23.1 Merchandise trade is a key component of Canada's current account. This information is used by governments in making economic policy decisions and international trade agreements. Merchandise trade data are also used at the detailed level by governments and business to make decisions on markets and products.

23.2 Statistics Canada relies largely on administrative data collected by the Department of National Revenue (Customs) in compiling merchandise import trade data. Under a Memorandum of Understanding with the United States Bureau of the Census, data on merchandise imports are exchanged between Canada and the United States. This information is used to determine exports to each other.

23.3 Satisfactory controls are in place for the collection of import documentation and the electronic transmittal of data from Customs to Statistics Canada, and from Canada to the United States. The usefulness of sending copies of import documents to Statistics Canada needs to be reviewed.

23.4 Customs manual verification of commercial entries often is not performed and summary results are not reported to Statistics Canada. Deficiencies also exist in Statistics Canada verification and imputation procedures. The impact of these deficiencies on the overall quality of merchandise trade data has not been determined.

23.5 Improvements in communications and co-ordination between Customs and Statistics Canada are needed to resolve current and emerging data quality issues.

23.6 Difficulties exist in measuring illegal, non-commercial postal and casual imports, and in defining their relationship to merchandise trade data.

23.7 Current Customs procedures do not ensure complete collection of export declarations, and there is ongoing under-reporting of merchandise exports to countries other than the United States.

23.8 Recent Customs initiatives to streamline operations, and changing technology and patterns of trade, will require Customs and Statistics Canada to review their working relationship in the collection, verification and analysis of merchandise trade data to ensure an uninterrupted flow of high-quality data.

23.9 Internationally, Canada is considered to have high-quality merchandise trade statistics but the maintenance of the system for collection and verification will require continuing vigilance and co-operation.

23.10 The findings of the United States General Accounting Office review of United States merchandise trade data are consistent with those of our audit of Canadian trade statistics.

Introduction

23.11 Canada's trade with the world consists of imports and exports of merchandise and services. Net changes in the level of imports and exports of merchandise and services are collectively recorded in Canada's current account. This account also includes financial flows such as dividends and interest paid to and received from foreign sources.

23.12 Accurate and timely statistics on trade are an important source of information about the pattern and pace of economic activity. The current account is a key indicator of how well the economy is performing in relation to those of its international trading partners. Such information is essential to private sector decision makers, as well as to government. In addition to supporting fiscal and monetary policy decisions, governments use trade data to support the negotiation of international trade agreements, such as the General Agreement on Tariffs and Trade, the Free Trade Agreement between Canada and the United States and the pending North American Free Trade Agreement.

23.13 Merchandise trade data, the largest component of the current account, are derived from the administrative records of goods imported and exported by Canada. Records are mainly collected by National Revenue (Customs) at all points where goods enter or leave the country legally. This information is transmitted to Statistics Canada for analysis and reporting of merchandise trade statistics.

23.14 Neither Canada nor the United States collects data on merchandise exports to the other. Instead, since one country's exports are another country's imports, they merely translate each other's import data into statistics on exports. In January 1990, Canada and the United States implemented a Memorandum of Understanding to govern the collection and exchange of merchandise trade data.

23.15 Our report focusses on the data used to report on Canadian merchandise imports, which in 1992 amounted to $148.0 billion, whereas Canadian merchandise exports to the world were reported as $157.5 billion. Historically, Canada has maintained a surplus in its merchandise trade with the world. In recent years this surplus has fallen as deficits in services and financial transactions have increased.

23.16 In 1992, Canada's surplus of $9.5 billion in merchandise trade could not compensate for a deficit of $38.0 billion in non-merchandise transactions, and a net deficit of approximately $29 billion was left in Canada's overall trade with the world. Net balances on trade in goods and services are an important factor in Canada's ability to service international liabilities over the long term. (see Exhibit 23.1 )

23.17 To address each country's interest in the quality of import and export data, the United States General Accounting Office and the Office of the Auditor General agreed to examine the systems, procedures and practices in place for the collection and reporting of merchandise import data in their respective countries.

Audit Scope

23.18 In Canada, Statistics Canada is responsible for the reporting of merchandise trade statistics. Responsibility for the collection and verification of data is shared by Customs and Statistics Canada. Customs is required to collect import documentation for all goods entering Canada, except for data transmitted directly to Statistics Canada by Canada's National Energy Board, and data on imports of fully assembled vehicles provided by the major North American automobile companies. Our audit examined:

the reliability of Customs systems, procedures and practices in the collection of all merchandise import documentation, along with some export data, at border and merchandise transshipment points within Canada;
the accuracy achieved by Customs procedures and practices in the processing and verification of trade data, and the transmission of electronic and paper documentation to Statistics Canada;
the reliability and accuracy achieved by Statistics Canada systems, procedures and practices for the verification, analysis and reporting of merchandise trade data; and
the appropriateness of actions taken by Statistics Canada and Customs to address changes in technology and other initiatives that could affect the collection and reporting of merchandise trade data.

23.19 The methodology of our audit was developed in conjunction with a parallel review of United States trade data conducted by the General Accounting Office. Methods included a literature review on the compilation and interpretation of trade statistics, interviews with employees of Customs and Statistics Canada, and site visits to four Customs regional offices and a number of border crossings between Canada and the United States. Work included the detailed charting of the trade data system to identify and assess controls on computer access, data completeness, and accuracy in the recording of transactions and verification procedures. Our examination of controls included the review of a representative sample of transactions used in the verification and imputation of import data.

