U.S. Merchandise Trade Statistics

A Quality Profile

Methods Research & Quality Assurance BranchForeign Trade DivisionBureau of the CensusDecember 19, 2002

2

3

Table of Contents

I. Introduction:

A. Purpose of the PaperB. Source of the StatisticsC. Recent Major Changes

II. Data Quality Assessment:

A. Undocumented Shipmentso Exports Through Canadao Reconciliationso Customs Compliance Studieso Northbound Shipmentso Foreign Trade Zones

B. Transiting GoodsC. Underestimation of Low Valued TransactionsD. Late FilingE. Reporting ErrorsF. Data Capture ErrorsG. Quality and Automated ReportingH. Disclosure AvoidanceI. Requests for Re-Verification of Trade Statistics

III. Adjustments:

A. Adjustments for Seasonal and Trading Day VariationB. Adjustment for Price Change

IV. Conclusion

V. Appendices:

A. Import Entry SummaryB. Shipper’s Export DeclarationC. Annotated BibliographyD. Glossary of Acronyms

4

I. INTRODUCTION

A. Purpose of the Paper The merchandise trade statistics measure goods tradedbetween the United States and other countries. They arethe official source of information about U.S. imports,exports and balance of merchandise trade. As a leadingeconomic indicator and a major component of the GrossDomestic Product (GDP), the statistics provide criticalinformation to a wide and varied group of users in thepublic and private sectors. The U.S. Census Bureau(Census) strives, with available resources, to provideaccurate, high quality statistics to meet these needs. However, to use the information wisely and appropriately,users need to understand the nature and limitations ofthe merchandise trade statistics program. This paperdiscusses issues affecting the quality of the statisticsand some known limitations. It also includes discussionsof recent and proposed improvements in the quality of thestatistics.

This paper covers only the statistics on trade in goodson a Census basis. The data on trade in services and ona balance of payments basis, published in the FT900, U.S.International Trade in Goods and Services, are compiledby the Bureau of Economic Analysis (BEA). Information onthe BEA data is available on the Internet (www.bea.gov)or by calling Christopher Bach at (202) 606-9545.

Comments or questions about the information in this paperor suggestions for future reports are welcome. Addressthem to David Dickerson, Chief of Methods Research andQuality Assurance Branch, Foreign Trade Division,(301)763-3080, David.M.Dickerson@census.gov.

B. Source of the Statistics

Census collects and compiles merchandise import andexport data from various electronic and manual (paper)sources. These data represent shipments through morethan 400 ports of entry and exit in the United States,Puerto Rico and the Virgin Islands.

5

Sources for import statistics include 1) the U.S. CustomsService (Customs) Automated Broker Interface (ABI), 2)paper import entry summaries (Appendix A), 3) paper orelectronic applications for foreign trade zone admission,and 4) Canadian data on exports of natural gas andelectricity to the United States provided by StatisticsCanada. Import documents are completed by the importersor their licensed brokers and are filed directly withCustoms. In 2001, Census collected trade statistics onmore than 33 million import transactions. Approximately99 percent of the transactions were receivedelectronically by Customs, the remaining one percent werefiled on paper documents. These paper documents accountfor roughly six percent of import value. Paper documentsare collected by Customs and mailed to Census' PuertoRico Trade Data Staff, which enters the statistical dataand transmits them to Census headquarters for furtherprocessing.

Sources for export statistics include 1) the AutomatedExport System (AES), 2) paper Shipper's ExportDeclarations (SEDs, see Appendix B), and 3) Canadian dataprovided by Statistics Canada. Export declarations,either paper or electronic, are required for shipments toall countries except Canada, and are completed byexporters (U.S. principal parties in interest) or theirduly authorized agents who submit them to the exportingcarrier who has the responsibility to submit them toCustoms at the time of exportation.

