AK Steel Corporation, Inland Steel Industries, Inc., Bethlehem Steel Corporation, LTV Steel Company, Inc., National Steel Corporation, and U.S. Steel Group (collectively “domestic producers” or “appellants”) appeal the judgment of the United States Court of International Trade in this anti-dumping duties case. The court upheld the International Trade Administration, United States Department of Commerce’s (“Commerce”) decision: (1) using a three-part test adopted informally in 1987 to determine whether certain sales to U.S. buyers of Korean steel by U.S. affiliates of the Korean producers 1 are properly classified as Export Price (“EP”) sales rather than Constructed Export Price (“CEP”) sales, and (2) declining to apply the “fair-value” and “major-input” provisions of 19 U.S.C. §§ 1677b(f)(2)-(3) (1994) to transfers among affiliated steel producers in Korea that it had treated as one entity for purposes of the anti-dumping determination. As a consequence of these methods and their manner of application, the duty rates were minimal. The domestic producers therefore filed suit in the trial court challenging these methods as contrary to the anti-dumping statutes. See AK Steel Corp. v. United States, 34 F.Supp.2d 756 (Ct. Int’l Trade 1998). We hold that the three-part test employed by Commerce is contrary to the express *1333terms defining EP and CEP .in the statute as amended in 1994 and therefore reverse-in-part and remand for a redetermination of the anti-dumping duties. As to the fair-value and major-input provisions, however, we hold that Commerce’s decision not to apply those provisions to the transactions was reasonable and within its discretion, and therefore we affirm-in-part.
BACKGROUND
In 1993 Commerce issued an order imposing anti-dumping duties on certain steel products from Korea. See Certain Cold Boiled Steel Flat Products from Korea, 58 Fed.Reg. 44,159 (Dept. of Commerce 1993) (hereinafter “Certain Steel Products from Korea ”). In August of 1995 both the domestic producers and the Korean producers requested an administrative review of that anti-dumping duty order. In its second administrative review of the anti-dumping duty order, Commerce classified all of the sales of the subject merchandise at issue in this appeal2 as EP sales rather than CEP sales pursuant to 19 U.S.C. §§ 1677a(a)-(b) (1994). See Certain Cold-Rolled and Corrosion-Resistant Carbon Steel Flat Products from Korea, 62 Fed. Reg. 18,404, 18,434 (Dept, of Commerce 1997) (hereinafter “Final Results ”). In addition, because Commerce had collapsed POSCO and its affiliates POCOS.and.PSI for purposes of assigning dumping margins, it opted not to apply the so-called “fair-value” and “major-input” provisions to transactions between those companies. Id. at 18,430.
I.
In calculating dumping margins, Commerce compares the “U.S. Price” to the “normal value” of the subject merchandise, imposing anti-dumping duties if, and to the extent, the former is lower than the latter. The U.S. Price is calculated using either the EP or CEP methodology. In general, Commerce uses the EP methodology for the U.S. Price when the foreign producer or exporter sells directly to an unrelated purchaser located in the United States. The CEP methodology is used when the foreign producer’s or exporter’s steel is sold to an unaffiliated' U.S. buyer by an affiliated company located in the United States. If the CEP methodology is used, additional deductions are taken from the sales price to arrive at the U.S. Price.3 The statute defines EP and CEP as follows:
(a) Export Price
The term “export price” means the price at which the subject merchandise is first sold (or agreed to be sold) before the date of importation by the producer or exporter of the subject merchandise outside of the United States to an unaffiliated purchaser in the United States or to an unaffiliated purchaser for exportation to the United States....
(b) Constructed Export Price
The term “constructed export price” means the price at which the subject merchandise is first sold (or agreed to be sold) in the United States before or after the date of importation by or for the account of the producer or exporter of such merchandise or by a seller affiliated with the producer or exporter, to a purchaser not affiliated with the producer or exporter....