Observations and Recommendations

Trade Data: Administrative Arrangements

23.20 Customs is responsible to ensure that all duties and taxes are assessed and collected on commercial goods entering Canada. It is also responsible for controlling the movement of people and goods to achieve compliance with legislation. Statistics Canada is responsible for defining the concepts of trade statistics and for the conversion of administrative data into trade statistics.

23.21 The Statistics Act requires the Minister of National Revenue to provide the Chief Statistician with the administrative data on Canadian imports and exports collected by Customs. The arrangement by which data are transmitted was agreed upon in a 1984 memorandum of understanding between the two departments. (see Exhibit 23.2 )

23.22 In January 1988, Customs implemented the Customs Commercial System, an automated system to process all commercial goods entering Canada. At the same time, Canada adopted the international convention on the Harmonized Commodity Description and Coding System based on a six-digit code to classify goods. However, Canada and the United States expanded the code to 10 digits. The first six digits are the Harmonized System Code for the commodity. The next two digits are to identify a particular tariff item and the last two digits are for statistical detail on the commodity. A revised memorandum of understanding between Customs and Statistics Canada was signed in 1993 to incorporate these changes and the data exchange agreement between the two countries.

Customs Collection of Documentation

Satisfactory controls are in place for the collection of import documents

23.23 In 1992, Customs processed over 40 million commercial documents, including cargo control documents, commercial invoices and import accounting coding forms. Most transactions are processed via the Customs Automated Data Exchange (CADEX). In our view, satisfactory controls are in place to ensure that import documents are collected for shipments released and electronic records of transactions are transmitted to Statistics Canada. ( see photograph )

Usefulness of sending copies of import documents to Statistics Canada needs to be reviewed

23.24 Customs keeps a copy of the commercial invoice and other documentation for all commercial entries processed and sends Statistics Canada copies for transactions of over $1200. Customs keeps the documentation for six years and Statistics Canada for two. Brokers and importers keep their own records. Current procedures for the storage and transport of commercial import documents involve manual collecting, sorting, dispatching, filing, storing and retrieving of documents. Statistics Canada estimates that approximately seven percent of the 750,000 documents sent to the agency each month cannot be retrieved. In testing the system to retrieve import documents for data verification, we noted that, while the electronic records of transactions were complete, a high proportion of the paper documents at Statistics Canada could not be found within the test period.

Statistics Canada response: The Agency's present quality assurance and verification procedures for imports depend upon direct access to the paper documents and the estimated non-retrieval rate of seven percent under the regular search program is acceptable. This is likely to become a moot issue, however, as Customs moves toward a paperless environment. Statistics Canada is prepared to work with Customs in developing on-line access to information as well as new quality assurance procedures.

Customs Verification Procedures

Often, required manual verification work is not performed

23.25 One Customs procedure involves manual review and adjustment of commercial entries to ensure the correct application of duty and taxes on imported goods. Such procedures also improve the reliability of data transmitted to Statistics Canada.

23.26 Merchandise import transaction data, input into the Customs Commercial System by Customs staff at ports or by brokers or importers via CADEX, are electronically validated. Only commercial entries without validation errors are accepted. A selection of validated commercial entries is subsequently reviewed by Customs commodity specialists. In 1992, approximately 1.8 million transactions, or eight percent of all transactions, were selected for review by approximately 263 commodity specialists. Commodity specialists verify details of selected transactions such as commodity classification, value, quantity and country of origin.

23.27 We noted that Customs commodity specialists made data adjustments in approximately six percent of the transactions selected for review. The remaining transactions not adjusted by a commodity specialist are supposed to be subject to a quarterly spot check, but often these checks are not performed, and error rates are not reported.

23.28 Our audit also noted that often transactions selected for manual review by commodity specialists are not reviewed; perhaps because of inadequate human resources to handle the volume of transactions selected. In other cases, the commodity specialist may decide on the basis of experience that a review is not necessary.

23.29 Customs uses electronic verification tests (called edits) designed by Statistics Canada to review all transactions with a value of more than $25,000. Transactions of more than $25,000 that do not fall within a predetermined unit value range are selected for detailed review since they are considered to be inaccurate; these account for approximately one quarter of all data entries selected for verification. Unit value ranges refer to the range of normal or expected values established by Statistics Canada for each type of merchandise. If these ranges are to serve their purpose, which is to identify imports where unit values fall outside these ranges and then to subject these transactions to further review, they must be both appropriate and current. Our audit noted that, in fact, unit values often are not updated on a timely basis, and many of the value ranges are too wide to allow the enforcement of meaningful standards. Unit values, which allow the computer to target these transactions for manual review, are necessary to validate data for commodity classification groups, for example, electronic goods and pharmaceutical products, that are subject to wide variation over time. About 4000, or 23 percent, of the 16,000 classification codes involving commodities without standardized units of measure do not have the unit value ranges needed to do electronic verification. Statistics Canada is attempting to reduce this percentage over a period of time.

Summary information on results of Customs review is not provided to Statistics Canada

23.30 Statistics Canada estimates that if Customs performs the verification procedures required by the memorandum of understanding between Customs and Statistics Canada, 75 percent of the value of each commodity class entering Canada will be verified. In fact, there is insufficient information on the frequency of verifications carried out by Customs to know whether this level of verification is appropriate and is actually achieved.

23.31 Customs, in selecting samples for verification, also applies selection criteria based on target groups and perceived risk. In addition, it makes random selections of transactions based on the commodity specialist workload. However, the random selection procedure, which accounts for approximately 25 percent of all transactions selected for verification, is not intended to provide a statistically valid sample or to be used to assess overall data quality.

