on the annual questionnaires sent to the various levels of Canadian government (including the government enterprises) and, beginning with the data for 1991, to corporations. Respondents are requested to correct or update the information as needed.

A separate data base with a similar degree of detail is maintained for domestically issued money market instruments held by nonresidents. The system also calculates interest, retirements, and the stock of debt outstanding.

Information on banking transactions is obtained from monthly reports filed by Canadian banks with the Bank of Canada on their foreign claims and liabilities. These reports are provided on magnetic tape to Statistics Canada, which processes the information to derive flows from changes in reported positions. Flows derived from foreign currency positions exclude valuation changes. Banks also report annually on write-offs of external claims booked in Canada; this information is used to exclude write-offs from the flow figures derived from changes in positions. In the balance of payments, the foreign currency transactions of Canadian banks are published on a net basis (liabilities net of assets), as is the case in the international investment position.

Information on Canadian nonbank deposits abroad comes from foreign banking data and is derived from changes in position figures. Data for other claims and liabilities are largely obtained from quarterly and annual surveys. Capital flows related to short-term claims are derived from changes in the amounts outstanding at the beginning and end of the reporting period. For long-term capital transactions, data on drawings and repayments are reported.

Existing Problems as Seen by National Compilers

Statistics Canada regularly adapts its statistical system in response to developments in international markets, aided by Statistics Canada’s close contact with major transactors in these markets. Relatively minor difficulties arise when foreign data are used for such activities as banking and portfolio transactions. For nonbank deposits abroad, the foreign data cannot be verified, so that information on major revisions and large shifts in positions is not available.

For portfolio transactions, the Canadian statistical system is believed to capture most of the foreign investment flows, but no benchmark surveys are conducted to check the accuracy of data on transactions in the domestic market. For borrowings in international bond markets, little is known on who holds Canadian securities (apart from holdings by Japan, which are reported by Japanese financial institutions).

Plans for Improvements

Statistics Canada is working in the following areas to maintain or improve the adequacy of the capital account statistical system: better monitoring of survey coverage; enhancing systems to process information; working with U.S. compilers to reconcile bilateral capital flows; and producing estimates of reinvested earnings on direct investment.

  • —Coverage: Quarterly data are initially obtained from sample surveys and then superseded by annual benchmark surveys. The quarterly coverage is supplemented by an economic intelligence system. Various newspapers, periodicals, and other public sources of information are systematically reviewed on a weekly basis to identify Canada’s international transactions. For direct investment flows, this information is scrutinized to ensure coverage of all large transactions. Since 1987, public information on direct investment transactions has been entered into a data bank containing the names of the Canadian transactors, survey codes to determine if the enterprise is already surveyed on the balance of payments questionnaires, the names of the foreign transactors, the value of the transactions, and other pertinent information. The data base ensures that large transactions are covered quarterly and is also used to update the annual census. In addition, the coverage of the annual direct investment surveys is validated against tax data.

  • —Data processing systems: A recent innovation in the processing of data on securities transactions reported by investment dealers and banks has been the replacement of questionnaires providing only aggregated data with detailed information on individual transactions that is reported on magnetic tape. The benefits of this approach have proved to be multifold. More detailed transactions information has allowed for the updating of the data base on Canadian bonds held by nonresidents (data previously entered manually). It has also substantially enhanced the quality of the data, which previously included classification and valuation errors, and has reduced the reporting burden. It is intended to exploit further the detailed information on trading in securities by setting up an inventory of Canadian equities held by nonresidents, similar to the system now being used for bonds. Moreover, it is now possible to identify the type of foreign securities in which Canadian residents invest and to survey foreign issuers periodically for information on Canadian investments. The data processing system for bonds has recently been enhanced to handle currency and interest rate swaps and, by extending the data base, to include long-term bank loans and other long-term debt. The latter information is obtained from annual international investment position questionnaires.

  • —Reconciliation of Capital account data with the United States: Canadian current account transactions with the United States have been reconciled on an annual basis since 1972. It is expected that data reconciliation will be extended to the capital account.

  • —Reinvested earnings on direct investment: Currently, steps are being taken to produce quarterly data on reinvested earnings on direct investment for inclusion in the balance of payments within the next few years.

France

Main Features of Capital Account Compilation

The French balance of payments uses two sources of data on capital movements:

  • —records of payments that are filed either by credit institutions on behalf of their customers or by enterprises that have large transactions with nonresidents or that have accounts abroad.

  • —monthly and quarterly reports of banks providing information on their stocks of claims and liabilities vis-à-vis nonresidents to the Banking Commission (Commission Bancaire). The banking sector’s capital transactions are derived from changes in the stocks of assets and liabilities denominated in francs and foreign currencies. For items denominated in foreign currencies, adjustments are made for changes in exchange rates.

Existing Problems as Seen by National Compilers

The existing problems facing national balance of payments compilers are attributable to a combination of several factors.

  • 1. The complete lifting of exchange controls at the end of 1989. Balance of payments compilation was once a by-product of the exchange control system, which was based on the completion of vouchers for each international payment. Since the lifting of exchange controls, the task of identifying capital transactions has become more difficult because banks are showing a certain reluctance to ask their customers to provide statistical information on transactions in the newly deregulated exchange market.

  • 2. The internationalization of markets. The internationalization of markets has had a pronounced effect on capital movements, which have greatly increased during the past two years. It has not been possible to compile statistics for these fast-growing transactions as rigorously as has been done in the past, a problem reflected in the increase in errors and omissions.

  • 3. Financial innovations. Innovations in the area of short-term credit facilities, in particular the expansion of the commercial-paper market and the introduction of financial derivatives, do not lend themselves readily to statistical measurement.

