The U.S. antidumping law enjoys broad political support in part because so few people understand how the law actually works. Its rhetoric of “fairness” and “level playing fields” sounds appealing, and its convoluted technical complexities prevent all but a few insiders and experts from understanding the reality that underlies that rhetoric. In this study we seek to penetrate the fog of complexity that shields the antidumping law from the scrutiny it deserves. Here we offer a detailed, step-by-step guide to how dumping is defined and measured under cur- rent rules. In addition, we identify the many methodological quirks and biases that allow normal, healthy competition to be stigmatized as “unfair” and punished with often crippling- ly high antidumping duties. The inescapable conclusion that follows from this analysis is that the antidumping law, as it currently stands, has nothing to do with maintaining a “level playing field.” Instead, antidumping’s primary function is to provide an elaborate excuse for old-fashioned protectionism. We illustrate the antidumping law’s serious methodological flaws in a variety of different ways. First, we use simplified examples to demonstrate how particular steps in the dumping calculation operate to generate phantom dumping findings. Next, we use actual case records from 18 different dumping determinations to quantify the effects of methodological dis- tortions in specific, real-life cases. Finally, we present a detailed hypothetical case study in which each step in the dumping calculation is explained and faithfully recreated. In that case study, we show how a foreign producer that sells widgets in the United States at net prices 13.95 percent higher than in its home market nonethe- less winds up with a dumping margin of 7.37 percent. Antidumping 101 The Devilish Details of “Unfair Trade” Law by Brink Lindsey and Dan Ikenson November 21, 2002 No. 20 Brink Lindsey is director of the Cato Institute’s Center for Trade Policy Studies. Dan Ikenson is a trade policy analyst with the center. Executive Summary
Transcript
The U.S. antidumping law enjoys broadpolitical support in part because so fewpeople understand how the law actuallyworks. Its rhetoric of “fairness” and “levelplaying fields” sounds appealing, and itsconvoluted technical complexities preventall but a few insiders and experts fromunderstanding the reality that underliesthat rhetoric.
In this study we seek to penetrate the fogof complexity that shields the antidumpinglaw from the scrutiny it deserves. Here weoffer a detailed, step-by-step guide to howdumping is defined and measured under cur-rent rules. In addition, we identify the manymethodological quirks and biases that allownormal,healthy competition to be stigmatizedas “unfair” and punished with often crippling-ly high antidumping duties. The inescapableconclusion that follows from this analysis isthat the antidumping law, as it currentlystands, has nothing to do with maintaining a
“level playing field.” Instead, antidumping’sprimary function is to provide an elaborateexcuse for old-fashioned protectionism.
We illustrate the antidumping law’sserious methodological flaws in a variety ofdifferent ways. First, we use simplifiedexamples to demonstrate how particularsteps in the dumping calculation operate togenerate phantom dumping findings.Next, we use actual case records from 18different dumping determinations toquantify the effects of methodological dis-tortions in specific, real-life cases. Finally,we present a detailed hypothetical casestudy in which each step in the dumpingcalculation is explained and faithfullyrecreated. In that case study, we show howa foreign producer that sells widgets in theUnited States at net prices 13.95 percenthigher than in its home market nonethe-less winds up with a dumping margin of7.37 percent.
Antidumping 101The Devilish Details of “Unfair Trade” Law
by Brink Lindsey and Dan Ikenson
November 21, 2002 No. 20
Brink Lindsey is director of the Cato Institute’s Center for Trade Policy Studies. Dan Ikenson isa trade policy analyst with the center.
Executive Summary
Introduction
For many years the U.S. antidumping lawhas enjoyed strong political support fromRepublicans and Democrats alike. Whataccounts for this enduring, bipartisan popular-ity? In the first place, the law’s rhetoric is com-pelling. After all, which members of Congresswould like to stand up and say they favor ille-gal and unfair dumped imports? Who could beopposed to a “level playing field”?
Meanwhile, the reality behind the rhetoric isobscure. The antidumping law is notoriouslycomplicated, and its inner workings are knownonly to a select handful of users, targets, bureau-crats, and lawyers. The jargon alone—EP, CEP,POI, POR, CONNUM, FUPDOL, TOT-PUDD, model match, cost test, arm’s-lengthtest, CV, facts available, DIFMER, and on andon—is enough to make a nonspecialist’s eyesglaze over. As a result, most supporters of thelaw simply take its appealing rhetoric at facevalue. If it sounds good, it must be good.
The purpose of this paper is to cut throughthe fog of technical complexity that surroundsthe antidumping law—to show how it reallyoperates, and show in particular that its actualoperation all too often has nothing to do withits fine-sounding rhetoric.
An earlier Cato Institute study by one of thecoauthors of this paper pursued a similar object.That study, titled “The U.S. Antidumping Law:Rhetoric versus Reality,” focused on the theoret-ical justifications that antidumping supportersoffer for the law and then examined antidump-ing practice in light of those justifications.1 Theconclusion: Antidumping measures seldom suc-ceed in targeting “unfair trade” as antidumpingsupporters define that term.
This paper’s focus is more practical and lesstheoretical. Here the emphasis is on the nutsand bolts of how the U.S. Department ofCommerce defines and calculates dumping. Inparticular, we expose the dirty little secrets ofantidumping—the methodological quirks andbiases that result in findings of dumping evenwhen the U.S. prices of imported goods areidentical to or even higher than the prices
charged by the foreign producer of those goodsin its own home market. Once those quirksand biases are understood, the illusion that theantidumping law in its current form restrictsonly “unfair trade” cannot be sustained.
We are not attempting here a comprehen-sive analysis or critique of the antidumping law.For antidumping measures to be imposedunder U.S. law, two major requirements mustbe satisfied: (1) the DOC must find thatimports are being dumped, and (2) theInternational Trade Commission must findthat dumped imports are causing or threaten-ing injury to a domestic industry. In this paperwe do not address the ITC’s injury analysis.Neither do we address various proceduralaspects of antidumping investigations—including standing requirements, initiationstandards, use of “facts available,” or the distri-bution of duties to domestic producers underthe so-called Byrd amendment.
Although much in these other elements ofantidumping practice is ripe for criticism, wechoose to home in on the central question ofantidumping policy: What constitutes dump-ing? How that question is answered determinesthe fundamental character of the antidumpinglaw. If it can be shown that normal, healthycompetition is regularly stigmatized as “unfair”dumping because of methodological flaws andbiases, then the law itself is fundamentallyflawed—not just according to the ivory-towerstandards of economists and policy wonks, butaccording to the very standards upheld by thelaw’s supporters.
Champions of the antidumping status quoargue that the law is not protectionist—that is, itdoes not discriminate against imports generally infavor of domestic production.Rather, its purpose ismerely to clear the way for normal, healthy inter-national competition by discouraging “unfair,”market-distorting trade. According to its support-ers, the antidumping law acts to preserve a “levelplaying field” on which domestic and foreign pro-ducers can compete on the basis of who makes thebest product at the lowest cost—as opposed to thesituation in which some foreign producers enjoy anartificial advantage because of government-causeddistortions in their home market. But if, in fact,
2
The antidumpinglaw is notoriouslycomplicated, and
its inner workingsare known only to a
select handful ofusers, targets,
bureaucrats, andlawyers.
dumping is defined in such a way that even per-fectly innocent import competition is classified asunfair, then the distinction between antidumpingmeasures and garden-variety protectionism col-lapses—and even antidumping supporters shouldadmit that there is a problem that needs fixing.
The Antidumping Process
Before critiquing the antidumping law’sflawed methodologies, it is necessary first tounderstand, at least in broad outline, how theantidumping process works. Following, there-fore, is a basic overview of current antidumpingpractice, with particular attention to theDOC’s rules and procedures for identifyingand measuring dumping.
Investigations and ReviewsDumping is defined as the sale of a com-
modity in an export market (i.e., the UnitedStates under U.S. law) at a price less than “nor-mal value.”2 Normal value is based either onthe price of the same or a similar product in acomparison market (normally the foreign pro-ducer’s home market) or on “constructedvalue,” the cost to produce the product plussome amount for profit.3 The extent of dump-ing is called the “dumping margin,” which iscalculated by subtracting the export price fromnormal value and dividing the difference(assuming it is positive) by the export price. Inother words, if a foreign producer sells widgetsfor 10 at home and for 8 in the United States,its dumping margin is (10 – 8)/8, or 25 percent.
For antidumping duties to be imposed, it isnot enough that dumping be found to exist. Inaddition, the authorities must determine thatthe domestic industry is materially injured orthreatened with material injury, or else that thedevelopment of an industry is materiallyretarded, by reason of dumped imports.
In the United States, an industry can seekrelief under the antidumping law by filing apetition with the ITC and the DOC.Domestic producers joining the petition arethus called “petitioners.” Producers accountingfor at least 25 percent of domestic production
(by volume) must support the petition, andopposition by other producers must constituteless than 50 percent of the total output of allproducers expressing an opinion one way or theother. Petitions are required to contain someevidence of dumping and injury in order to ini-tiate an investigation. However, the evidentiaryrequirements are quite modest.
Once an antidumping investigation is initi-ated, the ITC has 45 days to determine whetherthere is reason to believe that dumped importsare causing or threatening injury to a domesticindustry. Affirmative preliminary findings arerendered in about 79 percent of cases.4
Following an affirmative preliminary finding bythe ITC, the DOC issues questionnaires to“mandatory respondents”––the largest known for-eign producers and exporters of subject merchan-dise from the countries in question.5 Failure torespond to the DOC’s questionnaire or failure tocooperate fully in the proceeding typically results inthe assignment of an adverse rate of duty, which isusually based on allegations in the domestic indus-try’s petition. If a foreign producer has any hope ofmaintaining access to the U.S. market, it is forcedto respond to the DOC’s voluminous requests forinformation.Responding to an antidumping ques-tionnaire usually requires the diversion of signifi-cant company resources and retaining legal,accounting, and economic expertise.
The DOC normally makes its preliminarydetermination (prelim) within 140 days of theinvestigation’s initiation (although the deadlinecan be extended by 50 days). If the DOC pre-lim is affirmative, liquidation (i.e., final deter-mination of duties owed to the CustomsService) is suspended for all future subjectimports, and a bond must be posted to coverpossible antidumping duties at the rateannounced in the prelim. The DOC thenmakes its final determination (final) within 75days of the prelim (although, here again, thedeadline can be extended—this time by up to60 days). Between the prelim and the final, theDOC conducts an on-site verification of therespondents’ questionnaire data and considersfactual and legal arguments submitted by peti-tioners and respondents. After the final,assuming it is affirmative (an outcome that
3
Dumping isdefined as the saleof a commodity inan export market ata price less thannormal value.
occurs about 94 percent of the time),6 respon-dents must pay cash deposits on possibleantidumping duties at the rate announced inthe final.
Once the DOC issues its final determina-tion, the ITC normally has 45 days to make itsfinal injury finding. If that determination isaffirmative (an outcome that occurs about 83percent of the time),7 an antidumping order isissued, which subjects prospective imports toantidumping duty deposits equal to the calcu-lated rate of dumping.
The antidumping deposit rate is only anestimate of dumping liability. The rate is basedon dumping margins calculated for the periodof investigation (POI), normally the four mostrecently completed calendar quarters beforeinitiation—that is, a period that concludesmonths before duty liability generally begins.Final liability is determined by administrativereviews conducted later by the DOC. The firstperiod of review (POR) covers imports fromthe beginning of duty liability to the firstanniversary of the antidumping duty order.Subsequent PORs cover imports from oneanniversary date to the next. The DOC’s finaldetermination in a review settles final dumpingliability for imports during the POR. If theamount of dumping found is more than theamount of cash deposits, the respondent owesthe difference with interest. If, on the otherhand, the amount of dumping is less, therespondent gets a refund with interest. In addi-tion to settling final liability for past imports,the final determination in a review also estab-lishes a new deposit rate for future imports.
Dumping margins can thus be recalculatedannually through administrative reviews. Thecontinued existence of injury, on the otherhand, is revisited only once every five years inso-called sunset reviews. Under the sunsetreview provision, an antidumping duty order isterminated automatically after five years unlessa sunset review is requested. In this review, theDOC and the ITC determine whether termi-nation of the order would be likely to lead tothe continuation or resumption of dumpingand injury, respectively. If both make affirma-tive determinations, the order continues for
another five years. Between July 1998 andAugust 2002, 354 sunset reviews were initiat-ed, of which 265 were contested by petitioners.The outcomes in 2 of these 265 contested caseswere still pending.The DOC made affirmativesunset determinations to continue the order inall but 4 of the 263 decided cases, while theITC voted affirmative 72 percent of the time.8
The DOC QuestionnaireTo conduct its dumping investigation, the
DOC issues detailed questionnaires to the pri-mary foreign producers of the “subject mer-chandise,” as the investigated imports areknown. The questionnaire has four and some-times five distinct parts—Sections A, B, C, D,and sometimes E. Section A seeks informationon the company’s corporate structure and affil-iations, distribution process, sales process,accounting and financial practices, and theproducts under investigation generally.