23.32 Customs does not provide Statistics Canada with summary information on verification results relating to transactions selected for verification. Lacking an analysis of results, it is impossible to evaluate the effectiveness of Customs verification procedures and the usefulness of Statistics Canada's computer verification tests. In our view, Customs and Statistics Canada need to determine whether these edits are a cost-effective method of ensuring the accuracy of trade data.

23.33 Customs and Statistics Canada should develop appropriate performance indicators and review the effectiveness and workload implications of the procedures used to verify merchandise trade data.

Customs response: Agreed. We will, in consultation with Statistics Canada, work toward developing appropriate indicators within the context of current activities related to government restructuring and the implementation of the Customs new business re-engineering initiatives. Both of these can be expected to have a substantial impact on the collection and verification procedures for trade data.

Statistics Canada response: The Agency fully supports the recommendation that the two departments should develop appropriate performance indicators, and management information tools, to determine the cost-effectiveness of data verification procedures. However, joint consideration must be given to whether the investment required to undertake the task at this time is warranted, when Customs' re-engineering initiatives, such as audit verification and electronic release, will change the way in which Customs and Statistics Canada will control the quality of the data in the future.

Customs Compliance Verification and Audit

The cumulative effect of errors detected by Customs compliance verification and audit work has not been determined

23.34 In 1992, Customs set up compliance verification units in its regional offices to perform post-importation reviews of commodity declarations. These focussed on commodity classification, valuation and country-of-origin data. While this work resulted in additional revenue, a lack of summary information on adjustments made to the import data means that the impact of the reviews on merchandise trade statistics is unknown.

23.35 During the same period, Customs implemented a pilot program to audit importers' activities rather than focus on specific transactions. Pilot audits of two large companies were conducted in 1992-93, the purpose being to evaluate audits as a cost-effective way of verifying compliance with Customs accounting requirements. Both pilot audits revealed classification errors in the reporting of merchandise trade data. In our view, these errors would not have been detected by commodity specialists performing desk verifications.

23.36 Customs and Statistics Canada jointly should analyze the statistical implications of Customs compliance verification activities and audits in relation to merchandise trade data.

Customs response: Agreed. We will, in consultation with Statistics Canada, develop the means to do so within the context of departmental efforts in business re-engineering.

Statistics Canada response: The Agency concurs. A joint effort to analyze the statistical implications of Customs' initiatives to streamline operations is especially critical if Statistics Canada is to ensure an uninterrupted flow of high-quality merchandise trade data.

Customs Physical Examination of Commercial Shipments

Physical inspection confirms only the narrative description of goods provided on import documents

23.37 One aspect of Customs commercial enforcement is the physical examination of goods. Mandatory examinations involve the physical inspection of all shipments of a particular commodity. Selective referrals depend on the inspector's judgment as to whether a shipment should be examined or not. For random referrals, the Customs Commercial System identifies for examination a specified percentage of all imported shipments.

23.38 A physical examination is performed to check information on the cargo control document and commercial invoice - including a description of the goods, country of origin and quantity - against the actual shipment. Unfortunately, these documents do not indicate commodity classification codes. Therefore, a physical examination can only confirm the general narrative description of goods provided on the import documents, but cannot be used to verify the commodity classification. Customs indicates that the most effective way of physically verifying the commodity classification would be through post-audits at the importers' premises.

23.39 Customs and Statistics Canada, in their reliance on electronic and paper documentation, assume that the nature of trade merchandise entering Canada is reasonably reflected in the information on file. We believe that the verification of commodity classification, whether by physical inspection of goods at the time of release or through on-site inspections at importers' premises, is important to improving trade data.

Statistics Canada Data Verification and Imputation

23.40 To improve data quality, Statistics Canada conducts a variety of electronic verification tests, manually reviews selected transaction documentation, imputes data values where information is not available and, where warranted, makes various adjustments to the data. In attempting to enhance data quality, Statistics Canada risks the introduction of inaccurate adjustments and imputations; in terms of value for money, it may be that the cost of adjustments is not justified by improvements in data quality.

23.41 Statistics Canada has an ongoing concern with the quality of trade data. Its publications usually define the limitations of their data, though quantitative indicators can be given only in relation to appropriate assessment measures. Statistics Canada maintains that there are no simple and efficient ways to determine accurate measures of quality: "Attempts made in the past to obtain error rates for the coding of commodities have yielded inconclusive results primarily because the task of commodity coding can be quite subjective since the Harmonized System of Commodity Description and Coding (HS) has over 16,000 codes. Furthermore, the quality of trade data has more dimensions (due to the large number of variables involved) than the accuracy of commodity coding and this quality cannot be captured in a single quantitative measure." The Quality of Canadian Imports Data , Statistics Canada Catalogue, no. 65-001, September 1991

23.42 While the above problems are inherent in the system of collecting and analyzing trade data, our audit identified a number of less integral deficiencies that also affect data quality. These deficiencies were identified through testing the accuracy of computer verification and imputation procedures, updating a 1991 data quality analysis, and reviewing results with senior staff at Statistics Canada.

23.43 Statistics Canada conducts three types of tests: one test detects inadmissible values (for example, non-existent unit of measure); the second test cross-references two pieces of data (for example, country of origin and classification code, as in "oil from Switzerland"); the third test checks that the value of a commodity lies within an acceptable range. Where specific transactions do not pass the computer's checks, adjustments are imputed automatically by the computer for transactions of less than $50,000. Rejected transactions of more than $50,000 are manually reviewed by a commodity officer.