*133419 U.S.C. §§ 1677a(a) — (b).4
For the sales of steel produced by each of the appellees, Commerce calculated the U.S. Price using the EP methodology. In determining whether to classify the sales as EP or CEP, Commerce applied a three-part test it developed during a remand of a 1987 case, PQ Corp. v. United States, 652 F.Supp. 724, 733-35 (Ct. Int’l Trade 1987) (the “PQ Test”). Representing an interpretation of the above statute, the test has been applied when sales are made prior to importation to an unaffiliated U.S. purchaser by a foreign manufacturer’s affiliated entity in the United States, as in the case of the sales at issue here. Using the PQ Test, Commerce classifies sales made by U.S. affiliates as EP sales if the following criteria are met:
(1) the subject merchandise was shipped directly from the manufacturer to the unrelated buyer, without being introduced into the inventory of the related shipping agent;
(2) direct shipment from the manufacturer to the unrelated buyer was the customary channel for sales of this merchandise between the parties involved;
(3) the related selling agent in the United States acted only as a processor of sales-related documentation and a communication link with the unrelated U.S. buyer.
See, e.g., Certain Stainless Steel Wire Rods from France, Final Determination of Sales at Less than Fair Value, 58 Fed. Reg. 68,865, 68,868-69 (1993).
All of the sales at issue in the present case were “back-to-back” sales. That is: the Korean producer sold the steel to an affiliated Korean exporter; the exporter sold it to a U.S. affiliate; and finally, the U.S. affiliate sold it to the unaffiliated U.S. purchaser. In most cases, however, the steel was shipped directly to the unaffiliated purchaser without entering the inventory of the U.S. affiliate. In the second administrative review, whether the sales of steel manufactured by the Korean producers satisfied the third prong of the test was one of the principal factual issues in dispute. In classifying the sales at issue, Commerce rejected the domestic producers’ argument that the activities of the Korean producers’ U.S. affiliates failed the third prong of the test because they “exceeded] those of a mere communications link or processor of documents.” Final Results, 62 Fed.Reg. at 18,432.
II.
Commerce “collapsed” POSCO and its affiliates POCOS and PSI into one entity for purposes of the anti-dumping analysis and then levied a single anti-dumping duty on the entity. In the second administrative review, Commerce determined that “a decision to treat affiliated parties as a single entity necessitates that transactions among the parties also be valued based on the group as a whole ... [thus] among collapsed entities, the fair-value and major-input provisions are not controlling.” Final Results, 62 Fed.Reg. at 18,430. Therefore, in its 1995 review, Commerce declined to treat the transfers between the related companies as sales between affiliates but rather treated them as transfers between divisions of the same company and did not apply the fair-value and major-input provisions of 19 U.S.C. §§ 1677b(f)(2) — (3).
The domestic producers appealed the Final Results to the Court of International Trade, challenging the legality of the PQ Test and its application to the Appellees, the decision to collapse the POSCO affili*1335ates, and the determination that the fair-value and major-input -provisions did not apply to transfers among the collapsed companies. The Court of International Trade sustained Commerce’s Final Results, holding the PQ Test to be a reasonable interpretation of the statute and the application to be sustainable in this case. In addition the court held that the decisions to collapse the affiliated producers and not apply the fair-value and major-input provisions were within the discretion of the agency. See AK Steel Corp., 34 F.Supp.2d at 762, 764-66. The domestic producers timely appealed to this court those portions of the judgment based on statutory interpretation, challenging the legality of the PQ Test and the decision not to apply the fair-value and major-input provisions.
This court has jurisdiction under 28 U.S.C. § 1296(a)(5) (1994).
DISCUSSION
I. Standard of Review
The appellants challenge neither the application of the PQ Test to the sales by the various Korean producers at issue in this case nor the reasonableness of Commerce’s decision to “collapse” the affiliated Korean companies into one entity. Thus, no facts are in dispute and the only issues before us are: (1) whether' the Court of International Trade erred in concluding that Commerce’s PQ Test is a correct, or at least a reasonable, interpretation of ambiguous terms of the statute, despite the 1994 amendments; and (2) whether it erred in holding that the decision not to apply'the fair-value and major-input, provisions to transactions among related entities was within Commerce’s discretion.