Section B concerns the company’s sales of“such or similar” merchandise (i.e., products thatare identical or similar to the subject merchandiseunder investigation) in the comparison market.The comparison market is normally the foreignproducer’s home market, but a third-countryexport market may be used instead if the homemarket is deemed not “viable.”9 The DOC seeksinformation on products sold, selling prices,quantities sold, customer relationships, customerclass (i.e., distributor, wholesaler, retailer), trans-portation and warehousing expenses, insurancecosts, selling expenses,discounts and rebates,war-ranties, commissions, packing costs, and anythingelse related to home-market sales.This portion ofthe response is supplemented by a separate com-puter data file containing detailed information foreach sales transaction made during the period ofinvestigation and is used to calculate the “foreignmarket value” or normal value—in other words,the benchmark against which U.S. prices arecompared.
Section C is analogous to Section B butcovers sales made in the United States.Typically, more information is sought withrespect to U.S. sales because the process of sell-ing in an export market involves additionaltransportation and selling processes. Also, in
4
To conduct itsdumping investiga-
tion, the DOC issuesdetailed question-
naires to the primaryforeign producers
of the subjectmerchandise.
many cases, the U.S. importer is related to theforeign respondent and therefore the prices tothe importer are not used as the basis for thedumping calculation. In these cases—known asconstructed export price (CEP) transactions—the U.S. price is based on the price of the resaleby the related importer to the first unaffiliatedcustomer. When U.S. sales involve importationby a related company, expenses incurred by thatrelated importer are required to be reported inSection C.
Section D covers the cost of producing thesubject merchandise and the foreign such orsimilar merchandise. This portion of the ques-tionnaire seeks to obtain data on productionquantities, labor costs, materials costs, over-head, and general and administrative expensesassociated with production.The computer datafiles supplementing the narrative response toSection D must contain detailed unit cost ele-ments for every product subject to the investi-gation that was sold in both markets.
In cases where there is further processing ofsubject merchandise after importation by arelated company and before sale to an unaffili-ated customer, the DOC issues a Section E,which seeks information on the additionalmanufacturing processes and costs.
CONNUMs and Product DefinitionThe first step in comparing U.S. and foreign-
market prices consists of determining whichprices to compare to each other. If productsidentical to those sold in the United States arenot sold in the home market—which is the casein most antidumping investigations—the DOCcompares the average U.S. price to the averagehome-market price of the “next most similar”product. Among the most important aspects ofantidumping calculation methodology, there-fore, are product definition and determination ofthe next most similar products.
The product, as defined by the DOC, willalmost always be different from the product asdefined by the respondent. That is, the DOCcompares prices (and calculates costs) at theDOC-defined product level, not at the compa-ny-specific product code level. This departurefrom a respondent’s record-keeping protocol
can add significant complications to the prepa-ration of sales and cost records in response tothe DOC’s questionnaire.
Products are defined for antidumping pur-poses by the specific product characteristicsthat the DOC determines are necessary for“model matching”—that is, the characteristicsthat determine which products are more or lesssimilar to each other and thus which productsare compared with each other. For some prod-ucts, the relevant characteristics are few and theoptions within each characteristic are limited.For example, widgets may be classified by size(large or small) and material (rubber or plastic).Using these classifications, there are only fourpossible models (large rubber, large plastic,small rubber, small plastic). And since there areonly two possibilities within each characteris-tic, it is only necessary to determine “rankbetween” the characteristics, not “rank within”each characteristic. If size is considered a moreimportant matching characteristic than mater-ial, then the best match for a large rubber wid-get in the absence of an identical product in thehome market is a large plastic widget.Alternatively, if material is more significantthan size, the best match for a large rubberwidget would be a small rubber widget.
The DOC creates its own product code,known as the CONNUM, or control number,which reflects the relevant characteristics of theproduct. The CONNUM is constructed as aseries of the relevant product characteristics. So,for example, a small, rubber widget might beassigned a CONNUM of 11,where the first digitcorresponds to “size” (small = “1”; large = “2”) andthe second digit corresponds to “material” (rubber= “1”; plastic = “2”). This construction wouldreflect the decision that size is more importantthan material for this particular product becausethe “next most similar product” to an 11 is a 12,not a 21. Matching across sizes would occur onlyif there were no match across materials in thesame size.
If there were a third type of material (sayvinyl) and a third size (say medium), then therewould be nine possible products, and it wouldbe necessary to rank the order of similarity“within” each characteristic, as well as
5
The first step incomparing U.S.and foreign-marketprices consists ofdetermining whichprices to compareto each other.
“between.” Here things can get tricky. Whichis more similar to medium—large or small? Isrubber more similar to plastic or to vinyl?
In some cases there are only a few charac-teristics and a few alternatives within eachcharacteristic. But for others, particularly thoseinvolving steel, there are 10 to 20 characteris-tics, each with several or dozens of alternativevalues. For example, in the recent investigationof cold-rolled carbon steel from Germany,there were 14 relevant product characteristics,including carbon content, tensile strength,thickness, tolerance, surface quality, and hard-ening process. In this and other steel casesthere can be literally hundreds of thousands ofactual products.10
Price AdjustmentsDumping calculations are never made on
the basis of comparing actual sales prices.Instead, the DOC subjects actual sales pricesto a dizzying variety of adjustments. Dumpingmargins are then determined by comparingadjusted U.S. to adjusted home-market prices.
Selling and delivering products to differentmarkets entail incurring certain expenses thatare unique to each market.The DOC addressesthese issues by attempting to quantify allexpenses incurred after production and throughthe sale of the product in each market. Theseexpenses are then deducted from the gross sell-ing prices, yielding ex-factory prices (the pricesthe products would have fetched at the factorydoor), which presumably permit apples-to-apples comparisons. For example, it may be nec-essary to advertise in one market, where theproduct is less known or where greater competi-tion prevails, while it is unnecessary to advertisein another market. The sales operation in onemarket might involve selling agents, who receivecommissions on sales, whereas in another mar-ket, customers purchase directly from a cata-logue without any sales intermediary.
Selling in different markets, particularlymarkets at different distances from the point ofproduction, involves differences in transporta-tion costs. Selling in one’s home market mightinvolve some trucking fees and other localexpenses, but selling in an export market usu-
ally entails additional transportation costs suchas outbound freight to the port of exit, ware-housing fees, ocean shipping, insurance, bro-kerage and handling fees, customs duties, andinland freight costs.
Some of these expenses are precisely quan-tifiable; others can be estimated only roughly.The DOC seeks to obtain information on allsorts of expenses, direct and indirect, realizedand imputed. Many of the expenses are deduct-ed from gross selling prices, while others areused to offset or limit deductions made fromgross prices in a particular market.
Discounts and Rebates. Discounts and rebatesare the first class of expenses deducted from grossselling prices. Companies often provide incen-tives for customers in particular markets to enticeearly payment or large-volume purchases, or topass along savings afforded them by their ownsuppliers. All discounts and rebates are deductedfrom gross selling prices in both markets.
Movement Expenses. Movement expensescomprise all of the costs incurred by the sellerto transport merchandise from the factory tothe customer. Depending on the sales anddelivery terms specified on the invoice, theseexpenses could include freight from the facto-ry to a warehouse, warehousing costs, freightfrom the warehouse to the customer, freight tothe port of exit, marine shipping and insurance,brokerage and handling fees, customs duties,and inland freight in the export market. Allmovement expenses are deducted from grossselling prices in both markets.
Direct Selling Expenses. Direct selling expensesare characterized generally as expenses incurred tofacilitate specific sales. These expenses includeadvertising costs involved in promoting the sub-ject or “such or similar” merchandise, warrantyexpenses associated with materials and labor toservice defective merchandise, and commissionspaid on particular sales.Adjustments to gross sell-ing prices are made for all direct selling expensesin both markets.
Indirect Selling Expenses. Indirect sellingexpenses are costs incurred on behalf of a com-pany’s sales operation that are not directlyattributable to particular sales. These expensesinclude sales department overhead such as
6
Dumping calcula-tions are never
made on the basis ofcomparing actual
sales prices. Instead,the DOC subjects
actual sales prices toa dizzying variety of
adjustments.
rent, salaries, and supplies. They can alsoinclude advertising of a general nature, used topromote the company name or brand, but notspecific products. The overhead expenses of atechnical service department devoted to hon-oring warranties and to providing customersupport may also be considered indirect sellingexpenses. Unlike each of the previous expensegroups described, indirect selling expenses arenot given similar treatment in both markets.They are deducted from U.S. prices under cer-tain situations—so-called constructed exportprice (CEP) transactions—but are not alwaysdeducted, at least not entirely, from the pricesof the home-market products.
Imputed Expenses. Imputed expenses areimplicit costs that do not accrue in an actualaccounting sense but are presumed to affectprices. An imputed expense is not actuallyincurred; it is instead an opportunity cost. Inantidumping cases, the DOC calculatesimputed credit expenses and inventory carry-ing costs. Together, these expense adjustmentsreflect the opportunity cost of not receivingpayment immediately after a product is pro-duced. Inventory carrying costs impute thecosts associated with the period between pro-duction and sale; credit expenses impute thecosts associated with the period between saleand receipt of payment. Each is calculated typ-ically by multiplying the number of days in therespective period, divided by 365, times theprevailing local short-term interest rate, timesthe gross price (in the case of credit) or the costof production (in the case of inventory carryingcosts). Imputed credit expenses are treated asdirect selling expenses; inventory carrying costsare treated as indirect expenses.
Level-of-Trade Adjustments. Often, cus-tomers in the home market are of a different“class” than those in the U.S. market. Forexample, U.S. customers may be large whole-salers, while home-market customers are smallretailers or even end users. The common com-mercial practice in many industries is to offerdifferent prices to different classes of customersbecause their size or function may affect thevolume or variety of purchases. Sellers mayoffer different services or incentives to different
classes of customers to attract their business orcement their loyalty. When sales are made todisparate customer classes, any price differ-ences may simply reflect the different nature ofthe customers’ businesses. The DOC addressesthis issue by attempting to match prices of U.S.and home-market sales at the same “level oftrade.” When there are no home-market salesat the same level of trade, and comparisonsmust be made across such levels, the DOC willconsider making a level-of-trade adjustment toaccount for the difference. If it can be demon-strated that a consistent pattern of price differ-ences prevails between the levels of trade, andthat the seller performs distinctly differentfunctions on behalf of the customers in the dif-ferent levels, and that the home-market level ismore remote from the factory, an adjustmentwill be made to normal value.
Difference-in-Merchandise Adjustments. As theDOC attempts to account for inevitable dispari-ties in prices asked at different levels of trade, italso makes price adjustments when nonidenticalproducts are compared. When products identicalto the U.S. product are not sold in the home mar-ket, or are sold but deemed ineligible for pricecomparisons through the various tests and proce-dures described below, the prices of similar, butnot identical, products are compared. An elabo-rate imagination is not required to appreciate thatdifferent products sold in different markets mighthave different prices.
The DOC makes a difference-in-merchan-dise (DIFMER) adjustment when prices of non-identical products are compared. A DIFMERadjustment is calculated as the difference betweenthe variable costs of manufacturing the two dis-tinct products. So, instead of directly comparingthe net U.S. price and the net home-market priceof different products, the home-market price isadjusted by the difference in variable productioncosts first,presumably to countervail the price dif-ference arising from the product difference.
CEP Profit and CEP Offset. Sales in theUnited States fall into one of two classifications.Export price (EP) sales are transactions betweenthe exporter and an unaffiliated importer; con-structed export price (CEP) sales are transac-tions in which the importer is affiliated in some
7
When sales aremade to disparatecustomer classes,any price differ-ences may simplyreflect the differentnature of the cus-tomers’ businesses.
manner with the exporter, and thus the transac-tion is deemed unreliable for purposes of dump-ing calculations. Rather than use the informa-tion in this primary transaction, the DOC usesthe U.S. prices of the resales by that affiliatedimporter to the first unaffiliated customer.
In CEP transactions, the DOC deductsfrom U.S. prices not only the U.S. indirect sell-ing expenses but also the estimated profit onU.S. operations. The CEP profit is determinedby calculating total revenues (quantity timesprice) in both the U.S. and home markets andsubtracting from that figure the total cost ofproducing, selling, and transporting (quantitytimes [cost plus selling expenses plus move-ment expenses]) the merchandise in both mar-kets. This aggregate figure is allocated to eachCEP transaction on the basis of the ratio of thesales-specific CEP expenses to the total costs.
When indirect selling expenses are deducedfrom U.S. prices, the DOC calculates a CEPoffset that is deducted from the foreign-marketprice before the unit margins are calculated.The value of the offset is determined by fol-lowing a complicated set of computer instruc-tions to ultimately evaluate various expenses inboth markets. If certain conditions are met(e.g., commissions are paid in the home marketbut not in the U.S. market), the offset takes ona certain value. If the conditions are not metbut others are, the offset takes on a differentvalue. Although the offset can have a smallervalue than the U.S. indirect selling expenses, itcan never exceed the amount of U.S. indirectselling expenses. Finally, if the DOC is able tocalculate a level-of-trade adjustment, it doesnot calculate the CEP offset.