Statistics Canada verification and imputation procedures do not always improve data quality

23.44 Our test of Statistics Canada's verification and imputation system included a representative sample of 520 transactions selected from current trade data. While our test sample is too small to allow us to estimate total error, the results raise questions about the ability of the system to optimize data quality.

23.45 Of the 520 transactions in the sample, documents for only 233 transactions were retrieved from the Statistics Canada records room; this retrieval rate was less than Statistics Canada estimates of documents not received or misfiled. Many transactions rejected by the edits are not analyzed and may be incorrectly imputed because copies of documents cannot be found in Statistics Canada.

Many transactions with invalid codes or missing information are not analyzed or are incorrectly imputed

23.46 In comparing the computer records against the physical records of transactions in documents we did obtain and examine, we noted that where the transaction's commodity classification code cannot be matched to a table of valid codes or imputed, the computer allocates an "unassigned" code to the transaction. The total value of unassigned transactions, which mainly includes low-value items under $1200, is approximately $1.5 billion annually; but these transactions are not analyzed.

23.47 In comparing the commodity classification code with the country of origin, the computer incorrectly imputed the data in 32 out of 35 cases: the effect is to overstate the number of imports from the United States, our largest trading partner, and to understate the number of imports from Canada's smaller trading partners.

23.48 We noted that some of the unit values in our sample had been recently updated. We also found that the commodity officers were well informed of current developments in their areas of expertise although, because of the increasing volume of international trade, they were not able to review and adjust all selected transactions. Our review also indicated that errors in unit of measure (for example, imperial instead of metric) for specific commodity groups could be more easily detected by Customs at the data input stage.

23.49 Information on the mode of transportation for imports from countries other than the United States is derived from the cargo control document for the transaction. Our audit found that in 31 of the 46 transactions we examined the mode of transport was incorrectly derived. In fact, since imports from countries other than the United States tend to involve more than one mode of transport, it has been difficult to assign a single appropriate mode. However, as of 1 April 1993, it became mandatory for the importer to provide information on the principal mode of transport, so those data will no longer be derived. This will improve the quality of information required for the data exchange with the United States.

23.50 The size of the sample of transactions selected for review is too small to allow us to estimate the effect of these deficiencies on Statistics Canada's verification and imputation procedures. However, the results obtained do raise questions about the effect of current imputation procedures on improving data quality.

23.51 Statistics Canada verification and imputation procedures should be reviewed to determine their effectiveness in improving data quality. The review should include an assessment of opportunities to apply better, possibly automated, quality control techniques to update the unit values used to verify trade data.

Statistics Canada response: The Agency recognizes that its verification and imputation procedures require review and revision. Even more important, the changes taking place in the international trading environment and the implementation of the "New Business Relationship" initiatives at Customs will require revisions to the concepts and methodologies underlying the present trade statistics program.

In reference to the specific data quality issues related to unit value ranges, Statistics Canada and the U.S. Bureau of the Census are in the process of reviewing and harmonizing their editing strategies to alleviate inconsistencies and improve the effectiveness of these edit parameters. This harmonization process is being carried out under the direction of the Heads of Agencies who meet annually to review the status of the Canada-U.S. Memorandum of Understanding on the Exchange of Import Data. In addition, the recent recommendations of the Customs Co-operation Council in Brussels to internationally standardize units of quantity may facilitate the alignment process.

Data Transmission to the United States

Satisfactory controls exist for the transfer of data to the United States

23.52 Once verification and adjustments are complete, Statistics Canada transmits data on merchandise imports from the United States to the United States Bureau of the Census. This is done twice a month through a communications link-up to the Canadian Embassy in Washington. The Bureau reviews the data and, if questions are raised or changes are required, contacts Statistics Canada directly. Statistics Canada reconciles its monthly merchandise trade statistics with those of the United States Bureau of the Census.

Operating Relationships: Customs and Statistics Canada

23.53 Differences in the strategic objectives of Customs and Statistics Canada affect the collection and verification of merchandise trade data. While Customs is responsible for the initial collection of merchandise trade data, its primary objectives are to collect revenue and to control the movement of goods and people as required to achieve compliance with legislation. Statistics Canada is responsible for the analysis of data and reporting of statistics. Electronic verification procedures designed by Statistics Canada can increase the workload of Customs. For example, out-of-date unit values used in electronic verifications increase the number of selected transactions that require manual review by Customs commodity specialists. This practice can leave commodity specialists unable to perform the required verifications or reduce the time available for them to do other required work, such as examining high-risk commercial entries.

Extending direct communications at the operational level would promote timely resolution of problems

23.54 Maintaining good data quality depends on good communication between Customs and Statistics Canada. In fact, current operating procedures do not encourage direct communication between Statistics Canada and Customs regional offices on specific data quality issues.

23.55 Statistics Canada usually communicates directly with Customs headquarters rather than with regional offices, where at least some data inquiries could be answered directly. This increases the time required by both organizations to respond to data quality improvement initiatives. The lack of prompt action taken in response to "problem reports" and "case studies" is evidence of the communication problems between Customs and Statistics Canada.

23.56 "Problem reports", which are filed by Customs when regional commodity specialists note that a particular Statistics Canada validity check is out-of-date, are forwarded to Statistics Canada for review and update. "Problem reports" are relayed through Customs headquarters, which reviews and then transmits them to Statistics Canada: this transmission does not always occur in a timely manner. Most Customs commodity specialists, seeing little action result from their reports, do not bother to prepare reports when unit value problems are detected.