“In reviewing the Court of International Trade, this court decides de novo the proper interpretation of governing statutory provisions.” Koyo Seiko Co. v. United States, 36 F.3d 1565, 1570 (Fed.Cir.1994). When reviewing statutory interpretation by an agency charged with implémenting a statute by rulemaking and adjudication, as is the. Commerce Department here, see Suramerica de Aleaciones Laminadas v. United States, 966 F.2d 660, 665 & n. 5 (Fed.Cir.1992), we are guided by the Supreme Court’s decision in Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 842-43, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984), which requires the court to ask two questions. Under Chevron the court first asks whether Congress has made an express or implied delegation to the agency or if it has “directly spoken to the precise question at Issue.” Id. If it has not made any delegation to the agency, then courts and the agency “must give effect to the unambiguously expressed intent of Congress.” Id. If Congress has not spoken unambiguously on the subject, and thus made an express or implied delegation to the agency; then the court must ask Chevron’s second question and determine whether the interpretation of the statute by the agency charged with implementing it is a reasonable one. See Id'. The core issue in this appeal is whéther Congress itself discriminated between EP and CEP sales, or left it to Commerce to do so.
II. Classification as Export Price Sales
The Court of International Trade held that the PQ Test did not contradict the statute as amended. The court found that the test was “simply a means to determine whether the sale at issue for anti-dumping duty purposes is in essence between the exporter/producer and the unaffiliated buyer, -in which case the EP rules apply.” AK Steel, 34 F.Supp.2d at 762 (emphasis added). The domestic producers argue that Commerce’s PQ Test is in conflict with the unambiguously expressed intent of Congress because the statute and legis*1336lative history make clear that a sale by any affiliated seller in the United States to an unrelated U.S. buyer must be classified as CEP. The appellees argue, however, that the statute is ambiguous about how to classify those sales that occur before importation but are made by affiliated entities in the United States. Therefore, according to the appellees, the PQ Test is an appropriate methodology for determining whether EP or CEP classification is best applied to those sales.
The language of the statute must be viewed in context. The U.S. Price used in making anti-dumping determinations is meant to be the sales price of an arm’s length transaction between the foreign producer and an unaffiliated U.S. purchaser. The U.S. Price is derived from either EP or CEP sales. To isolate an arm’s length transaction under the current statute, Commerce looks to the first sale to a purchaser that is not affiliated with the producer or exporter. If the producer or exporter sells directly to the U.S. purchaser, that sale is used because it is considered an arm’s length transaction. In that situation the sale is classified as EP.5 If, however, the first sale to an unaffiliated purchaser occurs in the United States, then that sale must be used to determine the U.S. Price. Such a sale will be classified as a CEP sale and have additional deductions made to account for certain expenses of the seller in the United States.6 The purpose of these additional deductions in the CEP methodology is to prevent foreign producers from competing unfairly in the United States market by inflating the U.S. Price with amounts spent on marketing and selling their products based on costs that are in excess of what is spent in their home markets. In the administrative review process, the foreign producers submit to Commerce the information about sales to unaffiliated purchasers. Those sales must be classified as either: (1) between an unaffiliated U.S. purchaser and the producer or exporter, and thus EP; or (2) between the unaffiliated U.S. purchaser and another entity in the United States, that must, by definition, be related to the producer, and thus CEP. Sales in the United States between unaffiliated purchasers and unaffiliated sellers are never at issue; such a sale could never be the first sale to an unaffiliated purchaser.
The question at the root of this appeal is whether a sale to a U.S. purchaser can be properly classified as a sale by the producer/exporter, and thus an EP sale, if the sales contract is between the U.S. purchaser and a U.S. affiliate of the producer/exporter and is executed in the United States. The appellees argue that it can, if the role of the U.S. affiliate is sufficiently minor that the sale passes the PQ Test. The domestic producers argue that the plain language of the statute prevents such a classification.
We note first that Commerce’s three-part test and much of the Court of Inter*1337national Trade case law reviewing it were created before the enactment of the URAA in 1994. Prior to the URAA, “purchase price” (now EP) was described as:
the price at which merchandise is purchased, or agreed to be purchased, prior to the date of importation, from a reseller or the manufacturer or producer of the merchandise for exportation to the United States.
19 U.S.C. § 1677a(b) (1988). The “exporters sales price” (now CEP) was defined as:
the price at which merchandise is sold or agreed to be sold in the United States, before or after the time of importation, by or for the account of the exporter.