Arm’s-Length and Cost TestsAfter defining the products in both markets,
but before matching home-market sales to U.S.sales, the DOC filters out some (or maybe evenall) home-market sales with two separate tests:the arm’s-length test and the cost test. These fil-ters are used,ostensibly, to eliminate from consid-eration all home-market sales that may be aberra-tional, or outside “the ordinary course of trade.”
The purpose of the arm’s-length test is todetermine whether sales to affiliated11 cus-
tomers in the home market have been made atprices and on terms comparable with thosegranted to unaffiliated customers. The testinvolves comparisons of the average net sellingprices per product for each affiliated customerto the average net selling prices per product toall unaffiliated customers.
A ratio is calculated in which the numera-tor is the average net price per product peraffiliate and the denominator is the average netprice of that same product to all unaffiliatedcustomers. A similar ratio is calculated for eachunique combination of affiliated customer andproduct.12 Finally, a weighted-average ratio iscalculated for each affiliate. If that ratio is atleast 99.5 percent, then all sales to the affiliateare accepted as having been made at arm’slength. Otherwise, all sales to that affiliate areexcluded from the calculation of averagehome-market prices.13
The purpose of the cost test is to eliminatefrom consideration sales made in the home mar-ket at prices lower than the full cost of produc-tion. Like the arm’s-length test, the cost test(and the whole antidumping analysis, for thatmatter) is conducted at the CONNUM level.The selling price of each home-market transac-tion, net of all nonimputed expenses, is com-pared with the full cost of producing the CON-NUM sold in the respective transaction. Aftereach transaction is evaluated in this manner, asummary for each CONNUM is generated.
If 80 percent or more, by volume, of thesales of a specific CONNUM are made at netprices at or above the full cost of production,then all sales of that CONNUM are consid-ered to have passed the cost test. Subsequently,all of those sales enter into the pool of poten-tial matches for U.S. sales. If less than 80 per-cent of the sales of a specific CONNUM aremade at net prices at or above the cost of pro-duction, then all sales at below-cost prices areconsidered to have failed the cost test and areexcluded from that pool.
Conceivably, exercise of the arm’s-lengthand cost tests could cause all home-marketsales to become ineligible as matches for U.S.sales. If no comparable sales are found at arm’s-length prices and above the cost of production,
8
After defining theproducts in both
markets, the DOCfilters out some (or
maybe even all)home-market sales
with two separatetests: the arm’s-
length test and thecost test.
an alternative basis for normal value, known asconstructed value (CV), is used.
CV is a cost-based approximation forhome-market selling prices. When there are noeligible home-market sales comparisons, theDOC resorts to CV, which is calculated as thecost of producing the particular U.S. CON-NUM, plus an average amount for home-mar-ket expenses, plus an amount for profit.14 Thevalue is “constructed” as an estimation of whatthe product would have sold for if it had beensold in the home market.
Ineligibility of all home-market sales is notthe only basis for resorting to constructedvalue. CV can be used in situations wherehome-market sales still remain, but none of theeligible home-market CONNUMs are consid-ered appropriate matches for specific U.S.products. Following is a discussion of whenand whether a match is appropriate.
Model MatchingAfter net prices in both markets have been
calculated, and the home-market database hasbeen purged to exclude all sales that fail thearm’s-length and cost tests, the DOC deter-mines which products to compare. The prefer-ence is to compare the prices of identical prod-ucts. But given the differences in tastes andcustomer requirements that often existbetween markets, identical merchandise is notalways sold in each. This problem is com-pounded by the fact that the pool of eligiblematches is only a subset of the merchandiseactually sold in the home market, since thearm’s-length and cost tests tend to reduce thenumber of available sales.
In the absence of an identical productmatch, the next most similar product is soughtfor comparing prices. That home-marketproduct is the one most similar to the U.S.product according to the relevant productcharacteristics (as discussed earlier). In somecases, the ultimate match may have character-istics quite different from those of the U.S.CONNUM. As long as that product is themost similar available, and the difference invariable costs of manufacturing between thetwo does not exceed 20 percent of the total cost
of manufacturing the U.S. product (theDIFMER test),15 it will be the selected match.
It is possible that the most similar matchcan fail the DIFMER test. When this is thecase, the next most similar match is sought,and also subjected to the DIFMER test. If thismatch fails, the search continues. Ultimately,each U.S. product is compared with the mostsimilar home-market product that passes theDIFMER test. If no matches satisfy theDIFMER test, then the U.S. product is com-pared with constructed value.
Dumping CalculationIn an antidumping investigation, the dump-
ing margin is based on a comparison of the aver-age net U.S. price for each CONNUM16 duringthe period of investigation with its normal value.Normal value is either the average net price ofthe most similar home-market product duringthat same period or, in the absence of such orsimilar merchandise, constructed value. In eithercase, normal value is converted to U.S. dollars bymultiplying its foreign-currency-denominatedaverage by the average exchange rate in place onthe dates of all U.S. sales comprising the averageU.S. price.
Normal value expressed in dollar terms isknown as FUPDOL (foreign unit price in dol-lars) in the DOC’s dumping calcuation com-puter program. FUPDOL minus the averageU.S. price (USPR) equals the unit margin ofdumping (UMARGIN). The full impact ofthe unit margin is determined by the volume ofsales of the U.S. product in question (QTYU).The “extended” margin or full-dollar value ofthe incidence of dumping of that specific U.S.CONNUM, or EMARGIN, is the product ofUMARGIN times QTYU.
The calculations just described are undertakenfor each unique combination of U.S. CONNUMand sales type. If there are 50 such unique combi-nations, then 50 unique EMARGINs are calculat-ed. The total amount of dumping, which is alsoknown as the total potentially uncollected dumpingduties (TOTPUDD), is the sum of all positiveEMARGINs. All price comparisons that generatenegative dumping margins because FUPDOL wasless than USPR are effectively set equal to zero,
9
Given the differ-ences in tastes andcustomer require-ments that oftenexist between markets, identicalmerchandise is notalways sold in eachmarket.
regardless of the amount of “negative” dumping. Inother words, if 25 of the 50 unique combinationsgenerated positive dumping margins of an aggre-gate of, say, $10,000, and the other 25 generatednegative dumping margins of the same aggregateamount, $10,000, the total amount of dumpingwould be $10,000, not $0. The practice of disre-garding negative dumping margins is known aszeroing because the negative dumping amounts aretreated as equivalent to zero.
Ultimately, the level of dumping is expressedon an ad valorem basis to determine anantidumping duty rate.This percentage margin,which is identified in the DOC’s computer pro-gram as PCTMARG, equals the sum of thepositive TOTPUDDs divided by the total netvalue of all U.S. sales (USPR times QTYU). Forexample, if TOTPUDD equals $10,000, andthe total net U.S. sales value is $100,000, thenthe percentage margin is 10 percent.
Dumping vs. PriceDiscrimination
The U.S. antidumping law is ostensibly con-cerned with international price discrimination.17
In the paradigmatic case, a foreign producer issaid to be dumping when it charges lower pricesin the United States than it charges at home.This kind of price discrimination supposedlyreveals the existence of an unfair market distor-tion—in particular, a closed or “sanctuary” homemarket for the foreign producer. In a sanctuarymarket, trade barriers or other restrictions oncompetition allow the foreign producer tocharge artificially high prices and earn artificial-ly high profits with which it can cross-subsidizeartificially low prices abroad. And because com-petition is restricted in the home market, theprofit sanctuary cannot be arbitraged away byreimportation of dumped exports or retaliatorydumping by aggrieved foreign competitors.
The assumption that international price dif-ferences must be the result of underlying marketdistortions in the home market is highly ques-tionable.18 In fact, such price differences couldhave many innocent explanations that havenothing to do with “unfair trade” under any
plausible definition of that term. Nevertheless,for present purposes, we will take for grantedthat imports sold at lower-than-home-marketprices are indeed worth worrying about. Here,we will limit our analysis to this simple question:How well does the antidumping law actuallymeasure international price discrimination?
The answer, unfortunately, is not well at all.In a depressingly wide variety of circumstances,a foreign producer can charge prices in theUnited States that are identical to or even high-er than its home-market prices and still befound guilty of dumping. All too often, method-ological quirks and biases in the U.S. law workto conjure dumping margins out of thin air.
The Effect of Price FluctuationsAntidumping investigations compare aver-
age home-market and U.S. prices over thecourse of a year-long period of investigation. Ifprices fluctuate during that year, then differ-ences in sales volumes can generate differentaverage annual prices even if, at any given time,identical prices were charged in the two mar-kets. Consider the hypothetical example shownin Table 1. Here, a foreign company sells wid-gets from a published price list and offers iden-tical prices to all of its customers, domestic andforeign. For the first six months of the year, theprice for a particular widget is $2.00. The pricefalls to $1.00 during the second half of the year.At both prices, demand is greater in the homemarket. But because U.S. customers respondmore to the price decrease (i.e., U.S. demand ismore elastic), the weighted-average price islower in the United States ($1.33) than in thehome market ($1.45). Comparison of theseaverages leads to a dumping margin (equal tothe difference between home-market price andU.S. price divided by U.S. price) of 9.09 per-cent, even though there is no price discrimina-tion whatsoever.
Sales at identical prices generate dumpingmargins whenever a relatively larger volume ispurchased in the export market at the lowerprice. If demand surges at different times indifferent markets, dumping can be found easi-ly. Consider the example in Table 2, againshowing sales made from the same price list.
10
In a depressinglywide variety of cir-cumstances, a for-eign producer can
charge prices in theUnited States thatare identical to or
Table 2More Phantom Dumping Margins Arising from Price List Changes
Because demand surged in the home marketduring June and U.S. demand surged in July,dumping margins of 5.26 percent are the result.
In these hypothetical examples the foreignproducer does absolutely nothing wrong—itcharges the same prices to everybody. Butbecause its customers in the two different mar-kets react to those prices differently, the com-pany is found guilty of dumping. In these cases,then, determinations of dumping have nothingto do with unfair trade; they are simply artifactsof an imperfect methodology.
Arm’s-Length TestIn the preceding examples, dumping mar-
gins emerge from identical prices simplybecause of the vagaries of market reactions. Inmany other instances, however, the DOCactively skews the data to tilt the scale in favorof affirmative dumping determinations.
The arm’s-length test is premised on thenotion that affiliated customers may receivemore favorable sales terms than do unaffiliatedcustomers—thus the need to exclude home-market sales to affiliated customers who havepaid lower prices. But what about the just-as-likely possibility that prices to affiliates might behigher than those to unaffiliated customers? Ifthe affiliates are seeking to maximize their com-bined welfare (which is why sales to affiliates aresuspect in the first place), it might be optimal toshow higher revenues for the seller (to attractinvestors, improve the stock value, etc.) andhigher costs for the buyer (to avoid tax liabilities,etc.). Yet, only when sales to affiliates are lowerthan sales to unaffiliated customers are theyexcluded.This exclusion has the effect of raisingaverage prices in the comparison market, andhence raising dumping margins. Why shouldsales that are priced slightly lower be perceivedas outside “the ordinary course of trade,” whilethose that are priced much higher lead to nosuch perception?
This asymmetry in the current arm’s-lengthtest was found to be inconsistent with theWTO Antidumping Agreement in a disputesettlement case growing out of the antidumpinginvestigation of Japanese hot-rolled steel.19 Inresponse to the WTO’s adverse ruling, the
DOC is now proposing to change the test.Under the new approach, all sales under a 98percent threshold and above a 102 percentthreshold would be eliminated.20
The proposed new “band” approach elimi-nates the arm’s-length test’s asymmetry, but thefixed 98 percent and 102 percent thresholds,like the current 99.5 percent threshold, aresimply arbitrary. While in some industries or insome years there may be very little price varia-tion, prices may vary widely in others. Whatthis suggests is that the threshold for deviationsfrom the average price should be more liberalwhen there is wider price variation.
Consider the example in Table 3, whichdepicts four different price situations. In eachscenario, there are 15 sales to unaffiliated cus-tomers of one product for which the averageprice is $10. Under the current arm’s-length test,the threshold of $9.95 prevails under each sce-nario (99.5 percent of $10). If the average priceto an affiliated customer is below $9.95, then allsales to that affiliate are considered to be outsidethe ordinary course of trade and are subsequent-ly eliminated from further consideration.
Assume the average price to an affiliate is$9.50. Then, under each scenario, all sales tothat affiliate would be dropped. UnderScenario 1, that outcome is not necessarilyunreasonable. After all, all sales to unaffiliatedcustomers were $10 and there was no pricevariability whatsoever. Under Scenario 2,though, there was enough price variability thatthe average price to affiliates was higher thanthe price to unaffiliated customers 33 percentof the time. Is it reasonable to conclude thataffiliated sales are outside the ordinary courseof trade on account of unrepresentative pricingwhen those sales are priced higher than a siz-able portion of sales to unaffiliated customers?What if they are priced higher than 40 percentof unaffiliated sales (Scenario 3)? Or 93 per-cent (Scenario 4)? The point is that a constantbenchmark (or constant benchmarks, as in theproposed new test) makes no sense.
