Reconciliation of the U.S.-Canadian current account.

DiLullo, Anthony J. ; Laliberte, Lucie

This article presents a detailed description of the reconciliation of the bilateral current-account estimates of Canada and the United States for 1990 and 1991. Such reconciliations have been undertaken by Statistics Canada and the Bureau of Economic Analysis (BEA) each year since 1970.(1) They were initiated because of the extensive economic links between the two countries and the need to explain differences in the Canadian and U.S. published estimates of the bilateral current account. In principle, the bilateral current account of one country should mirror the bilateral current account of the other country.

The reconciliation process has resulted in greater accuracy of the published estimates of transactions between Canada and the United States and in increased efficiency in producing the estimates. These improvements have been accomplished through the exchange of data between the two countries and the development of improved estimating techniques.

Over 80 percent of the data used by Canada and the United States to compile Canadian-U.S. current-account estimates is now provided through the exchange of data such as the following: U.S. and Canadian merchandise imports (beginning January 1990); services such as travel, passenger fares, inland freight, and government expenditures; unilateral transfers such as pensions; and certain Canadian and U.S. data banking data used to estimate interest income. The exchange of data was undertaken in some instances because either the Canadian or the U.S. data were substantially more comprehensive and complete; an example is the exchange of each country's merchandise import data, which were found to be more complete than the other country's export data. In other instances, the exchange of data was aimed at eliminating duplication of effort, such as the use of the same source data to develop estimates of bilateral travel and passenger fare transactions. The elimination of the processing of export declarations resulted in reductions in some government processing costs, and the elimination of duplicate surveys of U.S. and Canadian travelers resulted in reductions in respondent burden.

The reconciliation process has also resulted in improvements of estimates that are not based on the exchange of data. For example, U.S. compilers revised the methodology for developing some U.S. estimates of portfolio income receipts from Canada, and Canadian and U.S. compilers each effected improvements in Canadian and U.S. source data for a number of service transactions.(2)

Revisions based on the reconciliations are incorporated into the Canadian And U.S. estimates as far as possible. Complete exchange of data or substitution of reconciled estimates for published estimates is not feasible because of definitional and methodological differences and because estimates of transactions with third countries would be affected in some cases. In addition, for some estimates, protecting the confidentiality of source data bars the exchange of data.

This article has two purposes. The first is to present a detailed analysis of the differences in published estimates and an explanation of the reconciliation adjustments. The reconciliation process has now evolved to the point where the causes of most differences in published estimates are understood in detail, and a process for reconciling the differences has been developed and tested. In the last few years, considerable progress has been achieved in developing the process, partly resulting from the increased importance given to the reconciliation process by Canadian and U.S. statistical agencies. The second purpose is to present the Canadian-U.S. current-account reconciliation process as a guide for balance, of payments compilers in their bilateral reconciliation efforts and to provide detailed information to users of Canadian and U.S. balance of payments statistics.

Major Types of Reconciliation Adjustments

Reconciliation adjustments to each country's published estimates fall into three broad categories - definitional, methodological, and statistical - which, reflect the differences that occur in the published estimates.

Definitional and methodological adjustments

Definitional and methodological adjustments are required because of differences in definitions and methodologies used to compile the international accounts in Canada and the United States. These differences may reflect each country's requirements to integrate the external accounts with domestic accounts. In other instances, there are differences in conventions used by Canadian and U.S. compilers that are largely due to institutional factors. Choice of one definition or methodology over another in developing reconciliation adjustments does not indicate agreement on what the correct definition should be nor on the most appropriate methodology. Often, the choice is based on Practical considerations, such as the availability of data.

Among definitional differences, the United States includes reinvested earnings as a component of direct investment income, whereas Canada does not; Canada records merchandise imports on a country-of-shipment basis, whereas the United States records them on a country-of-origin basis. To achieve reconciliation, a common definition is selected.

Among methodological differences, one country may classify a group of transactions in an account different from that of the other country, or it may record certain transactions on a gross basis rather than a net basis. To achieve reconciliation, transactions are reclassified to a common account and are adjusted to a gross or net basis, as required.

Statistical adjustments

Statistical adjustments mainly reflect differences in data sources and estimation techniques. In addition, differences in private accounting practices in Canada and the United States affect the type of data available from statistical surveys.