In 2001, Census collected trade statistics on more than22 million export transactions. Approximately 52 percentof export transactions are currently submittedelectronically through AES and 12 percent sent to Census'National Processing Center (NPC) where they are keyed andthe data transmitted to Census headquarters for furtherprocessing. The remaining 36 percent of the exporttransactions are acquired through a data exchange withStatistics Canada. The United States derives itsstatistics on exports to Canada from import data filedwith Canada. Likewise, Canada derives its exportstatistics for shipments to the United States from U.S.import data. For more information on the data exchangesee "U.S.-Canada Data Exchange," 1990-2001, (available onthe Internet at www.census.gov/foreign-trade or contactVanessa Ware at (301) 763-6984, Vanessa.Ware@census.gov).

6

C. Recent Major Changes

Prior to January 1, 2000, Census also received exportdata through the Automated Export Reporting Program(AERP). In 1998, approximately 32 percent of monthlyexport data were filed through AERP. The AES, which hasreplaced AERP, incorporates on-line edits and refersquestionable data back to the filer for verification orcorrection. A significant decrease in reporting errorrates on export transactions has occurred as more filersopt for automated reporting on AES. Prior to thewidespread use of AES, an average of 27 percent of allexport transactions contained one or more reportingerrors. Now that more filers have opted to use AES,export reporting errors have declined to an average of 17percent.

II. DATA QUALITY ASSESSMENT

Census conducts various research studies to assess the qualityof trade statistics and to identify quality issues related tothe data. Quality assurance procedures are performed at everystage of collection, processing and tabulation; however, thedata are still subject to errors. The following sectionsinclude an explanation of the most significant sources oferrors: undocumented shipments, transiting goods,underestimation of low valued transactions, late filing,reporting errors, and data capture errors. This section alsoincludes an assessment of the effects of automated reportingon the quality of the data, information on disclosureavoidance and requests for re-verification of tradestatistics.

A. Undocumented Shipments

Federal regulations require importers, exporters or theiragents to submit import and export information for allmerchandise shipments above established exemption levels. Exemption levels are $2,000 ($250 for quota items) forimports and $2,500 for exports to countries other thanCanada. Through comparisons of trade data with majortrading partners, audits of trade documentation, andother measures, Census has identified missing informationfor both import and export shipments. Missing

7

documentation is more common with exports than withimports since import information is subject to greaterscrutiny by Customs in relation to the administration oftariffs, quotas and other enforcement activities.

Exports Through Canada

The movement of U.S. goods northbound through Canadaen route to another destination can result inundercoverage in the U.S. export statistics. Mostshipments to Canada do not require an SED, as theyare collected through the U.S.-Canadian DataExchange. However, an SED is required for goodsthat cross the Canadian border and are shipped toanother country from a Canadian port. Exporters maynot comply with the reporting requirements forseveral reasons.

In some cases, exporters may mistakenly believe thatthey are not required to file an SED, as most goodsshipped to Canada do not require one. In othercases, shippers may choose not to file the SEDbecause of a lack of enforcement of the reportingrequirement. Aside from these likely causes ofnoncompliance, the lack of infrastructure (e.g.drop-off facilities) at the U.S.-Canadian border mayalso discourage carriers from filing the exportdeclaration. Increased border security efforts andthe future implementation of mandatory electronicfiling of truck manifests and export declarationsshould significantly reduce this problem.

Reconciliations (Trade Data Comparisons)

In the early 1970s, The United States and Canadabegan annual studies to reconcile differences inmerchandise trade statistics compiled by the twocountries. After adjusting for differences in tradedefinitions, valuation, timing and errors, thestudies uncovered a chronic understatement of eachcountry's exports. At one point, the understatementof U.S. exports to Canada was estimated to be ashigh as 20 percent. This high undercount, in largepart, reflected the difficulties of collecting dataon overland shipments moving across the open border.

8

Given the much higher quality of each partner’simport statistics, the United States and Canadasigned a Memorandum of Understanding. Under thatagreement, implemented January 1990, Census derivedits export statistics from Canadian imports andStatistics Canada derived its exports from U.S.import statistics. Both countries consequentlyeliminated the requirements for filing exportdocuments for goods destined to the other.