In preparing this paper, we were able to gainaccess to and analyze the proprietary records ofactual U.S. antidumping determinations—14in original investigations and 4 in administra-
12
The asymmetry inthe current
arm’s-length test was found to be
inconsistent with the WTO
AntidumpingAgreement.
tive reviews.21 For each of these 18 determina-tions, we were able to recreate the dumpingmargins determined by the DOC using theDOC’s own dumping calculation computerprograms. We then were able to alter thoseprograms to gauge the effect of various specif-ic methodological distortions on the ultimateoutcomes.
Table 4 reveals the significant effect of thearm’s-length test on dumping margins.Eliminating the test from the dumping calcula-tion affected the results in 8 of the 13 cases inwhich the test was used (affiliated sales were notan issue in 5 of the 18 cases), with an overall aver-age reduction in the dumping margin of 6.95 per-cent. In each of the 8 cases affected, the margindecreased. In an investigation of hot-rolled steelfrom Japan, the company’s dumping margindecreased by almost 16 percent.22 And the mar-gin decreased by more than 52 percent for a com-pany in a review concerning stainless steel sheetand strip from Japan.
Excluding Below-Cost SalesOne of the most egregious methodological
distortions in contemporary antidumpingpractice is the so-called cost test. The purposeof the cost test is to eliminate from considera-tion sales made in the home market at priceslower than the full cost of production. Whenbelow-cost sales are eliminated in this way, theresult is that all U.S. sales are compared withonly the highest-priced (that is, above-cost)home-market sales.
What possible purpose could be served byexcluding below-cost home-market sales fromnormal value? Remember that the main theo-ry behind the antidumping law is that the for-eign producer is enjoying an artificial advan-tage because of a sanctuary market at home.According to the theory, trade barriers or otherrestrictions on competition cause prices (andprofits) in the home market to be artificiallyhigh, thus allowing the foreign producer tocross-subsidize unfairly cheap export sales.
13
One of the mostegregious method-ological distortionsin contemporaryantidumping prac-tice is the so-calledcost test.
Consequently, price differences between theexport market and the home market are sup-posedly probative of unfair trade because theymight indicate the existence of a closed sanctu-ary market in the foreign producer’s homemarket. Whether those price differences exist,though, cannot be fairly determined if all thelowest home-market prices are excluded fromthe comparison.
Indeed, the existence of below-cost sales inthe home market is actually affirmative evi-dence of the absence of a sanctuary market. Asanctuary market, after all, is supposed to be anisland of artificially high prices and profits. Ifhome-market sales at a loss are found in signif-icant quantities, isn’t that a fairly compellingindication that there is no sanctuary market?But because of the cost test, it is precisely underthese conditions that dumping margins areboosted significantly higher than they other-wise would be.
The cost test is thus fundamentally miscon-ceived. And the way the test is applied onlymakes matters worse. Specifically, individualnet prices are compared to average annualcosts. But if unit costs fluctuate over the periodof investigation—and they always do, if for noother reason than that varying levels of outputcontinually change the denominator overwhich fixed costs are distributed—then thecomparison of individual prices to averagecosts can yield perverse results. A companycould make every single sale at prices above
transaction-specific costs, but still some ofthose prices could be below average annualcosts. Alternatively, a producer could recover allcosts and make money over the course of ayear, yet still charge below-cost prices on a sig-nificant fraction of sales.
The effect of the cost test on the dumping cal-culation can be dramatic. For example, in Table 5,there are five sales of widget product Code 1 inthe U.S. market at different prices ranging from$1.00 to $5.00. Likewise, in the home marketthere are five sales at the identical prices.Assuming the same volume is sold in each of the10 transactions, the weighted-average price forProduct 1 is $3.00 in both markets.The dumpingmargin for this comparison is zero. There is noprice discrimination whatsoever. However, this isnot how the calculation works.
The cost test imposes restrictions on the eli-gibility of home-market sales that factor into theaverage price. Sales made at prices below the fullcost of production are eliminated from consider-ation. In Table 5, the two home-market sales atprices below $2.50 are excluded, causing theaverage home-market price of Product 1 to riseto $4.00. This generates a dumping margin of33 percent despite the fact that there are no pricedifferences between markets.
Empirically, the cost test is among the mostsignificant causes of inflated dumping margins.In the 17 actual DOC dumping determina-tions that we examined in which the cost testwas applied, the dumping margin decreased
15
The existence ofbelow-cost sales inthe home market isactually affirmativeevidence of theabsence of a sanc-tuary market.
Net U.S. Net H.M. Unit Cost Net H.M.Price Price Cost Test Price (used)
each and every time—by an average of 60 per-cent—when the cost test was eliminated fromthe protocol23 (see Table 4). In two cases, oneinvolving DRAMs from Taiwan and the otherinvolving stainless steel sheet and strip fromJapan, the margins were reduced all the way tozero. Margins in a review of preserved mush-rooms from Indonesia and an investigation ofpolyester staple fiber from Taiwan decreased bymore than 90 percent each. All but three casesshowed double-digit declines and all but fivedecreased by more than 50 percent.
Use of Constructed ValueApplication of the arm’s-length and cost tests
systematically eliminates lower-priced sales inthe home market, increasing the likelihood offinding positive dumping margins for two rea-sons. First, the sales remaining in the home-market database comprise only a higher-pricedsubset of all home-market sales. This translatesinto higher normal values. Second, the likeli-hood of matching any given U.S. CONNUM toan identical or similar home-market CON-NUM is diminished because the universe ofpotential matches is smaller. This increases thechances that constructed value—a cost-basedapproximation of normal value—will be used.
Constructed value is calculated by adding tothe cost of producing the U.S. product an esti-mated amount for home-market selling expensesand profit. Both of these estimates are based onthe averages of only those sales that pass both thearm’s-length and cost tests. In other words, eventhough a substantial portion of the home-marketsales may have been made at a loss—a fact thatobviously tempers or even negates overall home-market profitability—only the expenses and prof-its of the profitable sales are used to calculate theaverages for constructed value.
A comparison of U.S. prices to constructedvalue cannot indicate anything about the pos-sible existence of price discrimination causedby a sanctuary market because constructedvalue is not based on price data. A cost-basednormal value could only be relevant for assess-ing whether U.S. prices are below cost—but inthat case, no profit amount should be includedat all. Accordingly, a finding of dumping based
on comparing U.S. prices to constructed valuedoes not show that the U.S. sales are belowcost; it shows only that they fall below some(arbitrary) benchmark for profitability. Such afinding has no relevance whatsoever to anyplausible theory of unfair trade.
Five of the 18 actual DOC dumping deter-minations examined involved comparing ofU.S. sales to constructed value. Four of thosefive experienced reductions in dumping mar-gins when the profit component was excludedfrom the calculation of constructed value (seeTable 4). Although the overall average reduc-tion was 11 percent, it was 22.8 percent in aninvestigation involving concrete reinforcingbars from Moldova, 18.25 percent for an inves-tigation of SRAMs from Taiwan, and13.91 percent in an investigation concerningDRAMs from Taiwan.
Model MatchingDumping margins can also emerge from
the technicalities of product definition andmodel matching. Specifically, the more poten-tially price-relevant product characteristics fora given product, the more intractable is thedilemma facing antidumping authorities.
On the one hand, if the DOC were to ignorecertain characteristics as irrelevant for productdefinition, physically different products might betreated as identical. The more broadly identicalproducts are defined, the greater the likelihoodthat phantom dumping margins could be gener-ated, or real dumping margins could be masked,simply because of differences between the prod-uct mix sold in the United States and that sold inthe home market. Consider, for example, if theDOC were to ignore material as a matchingcharacteristic for widgets in the example present-ed earlier—even though material does have somebearing on price. And let’s say that only relativelylow-priced rubber widgets are sold in the UnitedStates, while the mix sold in the home marketincludes rubber widgets as well as relatively high-priced plastic and vinyl widgets. If large plasticand vinyl widgets sold in the home market aretreated as identical to large rubber widgets sold inthe United States, and small plastic and vinylwidgets sold in the home market are treated as
16
A comparison ofU.S. prices to con-
structed value cannotindicate anythingabout the possible
existence of price dis-crimination caused
by a sanctuary mar-ket because con-
structed value is notbased on price data.
identical to small rubber widgets sold in theUnited States, price comparisons would showdumping simply because rubber widgets are lessvaluable merchandise. In other words, dumpingmargins would result simply because apples werecompared with oranges.
Two model-match scenarios are presented inTable 6. Each outcome is based on the identicalset of sales data. If both size and material areconsidered relevant product characteristics(Scenario A), there are no dumping marginswhatsoever. In this scenario, the average netprices of small rubber widgets sold in the UnitedStates are compared to the average net prices ofsmall rubber widgets sold in the home market.Likewise, large rubber widgets in the UnitedStates are matched to identical products in thehome market. The average net prices of smalland large rubber widgets are $1.25 and $2.55,respectively, in both markets, which generateszero dumping margins. All of the nonidenticalhome-market products were ignored.
Under Scenario B, only the size variable isconsidered a relevant product characteristic. Inthis case, the size variable has no effect on theaverage net prices in the United States becauseonly one material—rubber—is sold, so a smallrubber is equivalent to a small, and a large rub-ber is equivalent to a large.The average net pricefor small widgets is $1.25, and for large widgetsit is $2.55. However, results on the home-mar-ket side are much different under the one-char-acteristic specification. Now, rubber, plastic, andvinyl widgets of the same size group are aver-aged together. The average net price for smallwidgets in the home market is $1.38, and forlarge widgets it is $2.70. Although the data areidentical under both scenarios, the mere changein product specification generates an overalldumping margin of 7.46 percent.
This example would seem to argue for mak-ing product definitions as specific as possible.But that approach creates problems as well.Themore fine-grained the product definition, thehigher the likelihood that dumping margins arethe product of chance or arbitrary distinctions.
Table 7 demonstrates how subjective deter-minations about product definition affect theoutcomes in dumping calculations. The aver-
age net U.S. price of 12 sales of medium rubberwidgets is $2.25. In the home market, there are6 sales of large rubber widgets, averaging $3.25each, and 6 sales of small rubber widgets, aver-aging $1.25 each. Whether or not these salesare dumped hinges upon whether medium isdeemed to be more similar to large or to small.
In Scenario A, comparing U.S. mediumwidgets to home-market small widgets gener-ates a negative and therefore zero margin. Thenormal value (NV) is increased by the differ-ence in merchandise (DIFMER), which is thedifference in variable costs of manufacturingthe nonidentical products (VCOMU-VCOMH). The result is FUPDOL, fromwhich is subtracted USPR, generating a nega-tive $0.50 unit margin.
In Scenario B, however, large is deemed moresimilar to medium, and the result is an affirmativedumping margin of 22 percent. In this case, thehigher home-market price is slightly mitigated bya negative DIFMER of -$0.50, but the resultingFUPDOL is still $0.50 higher than the U.S.price.
Furthermore, when products are defined verynarrowly, the likelihood is greater that U.S. saleswill be compared to a very small (and possiblyunrepresentative) fraction of home-market sales.Thus, if all U.S. sales are of small rubber widgets,while small rubber widgets comprise only 1 per-cent of home-market sales, the existence ornonexistence of price discrimination will bedetermined on the basis of a very small sample ofhome-market sales. Those sales may haveuncharacteristically high or low prices and thusoffer a poor measure of overall price levels in thehome market. Nevertheless, all other sales in thehome market will simply be excluded from thedumping calculation and ignored.24 The situationgets worse when the sales of the U.S. product arecompared to a tiny sample of nonidentical (butnext most similar) sales in the home market.
Random chance and unavoidably arbitrarydistinctions can thus play a major role in deter-mining the final outcome of a dumping determi-nation. In the preceding examples, the foreigncompany was not engaging in price discrimina-tion, yet it was found guilty of dumping. Thedumping margins reflect, not unfair trade, butmethodological shortcomings.
17
Random chanceand unavoidablyarbitrary distinc-tions can thus play amajor role in deter-mining the finaloutcome of a dump-ing determination.