There are four broad types of statistical adjustments. First, some adjustments are based on knowledge about the quality and coverage of source data. When one country's source data are believed to be of better quality than the other country's source data, the better source data are used to develop the reconciled estimates. Second, some adjustments are made because the detailed data needed to make two estimates comparable may be available from one country but not the other. Third, adjustments are made to anticipate revisions in source data that become available after publication of the estimates. Fourth, many adjustments are based essentially on pragmatic factors. When no means of clearly establishing the merits of one country's data are available, reconciled values reflect compromises by the compilers, particularly when the compromise is within a reasonable range of error in measurement. For all types of statistical adjustments, the reconciled values represent the best possible compromise.

Reconciliation of Major Accounts

This section presents a discussion of the reconciliation adjustments made to the major accounts: Merchandise trade, services, investment income, and unilateral transfers. Although numerous adjustments are made, only those that involve important definitional or methodological differences or that are large in size are discussed in detail. Definitional adjustments, such as the exclusion of reinvested earnings and the reallocation of merchandise imports to the country of shipment, affect the reconciled current-account balance, as do the statistical adjustments. Methodological adjustments, such as reclassification and grossing or netting, are necessary to achieve common treatment, but because they are offsetting, they do not affect the current-account balance.

In this discussion, the term "northbound" refers to U.S. receipts, or Canadian payments. "Southbound" refers to Canadian receipts, or U.S. payments. All values are expressed in U.S. dollars.

For 1990, the difference between the latest U.S. and Canadian published estimates of the U.s.-Canadian current-account balance was $3.3 billion (table 1). After reconciliation, the difference was reduced to 0.4 billion. The U.S. reconciled balance was a surplus of 3.1 billion, and the Canadian reconciled balance was a deficit of $2.7 billion (chart 1). For 1991, the difference of $5.5 billion in the published estimates was virtually eliminated; the reconciled balance was a U.S. surplus, or Canadian deficit, of $3.6 billion. The large differences between the published and reconciled balances mostly reflect definitional adjustments made to merchandise trade, transportation and investment income (tables 2 and 3). Adjustments to southbound estimates are larger than adjustments to northbound estimates (tables 4 and 5).

[TABULAR DATA OMITTED]

Merchandise trade

Most of the differences between published U.S. and Canadian estimates of merchandise trade on a balance of payments basis stem from different treatment of the source data.(3) For reconciliation, the main task is resolving those differences in treatment (tables 6 and 7). Four major adjustments are required.

[TABULAR DATA OMITTED]

First, Canadian reexports are added to U.S. estimates of merchandise imports. In the U.S. published estimates, which are on a country-of-origin basis, these imports (Canadian reexports) are attributed to third countries rather than to Canada. A similar adjustment to Canadian imports is not required, because the Canadian published estimates, on a balance of payments basis, are allocated to the country of shipment and thus already include U.S. reexports.

Second, inland freight (freight charges on shipments of export, and imports from the plant to the border) is reclassified in the U.S. accounts from merchandise exports and imports to transportation; this aligns the U.S. treatment with the Canadian treatment.(4)

Third, the repair of equipment reported in the merchandise trade source data is reclassified from services to merchandise trade in the U.S. accounts to align with the Canadian treatment. In the U.S. published estimates, all equipment repairs are included in services.

Fourth, several other balance of payments adjustments are made to the estimates. In the Canadian accounts, the main ones reflect the exclusion, for reconciliation purposes, of valuation differences from Canadian estimates of exports of petroleum and natural gas to the United States and the reclassification of freight on exports of natural gas to transportation. Minor adjustments include the elimination of progress payments for certain military equipment from Canadian imports and the reclassification of processing fees from services to merchandise exports. In the U.S. accounts, most of the other balance of payments adjustments are for timing and valuation differences.

Services

Travel and passenger fares. - Canadian and U.S. published estimates of northbound and southbound travel and passenger fares are based on the same source data and common definitions. Thus, there is no need to reconcile the estimates except to account for timing differences that occur in the publication of revised estimates by Statistics Canada and BEA.