23.57 Statistics Canada case studies, which deal with specific data quality issues such as commodity classifications, value ranges or country of origin, also experience delays when they are forwarded to Customs headquarters for review and appropriate action. The inability of Statistics Canada to communicate directly with Customs regional offices to resolve regional data problems hinders the timeliness of data quality improvements. Both Customs and Statistics Canada have advised us that procedures are being developed to allow more flexibility and direct communication at the operational level.

Some duplication of effort between Customs and Statistics Canada

23.58 Current verification procedures sometimes result in a duplication of verification work by Customs and Statistics Canada. For example, transactions of more than $50,000 that fail the unit value verification edits are selected for review at Customs; if they are not adjusted at Customs, the transactions will undergo a similar review at Statistics Canada. As mentioned earlier, Statistics Canada is not informed of the summary results of Customs verification adjustments made as a result of Statistics Canada's edits. Hence, Statistics Canada is not able to analyze the effectiveness of its verification edits and the usefulness of the review procedures used to ensure data quality.

23.59 Customs and Statistics Canada should continue to improve communications with a view to achieving prompt action and enhancing data quality.

Customs response: We agree with the importance placed by the OAG on this activity. There has already been a marked improvement in communication between the two departments in the recent past. Customs and Statistics Canada have established a close working relationship and efforts have been ongoing to ensure that a high level of communication and co-ordination between the departments continues. Measures taken include the signing of a memorandum of understanding by the departments and the establishment of formal communication mechanisms such as the Data Quality/Classification Committee and the quarterly Working Committee for the memorandum of understanding on the exchange of import data. Statistics Canada is being consulted in the development of new departmental initiatives that may have an impact on statistical data collection, and is making direct contact with local Customs offices more frequently in order to promote the timely resolution of problems. Efforts in this regard will continue to be made under the terms of the memorandum of understanding between the departments.

Statistics Canada response: The Agency fully endorses this recommendation and believes it is especially critical at this time. The Agency is working with Customs to establish a closer working relationship and to effect the timely resolution of current and emerging data quality issues.

Measurement and Definitional Issues in Import Data

23.60 An accurate knowledge of what goods are included or omitted from the data collection process is an important element in the interpretation of international trade statistics. Transactions are often complex, and information on quantities, value or composition of shipments can be difficult to obtain. Knowledge of data limitations improves the usefulness of trade statistics.

23.61 In some cases, international conventions define trade categories and thus affect the classification of goods. For example, goods imported by travellers (such as used automobiles) are counted as services, whereas value added by services in the manufacture of goods and intellectual properties may be included in the price: for instance, royalties on any patented goods if the royalty payment is a condition of the sale.

23.62 Illegal imports, which are not included in the trade statistics, are difficult to measure. Any time there is an incentive to avoid taxes or duties, there is the possibility of under-reporting. With Canada's major trading partner, the United States, the incentive for misclassification of trade data is now diminishing as duties are lowered or eliminated under the Free Trade Agreement; however, it is not known if this decline will translate into increased accuracy in the value of trade reported in import documents.

23.63 In other cases, there may be an incentive for international firms to reduce taxes through "transfer pricing". This practice places an unrealistic value on exported or imported products transferred between two divisions of the same corporation, so that profits of the corporation are minimized in countries with high tax rates and maximized in countries with low tax rates. Another way to reduce the dutiable value of goods is to bill separately for services such as licensing, marketing, financial advice or data processing. Pricing issues are complex, and their full impact on trade statistics is unknown. These nuances in trade patterns and practices represent an ongoing challenge to producers and users of merchandise trade statistics.

Postal Imports

Non-commercial postal imports are based on estimates

23.64 Information on non-commercial postal imports is based on estimates. The estimated value of these transactions has increased from $541 million in 1988 to $1.5 billion in 1992. In July 1992, Customs implemented a Postal Import Control System to collect duties and taxes on postal imports with a value for duty over $20.

23.65 However, due to limitations in the Postal Import Control System, Customs is unable to provide specific information on postal items imported into Canada, though it estimated that postal imports accounted for approximately 2.5 million items in 1992-93. With the full implementation of the system, Statistics Canada expects to have better estimates of non-commercial postal imports to be included in the "Customs basis" merchandise trade data.

Casual Imports

Casual imports are not considered a component of merchandise trade data

23.66 In March 1991, Customs implemented the Travellers Entry Processing System, which by March 1993 was operating in 90 of Canada's 350 ports of entry. During 1992, Customs processed over 123 million travellers entering Canada and approximately 4.2 million casual import documents. However, because the system is not operational at all ports and is not designed to provide summary data, the total value of casual imports by travellers into Canada continues to be based on estimates.

23.67 To complement this system, Customs officers at ports collect data on the number of cars, residents and non-residents entering Canada. Statistics Canada has used this information, as well as quarterly samples collected in collaboration with Customs, to estimate the total value of casual imports at around $3 billion in 1992, though that value is not included in Statistics Canada merchandise trade data. Casual imports by travellers are defined by international convention as a component of the "service account" in Canada's balance of payments.

Merchandise Trade Data on Exports

Under-reporting of exports to countries other than the United States is an ongoing problem

23.68 Under the memorandum of understanding with the United States Bureau of the Census, Statistics Canada relies on U.S. import data to determine Canada's exports to the United States: these account for about 70 percent of Canada's merchandise exports worldwide.