  • 4. Challenges to the distinction between long-term and short-term instruments. Reporting enterprises have been finding it increasingly difficult to apply the putative distinction between short- and long-term instruments—particularly regarding new instruments (note issuance facilities, revolving underwriting facilities, and multioption facilities), which often come with guarantees of more than one year, and transactions in the interbank market, on account of the flexible and dynamic nature of this market’s cash management features (negotiability of underlying instruments, the option for early repayment, and securitization operations).

Plans for Improvements

Work is nearly finished on regulations setting out the statistical obligations of reporting enterprises, and steps are also being taken to develop the collection of data on stocks and to streamline certain reporting procedures. Expanded stock surveys are primarily designed to correct flow data, the quality of which has deteriorated.

Since 1989, stock surveys have been introduced primarily for French direct investment abroad and for portfolio investment. These surveys will be expanded to include foreign direct investment in France and other claims and liabilities of the private nonbanking sector, as well as commercial claims and liabilities. Surveys of direct investment will make it possible to identify reinvested earnings, which are currently not available from payments records. Data on changes in stocks of trade credits will replace the imperfect data derived from a statistical model for reconciling payments and shipments of merchandise. This model has been based on the financial arrangements reported on customs declarations, which will cease to be collected from the beginning of the unified European market in 1993.

Reporting procedures, specifically those for transactions passing through two resident intermediaries, are to be streamlined by making responsibility for the reporting rest with just one transactor. More precise statistics are to be introduced on transactions in financial derivatives (included under portfolio investment, as of 1991) and all negotiable credit instruments.

Germany

Main Features of Capital Account Compilation

Long-Term Capital Transactions

Direct Investment

Almost all German data are taken from the foreign payments statistics. Data on reinvested earnings, which are compiled from balance-sheet information, are an exception. In the broad sense used in the statistics, “payments” cover settlements on a gross basis and certain transactions not associated with a payment, such as the acquisition of equity capital through the provision of physical assets and rights. The figures are available for individual transactions, including those encompassed by a net settlement.

Portfolio Investment

The recording of securities transactions with nonresidents is based on transaction reports. Only transactions between a resident and a nonresident are recorded. Because domestic banks act as intermediaries in most securities transactions and because domestic customers may not know whether the party with whom they conduct a transaction is a resident of Germany, the banks must report all transactions with nonresidents, those for their own account and those for the account of their domestic customers. Domestic nonbanks have to report only those securities transactions with nonresidents that take place without the intermediation of domestic banks. Recorded are the transaction values according to the banks’ settlements. These figures include commissions and fees, as well as the accrued interest. In the case of portfolio investment, German investment abroad is attributed to the country in which the debtor or issuer of the securities is domiciled. Foreign investment in German securities is classified to the country of the partner in the purchase or sale transaction.

Other Transactions

Most of the figures on loans and other credits come from the foreign payments statistics. However, data from stock reports are used to derive the credits and deposits of banks. The data on other investment are chiefly derived from the foreign payments statistics.

Short-Term Capital Transactions

Banks have to report monthly the stocks of their external assets and liabilities, broken down by type, maturity, economic sector, partner country, and currency. Nonbanks—enterprises, individuals, and public authorities—have to report all external assets and liabilities (excluding securities) if the sum of the assets or the liabilities exceeds DM 500,000. Changes in stocks derived from these reports are adjusted for large changes in coverage and, in the case of banks’ external assets and liabilities, for changes resulting from exchange rate changes and other factors that do not reflect transactions.

Existing Problems as Seen by National Compilers

In principle, the German collection system is based on reports submitted by resident transactors, but in certain areas—for example, securities transactions and outgoing payments by payment order—domestic banks play an important role in transferring data on their customers’ cross-border transactions to the Deutsche Bundesbank (DBB). The reporting of international capital transactions of private individuals is incomplete. Specifically, the following problem areas can be identified:

  • —Transactions through new intermediaries, such as independent fund managers. Fund managing has until recently been effected predominantly by banks, whose statistical records were good. Currently, independent fund managers are attracting private customers.

  • —Transactions through foreign banks. Private and institutional investors have shifted their portfolio activities abroad and maintain custody accounts with banks in Luxembourg, London, and Zurich. German private investors, who are required to report balance of payments transactions affecting these accounts, often fail to report.

  • —Transactions in international financial centers. Large institutional investors purchase and sell securities in foreign markets. Depending on the settlement techniques, transactions may not be reported, and even double reporting is possible. Also, German investors use international financial centers, even when buying and selling German securities. Statistical distortions can easily occur.

  • —Financial innovations. In the beginning, the statistical coverage of new financing techniques might be partially incomplete. It usually takes extensive research to adjust the reporting rules to the new techniques.

Plans for Improvements

As a first step to improving balance of payments compilation, the DBB is reviewing new channels and instruments and encouraging institutional investors as well as banks to improve coverage. In addition, the DBB is adjusting its collection system to the newly emerging instruments and techniques, either by agreements with important institutional investors and banks or by changing reporting regulations (for example, in the case of financial derivatives). Third, the DBB is investigating whether statistics available from other countries or international institutions could be used to close gaps in its compilation system. In addition, the DBB plans to expand its activities in comparing balance of payments statistics at a bilateral level.

Italy

Main Features of Capital Account Compilation

The process of liberalizing capital movements, completed in 1990, was a major event for the balance of payments data collection system. Before the new exchange regime came into force in October 1988, balance of payments statistics in Italy were obtained as a by-product of the exchange control system. For control purposes, residents were required to channel transactions with nonresidents through the banking system, which in turn was required to transmit information on the transactions to the Italian Exchange Office (UIC). The latter maintained data banks, from which nonresidents could be identified and each transaction tracked from start to finish.