19 U.S.C. § 1677a(c) (1988). The amendments to the statute most relevant to this issue are the addition of the phrase “outside the United States” to the definition of EP, and “by a seller affiliated with the producer” to the definition of CEP.
While the statute has changed, Commerce’s test for interpreting it has not; Commerce still relies on the PQ Test. The function of this court is to determine whether the use of the test is appropriate in light of the current statutory language, not whether the test may have been a proper interpretation of the language effective as of the time the test was developed in 1987.
In PQ Corporation, the Court of International Trade focused on whether there was a relationship between the foreign producer and the U.S. importer as the primary factor enabling Commerce to differentiate between the two sales classifications in the old statute. In response to an argument by Commerce that there was no statutory requirement that “the importer must be an independent party in order to apply [EP],” the court held that:
[w]hile the statute does not state in so many words that [purchase price] and [exporter’s sales price] are to be distinguished by the relationship of. the foreign producer to the U.S. importer, the statutory definitions of [purchase price] and [exporters sales price] have been distinguished upon this basis from their inception .... The express terms of the statute make it clear that a U.S. importer’s relationship to a foreign producer will affect the determination of whether [purchase price] or [exporter’s sales price] will apply.
PQ Corporation, 652 F.Supp. at 732-33 (emphasis added). The test developed by Commerce after the remand in PQ Corporation does not directly examine the legal relationship between the producer and the importer, but rather looks at the role played by the importer in the transaction.
A. Amendment to the Constructed Export Price Definition
After the 1994 amendments the language of the statute became clear. Any ambiguity that might have necessitated a test to define what type of activity by an affiliated importer might give rise to a sale “in the United States” was eliminated by the addition of tfie phrase “by a seller affiliated with the producer.” 19 U.S.C. § 1677a(b). Like the language in PQ Corporation, the plain meaning of the language enacted by Congress focuses on whether the .foreign producer and the U.S. importer are related. The language added by the 1994 amendments makes clear that if a sale or.agreement to sell occurred in the United States and was made by a seller affiliated with the producer/exporter, then that sale must be classified as a CEP sale and the CEP deductions applied to the sale price. In contrast, an EP sale, may only be made by the producer or exporter, as the EP definition in the statute makes no provision for an EP sale by a U.S. affiliate of the foreign producer. Thus, the additional language highlights the difference between the two definitions: an EP sale can only be made by the pro*1338ducer/exporter while a CEP sale can be made by an affiliate:
Here the sales contracts in evidence clearly indicate that the sales to the unaffiliated U.S. purchasers were made by affiliates of the foreign producers that are located in the United States. If the importer and the producer are affiliated, then the first sale to an unaffiliated party is necessarily the sale between the affiliated importer and the unaffiliated purchaser (unless there is another intermediate U.S. affiliate involved). With the addition of the language in the 1994 amendment, the relationship that necessitated use of a sale between U.S. entities (CEP) rather than between a foreign and U.S. entity (EP) to establish the U.S. Price became even clearer: a U.S. seller to the ultimate buyer that is affiliated with the producer requires CEP classification, regardless of the specific activities or role of the U.S. affiliate, or whether the foreign producer was the party that initially negotiated the terms of the sale. Thus, the 1994 amendment rendered the PQ Test inappropriate because any arguable ambiguity was removed from the statute,» and the test conflicts with the plain meaning of the statute.
Despite the plain meaning of the language of the statute, the government and the Korean manufacturers cite to the Statement of Administrative Action (“SAA”) that accompanied the URAA as evidence of congressional intent to endorse the PQ Test as a proper interpretation of the new statutory language. The SAA declares that the “[n]ew section 772 retains the distinction in existing law between ‘purchase price’ (now called the ‘export price’) and ‘exporters sales price’ (now called the ‘constructed export price’).” That statement goes on to declare that “[Notwithstanding the change in terminology, no change is intended in the circumstances under which export price ... versus constructed export price ... are used.” H. Doc. No. 103-316, Vol. 1 at 822 (1994), reprinted in 1994 U.S.C.C.A.N. 3773, 4163. The appellants argue that this indicates congressional approval of the distinctions made by the PQ Test. We disagree.