Several immediate improvements to the U.S. exportstatistics resulted from the introduction of theU.S.-Canada Data Exchange. Most important, iteliminated undocumented shipments to Canada. Also,since the Canadian data are pre-edited, it improvedthe accuracy of the data reported to Census andincreased the inclusion of data in the correctstatistical month.

The success of the U.S.-Canadian reconciliationsprompted Census to seek reconciliation agreementswith other major trading partners. We haveconducted reconciliation studies with Japan, Korea,Mexico, Australia, and the European Community (EC). Although Census does not expect other data exchangesto result from these reconciliations, the studiesprovide valuable information about the quality ofthe U.S. merchandise trade statistics.

In the various reconciliation studies, the combinedeffect of proven nonreporting, underestimation oflow value trade, and unresolved discrepancies rangedfrom three to nine percent of the reported U.S.export value, with most of the studies falling inthe three to seven percent range. All of theseresults predate the AES, which is believed to haveincreased export coverage.

Census suspects that the nonreporting of exports toMexico is greater than that to overseas partners,however, we do not believe it to be as high as thatexperienced with Canada before the data exchange. One reason is that increased automated reporting ofexports has eliminated some of the errors resultingfrom the careless handling of paper documents at theborder crossings.

9

The reconciliation studies revealed thatunderestimation of low value trade accounted for upto three percent of the reported value of U.S.exports to those countries. However, this estimateis very rough since trading partners can definetheir reporting codes differently than the UnitedStates, thus creating more or less low valued traderelative to the United States. We were only able toobtain this information from three trading partners-Australia, Korea, and Mexico. The underestimationappears to differ significantly from country tocountry. So, while these comparisons support ourbelief that we underestimate low value trade, theydo not provide a sufficient basis for adjusting ourestimates.

Customs Compliance Studies

In 1999, Customs conducted an outbound compliancesurvey along the northern border of the UnitedStates. The survey was conducted to determine thelevel of compliance with U.S. export reportingrequirements for goods shipped by truck. The surveywas also designed to identify areas of potentialweakness in compliance with the collection ofaccurate data. For just 75 percent of the shipmentsrequiring an SED was an SED filed. Although thismeasure is not statistically reliable due to the lownumber of sampled shipments (since most shipments toCanada do not require an SED) it does suggest thatthere is a missing documentation problem along thenorthern border. Shipper’s Export Declarations areonly required for licensed goods which are includedin statistics via the data exchange and goodsshipped to other countries from Canadian ports.

Customs has also conducted outbound compliancesurveys for aircraft, vessel and commercial trucksalong the southwestern border. Based on thefindings of all these surveys, the survey committeesmade several recommendations to maintain or in somecases further increase compliance levels. Thoserecommendations included outreach efforts to educatethe export community, improving internal Customsprocedures, and to continue to conduct surveys

10

periodically to identify potential weaknesses incompliance with document submission and accuracy.

Foreign Trade Zones

One of the few known sources of import undercoverageinvolves foreign merchandise that enters foreigntrade zones (FTZs). Foreign merchandise enteringFTZs should be included in the import tradestatistics under general imports. Census oftenencounters difficulties receiving the informationrequired for including FTZ transactions instatistics. Even though there are no actualestimates of the undercounting of FTZ transactionsin the merchandise trade statistics, there is ampleanecdotal evidence to support the idea that aproblem exists.

When foreign merchandise enters an FTZ, the zone isrequired to file a Customs Form (CF) 214A (thestatistical copy of the Foreign Trade ZoneAdmission). In theory, when a zone is authorized,the operator should make arrangements with Census tofile CF 214As either electronically or on paper. Inpractice, zone operators often overlook thisrequirement and Census may not be able to detect theomission.