23.69 For exports to countries other than the United States, Customs is responsible for the collection, sorting and forwarding of export documentation to Statistics Canada. In addition, Statistics Canada receives about 50 percent of the export declarations from Customs-approved exporters, in the form of monthly summary reports. Statistics Canada is responsible for compiling export data and for developing, implementing and monitoring data verification procedures.

23.70 In fact, the collection of export forms is erratic. There is no penalty to exporters for not submitting the export forms and there are inadequate controls on the collection of export declarations at Customs border points. For example, marine carriers present bills of lading or equivalent printouts to Customs within five days after leaving a Canadian port. Customs then checks the bill of lading to determine if the export form is required, and if it has been submitted. If the form is missing, a tracing letter is sent to the exporter. After a month, if no export form has been received, there is no further administrative recourse and the file is closed. In one region, Customs estimates that approximately seven percent of the data on marine export trade, representing approximately $100 million annually, goes unreported to Statistics Canada.

23.71 As a result of the data exchange between Canada and the United States, Customs no longer requires Canadian exporters to complete an export declaration for goods entering the United States. However, goods entering the United States from Canada that are destined for other countries do require an export form to be completed; exporters often assume otherwise, since such forms are not required for goods destined for the United States, and do not comply.

23.72 Also, if Canadian goods are purchased by a company in the United States for eventual sale to a third country, such as Mexico, the Canadian exports are recorded as a sale to the United States and not to the final country of destination. Such practices in the international recording of data affect the interpretation of trade statistics by overstating exports to the United States and under-reporting exports to, for example, Mexico.

23.73 Customs and Statistics Canada recognize the problem of under-reporting of exports to countries other than the United States and have undertaken a number of studies to estimate the impact on trade data, specifically: seeking to assess trade by different modes of transportation; attempting to reconcile Canadian merchandise export trade data with the data of other countries; and establishing a joint Statistics Canada and Customs data committee to monitor the quality of data on Canadian exports. Customs has also taken steps to encourage exporters to participate in monthly summary reporting to Statistics Canada. The importance of good quality data on Canada's world trade patterns warrants increased efforts to improve current methods of collecting data on Canadian exports to other countries via the United States.

23.74 Customs and Statistics Canada should work with United States agencies to review current practices in the recording of trade data involving exports from Canada to other countries through the United States.

Customs response: Both Customs and Statistics Canada have been working with the United States agencies for some time to improve our trade data with respect to in-transit trade statistics. These efforts will continue as a result of the memorandum of understanding between the United States and Canada, which provides and sets out the terms and conditions for the exchange of import data.

S tatistics Canada response: The under-coverage of Canadian exports to countries outside the United States, for shipments directly to those countries and in transit through the United States, is a critical issue in Statistics Canada's measurement of export trade. The Agency is working with both Customs and the U.S. agencies to quantify this under-coverage and to increase the accurate filing of export data. Nevertheless, in the absence of Customs border controls and compliance sanctions, there is no definitive solution to the problem. The registration of new exporters, however, under the Single Business Registration Number initiative, will help to develop a statistical solution to the under-coverage problem.

Future Initiatives: Impact on Trade Statistics

23.75 In 1992, National Revenue announced a number of initiatives intended to streamline Customs operations. Two Customs initiatives in particular will have a major impact on the collection, verification and reporting of merchandise trade statistics. The primary objective of these initiatives, which are described below, is to increase the efficiency of Customs operations and to allow a speedy "no hassle" transfer of goods across Canada's borders.

Electronic Release and Data Interchange

Customs plans to release commercial shipments by transferring information electronically

23.76 The purpose of this initiative is to increase the use of electronic data interchange in effecting import clearances, and to move toward a paperless system of reporting on, and accounting for, imports. Under the new system, release information is electronically transmitted to Customs by the importer or broker. There it is reviewed, and the transporter is informed through electronic data interchange when the goods are cleared for import into Canada.

23.77 This initiative is directed to the 70 percent of the importing community already transmitting accounting information via CADEX. The program eliminates the need for importers to physically deliver paper documents to Customs, although hard copy documentation would be retained by the importers and could be forwarded to Customs on demand. Other government departments, including Statistics Canada, would receive information through access to the Customs Commercial System.

23.78 In March 1993, Customs began the process of identifying user requirements for electronic release, and announced in June 1993 that the automotive industry will be the first to use the new system to process the import of auto parts and supplies. The program will also be extended to other industries. As yet, there is no agreement between Customs and Statistics Canada on the best ways to maintain data quality.

Auditing Importers

Emphasis in verification is shifting to auditing of importers

23.79 In the streamlining of its operations, Customs indicated that in future it would turn toward a greater use of importer audits. To that end, Customs is now developing procedures that will eliminate "hard copy" invoice requirements for shipment clearance; instead, import business will be verified through audits of importers after the goods have entered Canada. This initiative will have a significant impact on Statistics Canada's merchandise data verification requirements since the data format will differ from that of the current import system.

23.80 The implementation of periodic verifications or audits could have a significant impact on the quality of merchandise trade data being reported: for example, commodity classification errors may not be detected in a timely manner. Depending on Customs capability and the number of audits to be carried out, a large number of importers may not be audited. In our view, continuing consultation between Customs and Statistics Canada is needed to assess how this initiative will affect the collection and reporting of merchandise trade data.

Statistics Canada: Alternative Data Sources

As systems for data collection change, new and broader sources of data are needed

23.81 In view of potential changes in the way import trade data are collected, Statistics Canada has been working since April 1991 to identify alternative data sources. Officials in Customs and Statistics Canada have met periodically to discuss matters of mutual interest in this area.