The addition of the word “affiliated” to the CEP sales definition is consistent with a definition of CEP sales based on whether the foreign exporter/producer and the U.S. importer are related, as the Court of International Trade said in PQ Corporation. There is nothing in the SAA to indicate that Congress was aware of Commerce’s PQ Test or intended the use of EP or CEP to be determined based on the activities of the importer rather than the relationship between the importer and the producer/exporter. Defining a sale by a U.S. affiliate as one that requires a CEP classification did not change the circumstances under which CEP and EP had historically been used. The addition of this term merely reinforced the language in the statute pri- or to the 1994 amendment and the holdings in the case law by emphasizing that the critical question was not the role of the affiliate in the sale but the relationship between the seller and the producer/exporter, i.e., whether the U.S. importer is an affiliate of the foreign producer.
In addition, the Korean producers argue that because the terms “seller” and “sold” are undefined in the statute, they are therefore ambiguous. Thus, they argue, Commerce developed the PQ Test to determine, based on the affiliate’s activities, if it is indeed a “seller.” When a word is undefined in a statute the agency and the reviewing court should give the undefined term its ordinary meaning. See Perrin v. United States, 444 U.S. 37, 42, 100 S.Ct. 311, 62 L.Ed.2d 199 (1979) (“A fundamental canon of statutory construction is that, unless otherwise defined, words will be interpreted as taking their ordinary, contemporary, common meaning.”). Black’s Law Dictionary (6th ed.1990) defines “seller” as “one who has *1339contracted to sell property ... the party who transfers property in the contract of sale.” As to “sold,” this court previously addressed the meaning of that term in the definition of the Exporter’s Sales Price (now CEP). See NSK Ltd. v. United States, 115 F.3d 965, 973 (Fed.Cir.1997). In that case we defined “sold” to require both a “transfer of ownership to an unrelated party and consideration.” Id. at 975 (emphasis added). We see no reason to question these definitions, and therefore hold that the “seller” referred to in the CEP definition is simply one who contracts to sell and “sold” refers to the transfer of ownership or title. Since there can be no real ambiguity about these terms, contrary to the assertions of the appellees, we are not required to do any analysis under the second part of the Chevron test.
The record in this appeal is not disputed; it was the U.S. affiliates of the Korean producers that contracted for sale with the unaffiliated U.S. purchasers. The title or ownership passed from the U.S. affiliate to the unaffiliated U.S. purchaser. There were no contracts between the Korean producers and the unaffiliated U.S. purchasers. Thus, U.S. affiliates were the “sellers,” as indicated by the plain language of the statute. Commerce does not require a cumbersome test, examining the activities of the affiliate, to determine whether -or not the affiliate is a seller, when the answer to that question- is plain from the face of the contracts governing the sales in question. If Congress had intended the EP versus CEP distinction to be made based on which party negotiated the terms of the deal or on the relative importance of each party’s role, it would not have written the statute to distinguish between the two categories based on where the sale was made and which party made the sale.
B. Amendment to the Export Price Definition
The addition of the words “outside the United States” to the EP definition in the 1994 amendments similarly clarified what could be an EP sale. An earlier decision of the Court of International Trade held these words to be ambiguous, finding that it was unclear whether they described the location of the sale or the location of the prodiicer/exporter. See Mitsubishi Heavy Industries, Inc. v. United States, 15 F.Supp.2d 807, 812 (Ct. Int’l Trade 1998). The language of the CEP definition leaves no doubt that modifier “in the United States” relates to first sold. Wé hold that the term “outside the Uhited States,” read in the context of both the CEP and the EP definitions, as it must be, applies to the entire transaction at issue. Therefore, the 1994 amendment makes clearer that the critical difference between EP and CEP sales is whethér the sale or transaction takes place inside, or outside the United States.
A sales contract executed in the United States between two entities domiciled in the United States cannot generate a sale “outside the United States.” Thus, if “outside the United States” refers to the sale, as the government argues in this appeal, one of the parties to the sale or the execution of the contract must also be “outside the United States” for an EP classification tó be proper.7 Accordingly, qontrary to the holding of the Mitsubishi court, the phrase does not ambiguously refer to either the sale or the producer/exporter; those two things are not separable in this context. The addition of the phrase “outside the United States” merely clarified what was already evident from the statutory definition prior to the amendments: a transaction in which both parties are located in the United States and the contract is *1340executed in the United States cannot be an EP transaction.