Foreign Trade Zone merchandise can also be a factorin export undercoverage. When goods are withdrawnfrom an FTZ for exportation, the shipment issupposed to be reported on an SED. However, theexported goods generally are shipped in-bond (on aCF 7512) to the port of exit. This is one of thefew situations where an SED is required for goodsmoving in-bond, so the carrier may not know tocollect the SED. The exporter, who often is not thezone operator, is required to file an SED withCustoms at the port of exit.

In an effort to improve reporting of FTZtransactions, Census has visited a number of zonesand Customs ports to educate operators and Customsstaff about statistical reporting requirements. Census has also participated in meetings andseminars sponsored by the National Association of

11

Foreign Trade Zones to give presentations to memberson statistical reporting requirements. Finally,Census has conducted informal surveys of zoneoperators to gauge their understanding and adherenceto statistical reporting requirements. Theseefforts have significantly improved the awareness ofthe zone operators and Customs staff of thestatistical reporting requirements and increasedfiling of both zone entries and zone exports.

B. Transiting Goods

Shipments of goods moving through the United States enroute from one foreign country to another can createcoverage problems that affect trade statistics. Whentransiting goods are shipped under bond, they are notsubject to duties and are excluded from the merchandisetrade statistics in accordance with the guidelines forinternational merchandise trade statistics established bythe United Nations.

As many U.S. tariffs have been reduced or eliminated, thebenefits of shipping goods under bond have decreased. Asa result, companies often enter transiting goods into theUnited States using the import entry summary and file anexport declaration when the goods leave the United Statesfor the final country of destination. This practice,though less burdensome for companies than the in-transitprocedure, results in the inclusion of the transitinggoods in the trade statistics, contrary to the UNguidelines.

While the practice of entering transiting goods does notaffect the total trade balance, it does affect bilateraltrade balances. For example, if goods shipped fromCanada destined for Mexico enter the United States andare then re-exported to Mexico, the U.S. statistics willshow an import from Canada (thus overstating our deficitwith Canada) and an export to Mexico (understating ourdeficit with Mexico). In addition, this practice maycreate discrepancies between U.S. statistics and those ofthe foreign countries shipping the goods, if the shippingcountry records the goods as exports to the country ofultimate destination, not to the United States.

12

The issue of transiting goods is especially problematicfor bilateral trade between Canada and the United States. While the implementation of the U.S.-Canada Data Exchangehas improved coverage of both countries’ exportstatistics, there is evidence that a substantial numberof Canadian goods transiting southbound through theUnited States are entered as U.S. imports from Canada.

In looking for solutions to undocumented transiting goodstrade, one option is to have U.S. Customs modify theimport entry summary document in the Automated CommercialEnvironment (ACE) to capture transiting goods data thatmay be used to compile supplementary statistics, possiblyon a post-publication basis. Implementing thisrequirement would require regulatory changes.

C. Underestimation of Low Valued Transactions

The United States does not require importers andexporters to file documents for shipments valued lessthan a specified exemption level. Currently, theexemption levels are $250 for imports of certain productsunder quota, $2,000 for other imports and $2,500 forexports. Census then estimates the total value of tradebeneath the exemption level for each country based uponhistorical patterns of trade. The data upon which thesefactors are based are very old and do not reflect recentshifts in trade patterns. For example, on exports, wehave not collected data on transactions valued below$1,000 since the mid 1980's and information ontransactions valued between $1,500 and $2,500 have notbeen reported since October 1989. As a result, thefactors do not adequately reflect the rise of air express(courier) trade, "just in time" processing and otherdevelopments that have likely increased the proportion oflow valued trade. Census has little information on whichto accurately assess the effects of changes in tradepatterns. The trade data reconciliations conducted inthe late 1980s and early 1990s provided some evidencethat the value of export transactions below the exemptionlevel is underestimated and the value of importtransactions below the exemption level is overestimated.

D. Late Filing

13

Late filing and other problems may result in some importor export shipments not being included in the correctmonth’s statistics. During the early 1980s, scarceresources caused Customs to relax its procedures forhandling statistical documents. These changes, coupledwith dramatic increases in the numbers of import andexport transactions, delayed the delivery of largenumbers of documents to the Census processing center. Often these late arrivals were too late for inclusion inthe proper month's statistics. Because these shipmentscarried over into a subsequent month's statistics, Censuscoined the term "carry-over."