Exchange control made it possible to maintain reliable statistics for balance of payments purposes. Nonetheless, there were some problems regarding capital movements:

  • —The covert export of capital. In the first half of the 1970s, large capital outflows in conjunction with an expected devaluation of the lira were disguised as current payments (for example, by over-invoicing tourism expenses and imports). Some of these outflows were in the form of Italian bank notes. Evidence suggests that they were used for financial investments abroad, especially in Switzerland, and were included in the balance of payments as capital flows.

  • —Capital flows associated with other illegal activities. No estimates were included in the balance of payments Most of these outflows were probably directed to Switzerland and various offshore centers.

  • —Reinvested earnings. Data were not available on a regular basis from the UIC, but occasional estimates were based on the outstanding stock.

After the liberalization process, the old system became inadequate and a new procedure for collecting data was implemented. The new procedure, known as Eleuteria, became operational in December 1990.

The new statistical system endeavors to overcome some of the drawbacks arising from the exchange control system, in particular, the requirement to channel all transactions through an Italian bank. The new legislation has three major prescriptions. First, for statistical purposes, all residents are required to transmit to the Foreign Exchange Office data on foreign transactions exceeding Lit 20 million. For nonbank residents this is done by filling out a foreign exchange statistical form. Second, the UIC may no longer maintain a central data bank that includes the name of the transactor. Instead, each statistical return is subject to a formal check of its completeness, correctness, and congru-ency. In the event of errors or omissions, the UIC will contact the bank or the transactor directly, but once the error or omission has been corrected the name of the transactor must be erased from the return. Third, no direct information on the holdings of foreign assets and liabilities of nonbank residents can be collected by the UIC.

The new legislation authorizes the UIC to collect information concerning settlements. The Bank of Italy compiles the balance of payments accounts on a transactions basis, the principal source for which is the data collected by the UIC.

The banking system provides information on both its assets and liabilities (broken down by currency, maturity, and partner country) and its customers’ transactions (broken down by type, currency, and partner country). The information conveyed through the banking channel is part of a more general reporting framework for banking operations. Information is collected from all banks and relates without exception to all transactions.

The exchange rate used for converting capital flows into lire is the average official foreign exchange rate for the month in which the flow takes place. Stocks are valued at market prices and converted at the prevailing exchange rate.

Important methodological changes introduced with the new collection system include:

  • —Direct investment. The transactor has to indicate whether a transaction is connected with a lasting interest or an active role in the management of the direct investment enterprise. The investor is also asked to specify the percentage of equity capital held as a result of the transaction; all investments resulting in a share of more than 20 percent are treated as direct investment, regardless of the investor’s reply to the first question.

  • —Trade credits. For a given trade credit, the new statistical form includes questions on the amount of future payments or receipts connected with that trade credit. When the replies are correlated systematically with settlement and transaction dates, it is possible to produce reliable and timely data.

Existing Problems as Seen by National Compilers

The new system does not fill all the gaps left by the old system of collecting data on capital movements. In fact, it creates a number of new ones.

  • —The new legislation on capital movements requires nonbank operators to fill in the statistical form only for operations exceeding Lit 20 million.

  • —Transfers of financial instruments carried out without an underlying payment (with the sole exception of compensations) are no longer recorded by the UIC.

  • —Information regarding stocks is collected by the UIC only from authorized intermediaries and the central bank. When exchange controls were in force, all Italian securities owned by nonresidents and foreign securities bought by residents had to be deposited in an Italian bank, so that it was possible to collect reliable data on stocks. Now that this regulation has been abolished, stock data have to be estimated by cumulating flows.

  • —Data on reinvested earnings are not collected by the UIC.

Since the new data collection system has been in operation only for a short period of time, it is not possible for the Italian compilers to assess all the problems still to be solved. At the moment three main issues are under discussion:

  • —Whether to classify new financial instruments (such as futures and options) as services or as capital flows.

  • —The classification of flows connected with Italian residents’ deposits with foreign banks. When deposits constitute working accounts opened to settle current payments, they should not be considered financial assets.

  • —Foreign securities recently admitted for trading on Italian exchanges. Certain conventional procedures are being adopted to measure the daily transactions between residents and nonresidents, but it is difficult to record them all.

Plans for Improvements

No major reassessments of the new system are envisaged in the near future. Significant improvements in the quality and comparability of data could be achieved by international agreements on the harmonization of statistical forms.

Japan

Main Features of Capital Account Compilation

Japanese balance of payments statistics are compiled under the Foreign Exchange and Foreign Trade Control Law. Data are derived from reports submitted by authorized foreign exchange banks and designated securities dealers, as well as by other residents who are subject to the reporting requirements of the law. The Ministry of Finance has authorized the Bank of Japan to collect the reports and to compile the balance of payments statistics.

In principle, settlements of cross-border transactions have to be made through the foreign exchange banks.

Residents who enter into transactions with nonresidents are required to report to the authorities by way of the foreign exchange banks, which are required to ascertain the purpose of a money transfer by residents.

Designated securities dealers and other authorized institutions are required to submit reports on portfolio transactions for their customers’ accounts and for their own accounts. Individual investors directly entering into transactions with foreigners must report each transaction.

For direct investment, investors (for outward investment) or their agents (for inward investment) have to submit to the authorities a notification of their intentions, which are used to verify the payments records.

In these varied ways, almost all capital transactions are captured and entered into the balance of payments statistics.