In the Final Results, Commerce attempted to circumvent this geographic restriction on the use of EP sales by stating that when the PQ Test was satisfied it “considered] the exporter’s selling functions to have been relocated geographically from the country of exportation to the United States, where the [U.S. affiliate] performs them.” The trial judge’s holding that the PQ Test does not contradict the statute because it is a means of defining whether a sale is “in essence” between a producer/exporter and the unaffiliated buyer appears to make the same point. This “relocation” concept is contrary to the plain language of the statute.
The statute distinguishes CEP and EP sales based on whether the transaction occurs inside or outside the United States. In addition, PQ Corporation precludes “relocation” of selling activity by holding that the “statute provides no mechanism for imputing actual sales by an importer to that importer’s related ‘foreign manufacturer or producer of the merchandise’ so that [purchase price (now EP) ] will apply.” PQ Corp., 652 F.Supp. at 733. Thus, Commerce’s decision to redefine the activities occurring inside the United States as occurring outside the United States makes an impermissible end-run around both the plain meaning of the statutory language and the mandate of the Court of International Trade. Whether this redefinition does or does not perfectly comport with Congress’s intent to isolate an arm’s length transaction is not for this court to decide. Where Congress has made a clear distinction between the two categories based on the geographic location of the transaction, the agency may not circumvent this geographic distinction by “relocating” the activities of the U.S. affiliate.
Thus, the 1994 amendments made even clearer that an EP sale (or agreement for sale) must occur outside the United States. When, as here, there are contracts showing that the sales at issue took place in the United States between two entities with United States addresses, it is contrary to the plain meaning of the statute for Commerce to continue to use the PQ Test to define the sale as occurring outside of the United States.
A sale can be defined as either an EP or a CEP sale; a sale can not be in both categories, nor is there a middle ground. Congress only provided for two categories and those two categories are mutually exclusive. Thus, when the EP and CEP definitions are read together, it is clear that Congress distinguished the categories in terms of the relationship of the parties and the location of the sales. In distinguishing the two, Congress opted for what can be seen as a structural approach to defining EP and CEP sales, not the function-driven approach of the PQ Test. The words chosen by Congress are clear and unambiguous words such as “affiliated,” “sold,” and “in” or “outside” the United States.
Congress’s intent to fully define EP and CEP without any delegation to Commerce is further evident when the sections are viewed in the context of the rest of the anti-dumping statute. It is common in the anti-dumping statute for Congress to leave decisions about how to make dumping calculations to Commerce’s discretion because of its expertise. (See, for example the discretionary use of the so-called fair-value provision discussed infra.) Accordingly, if Congress had intended for Commerce to use its discretion to determine whether the use of CEP or EP was appropriate, it would have explicitly left that task to the agency as it did with other calculations in the statute.
When Congress makes such a clear statement as to how categories are to be defined, neither the agency nor the courts *1341are permitted to substitute their own definition for that of Congress, regardless of how close the substitute definition comes to achieving the same result as the statutory definition.8 The job of revising statutes is for Congress, not the agency or the courts.