18
Scen
ario
A:A
bbre
viat
ed U
.S. S
ales
Lis
tSc
enar
io B
:Abb
revi
ated
H.M
. Sal
es L
ist
OB
SUSI
ZEU
MAT
ERIA
LUQ
TYU
USP
RO
BSH
SIZE
HM
ATER
IALH
QTY
HN
V1
Smal
lR
ubbe
r10
$1.2
01
Smal
lR
ubbe
r10
$1.2
02
Smal
lR
ubbe
r10
$1.2
02
Smal
lR
ubbe
r10
$1.3
03
Smal
lR
ubbe
r10
$1.2
03
Smal
lPl
astic
10$1
.35
4Sm
all
Rub
ber
10$1
.30
4Sm
all
Plas
tic10
$1.4
55
Smal
lR
ubbe
r10
$1.3
05
Smal
lV
inyl
10$1
.45
6Sm
all
Rub
ber
10$1
.30
6Sm
all
Vin
yl10
$1.5
57
Larg
eR
ubbe
r10
$2.3
07
Larg
eR
ubbe
r10
$2.5
08
Larg
eR
ubbe
r10
$2.4
08
Larg
eR
ubbe
r10
$2.6
09
Larg
eR
ubbe
r10
$2.5
09
Larg
ePl
astic
10$2
.60
10La
rge
Rub
ber
10$2
.60
10La
rge
Plas
tic10
$2.7
011
Larg
eR
ubbe
r10
$2.7
011
Larg
eV
inyl
10$2
.80
12La
rge
Rub
ber
10$2
.80
12La
rge
Vin
yl10
$3.0
0
SIZE
UM
ATER
IALU
QTY
UU
SPR
VALU
ESI
ZEH
MAT
ERIA
LHQ
TYH
NV
UM
AR
GIN
EMA
RG
INTO
TPU
DD
PCTM
AR
GSm
all
Rub
ber
60$1
.25
$75.
00Sm
all
Rub
ber
20$1
.25
$0.0
0$0
.00
$0.0
00.
00%
Larg
eR
ubbe
r60
$2.5
5$1
53.0
0La
rge
Rub
ber
20$2
.55
$0.0
0$0
.00
$0.0
00.
00%
$228
.00
$0.0
00.
00%
SIZE
UM
ATER
IALU
QTY
UU
SPR
VALU
ESI
ZEH
MAT
ERIA
LHQ
TYH
NV
UM
AR
GIN
EMA
RG
INTO
TPU
DD
PCTM
AR
GSm
all
N/A
60$1
.25
$75.
00Sm
all
N/A
60$1
.38
$0.1
3$8
.00
$8.0
010
.67%
Larg
eN
/A60
$2.5
5$1
53.0
0La
rge
N/A
60$2
.70
$0.1
5$9
.00
$9.0
05.
88%
$228
.00
$17.
007.
46%
Tabl
e 6
Mod
el M
atch
ing:
How
Dum
ping
Mar
gins
Can
Res
ult f
rom
Pro
duct
Def
initi
on
Scen
ario
AIf
bot
h ch
arac
teris
tics
are
deem
ed re
leva
nt, a
vera
ge n
et p
rices
and
dum
ping
mar
gins
are
as
follo
ws:
Con
clus
ion:
No
dum
ping
.
Con
clus
ion:
Dum
ping
mar
gins
of 7
.46%
.
Scen
ario
BIf
onl
y th
e si
ze c
hara
cter
istic
is d
eem
ed re
leva
nt, a
vera
ge n
et p
rices
and
dum
ping
mar
gins
are
as
follo
ws:
19
Scen
ario
A:A
bbre
viat
ed U
.S. S
ales
Lis
tSc
enar
io B
:Abb
revi
ated
H.M
. Sal
es L
ist
OB
SUSI
ZEU
MAT
ERIA
LUQ
TYU
USP
RV
CO
MU
OB
SHSI
ZEH
MAT
ERIA
LHQ
TYH
NV
VC
OM
H1
Med
ium
Rub
ber
10$2
.00
$1.5
01
Smal
lR
ubbe
r10
$1.0
0$1
.00
2M
ediu
mR
ubbe
r10
$2.0
0$1
.50
2Sm
all
Rub
ber
10$1
.10
$1.0
03
Med
ium
Rub
ber
10$2
.10
$1.5
03
Smal
lR
ubbe
r10
$1.2
0$1
.00
4M
ediu
mR
ubbe
r10
$2.1
0$1
.50
4Sm
all
Rub
ber
10$1
.30
$1.0
05
Med
ium
Rub
ber
10$2
.20
$1.5
05
Smal
lR
ubbe
r10
$1.4
0$1
.00
6M
ediu
mR
ubbe
r10
$2.2
0$1
.50
6Sm
all
Rub
ber
10$1
.50
$1.0
07
Med
ium
Rub
ber
10$2
.30
$1.5
07
Larg
eR
ubbe
r10
$3.0
0$2
.00
8M
ediu
mR
ubbe
r10
$2.3
0$1
.50
8La
rge
Rub
ber
10$3
.10
$2.0
09
Med
ium
Rub
ber
10$2
.40
$1.5
09
Larg
eR
ubbe
r10
$3.2
0$2
.00
10M
ediu
mR
ubbe
r10
$2.4
0$1
.50
10La
rge
Rub
ber
10$3
.30
$2.0
011
Med
ium
Rub
ber
10$2
.50
$1.5
011
Larg
eR
ubbe
r10
$3.4
0$2
.00
12M
ediu
mR
ubbe
r10
$2.5
0$1
.50
12La
rge
Rub
ber
10$3
.50
$2.0
0
SIZE
UM
ATER
IALU
QTY
UU
SPR
VALU
ESI
ZEH
MAT
ERIA
LHQ
TYH
NV
DIF
MER
FUPD
OL
UM
AR
GIN
EMA
RG
INTO
TPU
DD
PCTM
AR
GM
ediu
mR
ubbe
r12
0$2
.25
$270
.00
Smal
lR
ubbe
r60
$1.2
5$0
.50
$1.7
5-$
0.50
-$60
.00
$0.0
00.
00%
SIZE
UM
ATER
IALU
QTY
UU
SPR
VALU
ESI
ZEH
MAT
ERIA
LHQ
TYH
NV
DIF
MER
FUPD
OL
UM
AR
GIN
EMA
RG
INTO
TPU
DD
PCTM
AR
GM
ediu
mR
ubbe
r12
0$2
.25
$270
.00
Larg
eR
ubbe
r60
$3.2
5-$
0.50
$2.7
5$0
.50
$60.
00$6
0.00
22.2
2%
Tabl
e 7
Mod
el M
atch
ing:
How
Dum
ping
Mar
gins
Can
Res
ult f
rom
Ran
king
Hie
rarc
hies
Scen
ario
AIf
med
ium
is d
eem
ed m
ore
sim
ilar t
o sm
all,
aver
age
net p
rices
and
dum
ping
mar
gins
are
as
follo
ws:
Con
clus
ion:
No
dum
ping
.
Con
clus
ion:
Dum
ping
mar
gins
of 2
2.22
%.
Scen
ario
BIf
med
ium
is d
eem
ed m
ore
sim
ilar t
o la
rge,
ave
rage
net
pric
es a
nd d
umpi
ng m
argi
ns a
re a
s fo
llow
s:
Difference-in-Merchandise AdjustmentThe problems caused by comparing prices of
nonidentical products are amplified by theshortcomings of the DOC’s DIFMER adjust-ment. When comparing prices of nonidenticalproducts, the DOC adjusts normal value by thedifference between the two CONNUMs’ vari-able manufacturing costs (i.e., materials, directlabor, and variable factory overhead). Althoughthe basis for this adjustment is logical enough—products that are more costly to produce gener-ally have higher prices—the assumption thatprice differences exactly mirror cost differencesis nonetheless totally artificial.
In all too many cases, wide variations inprices between products exist alongside little orno differences in costs of production. A case inpoint from the early 1990s involved fresh cut-roses from Colombia. The CONNUMs in thatcase reflected stem size, bulb size, and bulb color.Although long-stemmed and large-bulbed redroses commanded the highest prices, there werevirtually no differences in the cost of productionbetween this variety and the short-stemmed andsmall-bulbed yellow roses, which were the low-est priced. Because demand is seasonal andpeaks at different times in each market (duringValentine’s Day in the United States andMother’s Day in Canada, the third-countrycomparison market used in this case), identicalproducts were often not available in the com-parison pool. As a result, the next most similarproduct was selected on the basis of the relevantcharacteristics. But since production of all roses,regardless of variety, entailed the same basiccosts for seed, fertilizer, land, and water, theDIFMER adjustment made for nonidenticalmatches was small and usually zero. As a result,low-priced, short-stemmed, small-bulbed yel-low roses in the United States were matched tohigh-priced, long-stemmed, large-bulbed redroses in Canada during May without any signif-icant adjustment. Although the oppositedynamic prevailed in February, the large nega-tive dumping margins associated with the high-er U.S. prices had no impact on the overall mar-gin because they were zeroed out.25
The inadequacy of the DIFMER adjust-ment combines with the unfairness of the cost
test to generate large phantom dumping mar-gins whenever second-quality merchandise issold in the United States. Such flawed or off-spec products sell for a fraction of the price ofprime merchandise, but they cost the same toproduce. As a result, sales of second-qualitymerchandise are almost always made at pricesbelow the cost of production. But because thecost test is administered on home-market salesonly, the necessarily low-priced U.S. sales ofsecond-quality merchandise must be comparedto the higher-priced sales of primary productsin the home market, or to constructed value.And because there is no appreciable cost dif-ference between second-quality and primemerchandise, there can be no significantDIFMER adjustment. Accordingly, high-priced prime merchandise sold at home iscompared to low-value second-quality mer-chandise in the United States—with big phan-tom dumping margins as the result.
Asymmetric Treatment of Indirect SellingExpenses
Calculation and deduction of indirect sellingexpenses introduce some potentially significantinequities into the process of determining dump-ing margins. Indirect selling expenses are expens-es that do not vary directly with the volume ofsales—sales staff salaries, sales department over-head, and so forth. In “export price” situations—that is, when the foreign producer sells directly toan unrelated purchaser in the U.S. market—noadjustment is made to export or home-marketprices for such indirect selling expenses. But in“constructed export price” situations—when theforeign producer sells to unrelated U.S. customersthrough a related reseller in the United States—certain indirect expenses are deducted.Specifically, all indirect selling expenses incurredwith respect to U.S. sales are deducted from theexport price, but the adjustment to home-marketprice for home-market-related indirect sellingexpenses is capped at the amount of the U.S. indi-rect selling expenses. All home-market-relatedindirect selling expenses in excess of the cap aresimply disregarded.
There is no possible justification for thisasymmetry. The policy of deducting U.S. indi-
20
The assumptionthat price differ-
ences exactly mirrorcost differences is
totally artificial.
rect expenses is apparently based on theassumption that resales by the reseller in theexport market are on a different level oftrade—and therefore that those resale pricesinclude additional expenses—than direct salesby the foreign producer in the home market.That assumption, though, is completely arbi-trary. It may be that the reseller’s U.S. cus-tomers are large national distributors, while theforeign producer sells directly to small localwholesalers at home—in which case the home-market price actually has more of the distribu-tion chain built into it than does the U.S. price.Why then should the adjustment for home-market indirect selling expenses be limited tothe amount of similar expenses incurred in theUnited States?
This asymmetry—known as the CEP offsetcap—skews dumping calculations in the direc-tion of higher dumping margins. If U.S. indirectexpenses are greater than equivalent home-mar-ket expenses, then both are fully taken intoaccount. If, however, the home-market expensesare greater, they are capped.The result in that caseis an artificially inflated normal value—and anartificially inflated dumping margin.
Of the 18 actual DOC determinations thatwere examined, 10 would have had lowerdumping margins if the asymmetry of theCEP offset cap had been eliminated (see Table4). For example, in two reviews involvingtapered roller bearings from Japan (one reviewinvolved “large” bearings; the other reviewinvolved “small” bearings), if no indirect sellingexpenses had been deducted on either side, thedumping margin in the large bearings casewould have been 25.55 percent lower, and themargin in the small bearings case would havebeen 15.84 percent lower.
In reality it makes no sense to adjust auto-matically for indirect selling expenses. No rea-sonable basis exists for assuming that suchoverhead costs are built directly into the sellingprice. Stripping them out of the price, ratherthan creating fairer price comparisons, insteadproduces price comparisons even moreremoved from the actual market reality of realsales prices. This artificiality is exacerbated bythe asymmetry of always deducting all U.S.
indirect selling expenses while only partiallydeducting equivalent home-market expenses.
CEP ProfitThe misguided policy of deducting indirect
selling expenses from the U.S. price in CEPsituations is exacerbated by the further deduc-tion of so-called CEP profit—the estimatedprofit attributable to U.S. selling operations.
What is the rationale for deducting such prof-it? As with the deduction of indirect sellingexpenses, the assumption apparently is that U.S.sales are on a lower level of trade—that is, closerto the ultimate end user—than the home-marketsales to which they are compared. But this is acompletely arbitrary assumption that may or maynot be valid in any particular case. Accordingly,why deduct profit from U.S. sales without mak-ing any corresponding deduction from normalvalue? This asymmetry can have no effect but todrive up dumping margins artificially.
In 5 of the 10 actual DOC dumping deter-minations that were reviewed and that con-tained CEP sales, the margins decreased whenthe CEP profit deduction was eliminated. Forthe other 5 there was no impact. Overall, theaverage decrease was 1.23 percent (see Table 4).In these particular cases, then, the impact wasrelatively minor. The effect in other cases,though, could be significantly greater and, in anyevent, further contribute to the serious inflationof dumping margins caused by all the rest of theantidumping law’s methodological distortions.