During the period 1984 to 1988, Census and Customsintroduced changes to collection and processingprocedures to reduce the effect of carry-over on thestatistics. Ultimately, Census delayed the merchandisetrade release to allow more time for receipt andprocessing of documents. In addition, it revised priormonth trade totals to credit the value of any remaininglate documents to the proper statistical month.

The increased use of automated reporting has alsoimproved the timeliness of the data and decreasedcarry-over. Import and export carry-over is usuallybetween 0.5 and 1.5 percent of the total value for themonth. In 2000, for exports, however, carry-over reacheda high of 2.0 percent of the monthly value. The higherthan normal percentages were due to several new filersusing AES and developmental problems on AES. Theseproblems have been corrected and the carry-overpercentages in 2001 were between 0.1 and 1.0 percent.

As a result of increased automation and otherimprovements, beginning with January 2003, the statisticswill be released roughly seven days earlier than theyhave been released in recent years, generally withinforty-five days of the end of the reference month. Weanticipate only a slight rise carry-over as a result ofreleasing the data earlier.

E. Reporting Errors

Reporting errors are mistakes or omissions made byimporters, exporters or their agents when reportingimport or export shipments. Most errors involve missing

14

or invalid commodity classification codes and missing orincorrect quantities or shipping weights. They have anegligible effect on import, export and balance of tradestatistics. However, they can affect detailed statisticssignificantly if not corrected or corrected inaccurately.

Data from 2001 show that an average of 17 percent of allexport transactions and eight percent of all importtransactions contain one or more reporting errors,although error rates for transactions reported on paperdocuments are much higher (see section on quality andautomated reporting). Most of these errors are easilyand accurately resolved with no significant effect on thepublished statistics.

The data are subjected to several types of processes todetect and correct these errors. The AES and ABI containon-line validation checks that immediately detectreporting errors and refer these errors back to the filerfor correction.

In addition, data from all sources are edited and errorscorrected through clerical or electronic means. Clericalcorrection is time consuming and expensive. Defectivedata referred for clerical correction costs more toprocess than data that successfully pass the edits. Inaddition, data referred for correction may not get intothe correct statistical month. Electronic corrections orimputes, which are less expensive, may introduce someerror into the statistics. Therefore, the most accurateand cost-effective means of resolving reporting errors isto prevent them from occurring.

Beginning in 1991 Census introduced programs to reducereporting errors by educating Customs staff, exportersand forwarding agents about the statistical reportingrequirements.

Customs Port Briefings

Port briefings, conducted in the 1990s, consisted ofa review of Customs statistical collectionactivities followed by briefings for Customs staff. These briefings helped increase Customs staffawareness of their role in the collection ofmerchandise trade statistics. The review, conducted

15

before the briefing, provided background informationfor the briefings and frequently led to suggestedimprovements in Customs processes.

Exporter Education Program

The goal of the Exporter Education Program (EEP) wasto improve the quality of statistics by reducing theSEDs submitted with missing, obsolete, or invalidinformation. The EEP identified exportersresponsible for reporting errors, educated them inproper reporting, tracked their improvement, andperformed follow-up as appropriate.

During the initial two phases of EEP, in 1991 and1992, we concentrated on contacting exportersresponsible for specific types of reporting errors. We mailed more than 80,000 educational packetsconsisting of letters, other educational materials,and samples of defective documents filed by thepertinent exporter. Follow-up analysis showed theprogram reduced reporting errors on SEDs about 20percent.