Finally, the appellees argue that the PQ Test has been upheld repeatedly by the Court of International Trade. We are not swayed by that argument. First, this court is, of course, not bound by those decisions. Second, the Federal Circuit itself has never addressed the question of whether the PQ Test is consistent with the statute. Third, we are not persuaded that the decisions in those cases have any bearing on the question before this court. Cases decided based on the prior, statute, such as Outokumpu Copper Rolled Products v. United States, 829 F.Supp. 1371, 1378 (Ct. Int’l Trade 1993), E.I. DuPont de Nemours v. United States, 841 F.Supp. 1237, 1249 (Ct. Int’l Trade 1993),.and Independent Radionic Workers of America, 19 CIT 375 (1995), cannot enlighten our analysis of the new statutory language. In Mitsubishi Heavy Indus., 15 F.Supp.2d at 813-14, a post-amendment decision, the court affirmed a Commerce decision to classify sales as CEP sales after applying the PQ Test. Thus, while Mitsubishi did uphold the test, it did not uphold its use in a way that was contrary to the statutory language. If there was ever any doubt about how to define EP and CEP sales, Congress’s amendments to the statutory language make clear that it is the identities of the parties to the actual sale or agreement to sell, and the geographic location of those parties, that determine whether EP or CEP should be used to calculate the U.S. Price. , It is no longer necessary or permissible for Commerce to use the PQ Test to determine if .the sales activity has been “relocated” or if the sale was “in essence” between the producer and the unaffiliated purchaser because the U.S. affiliate acted only as a “communications link” between them. In light of the amended language Commerce must now base its decision on the identity and,geographic location of the parties to a sales contract. If the contract for sale was between a U.S. affiliate of a foreign producer and an unaffiliated U.S. purchaser, then the sale must be classified as a CEP sale. Stated in terms of the EP definition: if the sales contract is between two entities in the United States, and executed in the United States,it cannot be said to have been outside the United States; therefore the sale cannot be an EP sale. Similarly, a sale made by an affiliate or another party other than the producer or exporter cannot be an EP sale. Thus, we reverse the decision of the Court of International Trade and remand to that court for remand to the Department of Commerce for a redetermination of anti-dumping duties that is consistent with thi's holding.
III. Application of the Fair-Value and Major-Input Provisions
Commerce deemed three affiliated Kore^ an producers, POSCO, POCOS and PSI, a *1342single entity for its dumping analysis and levied a single anti-dumping duty on the entire group of related companies (a process referred to as collapsing). See Final Results, 62 Fed.Reg. at 18,430. Because it considered the separate companies as one entity, rather than separate though affiliated companies, Commerce did not “disregard” the transaction prices among the group members and apply “fair-value” prices instead when determining the constructed value.9 The so called “fair-value” provision is described in 19 U.S.C. § 1677b(f)(2):
A transaction directly or indirectly between affiliated persons may be disregarded if, in the case of any element of value required to be considered, the amount representing that element does not fairly reflect the amount usually reflected in sales of merchandise under consideration in the market under consideration. If a transaction is disregarded under the preceding sentence and no other transactions are available for consideration, the determination of the amount shall be based on the information available as to what the amount would have been if the transaction had occurred between persons who are not affiliated.
19 U.S.C. § 1677b(f)(2) (emphasis added).
Similarly, because the POSCO companies were “collapsed” into a single entity, Commerce did not apply the major-input provision of 19 U.S.C. § 1677b(f)(3):
If, in the case of a transaction between affiliated persons involving the production by one of such persons of a major input to the merchandise, the administering authority has reasonable grounds to believe or suspect that an amount represented as the value of such input is less than the cost of production of such input, then the administering authority may determine the value of the major input on the basis of information available regarding such cost of production, if such cost is greater than the amount that would be determined for such input under paragraph (2).
19 U.S.C. § 1677b(f)(3) (emphasis added).
While the domestic producers contested the collapsing of POSCO and POCOS below, they do not do so here. Thus, the only question before this court is whether, once collapsed, it was permissible for Commerce, in the exercise of its discretion, not to apply the provisions of 19 U.S.C. §§ 1677b(f)(2) & (3) to underlying transactions between those companies. In the earlier administrative review of this anti-dumping action and in other similar actions, Commerce applied the fair-value and major-input provisions despite collapsing several entities into one.10 In this second review, however, Commerce concluded that application of those provisions was unwarranted, if not unlawful, because “a decision to treat affiliated parties as a single entity necessitates that transactions among the parties also be valued based on the group as a whole.” Final Results, 62 Fed.Reg. at 18,430.
“Affiliated” persons are defined in the statute as those directly or indirectly owning five percent or more of the voting shares of an organization or two or more persons controlled by or controlling a common person. See 19 U.S.C. §§ 1677(33)(E) & (F) (1994). The domestic producers argue that collapsing for purposes of levying a single anti-dumping duty does not erase corporate form; thus POSCO and POCOS *1343are “affiliated persons” under the statute and Commerce should still have applied the fair-value and major-input provisions. The Appellees argue,' and the Court of International Trade upheld Commerce’s determination, that once they have been collapsed, POSCO, POCOS and PSI need no longer be treated as affiliated companies, but rather should be treated as one entity for all anti-dumping determination purposes. We may or may not agree that to do so is necessary or wise, but it cannot be fairly said to be an abuse of discretion.