Existing Problems as Seen by National Compilers

Compilers in Japan generally follow the concepts and principles of the Fund’s Balance of Payments Manual but deviate from these concepts and principles in some instances.

  • —The Japanese balance of payments accounts do not fully capture the activities of direct investment enterprises, including their reinvested earnings. In order to compile data on these activities, it would be necessary to impose additional reporting requirements, but, given the trend toward the liberalization of transactions and simplification of reporting forms, it is not easy to introduce more requirements.

  • —Borrowings by Japanese banks and other parent companies from their overseas subsidiaries are classified as increases in “other long-term liabilities” not as reductions in direct investment. Similarly, short-term capital flows between parent companies and their subsidiaries are recorded as “other short-term capital.”

  • —There are large discrepancies in the figures for nonbank deposits in the Japanese balance of payments and in the banking statistics of the Bank for International Settlements (BIS) and the IMF.

  • —For the geographical allocation of outward portfolio investment, the Japanese compilers do not use the debtor principle because they find it difficult getting information about the ultimate debtor.3 It would impose a great burden on securities dealers and other reporters of portfolio transactions to compile securities information on this basis.

Plans for Improvements

Given the trend toward the liberalization of transactions and simplified reporting, it is difficult to obtain from additional surveys data on the reinvested earnings of direct investors. In 1987, the Bank of Japan conducted a small survey to estimate these data. The Ministry of International Trade and Investment (MITI) has been conducting annual surveys, but they do not cover all industries. The Ministry of Finance and the Bank of Japan are planning to use the MITI data, together with data from other sources, to derive estimates of reinvested earnings.

The discrepancies between the figures for nonbank deposits published in the Japanese balance of payments statistics and those published in the banking statistics of the BIS and the IMF have been examined. Possible reasons for these discrepancies are (i) the inclusion in the BIS and IMF statistics of deposits related to gen-saki (repurchase) transactions by securities dealers, which are omitted from the Japanese statistics on non-bank deposits and (ii) the recording in the Japanese balance of payments of some nonbank deposits according to their ultimate purpose, for instance, as direct investment. However, there may be other reasons for the discrepancies, and the Bank of Japan will continue to investigate them.

Thus far, the Bank of Japan has not reported to the BIS data on the stock of portfolio investments because of the difficulty of their disaggregation. However, the Bank of Japan is reviewing the possibility of improving the reporting system without additionally burdening the reporting banks.

In addition, the Ministry of Finance and the Bank of Japan have made efforts to reduce the size of the errors and omissions that have emerged in the Japanese balance of payments statistics for recent years. They have done so by monitoring the reporting institutions and by clarifying the treatment of new financial products.

Netherlands

Main Features of Capital Account Compilation

All components of the Dutch balance of payments, including the capital account, are compiled from information the Nederlandsche Bank receives about transactions between residents and nonresidents. For this purpose, all residents (both banks and nonbanks) are obliged to report on accounts held with nonresidents. This reporting obligation was laid down in Section 13 of the External Financial Relations Act.

The system requires monthly information to be given about the opening and closing balances on external accounts as well as about the underlying transactions that explain the changes in these balances. Both balances and transactions are denominated in the transaction currency. Since the total change in the external account is entered in the balance of payments, a contra-entry for the underlying transactions is also made. Therefore, the Dutch balance of payments derived from these data does not include an item for errors and omissions.

In order to keep the system manageable, banks are asked to report the total only of transactions that have a value below the reporting threshold (currently f. 25,000) and that have been settled through banks. This total is allocated by the Nederlandsche Bank to individual balance of payments items by using a fixed distribution scheme.

The system is supplemented by some specific provisions. One provision concerns over-the-counter transactions in foreign currency. Banks report the total of sales and purchases of foreign bank notes. For transactions above a threshold of f. 5,000 the nature of the transaction is specified. All other transactions are classified as travel. In addition, details of transactions between residents and nonresidents that do not involve actual payment must be reported by the resident concerned, who must specify the transactions that have been netted out.

This type of reporting makes it possible to exercise tight control over reporting, since the opening balance on residents’ accounts abroad plus receipts minus payments always has to match the closing balance in these accounts. Moreover, the opening balance of accounts for a given period must be equal to the closing balance for the immediately preceding period.

Further checks are made for direct and portfolio investment by comparing stock data, collected by an annual survey, with flow data included in the balance of payments. The direct investment survey also provides information on reinvested earnings. Checks against stock data are also made for external transactions of domestic banks. The latter submit position reports for each currency on a quarterly basis. The reports show banks’ external claims and liabilities broken down by item, like nostro and loro accounts, time deposits, and loans.

Existing Problems as Seen by National Compilers

The Dutch compilation system has proved to be well equipped to handle increasingly complex and volatile financial flows into and out of the economy. This is basically because the reporting system in fact covers every type of account that residents—banks and non-banks—maintain abroad in a way that permits tight control. The excellent contacts the Nederlandsche Bank maintains with reporting banks and nonbanks also support the high standard of reporting. Serious problems are not perceived.

Plans for Improvement

Improvements in the reporting system are constantly being pursued. For instance, the Nederlandsche Bank is investigating whether ISIN (International Securities Identification Number) codes can facilitate the reporting of detailed information on portfolio investment and whether it is feasible to collect balance of payments data by domestic industrial sector. Furthermore, the process of European integration will require the harmonization of both the compilation and collection of balance of payments data. The Nederlandsche Bank argues that the Dutch compilation system will be able to make an important contribution to the harmonization because of its completeness and flexibility.