ZeroingThe final step in the dumping calculation
includes one of the antidumping law’s mostegregious distortions: the practice of zeroing.By ignoring “negative” dumping margins (i.e.,instances in which U.S. prices are higher thanhome-market prices), the DOC employs a“heads I win, tails you lose” strategy for maxi-mizing dumping margins.
Consider the simple example in Table 8. Eachproduct in this example is sold at identical netprices in both markets with the exception ofProduct 1 and Product 5. Product 1 is sold for$0.50 less in the home market than in the U.S.market, and Product 5 is sold for $0.50 more in
21
By ignoring “nega-tive” dumping margins, the DOCemploys a “heads I win, tails you lose” strategy for maximizing dump-ing margins.
the home market than in the U.S. market. Thearithmetic sum of the individual dumping mar-gins (Total Margin) is zero because the price dif-ferences for Products 1 and 5 cancel each otherout. Under U.S. antidumping rules, however, thenegative dumping margin on Product 1 is givenshort shrift by denying it any impact on the over-all margin. The total amount of negative dump-ing is always set equal to zero. Accordingly, in thiscase the DOC would find a dumping margin of10 percent despite the lack of any difference inoverall price levels between the two markets.
Zeroing, as this practice has come to beknown, has been found to violate the WTOAntidumping Agreement in a case involvingbed linen brought by the Indian governmentagainst the European Union.26 The EU hassince changed its practice, but zeroing remainsa controversial and significant component ofU.S. antidumping methodology.
In the 18 actual DOC determinations thatwere examined, zeroing was the most signifi-cant cause of dumping margins. It affected theoutcomes in 17 of the 18 cases. On average,eliminating the practice of zeroing caused themargins to decrease by 88.65 percent in these18 cases27 (see Table 4). The margins wouldhave been entirely eliminated in 5 of the 18cases had zeroing not been practiced. In a sixthcase, the margin was reduced by 96.84 percent.
A Hypothetical Case Study
To put all the methodological issues in con-text and to show how they interact in practice,we have constructed a detailed, hypotheticalcase study. What follows is a thorough assess-ment of the entire dumping calculation process,using a fictitious widget case as an example.
Appendix 1 provides an abbreviated sample ofan imaginary Section B home-market sales list. Itis abbreviated because many of the expense fieldshave already been consolidated into aggregateexpense groups like movement expenses andindirect selling expenses.This type of aggregationoccurs during the course of running the marginprogram on the itemized database submissions,but to keep these examples relatively simple andcompact, identification of each specific field isforgone. The realism of the case study, though, isnot compromised by this shortcut.
Appendix 1 is a sample home-market salesdatabase for a fictitious foreign widget produc-er. Each of the fields ends with the letter H,which indicates that the field is a home-marketitem.The first field, OBSH, corresponds to thesales observation and is sequential. There are45 observations.28
The second field, CONNUMH, is the DOCcontrol number, the significance of which has
22
Zeroing has beenfound to violate
the WTOAntidumping
Agreement.
Product Net U.S. Net H.M. Unit U.S. Total Total TotalCode Price Price Margin Quantity Margin PUDD Value
been discussed previously. In this example, theCONNUMH reflects the product characteris-tics, SIZEH and MATERIALH. The first digitin the CONNUMH reflects the size of the wid-get, where a value of “1” corresponds to “small,”and “2” corresponds to “large.” The second digitreflects the widget’s material content, where “1”equals “rubber,” “2” equals “plastic,” and “3” equals“vinyl.” Under this CONNUMH construction,there are six products, each of which is sold in thehome market.
The field PRIMEH indicates whether theproduct sold was “prime” or “secondary” mer-chandise. As explained earlier, prime merchan-dise is output that was manufactured withoutsignificant deviations from the intended qualityspecifications. Secondary merchandise is outputthat, because of a flaw in the production processor in one or more of the input materials, exhibitssignificant deviations from the intended qualitystandards. As a result, secondary merchandise isnot as desirable to the customer because it usu-ally cannot be used for its intended purpose.Although demand may still exist for the lesser-quality secondary merchandise, it is usually soldat prices well below those for prime.
The field CUSCODH is the customercode. Rather than use actual names, each cus-tomer is given a numerical code. CUSRELHindicates whether the customer is affiliatedwith the foreign producer. A value of “1” indi-cates no affiliation and a value of “2” indicatesthat the customer is affiliated. In this example,customer codes 10 and 20 are unaffiliated, andcustomer codes 30 and 40 are affiliated withthe producer.
QTYH represents the number of units sold inthe transaction. GRSUPRH is the gross sellingprice per unit. DISCREBH is the aggregatedunit value of all discounts and rebates awarded onthe transaction. Companies often grant volumediscounts, early payment discounts, or rebate pro-grams to their customers. MOVEH is the aggre-gated unit value of all transportation expensesincurred to get the merchandise from the factoryto the customer. DIRSELH is the aggregatedunit direct selling expenses associated with thetransaction. It typically reflects commissions,war-ranties, advertising, technical assistance, and any
other expenses associated directly with the sale inquestion. CREDITH is the unit credit expenseassociated with the specific transaction. It is animputed expense, which ostensibly measures theopportunity cost of maintaining accounts receiv-able. When a sale is made, the merchandise usu-ally leaves the producer’s warehouse for the cus-tomer’s destination. However, payment is oftennot received for some time.CREDITH attemptsto measure the opportunity cost of not havingreceived payment when the merchandise wasshipped to the customer. Its value is based on theprevailing short-term interest rate, the number ofdays between shipment and receipt of payment,and the sales price.
INDSELLH represents the aggregated unitindirect selling expenses, which typically includegeneral and administrative expenses incurred bythe sales department that are not associated withany particular sale. Office supplies, managers’salaries, postage, rent incurred by the salesdepartment, and other such items make up indi-rect selling expenses. INVCARH is anotherimputed expense, but unlike CREDITH, it isconsidered an indirect expense. It attempts tomeasure the opportunity cost of not selling mer-chandise on a just-in-time basis. Holding inven-tory is more expensive than selling straight offthe production line because there is an opportu-nity cost of not having the cash available imme-diately. Like CREDITH, the INVCARHexpense is based on the prevailing short-terminterest rate, but here the cost of production(rather than sales price) and the days betweenproduction and sale (rather than between saleand receipt of payment) are used as the compo-nents of the calculation.
PACKH is the cost of packing materials,labor, and overhead. Many products are placedin some sort of packaging before they areshipped to the customer. PACKH measuresthe aggregated per unit cost of packaging.
VCOMH is the unit variable cost (materials,labor, and variable overhead) of producing themerchandise. It is included in the Section B data-base because it is relevant to the model-matchingprocedure, which will be discussed below.
The fields in Appendix 1, although not acomplete list of all the data typically reported in
23
To put all themethodologicalissues in contextand to show howthey interact inpractice, we haveconstructed adetailed, hypotheti-cal case study.
a Section B response, suffice to provide realisticexamples of dumping calculation methodology.
Appendix 1 also contains some additional fieldsthat are calculated for future reference.NETPRIHis the net home-market sales price, calculated asGRSUPRH-DISCREBH-MOVEH-DIRSELH-CREDITH-PACKH. The home-market net price equals the gross selling priceminus discounts and rebates, movement expenses,direct selling expenses,credit,and packing.Indirectselling expenses are not deducted in the DOC’scalculation of home-market net price.NETPRIHis the basis for normal value, which is compared tothe average export price in the margin calculation.
NPRICOP is a different expression of thenet home-market sales price, one which is netof all nonimputed selling expenses, both directand indirect, and is used for the cost test to bediscussed. It is calculated as GRSUPRH-DIS-CREBH-MOVEH-DIRSELH-IND-SELH-PACKH.TOTCOP is the total cost ofproduction (materials + labor + overhead +general and administrative expenses + interestexpenses) for each CONNUMH reported inSection D of the questionnaire response. It ismerged into the home-market sales database toperform the cost test and to calculate profitrates, as follows. REVENUE is the total netrevenue, NPRICOP * QTYH. COST is thetotal cost, TOTCOP * QTYH. PROFIT isREVENUE – COST.
Note that the overall profit rate on home-market sales (the figure in the bottom rightcorner of the table in Appendix 2) is a negative7.36 percent—in other words, a loss of 7.36percent. The fact that the company lost moneyon its home-market sales should undercut anyreasonable conclusion that it is dumping, sincethere is evidently no sanctuary market of highprices and high profits.The DOC’s antidump-ing procedures, however, are not designed toreach reasonable conclusions.
Appendix 2 provides a summary of theweighted-average net prices per product(where product is a unique combination ofCONNUMH and PRIMEH), as well as therevenue, costs, and profit. This table is provid-ed for reference as an unadulterated summaryof the home-market sales situation. It is the
true picture of average prices, costs, and profits,which ultimately become skewed as a result ofthe DOC’s procedures.
The first such procedure performed onhome-market sales is the arm’s-length test.Appendix 3 derives from the previous exhibitsand lists only the fields necessary to conductthe test, which involves a comparison of netprices between each affiliated customer and allunaffiliated customers. Note that the customercode is relevant only for the affiliated cus-tomers, since each affiliate is compared to allunaffiliated customers.
Appendix 4 provides the results of thearm’s-length test. Each line represents a uniqueproduct and contains an average net price to allunaffiliated customers, and average prices,quantities, and ratios for each of the two affili-ated customers, when available. For example,CONNUMH 11 / PRIMEH 1 was sold toCustomer 40 and to at least one unaffiliatedcustomer, but was not sold to Customer 30.The ratio of 89.69 percent is the average priceto Customer 40 ($3.25) divided by the averageprice to all unaffiliated customers ($3.62).Individual ratios are calculated whenever thereare prices to compare, and then those ratios areweight-averaged by quantity to generate anoverall ratio for each affiliate.
On average, the net prices to affiliatedCustomer 30 exceed the net prices to unaffili-ated customers by 17.39 percent. Sales to thiscustomer therefore pass the arm’s-length testbecause the ratio exceeds 99.5 percent.Consequently, all sales to this customer remainin the home-market database—for the timebeing, at least. On the other hand, prices toCustomer 40 were only 93.58 percent of pricesto unaffiliated customers. Customer 40 there-fore fails the arm’s-length test and all sales to40 are excluded from further consideration.
Appendix 5 represents the post-arm’s-lengthtest home-market database. It has 11 fewer salesthan the original database because all sales toCustomer 40 have been eliminated. Appendix 6provides a summary of the post-arm’s-lengthtest data and reveals that the weighted-averagehome-market price increased from $5.80 to$5.90, and the profit rate increased from –7.36
24
The DOC’santidumping
procedures are notdesigned to reach
reasonable conclusions.
percent to –4.95 percent.The asymmetric appli-cation of the arm’s-length test invariably leads tohigher average home-market prices.
The cost test is the second systematicreview of the usability of home-market sales.Sales at prices below the full cost of produc-tion, if they are made in sufficient quantities,are deemed outside the ordinary course oftrade and eliminated from further considera-tion. The net selling price, as defined byNPRICOP, is compared to the cost of produc-tion, TOTCOP. If NPRICOP is less thanTOTCOP, the selling price is below the cost ofproduction. That sale does not necessarily failthe cost test, however. If the volume of all salesfound to be priced below cost comprises lessthan 20 percent of the volume of all sales ofthat product (CONNUMH / PRIMEH com-bination), then despite being priced below cost,those sales pass the cost test.
Appendix 7 shows how the cost test is per-formed. The field TOTCOP, or total cost ofproduction, comes from the Section D costdatabase. Typically, the DOC will merge theSections B and D databases by CONNUMHto extract the unit cost information, which isused for the cost test and to calculate profitrates for constructed value comparisons as wellas for the CEP profit deduction.
As shown in Appendix 8, the cost test inthis hypothetical case results in the exclusion ofyet more home-market sales. Note that eventhough Observation 11 is priced below cost,the collective volume of above-cost sales of thatproduct (CONNUMH 13 / PRIMEH 1)exceeds 80 percent.Thus, all sales of that prod-uct pass the cost test. Nonetheless, a significantportion of the home-market sales database (17of the remaining 34 sales) fails the cost test andis dropped from further consideration.Appendix 9 shows the bottom line: The aver-age home-market price rises from $5.90 to$6.80, and the profit rate increases from –4.95percent to 10.75 percent.
The three calculated variables presented atthe bottom of Appendix 9, DSELLH,ISELLH, and CVPROF, are used in the cal-culation of constructed value, which serves asnormal value when no viable price comparisons
are available. DSELLH is the average amountof home-market direct selling expensesincurred on sales made in the ordinary courseof trade. ISELLH is the average amount ofindirect selling expenses incurred on salesmade in the ordinary course of trade.CVPROF is the profit rate on sales made inthe ordinary course of trade.