23.82 The primary focus of the project is to identify the best methods for surveying importers, exporters and carriers so as to obtain high-quality import data. To begin with, such a survey requires an up-to-date inventory of Canadian importers and exporters.

23.83 While it has been concluded that much of the required trade data is available from Canadian importers, there are many concerns about the effects of using survey techniques in the collection of trade data, specifically: the response burden on Canadian business; whether data gathered by means of surveys would meet the requirements of Statistics Canada's data exchange agreement with the United States; and, finally, the role of the 350 Canadian customs brokers, as opposed to that of some 150,000 importers, in the retention of documentation required for data verification. Under the Free Trade Agreement, duties and tariffs between Canada and the United States will be gradually eliminated, but the complexities of compliance with international trade rules, particularly with regard to country of origin, will increase. A new role for customs brokers in serving these needs and in assisting the data collection process has not yet been established.

23.84 Statistics Canada and Customs, in consultation with other users of trade data, should take steps to ensure that, as new data collection and verification techniques are implemented, the quality of Canadian merchandise trade statistics is maintained.

Customs response: Both National Revenue and Statistics Canada are committed to the maintenance of quality statistics and will take appropriate steps to ensure that the quality of merchandise trade statistics is maintained when any new data collection or verification techniques are implemented.

Statistics Canada response: The Agency is pleased with the acknowledgment given to the high quality of its merchandise trade statistics and to the importance of accurate and timely trade statistics, to decision makers in both government and the private sector. As Customs proceeds with the implementation of its initiatives to streamline operations, the requirement for a joint consultative process becomes especially critical if Statistics Canada is to ensure an uninterrupted flow of high-quality merchandise trade data.

Conclusion

Continuing vigilance is required to maintain high-quality merchandise trade data

23.85 Statistics Canada generates merchandise trade statistics from collecting data on millions of individual transactions. Simply adding the results of all import transactions should, in theory, produce a precise statement of exactly what entered the country. In practice, however, there are bound to be inaccuracies.

23.86 Several factors contribute to the reporting of inaccurate merchandise trade data: human error, incomplete documentation, under- or over-reporting of imports to avoid taxes, deliberate or accidental misclassification to avoid duties, or wrong indication of country of origin to gain preferential duty treatment. In addition, there are legitimate differences of opinion on the interpretation of the commodity classification, country of origin and mode of transportation. Also, there are illegal imports. The challenge is to understand the limitations of the data, measure their impact and, whenever possible, take steps to overcome them.

23.87 Internationally, Canada is considered to have high-quality merchandise trade data and, indeed, the current system of collecting data is working well. However, to maintain that system in light of changing patterns of trade and technology will require continuing vigilance and responsiveness. Measurement issues and questionable categories of trade, when added together, substantially affect the interpretation of Canada's reported 1992 merchandise trade surplus of $10 billion: for example, an estimated $3 billion in casual imports by travellers, currently classified as services, is not included in the amount of merchandise imported into Canada; furthermore, trade data do not include estimates of illegal imports. At a more detailed level, $1.5 billion worth of goods that fall in the unassigned commodity code could, if assigned codes, substantially affect trade data for specific commodity groups, as could the estimated $1.5 billion in non-commercial postal imports. While small in relation to Canada's total trade with the world, these amounts could affect the picture for particular sectors of the Canadian economy.

23.88 Improving the accuracy of statistical measurements, or even maintaining it, will continue to be challenged by changes in technology, transnational transactions and budgetary restraint. Nevertheless, the demand for high-quality, detailed trade statistics will continue. Administrative records may still be the primary source of information, but methods of collection and verification will change. In managing this change, Customs and Statistics Canada must work together to ensure that there is an uninterrupted flow of data, and to implement improvements that meet the needs of governments and other users of trade data for accurate and timely statistics.

Appendix

United States General Accounting Office
Report to Congress

NOVEMBER 1993 MEASURING U.S.-CANADA TRADE

Shifting Trade Winds May Threaten Recent Progress
(GAO/GGD - 94-4)

EXECUTIVE SUMMARY

PURPOSE

The United States and Canada are the world's top trading partners. The free trade agreement between the two countries, which began in 1989 and is gradually phasing out duties on bilateral trade, is expected to further enhance their trade relationship. Considering the importance of this relationship to the economies of the two nations, accurate data on its nature and extent are vital. For this reason GAO and the Office of the Auditor General of Canada reviewed the capacity of the statistical systems of their respective countries to produce accurate and complete trade data for both the present and the future. This report focusses primarily on merchandise trade data, but also recognizes the importance of data on international transactions of services, investment income and capital.

BACKGROUND

Trade data have many uses. The balance of trade, the difference in the value of a country's imports and exports, has become an increasingly important economic measure because of the emerging global economy as represented by growth in both exports and imports. Trade data are indicators of the effect of a nation's trade policies and play a significant role in the negotiation of trade agreements, such as the U.S.-Canada Free Trade Agreement and the North American Free Trade Agreement. Because the U.S. and Canadian economies are so intertwined, an accurate measurement of their bilateral trade is important for understanding each country's economy.

Although there are many aspects of U.S. international trade, data on merchandise trade are the most closely watched. Data on every declared import and export of merchandise are to be collected by the U.S. Customs Service at ports of entry as part of its trade administration efforts. The U.S. Bureau of the Census then compiles this information and publishes it on a monthly basis. Merchandise trade data are also used by the Bureau of Economic Analysis in constructing the balance of payments accounts. These accounts are the statistical summary of all of the United States' international transactions and include information on transfers of services, interest income and capital.