The final phase of EEP, that began with January 1993statistics, took a different approach. Instead oftargeting exporters making certain types of errors,this phase targeted 16 exporters who accounted forthe most reporting errors on SEDs and profiled theirperformance. Census analysts contacted the exporterby telephone and discussed the reporting errors withthe parties responsible for document preparation. The telephone conversation was confirmed with aletter. Virtually all 16 exporters we selectedshowed immediate significant improvement. Follow-upanalysis of the final phase showed that the programreduced reporting errors on SEDs by 33 percent andthe associated value by 38 percent for targetedexporters. Also 11 exporters eliminated theirreporting errors entirely. This processsuccessfully reduced reporting errors by encouraginga personal and interactive relationship betweenanalysts and document preparation personnel at theexporting company.

16

The Exporter Education Program ended in 1994,however, we continue to identify, contact, educate,and conduct follow-up with filers making reportingerrors on SEDs. For example, the NPC sendsinaccurate or incomplete paper SEDs to Censusheadquarters so that staff can contact the qualifiedexporter or freight forwarder to acquire the correctinformation. The correct data are then entered intoAES. This process is successful in two ways. First, the company filing the data is educated as tothe correct way to complete an SED. Second, Censushas an opportunity to market AES to companies stillfiling data via paper documents.

Foreign Trade Export Seminar Program

In September 1999, Census initiated a new seminarseries designed to inform export data providers ofexport regulations and requirements. The seminarsexplain to participants the activities of variousfederal agencies involved in the export process; howto correctly complete and file SEDs; AES and itsadvantages; terms used by the foreign tradecommunity; and the Foreign Trade StatisticsRegulations, 15 CFR, Part 30.

Census holds between 10 and 12 seminars per yearwith 50 to 200 attendees at any one seminar. Speakers for the seminars include Census staff,Customs staff, and trade specialists from othergovernment agencies and private organizations. Using participant feedback as a measure, thisseminar series has been very successful in providingexport data providers with the information necessaryto file timely and accurate statistics.

AES User Meetings

Census and Customs jointly hold about six meetingsfor AES users each year. The meetings are designedto inform the trade community of the latesthappenings in AES including information on systemenhancements; common reporting problems; and ForeignTrade Statistics Regulations. The goal is to hosttwo meetings each year on the east and west coastand one in a southern and central state.

17

Participation averages between 50 to 125 people. These meetings have been so successful that thetrade community requested an additional meeting inNew York City, a heavy volume exporting port.

The AES User Meetings educate the trade community sothat their electronic reporting is timely andaccurate. By hosting user meetings throughout thecountry, Census reaches a large segment of the tradecommunity, readily responding to their needs andimproving export trade statistics.

F. Data Capture Errors

As explained earlier, Census captures import and exportinformation from administrative documents and throughvarious automated collection programs. It subjects datafrom each source to various checks to ensure accuratecapture of the information.

Automated Submissions

Data received through automated collection programs(ABI, AES, and U.S.-Canada Data Exchange) arescreened upon receipt to ensure completeness and anacceptable level of accuracy. On-line validationsand commodity specific edits are applied to datareported electronically. Other security proceduresguard against file damage that may compromise theintegrity of the data.

Paper Documents

Census uses a combination of data entry edits andkeying verification to ensure the accurate captureof information from the SEDs and Customs EntryForms. Automated data entry checks validate allcritical data fields and signal the keyer when theinformation is invalid. This process prevents manykeying errors and intercepts many reporting errors.

After data entry, the keyed information is subjectedto sample verification. Certain information commonto all transactions and all information fortransactions valued over $1 million for imports andfor exports is 100 percent verified. Lower valued

18

transactions are verified from a one percent sampleof documents. This provides the information toassess the quality of keying and to promote ordemote keyers from sample verification. Only fullyqualified keyers may key documents valued over $1million, and may key other documents at the onepercent sample rate. These keyers maintain a fieldkeying error rate of one half of one percent orless.