Together, the arm’s-length test and the costtest make a mockery of the presumption thatdumping margin calculations reflect a truecomparison of U.S. and home-market prices.As has been demonstrated, the home-marketdata are subjected to procedures that unam-biguously cause average net prices on that sideof the equation to be exaggerated.
Appendix 10 is an abbreviated sample of theSection C U.S.sales list for the same fictitious for-eign widget producer. As in the home-marketexamples, certain fields have been consolidated tofacilitate illustration of the relevant points. In thistable, which reflects the company’s sales in theUnited States, there are 45 sales observations ofsubject merchandise. These observations includesales of a unique product, CONNUMU 34,which is a large aluminum widget.
Most of the fields represent the same vari-ables as in the home-market exhibits (althoughthey each end with a U for U.S., rather than anH for home market). Fields not appearing onthe home-market side, which are important tothe dumping calculations, include TCOMUand SALEU. In addition, in the U.S. database,when there are CEP sales, indirect expenses aredivided into those incurred domestically andaccruing to all sales (DINDIRSU and DIN-VCARU) and those incurred in the UnitedStates as well as accruing specifically to U.S.sales (INDIRSU and INVCARU).
TCOMU represents the total cost of manu-facturing (not to be confused with the moreinclusive TOTCOP or total cost of production).Although a cost variable, TCOMU is relevantto the sales list because it comes into play whenselecting the appropriate products to compare.Ideally, identical CONNUMs are available inboth markets, which allows for the most reason-able price comparisons. However, because of theproduct exclusions caused by the arm’s-length
25
Together, thearm’s-length testand the cost testmake a mockery ofthe presumptionthat dumping mar-gin calculationsreflect a true com-parison of U.S. andhome-marketprices.
and cost tests, as well as the fact that differentmarkets demand different products, U.S. prod-ucts frequently do not have identical home-mar-ket matches. Under these circumstances, themost similar, nonidentical product is selectedfrom the remaining home-market options forprice comparison. Under DOC procedures, anyproduct is considered similar provided that thedifferences in variable costs (VCOMU-VCOMH) do not exceed 20 percent of the totalcost (TCOMU). Hence, TCOMU serves as abenchmark in determining whether productcomparisons are appropriate.
SALEU indicates whether the sale is anexport price (EP) or constructed export price(CEP) transaction. An EP sale is generallydefined as an arm’s-length transaction betweenthe exporter and an unaffiliated importer withthe terms of the sale (price, quantity, etc.) hav-ing been agreed on before importation of themerchandise. A CEP sale is generally a salemade after importation to an unaffiliated cus-tomer by an importer or reseller that is relatedto the exporter. The terms of CEP sales aregenerally agreed upon after importation of themerchandise. Big exporters with subsidiaryoperations in the United States often conductbusiness this way. Their U.S. subsidiaries areoften the importers of record and maintainwarehouse facilities, sales operations, and man-ufacturing capabilities in the United States.Transactions between the exporter and a relat-ed importer are considered to be unreliablebenchmarks for export price because the pricesmay not reflect true market prices.
Under U.S. antidumping rules, EP andCEP sales are treated differently in the processof calculating net prices. Net prices on EP salesare calculated by deducting only discounts,rebates, and movement expenses from thegross selling price. The selling expenses andpacking costs that are still reflected in the U.S.net prices are added back to the net prices onthe home-market side, which are already net ofthe home-market selling and packing expens-es. Indirect selling expenses are not deductedfrom either side in the dumping calculationsfor EP transactions. In CEP transactions, onthe other hand, home-market and U.S. move-
ment, packing, and direct selling expenses arededucted from each side, respectively. In addi-tion, indirect selling expenses are deductedfrom U.S. prices—and also from home-marketprices, up to the amount of U.S. indirectexpenses (the so-called CEP offset cap). CEPprofit is deducted as well.
To calculate CEP profit, the DOC talliesprofits on home-market and U.S. sales by cal-culating the revenues and expenses on allarm’s-length transactions, as depicted inAppendix 11. This table shows that the overallCEP profit ratio is 1.45 percent, which isapplied to all CEP selling expenses and thendeducted from the gross U.S. price.
Appendix 12 is derived from the originalU.S. sales database and shows the net prices(NETPRIU) along with total amounts fordirect selling expenses (DIREXPU) and pack-ing (PACKINGU).These values are carried for-ward to conduct the dumping margin calcula-tions. Note that on EP sales, the net price equalsthe gross price (GRSUPRU) minus discountsand rebates (DISCREBU) and movementexpenses (MOVEU). On CEP sales, the netprice equals the gross price minus discounts andrebates, movement expenses, direct sellingexpenses (DIRSELLU), credit expenses(CREDITU), U.S. indirect selling expenses(INDSELU and INVCARU), and CEP profit(CEPPROFIT). As a final step before conduct-ing the margin calculations, the net prices, directselling expenses, and packing costs are weight-averaged by CONNUMU, PRIMEU, andSALEU. Those figures appear in Appendix 13.
With average prices and expenses calculated inboth markets, the DOC determines which pricesto compare by following its model-matching pro-cedure. For each U.S. CONNUMU, the mostsimilar home-market CONNUMH is selected asthe basis for the comparison. Although identicalmatches are always the most similar, they are notalways available for reasons already discussed.Appendix 14 summarizes the outcome of themodel match as well as the calculated dumpingmargins for this hypothetical example.
Note that most of the U.S. CONNUMUswere matched to identical home-marketCONNUMHs. Despite the arm’s-length and
26
Under U.S.antidumping rules,EP and CEP salesare treated differ-
ently in the processof calculating
net prices.
cost tests, at least some home-market sales ofthese products (namely, the higher-pricedsales) remained in the database. CONNUMU34, which was unique to the U.S. market, didnot match an identical home-market product.The most similar home-market product wasCONNUMH 33, a large vinyl widget. Thedifference in variable costs, or DIFMER(VCOMU-VCOMH), was less than 20 per-cent of the total cost to manufacture CON-NUMU 34 (TCOMU), so the match passedthat requirement.
U.S. CONNUMU 32, however, found noeligible matches from the available pool ofhome-market CONNUMHs. All home-mar-ket sales of the identical CONNUMH weredropped after the arm’s-length and cost tests,and none of the remaining CONNUMHspassed the cost requirement. The closestmatches physically, CONNUMH 31 and 33,have variable costs of $5.30 and $8.10, respec-tively (see VCOMH in Appendix 1). With avariable cost of $6.70 and a total cost of $6.90for CONNUMU 32 (see VCOMU andTCOMU in Appendix 10), the differences invariable costs between the U.S. model and thetwo home-market models exceed 20 percent ofthe total cost of CONNUMU 32. As a result,the price of this CONNUMU was matched toconstructed value, as indicated by CV in thefield CONNUMH.
The dumping margin is calculated by com-paring the net U.S. price (NETPRIU) to theforeign unit price in dollars (FUPDOL). Forprice-to-price comparisons, FUPDOL isbased on the net home-market price (NET-PRIH). For price–to–constructed value com-parisons, FUPDOL is based on cost (TOT-COP). In either case, FUPDOL contains ele-ments in addition to these main components.
For price-to-price comparisons involvingEP sales, FUPDOL equals NETPRIH +DIFMER + DIREXPU + PACKU. DIFMERis the difference-in-merchandise adjustment orthe difference in variable costs, which is zerofor identical matches. DIREXPU refers to theaverage U.S. direct selling expenses, andPACKU is the average U.S. packing expense.For price-to-price comparisons involving CEP
sales, FUPDOL equals NETPRIH +DIFMER + PACKU.
For price–to–constructed value comparisonsinvolving EP sales, FUPDOL equals CV +DIREXPU, and for price–to–constructed valuecomparisons involving CEP sales, FUPDOLequals CV. In either case, CV equals TOTCOP+ HMISEL + PACKU + CVPROF. HMISELis the average amount of home-market indirectselling expenses on sales made in the ordinarycourse of trade. CVPROF is the estimated prof-it, which is based on the profit rate of sales madein the ordinary course of trade, converted to anabsolute figure by multiplying by TOTCOP.
Ultimately, the unit margin (UMARGIN) isthe difference between FUPDOL and NET-PRIU.That unit margin is multiplied by the quan-tity sold of each U.S. product (CONNUMU /PRIMEU / SALEU combination) to calculate theextended margin,or EMARGIN. If EMARGINis less than zero, it is set equal to zero in the fieldTOTPUDD, which stands for total potentiallyuncollected dumping duties.This practice of zero-ing, as discussed previously, contributes, in somecases substantially, to the exaggeration of dumpingmargins.The summation of TOTPUDD is morethan twice as large as the summation of EMAR-GIN, which is what the amount of dumpingwould equal if the negative margins were consid-ered. The field VALUE is the total net value ofU.S. sales,which equals NETPRIU times QTYU.The bottom-line dumping margin is calculated asthe sum of the TOTPUDD field ($184.31) divid-ed by the sum of the VALUE field ($2,500.31),which in this case equals 7.37 percent.
Table 9 shows what the dumping marginswould be if various methodological distortionswere corrected. If the arm’s-length test were elim-inated, NETPRIH values would be lower andthe margin would drop to 7.19 percent. If therewere no cost test, NETPRIH values would belower, and there would have been no CV com-parisons.The dumping margin would fall to zero.(Appendix 15 provides the documentation forthese calculations.) If indirect selling expenseswere not deducted in CEP transactions, marginswould fall to 7.14 percent. If no CEP deductionwere made, the dumping margin would decreaseto 7.35 percent. If the practice of zeroing were
27
Findings of dump-ing under the current U.S.antidumping laware barely worththe pages of theFederal Registerthey are printed on.
eliminated, the dumping margin would fall to3.21 percent. And if all these methodologicalflaws were eliminated, this fictitious widgetmaker would have a dumping margin of –13.95percent. In other words, it would be cleared of allcharges of dumping.
This example illustrates how a fictitious com-pany that sells at significantly higher prices in theUnited States than in its home market cannonetheless be found to be dumping under U.S.rules. In short, findings of dumping under thecurrent U.S. antidumping law are barely worththe pages of the Federal Register they are printedon. All too often they convey no useful informa-tion about a company’s selling practices for thesimple reason that the underlying procedures forevaluating those practices are fatally flawed.
Conclusion
This introduction to the antidumping law’s“tricks of the trade” reveals a sharp divergencebetween the law’s inner workings and itswholesome public image. The antidumpinglaw is hailed by supporters across the politicalspectrum as the guarantor of a “level playingfield” for U.S. industry and import competi-tion. In fact, however, the law systematically
discriminates against foreign goods withskewed rules that generate dumping marginsout of thin air.
The antidumping law imposes trade-restrictive duties, yet its supporters claim that itis not protectionist. They contend thatantidumping measures target only “unfairtrade”—in particular, artificially low-pricedproducts exported from sanctuary marketswhere normal competitive forces are sup-pressed. Because of the underlying distortionsof a sanctuary market, prices in the home mar-ket are abnormally high, which then allows for-eign producers to sell here in the United Statesat abnormally low prices. The antidumpinglaw, by imposing duties on imports sold atprices lower than those charged in the homemarket, counteracts the distortions caused bysanctuary markets without interfering in nor-mal, healthy foreign competition.
Or so the story goes. The reality, though, isvery different. In this study, we put aside thedeeper question of whether price differencesbetween national markets are, in and of them-selves, reliable evidence of market distortions.Instead, we examine the more basic issue ofwhether the antidumping law accurately mea-sures differences between U.S. prices and for-eign-market prices. Unfortunately, the law as it
28
The law systemati-cally discriminates
against foreigngoods with skewedrules that generatedumping margins
out of thin air.
Dumping Margin
Using DOC methodology 7.37%
Exclude arm's-length test only 7.19%Exclude cost test only 0.00%Exclude both "ordinary course" tests 0.00%
Exclude CEP indirect selling expense deduction only 7.14%Exclude CEP profit deduction only 7.35%Exclude both CEP asymmetric adjustments 7.13%
Exclude zeroing only 3.21%
Exclude all methodological distortions -13.95%
Table 9Hypothetical Case Study: Effect of Methodological Distortions on Dumping Margins
currently stands fails to meet that basic test—and fails badly. Comparisons of average pricescan be skewed by something as simple andcommon as price fluctuations over the periodof investigation. The existence or nonexistenceof dumping margins can turn on unavoidablyarbitrary definitions of the products beingcompared. Comparisons of nonidentical prod-ucts, even with difference-in-merchandiseadjustments, can easily produce dumping mar-gins that reflect nothing more than differentcommercial values.