RESULTS IN BRIEF

The United States and Canada have made great strides in improving the collection of the data on merchandise trade. For years, U.S. and Canadian government officials were aware of a serious undercount of U.S. exports to Canada mainly because of the failure of exporters to file export documents with U.S. Customs. To alleviate this problem, the United States and Canada agreed to develop a program to exchange their more accurate administrative records on imports and use this information to determine each country's exports to the other. This data exchange program has significantly improved the quality of U.S.-Canada merchandise trade data, although some issues concerning data collection and import classification still must be resolved.

Despite the success of the data exchange program, problems with collecting and processing import data could adversely affect to an unknown extent the accuracy of U.S. data on merchandise trade with Canada. Some of these problems, such as the lack of control of import documents sent from Customs to Census, will be corrected by the automation of Customs processes. Another problem is Customs' lack of assurance that its cargo examination and import document review procedures are effective in uncovering violations of Customs laws. Such violations, if undetected, could affect the accuracy of trade data. Customs, however, is making substantial efforts to improve its trade enforcement programs.

Even with their shortcomings, merchandise import data are still considered some of the most accurate trade data produced in the United States. Unlike data on other forms of trade, such as service transactions, which are collected through sample surveys that are limited in coverage, Customs attempts to collect data on all imports.

Changes in the international trade environment, however, could significantly affect the quality of merchandise import data. For example, when free trade with Canada is fully implemented in 1998, there will be little need for Customs' entry documentation beyond the collection of trade data. These changes may require developing new methods for collecting these data. Alternative methods for the United States to consider include the use of surveys and the direct reporting of merchandise trade data by businesses to Census. However, these methods have potential drawbacks, such as a loss of detailed product information and diminished statistical accuracy, that would have to be addressed.

GAO's ANALYSIS

U.S.-Canada Trade Data Exchange Successful, but Problems Remain

The federal government has long suspected that U.S. exports have been undercounted. In 1986, a reconciliation of U.S.-Canada merchandise trade data indicated that reported U.S. exports were 20 percent lower than Canada's recorded imports from the United States. The undercount was primarily due to the failure of exporters to report exports to Customs. To deal with the export undercount, the United States and Canada decided to exchange their more accurate administrative records on imports and use them to determine each country's exports to the other. U.S. and Canadian officials agree that the exchange, which began in 1990, has reconciled most of the differences in U.S.-Canada merchandise trade data. However, the countries are still working to resolve some problems, such as the undercounting of exports to third countries that go through the United States or Canada, and the failure of one country's import data to adequately capture some data elements previously available in the other's export data, such as the methods used to transport imports.

Although import data are generally considered to be more accurate than export data, they too are flawed. Recent evaluations by the National Research Council and GAO revealed that flaws in compliance and quality control procedures could affect the accuracy of import as well as export data. One problem, the lack of control over manually filed import documents, is diminishing as Customs continues to implement an electronic filing system. A problem that is potentially more serious is that Customs' cargo examinations and import document reviews lack effectiveness in identifying violations of Customs laws. The extent to which trade data have been affected by importer noncompliance is unknown. Since a September 1992 GAO study that identified these problems, Customs has been working to improve its trade enforcement efforts.

Changing Trade Environment Will Challenge Merchandise Trade Data System

Simply improving the current systems and procedures may not be enough to ensure the long-term quality of merchandise trade data, particularly that of import data. One reason is that Customs may end up giving less emphasis to traditional Customs documentation as the free trade agreement between the United States and Canada and other potential free trade agreements eliminate duties on imports. Customs also plans to further automate its cargo processing to deal more efficiently with the increasing trade volume. This planned automation could limit the amount of information provided by businesses to monthly summaries instead of single-transaction entries. These changes could significantly affect the quality of import data.

Problems With Service Transaction and Investment Income Information Limits Trade Data Quality

The monthly merchandise trade balance reports issued by Census are a closely watched indicator of the country's international economic competitiveness. However, merchandise trade accounts for only a part of United States' international economic activity. Service transactions, such as international long distance telephone calls, and interest paid on investments by foreign citizens constitute an important and growing part of the United States trade relationship with Canada and other countries. In fact, when service transactions and investment income are included with merchandise trade, the U.S. trade balance improves. For example, in 1992, this country had a merchandise trade deficit with Canada of $8.0 billion. However, when services and investment income are factored in, the balance with Canada became a $4.5 billion surplus.

Data for the service and investment income components of this balance are collected mainly through surveys that are limited in frequency, level of detail and coverage. Therefore the level of quality of the data on services and investment income are lower than that for merchandise trade. Initiatives have been proposed by the current and previous administrations to improve these data, but Congress has approved limited funding for these initiatives.

RECOMMENDATIONS

GAO recommends that Census and Customs form an interagency task force to study how U.S.-Canada merchandise trade data should be collected in the future trade environment. (The study should be expanded to include U.S.-Mexico trade data in the event that the North American Free Trade Agreement is ratified and implemented. Note: Congress may decide on NAFTA before report is issued). Census should consider joining Customs with their Canadian counterparts to form a bilateral task force to address these issues co-operatively. GAO further recommends that the work of the interagency task force be done in the context of broader efforts to improve the measurement of all forms of U.S. international trade.

1993 Report of the Auditor General of Canada

Chapter 23—Merchandise Trade Statistics