Transportation. - The main differences between Canadian and U.S. published estimates of northbound and southbound transportation transactions are the classification of inland freight and of certain services procured by airlines and railroads.(5) Canadian published estimates of transportation transactions include inland freight and exclude services procured in port by airlines and railroads. U.S. published estimates reflect the opposite treatment: Inland freight is excluded from transportation, and port services of airlines and railroads are included. For reconciliation, inland freight is reclassified from merchandise trade to transportation in the U.S. accounts, and port services of airlines and railroads are reclassified from business services to transportation in the Canadian accounts (tables 8 and 9). Two remaining classification adjustments - both to the Canadian published accounts - include the reclassification of inland waterways tolls from the Canadian northbound transportation account to the Canadian government services account and the reclassification of estimates of aircraft leasing (northbound and southbound) from business services to transportation.

[TABULAR DATA OMITTED]

The single definitional adjustment is the deletion of Canadian estimates of freight receipts southbound) on U.S. exports in transit in Canada. These receipts are reallocated to third countries. In the U.S. treatment, such charges are viewed as payable by the foreign importer (under the balance of payments convention that goods are the property of the importer after they leave the customs frontier of the exporting country) and thus are transactions between Canada and third countries.

The largest statistical adjustment is the reduction of U.S. published estimates of northbound inland freight. The estimates in the Canadian published accounts are believed to be more accurate. In addition, estimates of pipeline freight charges on U.S. imports of natural gas are added to U.S. estimates of southbound inland freight because of undercoverage in the U.S. source data. Finally, U.S. published estimates of freight receipts (northbound) and of payments for port services (southbound) are adjusted to match the Canadian estimates. U.S. published estimates of these transactions are developed on a global basis and then allocated by country and region using indicators such as volume of trade and passenger traffic. For reconciliation (and later for revised U.S. published estimates), Canadian estimates are used because the Canadian source data reflect these bilateral transactions with the United States more accurately than the U.S. source data.

Other services. - Other services are categorized as affiliated transactions, unaffiliated transactions, and government transactions. Affiliated transactions cover transactions between parent companies and affiliated companies. Unaffiliated transactions cover transactions between an other private residents of Canada and of the United States. Canadian government transactions Federal and Provincial) cover transactions with the U.S. Government and U.S. private residents, and U.S. government transactions include transactions with the Canadian governments and Canadian private residents. In addition, royalties and license fees, both affiliated and unaffiliated, are combined with other service charges for reconciliation because the Canadian definition of royalties and license fees is broader than the U.S. definition.

Affiliated transactions are adjusted to remove taxes from the Canadian published estimates (tables 10 and 11). Statistical adjustments to the U.S. published estimates include an increase in receipts and payments of U.S. parent companies for service charges. Statistical adjustments to the Canadian published estimates reflect reductions of the estimate for undercoverage of service charges.

[TABULAR DATA OMITTED]

Unaffiliated transactions are adjusted for reclassification, netting and grossing, and statistical differences. In the U.S. accounts, installation, maintenance, and repair of equipment are reclassified to merchandise trade, and U.S. receipts for medical services covered by Canadian Provincial health insurance programs are reclassified to Canadian government services. In the Canadian accounts, airline and railway port services and aircraft leasing estimates are reclassified from services to the transportation account, and commissions on merchandise trade are removed.(6) For communications, northbound estimates are netted against southbound estimates for comparison and reconciliation. Detailed comparisons cannot be undertaken because of the confidentiality of source data. In addition, Canadian estimates of insurance transactions are adjusted to a net basis: Receipts reflect premiums received less losses paid, and payments reflect premiums paid less losses recovered. Canadian published estimates include premiums received plus losses recovered as receipts, and premiums paid plus losses paid as payments.

Statistical adjustments - mainly to business, professional, and technical services and to sports and entertainment - reflect pragmatic adjustments made to Canadian and U.S. estimates to achieve reconciliation. In most cases, the adjustments involve raising either the Canadian or the U.S. published estimates, whichever are smaller, to the level of the other country's estimates; this approach is taken because the smaller estimates often reflect a narrower coverage in source data.

The differences that remain after reconciliation of unaffiliated services estimates are related to transactions of insurance companies. These transactions could not be reconciled, because of basic differences in accounting conventions and data collection methods for this industry in the United States and Canada.

Reconciliation of government services transactions requires only statistical adjustments that reflect timing differences in the publication of revised estimates. Estimates of Canadian government and U.S. government transactions are exchanged by Statistics Canada and BEA.