Switzerland

Main Features of Capital Account Compilation

Swiss data are taken from regular surveys of bank and nonbank companies. Banks are obliged by the federal law on banks and saving banks to report regularly on their operations with nonresidents, but nonbank companies are not obliged to participate in the survey. Experience shows that the voluntary character of reporting is often a drawback, affecting the coverage of the survey.

For the purpose of compiling the capital account of the Swiss balance of payments, it was decided to treat natural and legal entities whose center of interest lies in Switzerland or in Liechtenstein as residents. That decision was made after Liechtenstein declared the Swiss franc its official currency, following the currency agreement of 1980 between the two countries. Banks in Liechtenstein have to report to the Swiss National Bank in the same way that Swiss banks do. However, nonbank companies in Liechtenstein are not included in the balance of payments surveys because the Swiss National Bank is not authorized to include them.

Direct Investment

Banks and nonbanks are required to report their direct investment transactions if the value of their investment capital exceeds Sw F 10 million. The outward direct investment survey covers 300 companies; the inward survey uses data from 600 companies. Neither of the two surveys includes all companies that would be required to report on the basis of the Sw F 10 million threshold. Some companies refuse to answer because of the voluntary character of the survey; others are not known to the compilers.

Transaction values for short- and long-term intercompany credit are derived from amounts outstanding at the beginning and end of the reporting period. Short-term credits are equal to the difference between the two outstanding amounts; long-term credits are equal to the difference between the valuation-adjusted amounts outstanding. Bilateral comparisons of stocks of direct investment show lower values for assets and liabilities in the Swiss statistics than in foreign sources.

Portfolio Investment

The primary data sources for portfolio investment are reports of banks on the long-term security holdings (in their security and custody accounts) that they manage for their clients. Management responsibilities may vary from bookkeeping to the purchase and sale of securities. Reports of banks and nonbank companies on their own holdings of securities form a secondary data source.

Plausibility checks show that the total stock of securities issued by Swiss residents to nonresidents were reported fairly comprehensively. No such checks can be applied to securities that are issued by foreigners and acquired by Swiss residents.

Securities holdings are usually reported at their market value. If a market value is not available, the face value is reported. Portfolio flows are calculated by adjusting the change in holdings for changes in market prices and in foreign exchange rates.

Transactions of Commercial Banks

Information on short- and long-term credit transactions of banks and on fiduciary operations on behalf of residents is drawn from monthly balance-sheet and off-balance-sheet reports by resident offices of banks in Switzerland and Liechtenstein whose balance-sheet total exceeds Sw F 100 million. Fiduciary transactions are an off-balance-sheet business, whereby resident banks channel resident and nonresident funds abroad in the name of the bank. Risk and return are entirely for the account of the nonbank investors. Changes in the fiduciary accounts of nonresidents reflect transactions between nonresidents and are therefore excluded from the Swiss capital account. Fiduciary investments are usually made in the form of time deposits. At the end of 1990, the value of the stock of fiduciary accounts was Sw F 377 billion, of which Sw F 94 billion were held in the accounts of residents.

Assets and liabilities of banks as well as those held in fiduciary accounts are reported in Swiss francs. Assets and liabilities in foreign currencies are converted at the prevailing exchange rate. Generally, assets and liabilities are recorded at face value. However, the banks are allowed to offset provisions for bad debts directly against the assets concerned. Credit and fiduciary capital flows of assets and liabilities denominated in foreign currencies are calculated by adjusting the change in holdings for changes in foreign exchange rates.

Data on cross-border trade in gold and silver bullion are taken from customs records. In the Swiss balance of payments, purchases and sales of gold and silver bullion by banks are treated as financial transactions and are therefore included in the capital account.

Transactions of Nonbank Companies

Data on the capital transactions of nonbank companies are taken from company reports on foreign assets and liabilities outstanding. As mentioned above, flows of assets and liabilities denominated in foreign currencies are calculated by adjusting the change in holdings for changes in foreign exchange rates. Companies are required to report their position if the sum of their holdings of foreign assets and liabilities exceeds Sw F 10 million.

Existing Problems as Seen by National Compilers

Method for Compiling Flow Data

For the purpose of deriving flow data on assets and liabilities denominated in foreign currencies from the stock data, Swiss compilers make assumptions that affect the calculation. The assumptions concern the composition of the stocks, the exchange rates and stock exchange indices to be used for adjusting the stock data, and the comparability over time of the coverage of the reported data. If the assumptions are not realistic, the results will be distorted.

Coverage

Nonbank companies in Liechtenstein are, for reasons mentioned above, not included in the surveys. In addition, as a result of the absence of exchange control, many companies holding foreign assets and liabilities are not known to the compilers and are therefore not included in the surveys.

Direct Investment

Transactions between Swiss insurance companies and their nonresident branches are not included in the capital account because the insurance companies cannot distinguish between reserves attributable to their own operations and those attributable to their branches. Furthermore, experience has shown that reinvested earnings are sometimes distorted by valuation changes. Data on the purchases and sales of real estate abroad by Swiss residents are not available.

Commercial Banks

Transactions between Swiss commercial banks and the Bank for International Settlements (BIS) are treated as transactions between two residents. They are, therefore, not recorded in the capital account. In addition, information on transactions in precious metals between Swiss commercial banks and nonresidents is not available if the metal does not cross the Swiss border.

Other Transactions

Data on the transactions in short-term securities of private households are not available nor are data on changes in the holdings of Swiss franc bank notes by nonresidents.

Plans for improvements

Balance of payments compilers plan to ask companies to report data on transactions for all components of direct investment. The Swiss National Bank plans to collect data on transactions in portfolio instruments. Plans also include treating the BIS as a nonresident bank.