G. Quality and Automated Reporting

The increased use of automated reporting has generallyimproved the quality of the merchandise trade statistics. In 2001, Census received 99 percent of importtransactions and 85 percent of export transactionselectronically. As explained earlier, automatedreporting allows Census to receive and compile the dataquickly and include almost all shipments in the correctstatistical month. Automated reporting minimizes lostdata and provides better control over non-filing. Finally, since Census receives pre-edited data throughthe U.S.-Canada Data Exchange, ABI, and AES, theinformation arrives at the Census with fewer reportingerrors. Recent studies show that 56 percent of exportrecords captured from SEDs (export paper documents)contain an error as compared to about ten percent of pre-edited data records. Likewise, about 37 percent of datarecords captured from Customs Entry Forms 7501 (importpaper documents) contains an error as compared to eightpercent of automated records.

The enhanced quality of automated submissions resultsfrom the high standards required of automated reporters. Customs and Census test and approve all ABI brokers andAES participants before they can submit shipmentselectronically. In addition, many statistical Censusedits reside in the Customs computer system that housesABI and AES. These edits intercept data problems forreturn to the ABI broker or AES participant forresolution. Census and Customs client representativestrack and assist filers who frequently report problemdata; however, data filers who fail to resolve problemsor who repeatedly make the same errors face possiblefines or penalties.

19

Furthermore, Census edits also reside in the StatisticsCanada import processing system to meet U.S. exportneeds. Representatives from the statistical and customsagencies of each country meet regularly to resolve anyproblems and keep abreast of any changes that may affectthe export programs.

H. Disclosure Avoidance

For published reports, no disclosure analysis isperformed. Anyone believing that their data may bedisclosed in the published statistics at a commoditylevel, commodity by country level, or commodity bycountry by district level may send a written inquiry toMr. C. Harvey Monk, Jr., Chief, Foreign Trade Division. Requests may be sent via facsimile at (301) 457-2645, e-mailed to C.Harvey.Monk.Jr@census.gov or mailed to Chief,Foreign Trade Division, U.S. Census Bureau, Washington,D.C. 20233.

I. Requests for Re-Verification of Trade Statistics

Data users may request a re-verification of recentlyreleased trade statistics. To request re-verification,data users submit formal, written requests to theCommodity Analysis Branch of the Foreign Trade Division. These requests include the Harmonized CommodityClassification Code, which is the Harmonized TariffSchedule (HTS) number for imports and the Schedule Bnumber for exports, the reason for requesting re-verification and the month(s) or year(s) in question. Ifthe inquiry is restricted to trade with specificcountries, the countries must be identified in therequest. All requests for data re-verification aredirected to the Chief of the Commodity Analysis Branchvia facsimile at (301) 457-1158, e-mailed toPaul.E.Herrick@census.gov or mailed to Chief, CommodityAnalysis Branch, Foreign Trade Division, U.S. CensusBureau, Washington, D.C. 20233.

III. ADJUSTMENTS

Census adjusts merchandise trade data to remove the effects ofseasonal influences and price shifts. Adjusted data appear inthe monthly FT900, U.S. International Trade in Goods and

20

Services, released by the Department of Commerce. Theadjustments remove known or predictable influences that mayobscure true changes in the volume of merchandise traded.

A. Adjustment for Seasonal and Trading Day Variation

These adjustments, computed using the Census X-12-Arimaprogram (a variation of X-11 with Arima modeling), removepredictable calendar effects from the data. X-12develops seasonal factors from historical series of data. The merchandise trade series exhibit a high level ofirregular variation. To help overcome this problem,Census uses the mathematical modeling of X-12 to adjustfor level shifts and outliers in the data series. Thechoice of model can affect the calculation of factors andwill change over time as more data are added to theseries. Generally the use of longer series will helpclarify the seasonal patterns.

However, the adjustments will not remove all seasonalvariation. If the timing and magnitude of the seasonaleffect is not predictable or stable, Census cannot adjustthe series. As working day effects (the differencesbetween the level of trade on specific days of the week)are generally small, the high irregular variation of manyseries prevents accurate identification of thesemovements. Thus, while the final published series showno clear evidence of working day effect, some residualinfluence may be present.