When analyzing transfers between divisions ' of the same company Commerce need not and does not apply the fair-value or major-input provisions. See Certain Forged Steel Crankshafts from the United Kingdom, 61 Fed.Reg. 54,613, 54,614 (Oct. 21, 1996). Thus, it is not unreasonable for Commerce to decide not to apply those provisions to affiliates that are properly treated as one company for the balance of the anti-dumping analysis. Both provisions only apply to transactions “between ... persons”; once Commerce has decided to treat the companies as one “person” for. purposes of the anti-dumping analysis, it is not statutorily required to apply the provisions.
Indeed, the Domestic Producers’ argument that the statute requires the application of the provisions even to affiliates that were not treated as one entity is contrary to the plain language of the statute, which merely provides that Commerce “may” determine the values in a manner other than the use of the transfer price. Thus, the statute leaves possible application of the fair-value and major-input provisions to the discretion the agency, such that Commerce could decline to apply those provisions even if the POSCO producers were considered separate affiliates rather than one entity. We therefore affirm the decision of the Court of International Trade insofar as it upholds Commerce’s decision not to apply the fair-value and major-input provisions.
CONCLUSION
The judgment of the Court of International Trade is, accordingly
AFFIRMED-IN-PART, REVERSED-IN-PART and REMANDED.
COSTS
Each party to bear its own costs.
. Certain other sales were classified as CEP sales. No challenge was raised to those classifications in the Court of International Trade.
. The classification of the sales impacts the determination of the dumping margin, because the statute provides for certain deductions from CEP that are not deducted from EP. Specifically, commissions for selling, any expenses from the sale (such as credit expenses), the cost of further manufacture, and the profit allocated to those costs and expenses must be deducted from CEP sales. See 19 U.S.C. § 1677a(d).
. The administrative review at issue in this appeal was initiated after effective date of the 1994 amendments to the anti-dumping laws contained in the Uruguay Round Agreement Act ("URAA”). Thus, the statute as amended by the URAA applies to this case.
. We note that the statute appears to allow for a sale made by the foreign exporter or producer to be classified as a CEP sale, if such a sale is made "in the United States.” See 19 U.S.C. 1677a(a).
. The text of the statute provides that the CEP price should be reduced by the amount of any of the following expenses generally incurred by or for the account of the producer or exporter, or the affiliated seller in the United States, in selling the subject merchandise (or subject merchandise to which value has been added)—
(A) commissions for selling the subject merchandise in the United States;
(B) expenses that result from, and bear a direct relationship to, the sale such as credit expenses, guarantees and warranties;
(C) any selling expenses that the seller pays on behalf of the purchaser; and
(D) any selling expenses not deducted under subparagraph (A), (B), or (C).
19 U.S.C. § 1677a(c) (1994).
. While we can imagine a hypothetical sales contract between two U.S. domiciled entities that is entirely executed outside the United States, we make no determination regarding whether‘such á salé would be classified as an EP or CEP sale.
. Whether, in the absence of the two statutory definitions, the PQ Test does or does not allow Commerce to make accurate dumping determinations is not something this court need address. We do note, however, that in this case Commerce appears to have used the test to go far beyond not only the intent of Congress but even the language of Commerce’s own test. According to the Final Results and the lower court decision, warehousing, marketing, extension of credit to customers, and the processing of warranty claims were all considered "acting as a processor of sales related documentation” in accordance with part three of the PQ Test. Thus, under the current use of the PQ Test, significant expenses incurred in the U.S. are being included in the U.S. Price used in dumping calculations. This demonstrates a further flaw in the test; because it is poorly defined, it allows sales that are far beyond anything contemplated by Congress to be classified as EP sales.
. The constructed value is meant to create a proxy for the foreign market price by summing up the cost of inputs, processing and other relevant factors.
. That Commerce changed its interpretation, however, need not change the court’s analysis. See Chevron, 467 U.S. at 863, 104 S.Ct. 2778 (initial agency interpretation is not "carved in stone”).