Finally, plans are to include in the capital account those changes in assets and liabilities of resident non-banks Vis-à-vis nonresident banks that can be derived from the Eurocurrency statistics of the BIS.

United Kingdom

Main Features of Capital Account Compilation

The U.K. balance of payments account is conceptually identical to the overseas sector account within the framework of the national economic and financial accounts. Much of the information used for the overseas sector account is prepared in conjunction with that for the domestic accounts. This is both a strength and a weakness. It is a strength because overseas transactions are seen in conjunction with domestic transactions as part of an overall economic and financial framework. It is a weakness to the extent that the domestic economic and financial accounts use a variety of data sources with no single unifying principle.

Although the U.K. system has been characterized as a “survey” system, in contrast to the international transaction reporting systems (ITRS) used in many other countries, there is considerable variety in its data sources, some of which may not be recognized as surveys. Also, it needs to be recognized that the United Kingdom has not (or at least not yet) pursued the system to its logical conclusion and has accepted a degree of incompleteness, at a time when financial markets have become increasingly sophisticated, both in its coverage of some overseas transactions (such as trade credit) and the surveys it has carried out.

The guiding principles of the U.K. system, which may not yet be observed fully, are

  • —Using general government accounting records on holdings of or transactions in U.K. public sector securities maintained by the central bank;

  • —Requiring U.K. corporate bodies to report transactions in foreign assets and nonsecuritized liabilities with nonresidents;

  • —Using counterpart data from other countries on transactions of U.K. residents (other than the general government and banks) with nonresident banks;

  • —Dealing with transactions in the security liabilities of U.K. companies by a combination of reports from corporate bodies (direct investment), surveys of registers (for registered securities, such as equities), and by residual estimates (for bearer securities, such as bonds).

Table 1 identifies the key features of each part of the accounts.

Table 1.U.K. Statistical System for Recording Assets and Liabilities for Capital Account Categories, by Domestic Sector
Other capital
Portfolio investmentOther domestic sectors
Direct investmentLiabilitiesU.K.Vis-à-visOther
SectorAssetsLiabilitiesAssetsSharesBondsbanksforeign banksAssetsLiabilities
Central governmentBCDDDF
Local authoritiesDDDDF
Public corporationsADAA
Banks and building
societiesAAABCFABF
Other financial institutionsAAABCFEAAF
Industrial and commercial
companiesAAB/ABCFEBF/A1AF1
Personal2D3BE44
Guide to sources:A: Survey of U.K. enterprises (for own account)B: Survey of U.K. enterprises (as agent)C: Register survey D: Government accountsE: Non-U.K. sourceF: Derived as residual, using market data on net issues and transactions by domestic sectors.Note: Where two indicators are shown separated by a slash (/), the second is in the course of replacing the first.

May remain incomplete with respect to much trade credit.

Present estimate for property is speculative. Future estimates may rely on B (estate agents) or household surveys.

Based on taxation records.

Assumed to be negligible.

Guide to sources:A: Survey of U.K. enterprises (for own account)B: Survey of U.K. enterprises (as agent)C: Register survey D: Government accountsE: Non-U.K. sourceF: Derived as residual, using market data on net issues and transactions by domestic sectors.Note: Where two indicators are shown separated by a slash (/), the second is in the course of replacing the first.

May remain incomplete with respect to much trade credit.

Present estimate for property is speculative. Future estimates may rely on B (estate agents) or household surveys.

Based on taxation records.

Assumed to be negligible.

Existing Problems and Plans for Improvements

The main problem areas in the U.K. accounts, some of which are in the process of being resolved, are the following.

Incomplete Reporting by the Corporate Sector

The long-term objective of balance of payments compilers is to cover all parts of the corporate sector using reports on overseas transactions in which U.K. companies act as principals, thus reducing reliance on suspect data from intermediaries. Substantial extensions of coverage have been made recently with regard to securities dealers (from the end of March 1989) and nonfinancial companies (from the end of December 1990). Information from the latter led to substantial revisions in the estimated holdings of overseas portfolio assets by nonfinancial companies in the 1991 United Kingdom Balance of Payments (or the Pink Book). Further extensions in the financial sector are planned. Also, improvements are being made in some financial sectors already covered, such as private pension funds. Some components, like trade credit between unrelated companies, are not yet covered.

Personal Financial Transactions

So far, obtaining data from surveys of individuals, possibly as part of wider household surveys, has not been regarded as cost effective, and it has been assumed that personal transactions are either negligible or can be measured by surveys of intermediaries. Some work is being considered to improve the estimates of overseas property transactions by individuals through surveys of estate agents. Also, the use of a household survey to measure the overseas transactions of individuals, including transactions in securities and property, is being reconsidered. Transactions in overseas securities (at present measured very imperfectly via surveys of securities dealers and banks) may be more effectively (although still imperfectly) measured through surveys of fund managers’ private client business, which are under consideration.

Definition of the United Kingdom

As transactions by residents in offshore locations in the United Kingdom (Channel Islands and Isle of Man) are very poorly covered outside banking, and constitutional arrangements limit severely the scope for improvements, the practical implications of redefining the United Kingdom, for balance of payments purposes, to exclude offshore United Kingdom are now being explored.

International Banking Statistics and Domestic Data

The United Kingdom uses the IMF’s international banking statistics to measure U.K. nonbank private sector transactions with banks abroad and it uses domestic sources on the transactions of U.K. banks and U.K. public sector entities with banks abroad. Differences in the scope and coverage of international versus domestic banking statistics—for example, the treatment of nonbanking affiliates of banking groups—are potential sources of imbalances in the account. This may be an area where the Report on Capital Flows can play a part in producing more consistent data.