B. Adjustment for Price Change

Besides seasonally adjusting the data, Census alsopublishes trade data on a constant dollar basis. Theadjusted data provide an estimate of the change in thevolume of trade, as distinct from value changes resultingfrom price shifts. Census bases the adjustments on theInternational Price Indexes (IPI) published by the Bureauof Labor Statistics (BLS), supplemented with deflatorsproduced by the Bureau of Economic Analysis (BEA).

Census deflates trade in computers and peripherals usingthe quality adjusted hedonic deflator computed by BEA. The BEA uses the attributes (speed, memory capacity,etc.) of the computer equipment to estimate the value of

21

quality improvements over time. Prices relative toquality have dropped rapidly for this product grouping. As a result, computers and peripheral equipment comprisea larger part of total value on a constant dollar basisthan on a current dollar basis. Users may wish toexamine this area of trade separately from total trade.

Currently, the Census trade data are computed on acurrent-weighted basis. We are working to convert to thechain-weighted Fisher basis used by BEA in the NationalAccounts. We expect to implement this change in April2003.

IV. Conclusion

The merchandise trade statistics program was originallydesigned to meet the basic needs of policy makers and providesreasonable estimates of total trade with other countries. However, the program has grown in scope and complexity,attempting to meet the needs of many different data usergroups. For example, the program collects transportation datafor air and vessel carriers, port authorities and theDepartment of Transportation. It also collects detailedcommodity information for trade associations, marketinganalysts and government agencies investigating the impact ofinternational trade on local industry and economies.

The needs of data users expand and change continuously. Census and Customs strive to provide accurate and completeinformation for all data users. Realistically, however, theprogram meets the needs of some data users better than others.

To address the concerns of these many data users, Census willcontinue to release periodic quality profiles. These profileswill include the results of studies conducted to assess thequality of the data and any reports on the results of qualityassurance programs. The profiles also will describe anychanges in the programs that may effect the quality of thestatistics. In this way, Census hopes to provide theinformation and knowledge needed to make its many data users"educated consumers" of the merchandise trade data.

22

Appendix A

23

Appendix B

24

Appendix CAnnotated Bibliography

Bureau of the Census, 1992, "Adjustment of Foreign Trade DataforSeasonality and Price Change." Foreign Trade Division, Bureauof the Census, U.S. Department of Commerce, Washington, DC.

This technical paper explains the origin of the constantdollar series and the methods used to seasonally adjust anddeflate the merchandise trade data. Contact: Kathy Puzzilla(301) 763-7003.

Bureau of the Census, 1998, "Understatement of ExportMerchandise Trade Data." Foreign Trade Division, Bureau ofthe Census, U.S. Department of Commerce, Washington, DC.

This technical paper explains reasons for the understatementof the value of U.S. exports and proposed solutions to correctthis understatement. Contact: David Dickerson (301) 763-3080

National Research Council, Panel on Foreign Trade Statistics,1992, Behind the Numbers: U.S. Trade in the world Economy. Washington, DC: National Academy Press.

This report covers issues related to the merchandise tradestatistics and the need for information to guide the publicand private sectors in the current world economy. From aquality standpoint, it focuses on improving data collection,accuracy and analysis.

Mozes, Steven and Oberg, Diane, 1990-2001 “U.S.-Canada DataExchange.” Statistics Canada and Bureau of the Census.

This technical paper gives detailed information on the U.S.-Canada Data Exchange. Contact: Diane Oberg (301) 763-2223

25

Appendix DGlossary of Acronyms

ABI - Automated Broker Interface

ACE - Automated Commercial Environment

AERP - Automated Exporter Reporting Program

AES - Automated Export System

BEA - Bureau of Economic Analysis

BLS - Bureau of Labor Statistics

CF - Customs Form

EC - European Community

EEP - Exporter Education Program

FTZ - Foreign Trade Zone

GDP - Gross Domestic Product

HTS - Harmonized Tariff Schedule

IPI - International Price Indexes

NPC - National Processing Center

SED - Shipper's Export Declaration