Compounding these problems, the law thenproceeds to make matters worse—much worse—with rules that skew the calculation and compar-ison of net prices in the direction of higher dump-ing margins. The arm’s-length test eliminateslow-priced sales to home-market affiliated com-panies, but not high-priced sales, thus inflatingnormal value and, consequently, dumping mar-gins. The cost test eliminates low-priced sales inthe home market when they are found to bebelow cost, thus boosting dumping margins inprecisely those situations where the evidence for asanctuary market (which is supposedly highlyprofitable) is weakest. Indirect selling expenses arefully deducted from U.S. prices in CEP transac-tions, but not from the home-market prices towhich they are compared. CEP profit is also
deducted from U.S. prices in these situations, butno corresponding deduction is made on thehome-market side. Constructed value, whichsometimes serves as the benchmark for judgingthe fairness of U.S. prices, is inflated by the unjus-tifiable addition of an amount for profit—anamount that is then exaggerated by being basedonly on above-cost home-market sales. Finally,the practice of zeroing ensures that dumpingmargins are routinely much higher than the actu-al differences in net-price levels.
In sum, the antidumping law does not dowhat its supporters say it does. It does not accu-rately measure the differences between U.S.pricesand foreign-market prices. It is therefore inca-pable of distinguishing between “unfair trade”and normal,healthy competition.As a result,nor-mal, healthy competition from abroad is all toooften stifled in the name of fighting dumping. Inother words, the antidumping law,while pretend-ing to secure a “level playing field,” in factindulges in old-fashioned protectionism. Thatprotectionism is no less real because it is obscuredby a fog of technical complexity. Indeed, its com-plexity makes the law all the more effective as aprotectionist vehicle by shielding it from scrutiny.The aim of this study is to penetrate that fog andthereby expose the antidumping law to the criti-cal scrutiny it so richly deserves.
29
The antidumpinglaw, while pretend-ing to secure a “levelplaying field,” infact indulges in old-fashioned protectionism.
Notes1. Brink Lindsey, “The U.S. Antidumping Law:Rhetoric versus Reality,” Cato Trade Policy Analysisno. 7, August 16, 1999.
2. The antidumping statute is codified at 19U.S.C. §§ 1673–1677n. The DOC’s antidumpingregulations may be found at 19 C.F.R. § 351.
3. In antidumping cases against so-called nonmar-ket economies, the DOC calculates constructedvalue, not with the foreign producer’s own costdata, but with data from “surrogate” marketeconomies. See Lindsey, p. 5.
4. In the three-year period between April 1999 andMarch 2002, the ITC rendered 147 preliminarydeterminations, of which 116 were affirmative.These figures were compiled from data availableat the ITC website (www.usitc.gov).
5. Typically, the DOC attempts to send question-naires to all known foreign producers. If, however,the number of foreign producers is too large, theDOC will send questionnaires to only the largestforeign producers.
6. There were 125 specific final margins assignedin original investigations during 2001. Of those125, 117 were higher than de minimis—or at least2 percent. These figures were compiled from pub-lic data available in Federal Register notices.
7. In the three-year period between April 1999 andMarch 2002, the ITC rendered 58 final determina-tions, of which 48 were affirmative. These figures werecompiled from data available at the ITC website(www.usitc.gov).
8. These figures were compiled as of August 14,2002, from data available on the ITC website(www.usitc.gov).
9. Generally, the foreign producer’s home marketis deemed viable if sales volume in that market ofthe product under investigation exceeds 5 percentof its sales volume in the United States (althoughother factors can render a home market notviable). For the remainder of this paper and forthe sake of simplicity, the home market isassumed to be the comparison market.
10. Theoretically, there could be hundreds of thou-sands and even millions of products. The numberof possible products, theoretically, equals the prod-uct of multiplying the number of options withineach characteristic. So, if there are 10 characteris-tics, each with five alternative values, the number oftheoretical products equals 5 x 5 x 5 x 5 x 5 x 5 x 5x 5 x 5 x 5, or 9,765,625 products.
11. The definition of an affiliated person, as pro-vided in the DOC’s standard antidumping ques-tionnaire, is as follows:
Affiliated persons (affiliates) include (1)members of a family, (2) an officer or director ofan organization and that organization, (3) part-ners, (4) employers and their employees, and (5)any person or organization directly or indirect-ly owning, controlling, or holding with powerto vote, 5 percent or more of the outstandingvoting stock or shares of any organization andthat organization. In addition, affiliates include(6) any person who controls any other personand that other person, or (7) any two or morepersons who directly control, are controlled by,or are under common control with, any person.“Control” exists where one person is legally oroperationally in a position to exercise restraintor direction over the other person [section771(33) of the act].
12. In some cases, there are no sales to unaffiliatedcustomers of products sold to affiliates, so thecomparison cannot be made. As long as there is asufficient basis for comparing prices (normally justone common product), then the arm’s-length testis performed. If there are no common products,the affiliated customer generally fails the test.
13. Because of a World Trade Organization rulingthat the current U.S. arm’s-length test violates theWTO Antidumping Agreement, the DOC is nowconsidering a change in the test. See below for adiscussion of the proposed change from the cur-rent 99.5 percent standard to a band in whichsales below a 98 percent threshold and above a102 percent threshold are excluded.
14. Average expenses and profit for purposes ofCV are normally based on the experience of home-market sales in the ordinary course of trade.However, when there are no such sales becausethey have all failed the arm’s-length or cost test,this information is estimated from other sources.
15. This 20 percent threshold is known as the dif-ference-in-merchandise test.
16. Actually, the averaging process is even more refined.An average price is calculated for each unique combi-nation of U.S. CONNUM and sales type (EP or CEP).
17. Antidumping supporters frequently contend thatdumping can take the form of either internationalprice discrimination or below-cost export sales. Inpractice, however, the antidumping law does notattempt to measure whether subject imports are soldbelow their cost of production. The closest that it evercomes is when U.S. prices are compared with con-structed value—which equals cost of production plussome amount for profit. This artificial price is used as
46
a surrogate for normal value when comparison-mar-ket prices are unavailable or have been deemed unus-able; it is thus used as a “filler” and seldom serves asthe exclusive basis of normal value. And even when allU.S. prices are compared with constructed value, whatis measured is not whether the U.S. sales are belowcost; what is measured, rather, is whether U.S. sales arebelow some designated benchmark of profitability.See Lindsey, pp. 2–3, 7.
18. Ibid., pp. 11–15.
19. Appellate Body Report on United States–Anti-Dumping Measures on Certain Hot-Rolled Steel Productsfrom Japan, WT/DS184/AB/R, July 24, 2001.
20. See the DOC’s request for public comment, 67Fed. Reg. 53339, August 15, 2002.
21. These determinations were made in nine sepa-rate original investigations and three separateadministrative reviews. In other words, in some caseswe were able to obtain determinations for multiplecompanies in a single investigation or review.
22. This investigation gave rise to the dispute thatresulted in the WTO’s ruling that the U.S. arm’s-
length test is WTO-inconsistent.
23. The one DOC determination examined in whichthe cost test did not inflate margins was a nonmarketeconomy case. The cost test is not used in nonmarketeconomy cases because normal value is based on cost,and home-market sales are never considered.
24. In addition, narrower product definitionsheighten the distortions caused by the practice ofzeroing. See below for a detailed discussion.
25. See below for a discussion of the DOC’s prac-tice of zeroing when calculating dumping margins.
26. Appellate Body Report on European Communities—Anti-Dumping Duties on Imports of Cotton-Type Bed Linenfrom India, WT/DS141/AB/R, March 1, 2001.
27. The 88.65 percent figure is based on allowingthe full effect of negative dumping margins. Inpractice, a negative dumping margin is equivalentto a 0 percent margin.
28. Typically, there are hundreds, thousands, oreven tens of thousands of individual sales obser-vations in a Section B response.
47
Trade Policy Analysis Papers from the Cato Institute
“Willing Workers: Fixing the Problem of Illegal Mexican Migration to the United States” by Daniel T. Griswold (no. 19, October 15,2002)
“The Looming Trade War over Plant Biotechnology” by Ronald Bailey (no. 18, August 1, 2002)
“Safety Valve or Flash Point? The Worsening Conflict between U.S.Trade Laws and WTO Rules” by Lewis E. Leibowitz (no. 17,November 6, 2001)
“Safe Harbor or Stormy Waters? Living with the EU Data Protection Directive” by Aaron Lukas (no. 16, October 30, 2001)
“Trade, Labor, and the Environment: How Blue and Green Sanctions Threaten Higher Standards” by Daniel T. Griswold (no. 15,August 2, 2001)
“Coming Home to Roost: Proliferating Antidumping Laws and the Growing Threat to U.S. Exports” by Brink Lindsey and DanIkenson (no. 14, July 30, 2001)
“Free Trade, Free Markets: Rating the 106th Congress” by Daniel T. Griswold (no. 13, March 26, 2001)
“America’s Record Trade Deficit: A Symbol of Economic Strength” by Daniel T. Griswold (no. 12, February 9, 2001)
“Nailing the Homeowner: The Economic Impact of Trade Protection of the Softwood Lumber Insudstry” by Brink Linsey,Mark A. Groombridge, and Prakash Loungani (no. 11, July 6, 2000)
“China’s Long March to a Market Economy: The Case for Permanent Normal Trade Relations with the People’s Republic ofChina” by Mark A. Groombridge (no. 10, April 24, 2000)
“Tax Bytes: A Primer on the Taxation of Electronic Commerce” by Aaron Lukas (no. 9, December 17, 1999)
“Seattle and Beyond: A WTO Agenda for the New Millennium” by Brink Lindsey, Daniel T. Griswold, Mark A.Groombridge and Aaron Lukas (no. 8, November 4, 1999)
“The U.S. Antidumping Law: Rhetoric versus Reality” by Brink Lindsey (no. 7, August 16, 1999)
“Free Trade, Free Markets: Rating the 105th Congress” by Daniel T. Griswold (no. 6, February 3, 1999)
“Opening U.S. Skies to Global Airline Competition” by Kenneth J. Button (no. 5, November 24, 1998)
“A New Track for U.S. Trade Policy” by Brink Lindsey (no. 4, September 11, 1998)
“Revisiting the ‘Revisionists’: The Rise and Fall of the Japanese Economic Model” by Brink Lindsey and Aaron Lukas (no. 3,July 31, 1998)
“America’s Maligned and Misunderstood Trade Deficit” by Daniel T. Griswold (no. 2, April 20, 1998)
“U.S. Sanctions against Burma: A Failure on All Fronts” by Leon T. Hadar (no. 1, March 26, 1998)
Board of Advisers
James K. GlassmanAmerican EnterpriseInstitute
Douglas A. IrwinDartmouth College
Lawrence KudlowKudlow & Co.
José PiñeraInternational Center forPension Reform
Razeen SallyLondon School ofEconomics
George P. ShultzHoover Institution
Walter B. WristonFormer Chairman andCEO, Citicorp/Citibank
Clayton YeutterFormer U.S. TradeRepresentative
Nothing in Trade Policy Analysis should be construed as necessarily reflecting the views of theCenter for Trade Policy Studies or the Cato Institute or as an attempt to aid or hinder the pas-sage of any bill before Congress. Contact the Cato Institute for reprint permission. Additionalcopies of Trade Policy Analysis studies are $6 each ($3 for five or more). To order, contact theCato Institute, 1000 Massachusetts Avenue, N.W., Washington, D.C. 20001. (202) 842-0200,fax (202) 842-3490, www.cato.org.
The mission of the Cato Institute’s Center for Trade Policy Studies is to increase publicunderstanding of the benefits of free trade and the costs of protectionism. The center
publishes briefing papers, policy analyses, and books and hosts frequent policy forums andconferences on the full range of trade policy issues.
Scholars at the Cato trade policy center recognize that open markets mean wider choicesand lower prices for businesses and consumers, as well as more vigorous competition thatencourages greater productivity and innovation. Those benefits are available to any countrythat adopts free-trade policies; they are not contingent upon “fair trade” or a “level playingfield” in other countries. Moreover, the case for free trade goes beyond economic efficiency.The freedom to trade is a basic human liberty, and its exercise across political borders unitespeople in peaceful cooperation and mutual prosperity.
The center is part of the Cato Institute, an independent policy research organization inWashington, D.C. The Cato Institute pursues a broad-based research program rooted in thetraditional American principles of individual liberty and limited government.
For more information on the Center for Trade Policy Studies,visit www.freetrade.org.
Other Trade Studies from the Cato Institute
“Willing Workers: Fixing the Problem of Illegal Mexican Migration to the United States” byDaniel T. Griswold, Trade Policy Analysis no. 19 (October 15, 2002)
“The Looming Trade War over Plant Biotechnology” by Ronald Bailey, Trade Policy Analysisno 18 (August 1, 2002)
“Rethinking the Export-Import Bank” by Aaron Lukas and Ian Vásquez, Trade BriefingPaper no. 15 (March 12, 2002)
“Steel Trap: How Subsidies and Protectionism Weaken the U.S. Steel Industry” by Dan Ikenson,Trade Briefing Paper no. 14 (March 1, 2002)
“America’s Bittersweet Sugar Policy” by Mark A. Groombridge, Trade Briefing Paper no. 13(December 4, 2001)
“Safety Valve or Flash Point? The Worsening Conflict between U.S. Trade Laws and WTORules” by Lewis E. Leibowitz, Trade Policy Analysis no. 17 (November 6, 2001)