Inward Portfolio Investment

The compilation of data on overseas investment in U.K. equities has been improved by the introduction of an annual share register survey (from the end of 1989). The survey has helped correct biases apparent in earlier data provided by financial intermediaries. Investigations are continuing into ways of improving estimates by inquiries to fund managers.

Overseas investment in U.K. company securities other than equities has been derived, since March 1990, as a residual from total net new issues obtained from market sources and net acquisitions reported by domestic sectors. These estimates should be improved as reporting by corporate sectors improves. Improvement in the recording of overseas investment in government securities through further refinement of register information is being examined.

Sectorization of Short-Term Liabilities

The extension of corporate reporting has improved the coverage of nonsecuritized short-term liabilities to overseas residents. The sectorization of short-term securities—such as U.K. banks’ certificates of deposits and commercial paper—remains a problem area. The examination of residuals between net new issues and net acquisitions by domestic sectors is being explored and has led to substantial revisions in the 1991 Pink Book.

Chancellor’s Initiative on Economic Statistics

The report of a year-long review of balance of payments statistics, carried out as part of the Chancellor’s Initiative on Economic Statistics, was published in November 1991. In general, the developments described above are reinforced by its recommendations. An action plan has been prepared and is currently being implemented. Among the items being considered are

  • —The wider use of existing statutory powers to secure a full response to surveys;

  • —The transfer to the Central Statistical Office of some data collection previously undertaken by the Bank of England;

  • —An investigation into the scope for publication of a supplementary statement showing balanced accounts; and

  • —A study to determine the possibility of estimating external trade credit.

United States

The U.S. capital account data are mainly collected via quarterly surveys of transactors for direct investment and monthly data from financial intermediaries. The system relies on mandatory reporting for large transactors and intermediaries on the assumption that the total transactions of nonreporters do not constitute a significant proportion of the universe of transactions. Because capital flows are quite volatile, with frequent sign reversals, quarterly estimates of capital flows are based on reported data only. Benchmark surveys, which use lower reporting thresholds and more aggressive compliance efforts than the quarterly surveys, are conducted periodically to assess sample reporting relative to the universe. Results from the benchmarks are used to change instrument coverage, reporting thresholds, and sample selection.

Until recently this system of mandatory reporting by major transactors worked relatively well, but growth, innovation, and integration in international capital markets, combined with a reduction in resources, has resulted in a significant decline in the quality of U.S. Capital flow data. At the same time, increased integration of capital markets has underscored the importance of accurate data for economic policy.

Direct Investment

The Bureau of Economic Analysis (BEA) collects data on direct investment flows through various mandatory surveys. Benchmark surveys are conducted about every five years to supplement quarterly sample surveys. Annual and benchmark revisions of data on direct investment suggest that unreported quarterly flows are positive and statistically significant, and may have contributed to the persistent, unrecorded net Capital inflows that occurred in the United States during the 1980s.

The persistent underestimation of foreign direct investment in the United States reflects several factors:

  • —The large successive upward revisions to the first quarterly estimate resulting from nonresponses at the time of the original estimate and the longstanding policy of not providing estimates on capital flows for nonreported data;

  • —The absence of Capital flow estimates for those firms not required to report on a quarterly basis; and

  • —The difficulty of identifying changes in the universe on a timely basis.

Similar problems affect U.S. direct investment abroad, but the magnitude of the resulting error is much smaller. The following are the more important methodological and statistical problems the BEA is trying to address in the direct investment area.

Benchmarks

Preliminary comparisons between data reported to U.S. tax authorities and to the U.S. Bureau of the Census suggest that the coverage of the benchmark survey is fairly complete. The BEA intends to develop more complete reconciliations with data reported to the tax authorities and the Census Bureau in benchmark years and is already working on more comprehensive tracking systems for new investments.

The BEA is developing quarterly estimates at the universe level for the capital flows of those firms exempt from quarterly reporting. It does so by extrapolating benchmark position data forward based on data for those firms reporting quarterly. If tests of these techniques are satisfactory, these estimates will be included in the published quarterly capital flow estimates. Similarly, the BEA is developing quarterly estimates for enterprises that do not respond in time by using data reported for previous quarters and benchmark data.

Quarterly Surveys

The BEA has a program under way to improve compliance with the quarterly and annual surveys. Also, the BEA is improving on-line computer processing procedures in order to expedite processing and allow for the inclusion of data that are received too late to be processed under the current system.

Problems with the less than complete coverage of the quarterly surveys may be especially acute for real estate investments. Personal real estate investments are not covered at all and many commercial real estate transactions may not be reported, either because they are through limited partnerships, which are not covered, or they may be too small, and therefore exempt from reporting. The BEA hopes to explore alternative data sources and methods for real estate investment.

Portfolio and Other Investment

Most of the data on portfolio and other capital flows are collected through the Treasury International Capital (TIC) reporting system. TIC data are collected on purchases and sales of long-term U.S. and foreign securities, banks’ claims and liabilities and banks’ custody transactions, and nonbanks’ claims on and liabilities to unaffiliated foreigners. (Other data on official reserve assets and data on other U.S. government assets and liabilities are collected by the BEA from U.S. government agencies.)

Data are collected by the Department of Treasury in three separate sets of mandatory surveys. In general, the TIC system concentrates on collecting data from a limited number of major reporters on the assumption that most transactions flow through established financial channels. No estimates are made for nonreporters or for reporters whose transactions are below the mandatory reporting thresholds. Only the survey of long-term security transactions is subject to regular benchmarking.