Investment income

Direct investment income. - To achieve a common definition of direct investment income, the U.S. published estimates are adjusted to exclude reinvested earnings of incorporated affiliates, and the Canadian published estimates are adjusted to exclude taxes (tables 12 and 13). Other definitional adjustments to Canadian published estimates include changing the estimates of income of insurance affiliates to an accrual basis from a cash basis and eliminating dividends that in the U.S. accounts are recorded as payments to holding companies in third countries but in the Canadian accounts are recorded as receipts from the United States. In addition some dividends are reclassified in the Canadian accounts from portfolio income to direct investment income. Finally, Canadian estimates of interest receipts and payments are adjusted to a net basis for comparability with the U.S. estimates.

[TABULAR DATA OMITTED]

When final data become available, statistical adjustments to published northbound distributed earnings (dividends) are usually not required or are small. When only preliminary data are available, the U.S. published estimates are often raised to account for underestimation of data not yet reported, and the Canadian published estimates are reduced to account for overestimation of such data.

Statistical adjustments to southbound distributed earnings (dividends) are confined to the U.S. estimates, which are usually raised to match the Canadian estimates. Canadian estimates include a number of small dividends that are not reported in the U.S. source data, possibly because of reporting exemptions.

Statistical differences in published estimates of northbound and southbound income of unincorporated affiliates (branches) are mainly in the banking, insurance, and real estate industries. Canadian estimates of income from bank affiliates are adjusted to reflect the estimated effects of changes in loan loss reserves on income; the U.S. estimates already reflect such changes. Reconciliation of income from affiliates in the insurance industry is not, possible, because of differences in accounting practices and statistical surveys in Canada and the United States; a partial reconciliation of the northbound estimates is possible by comparing the combined income of incorporated and unincorporated affiliates (subsidiaries and branches) and converting the Canadian estimates to an accrual basis (inclusive of reinvested earnings of incorporated affiliates). Differences in income of affiliates in the real estate industry are traceable to problems in source data.

Other private investment income (portfolio). - Most reconciliation adjustments made to other private investment income are to account for differences in gross or net treatment of certain estimates of interest income of banks and to compensate for differences in source data (tables 14 and 15).(7)

Canadian estimates of income received and paid on bank claims and liabilities are published on a net basis (receipts on claims less payments on liabilities), whereas the U.S. estimates are published on a gross basis. With one exception, the Canadian estimates are converted to a gross basis for reconciliation. On a gross basis, the Canadian estimates of receipts and payments of income between affiliated U.S. and Canadian banks substantially exceed the U.S. estimates. On a net basis, these U.S. and Canadian estimates are almost identical, so for reconciliation the U.S. estimates are adjusted to a net basis. The reason for the pattern, while unclear, may be due to differences in reporting definitions. Income on other bank claim - U.S. banks' income from unaffiliated Canadian banks and nonbank Canadian residents and Canadian banks' income from unaffiliated U.S. banks and nonbank U.S. residents - are reconciled for statistical differences.

Most adjustments to income on securities are made to the U.S. estimates. First, U.S. estimates of income receipts from holdings of Canadian bonds (northbound) are aligned with the Canadian estimates. The U.S. estimates are based on estimated U.S. holdings that are initially developed from partial information on placements and retirements of Canadian bonds. For reconciliation, the Canadian estimate of U.S. receipts is used because Statistics Canada calculates U.S. receipts from a detailed inventory of individual Canadian bonds held by nonresidents that reflects actual placements and retirements. Because of unexplained discrepancies in source data, both Canadian and U.S. estimates of income payments on U.S. corporate bonds (southbound) are arbitrarily adjusted to a common level.

Second, U.S. estimates of northbound dividends are adjusted to the Canadian level on the assumption that the source data on dividend payments is more comprehensive than the source data on receipts. Canadian estimates of southbound dividends are adjusted to the U.S. level for the same reason.

Third, Canadian and U.S. estimates of U.S. income payments on U.S. Government liabilities are adjusted to a common level. The Canadian published estimates are smaller than the U.S. published estimates. Canadian compilers assume that a portion of U.S. Government securities purchased by Canadians are for the accounts of U.S. insurance affiliates of Canadian parent companies. Under this assumption, the assets and the income are for the accounts of U.S. residents (U.S. branches of Canadian companies) and are domestic U.S. transactions. The U.S. estimates are based on the assumption that the purchases are for Canadian accounts. For reconciliation, the published estimates are adjusted to an arbitrary midpoint.

Finally, U.S. estimates are adjusted to include U.S. interest payments on foreign currency deposits in U.S. banks to reconcile with the Canadian estimates. It is unclear whether all these deposits are interest bearing. In addition, some small statistical adjustments are made to the Canadian and U.S. accounts to reconcile miscellaneous commercial transactions.

Unilateral transfers

The largest adjustment in reconciling unilateral transfers is the removal of taxes from Canadian published estimates. In addition, U.S. estimates, which are published on a net basis, are converted to a gross basis to align with the Canadian treatment. Finally, small statistical adjustments are made to U.S. northbound estimates and to Canadian southbound estimates to compensate for lack of coverage in source data.

Summary

The reconciliations have resulted in a detailed understanding of most of the differences in the published bilateral estimates; U.S. and Canadian compilers have used this knowledge to increase the accuracy of published estimates and to improve estimating techniques. Most of the differences in the Published estimates of the current-account balance now reflect a few definitional and statistical differences. Recommendations of the forthcoming Balance of Payments Manual (fifth edition) of the International Monetary Fund may result in adjustments to published estimates that would resolve some of the definitional differences; further research and reconciliation of capital-account differences may resolve some of the statistical differences.

(1.) Summary results of the current-account reconciliations were published in the United States in the following issues of the Survey of Current Business: June 1975, September 1976 and 1977, December 1979, June 1981, and December 1981 through 1991. In Canada, the results were published in the following issued of Canada's Balance of International Payments (catalogue 67-001), a publication of Statistics Canada: Fourth Quarter 1973. Second Quarter 1976 and 1977, Third Quarter 1978 and 1979. First Quarter 1981, and Third Quarter 1981 through 1991. (2.) Details of recent or planned major improvement to the U.S. international accounts were published in the June 1992 issue of the Survey of Current Business, and to the Canadian international accounts, in the Fourth Quarter 1991 issue of Canada's Balance of International Payments (catalogue 67-001). (3.) Beginning in 1990, the source data are the same for both countries, except for the source data used by Statistics Canada for petroleum exports. The data, except as noted, are compiled form U.S. and Canadian customs documents filed by U.S. and Canadian importers. U.S. merchandise imports are compiled from U.S. customs documents, and U.S. exports (Canadian imports) are compiled from data provided from Canadian customs documents. Similarly, Canadian merchandise imports are compiled from Canadian import documents, and Canadian exports (U.S. imports) are compiled from data provided by the Bureau of the Census from U.S. customs documents. (4.) In the U.S. published estimates, inland freight is included in merchandise trade because BEA follows the convention of including the charges for transporting goods to the port of export as part of the value of merchandise trade regardless of the method of shipment. The Canadian Balance of Payments Division follows the practice of valuing merchandise trade at the plant and including freight charges in transportation. (5) Beginning with 1990, the estimates of southbound inland freight have been developed by the Bureau of the Census from data collected on U.S. customs documents; these estimates are used by Statistics Canada and BEA. Different estimates of northbound inland freight are used by Statistics Canada and BEA. The estimates included in the Canadian published accounts are produced by Statistics Canada's Balance of Payments Division from data of the Transportation Division. The estimates included in the U.S. published accounts, which are larger than those in the Canadian accounts, are developed from Canadian customs documents. For reconciliation, the Canadian estimates are used because they are believed to be more accurate and to conform more closely to the estimates produced by BEA from Census Bureau data until 1990. (6.) Commissions on merchandise trade were omitted from the reconciliation pending completion of a valuation study by Statistics Canada. Similarly, estimates of Canadian mail-order imports - which are included in the Canadian accounts, but not the U.S. accounts, as a balance of payments adjustment to customs trade data - were omitted from the reconciled estimates pending further review by BEA. (7.) Some adjustments could be developed further if the related capital positions were reconciled. For example, differences in estimates of income may be due to differences in assumptions about yields or in reported income, as well as differences in the estimates of the level of claims and liabilities.