Problems in the U.S. estimates of portfolio capital flows began to emerge in the 1980s, as an increasing volume of portfolio transactions flowed outside established financial channels and involved innovations, such as around-the-clock trading in foreign securities markets, reliance on international fund managers, direct foreign placements, and new financial instruments. Many of the new participants were difficult to identify or were unfamiliar—or unwilling to fully comply—with U.S. statistical reporting requirements for the increasingly complex international transactions. Toward the latter part of the 1980s and the early 1990s, problems in the U.S. financial system may have exacerbated these reporting problems. Troubled banks and financial firms, especially those involved in bankruptcy procedures, place statistical reporting far down on their list of priorities.

In addition to the effect of innovation, the sheer growth in U.S. gross financial transactions relative to the net financing needs of the current account have probably contributed to the statistical discrepancy between the current and the capital accounts. By 1989, foreign purchases and sales of long-term securities in U.S. markets were $5,000 million compared with current account purchases and sales of less than $1,000 million. Even if all other capital account transactions had been recorded perfectly, a 1 percent error in recording foreign purchases and sales of long-term securities would have caused a $50 billion statistical discrepancy between the U.S. current and capital accounts.

The following are the more important methodological problems the BEA is trying to address in the portfolio investment area.

Benchmarks

The last benchmark survey of U.S. portfolio investment abroad was conducted in 1943. The Treasury Department and the BEA are planning to conduct another benchmark survey of U.S. portfolio investment abroad in 1993. In order to collect accurate data on the debtor-creditor basis and to capture direct placements, the United States is attempting to coordinate this benchmark survey with several of its major investment partners.

Because of resource constraints, the Treasury Department has been slow to process benchmark surveys of foreign portfolio investments in the United States. The first preliminary data from the 1984 benchmark survey were not incorporated until the balance of payments data were revised in June 1990. In order to expedite the benchmarking process, the BEA has provided funds to the Treasury Department to help with the 1989 benchmark survey of foreign investment in long-term U.S. securities. Benchmark surveys of stocks of foreign investment in long-term securities in the United States are conducted roughly every five years and are based on surveys of U.S. issuers of securities and U.S. “holders of record.” The surveys are quite comprehensive but still focus on a relatively limited number of nominees and do not constitute a complete census.

Quarterly Surveys

Data on the transactions of U.S. nonbanking concerns with unaffiliated foreigners are collected mainly through the Treasury Department’s quarterly TIC-C survey (a special form of the TIC survey). No estimates are made for nonreporters; as a result, there may be significant errors and omissions in measured capital flows in this area. Coverage of nonbanking firms’ transactions with unaffiliated foreigners appears to be particularly low; in 1988, only 475 reporters filed quarterly TIC-C forms. Given the universe of exporters and importers that conduct direct transactions with foreigners, and the prevalence of direct placements of foreign financial paper in the United States, this figure seems quite small. Based on transactions reported in the BEA’s surveys, adequate coverage on the TIC-C survey would require more than 5,000 reporters. Enterprises that are not being surveyed may account for sizable transactions. Altogether, the assets of U.S. private nonbanks in foreign banks may be underestimated by as much as $220 billion, and U.S. nonbank liabilities to foreign banks may be underestimated by as much as $160 billion, based on comparisons with international banking data.

The BEA and the Treasury Department are working on a data-sharing agreement that will allow the BEA to provide information on direct investors—mainly large firms with extensive international transactions with unaffiliated as well as affiliated foreigners—to the Treasury Department with the aim of improving the coverage of the TIC-C form. Until such time as improvements in the TIC are in place, the BEA is exploring the use of bilateral banking data to estimate U.S. nonbank claims on and liabilities to foreign banks in selected countries.

Data on banking transactions between U.S. private residents and nonresidents are collected through the TIC-B survey of month-end closing balances for claims on, or liabilities to, foreigners. Overall coverage of U.S. banks appears good. Two difficulties with the banking reports are the poor quality of some responses and inadequate reporting of custody accounts. Also, periodic reconciliations of TIC-B data on foreign-owned banks with similar Federal Reserve regulatory data suggest increasingly large discrepancies.

In recent years, foreign-owned banks operating in the United States have played an increasing role in financial markets in the United States and abroad. There is some concern that the quality of reporting by foreign-owned banks is not as high as that of U.S.-owned banks. One of the proposed joint projects of the BEA and the Treasury Department is a series of on- site “educational” visits to inform and educate foreign-controlled banks on reporting requirements, to assess past reporting, and to encourage better compliance.

Data on transactions in long-term securities between U.S. private residents and unaffiliated nonresidents are collected through the monthly TIC-S survey of U.S. broker-dealers. Problems in recording securities transactions include the coverage of direct purchases and sales of securities through foreign rather than U.S. financial intermediaries, the growth in new instruments, the channeling of transactions through new intermediaries, and the identification of nonresidents.

The BEA and the Treasury Department have initiated a project to improve coverage of transactions in short-term commercial paper and other instruments, as well as a project to improve coverage of pension and mutual funds engaged in direct portfolio transactions with nonresidents.

The net effect of these statistical problems may be quite large, but the absence of benchmark surveys on U.S. portfolio investment abroad and the lack of a full census for the benchmark surveys of foreign portfolio investment in the United States make it difficult to assess the magnitude of the errors and omissions. So far, the annual portfolio flows for U.S. holdings of foreign assets in the 1980-90 period were revised by $5.4 billion. The revisions reflect the incorporation of late reports, the correction for errors and omissions in the reported data, and the incorporation of the results of two benchmark surveys of foreign portfolio investment in the United States. However, these were partial revisions and may still not fully represent the true values.

The country reports were edited by Arie C. Bouter and Patterson, who both served as members of the technical staff.

    Other Resources Citing This Publication