CP Kelco Oy v. United States

978 F. Supp. 2d 1315, 2014 CIT 42, 36 I.T.R.D. (BNA) 212, 2014 Ct. Intl. Trade LEXIS 51, 2014 WL 1707213

• Court: United States Court of International Trade
• Decided: April 15, 2014
• Docket: Slip Op. 14-42; Court 13-00079
• Status: Published

Opinion

OPINION AND ORDER

GOLDBERG, Senior Judge:

This is a trade case brought under Section 201 of the Customs Courts Act of 1980, 28 U.S.C. § 1581(c) (2006). Plaintiffs CP Kelco Oy and CP Kelco US, Inc. (collectively “Kelco”) challenge the dumping margin the U.S. Department of Commerce (“Commerce” or “the agency”) assigned their goods during the 2010-2011 administrative review of an antidumping order on carboxymethylcellulose. Specifically, Kelco claims that the targeted dumping inquiry Commerce conducted before calculating the margin was neither in accordance with law nor based in substantial evidence.

The court holds that Commerce was permitted by law to conduct a targeted dumping inquiry during the contested review. The court also finds that Commerce’s method for discovering targeted dumping and the application of that methodology to Kelco generally accorded with law and the evidence. Nevertheless, the court concludes that an element of the agency’s targeted dumping analysis — the de minimis test — was neither grounded in substantial evidence nor in accordance with law. The court remands to Commerce to conduct the targeted dumping inquiry afresh and to recalculate Kelco’s dumping margins consistent with that inquiry.

PROCEDURAL BACKGROUND

In August 2011, Commerce initiated an administrative review of an antidumping order on carboxymethylcellulose from Finland. Initiation of Antidumping and Countervailing Duty Administrative Reviews, 76 Fed.Reg. 53,404, 53,405 (Dep’t Commerce Aug. 26, 2011). 1 The next year, the Aqualon Company, a petitioner, alleged Kelco had sold its goods for less-than-fair value at prices that differed “significantly among purchasers, regions, and periods of time.” Letter from Haynes & Boone LLC to Hon. John Bryson, PD 56 at bar-code 3077453-01 (May 25, 2012), ECF No. 30 (July 2, 2013) (“Pet’r’s Allegation”). This practice is known as “targeted dumping.” In view of its allegation, Aqualon asked Commerce to compute Kelco’s margins using a methodology that accounts for targeted dumping among an exporter’s sales. Id. at 1-2. 2

Commerce initially declined to conduct a targeted dumping inquiry when calculating Kelco’s margins. Instead, the agency followed standard procedure and assigned Kelco a 5.86% dumping margin in the preliminary results. See Purified Carboxymethylcellulose from Finland, 77 Fed.Reg. 47,036, 47,038, 47,042 (Dep’t Commerce Aug. 7, 2012) (“Preliminary Results ”). Later, however, Commerce inquired whether Kelco had engaged in targeted dumping and discovered targeted dumping by time period. Commerce then reassessed Kelco’s margins using an alternative methodology and assigned an 11.62% rate. Post-Prelim. Targeted Dumping Analysis Mem. at 3-4, PD 68 at bar-code 3112119-01 (Dec. 21, 2012), ECF No. 30 (July 2, 2013) (“Post-Preliminary Analysis”). The- agency confirmed its findings from the targeted dumping inquiry in the final results, settling on a 12.06% dumping margin. Purified Carboxymethylcellulose from Finland, 78 Eed.Reg. 11,817, 11,817 (Dep’t Commerce Feb. 20, 2013) (‘‘Final Results ”).

Kelco filed a summons with this court to challenge the margin. Summons, ECF No. 1. In the accompanying complaint, Kelco alleges Commerce’s targeted dumping inquiry was neither in accordance with the law nor grounded in substantial evidence. Compl. ¶¶ 19-27, ECF No. 4. Kelco implies that Commerce should use its normal methodology to calculate the dumping margin on remand. See id. at 7 (prayer for relief).

LEGAL BACKGROUND

To understand how Commerce’s targeted dumping inquiry shaped Kelco’s mar gins, some legal table-setting is needed. In general, Commerce calculates dumping margins by comparing a good’s “export price” to its “normal value.” See 19 U.S.C. § 1677(35). Commerce makes this comparison using one of three methods: the average-to-average methodology (“A-A”), the transaction-to-transaction methodology (“T-T”), or the average-to-transaetional methodology (“A-T”). Commerce’s preferred method in both investigations and administrative reviews is A-A. 19 C.F.R. § 351.414(b)-(c) (2013). Under this approach, the agency adopts the good’s weighted-average U.S. price as the export price. Id. § 351.414(d). It then subtracts the export price from the good’s weighted-average price in the exporter’s home market (i.e., the normal value), yielding a dumping margin. Id. Commerce used the A-A methodology to calculate Kelco’s margins in the Preliminary Results. Preliminary Results, 77 Fed.Reg. at 47,042 n. 39. 3

Commerce may also use the A-T methodology to set margins, but in limited circumstances. In investigations, Commerce may apply A-T only if it finds a “pattern of export prices ... for comparable merchandise that differ significantly among purchasers, regions, or periods of time,” and alternative methodologies inadequately explain the pattern. 19 U.S.C. § 1677f-l(d)(l)(B)(i)-(ii). If an exporter’s sales meet these criteria, that exporter engaged in targeted dumping. Commerce may then use A-T to compute the exporter’s dumping margins, comparing weighted-average normal values to export prices from individual sales. See 19 C.F.R. § 351.414(b)(3). Commerce does not offset non-dumped sales against dumped sales when using the A-T method. See Issues & Decisions Mem. at Issue 1, PD 80 at bar-code 3118300-01 (Feb. 5, 2013), ECF No. 30 (July 2, 2013) (“I & D Mem.”). As a consequence, margins calculated under A-T can be significantly higher than those computed under A-A.

Hence Commerce’s method for discovering targeted dumping bears critically on an exporter’s margins. The method, widely known as the “Nails test,” proceeds as follows. In the first step, called the “standard deviation test,” Commerce determines “the volume of the allegedly targeted group’s (i.e., purchaser, region, or time period) sales of subject merchandise that are at prices more than one standard deviation below the weighted-average price of all sales under review, targeted and non-targeted.” Id. at Issue 2. Standard deviations are calculated on a product-specific basis by control number (“CONNUM”). If more than thirty-three percent of allegedly targeted sales are at least one standard deviation below the average price of all reviewed sales in a given CONNUM, Commerce moves to step two. Id.

In step two, the “gap test,” the agency considers by CONNUM the sales that passed the standard deviation test. Commerce first calculates the difference between the weighted-average price of allegedly targeted sales and the next higher weighted-average price of sales to a non-targeted group (the “target gap”). Next, Commerce calculates the average difference, weighted by sales volume, between prices to non-targeted groups (the “non-target gap”). Finally, the agency compares the target gap to the non-target gap. 4 If the target gap exceeds the non- target gap for more than five percent of the exporter’s sales to the alleged target by volume, Commerce finds that targeted dumping occurred. The agency may then use A-T to calculate the exporter’s margins, but only if Commerce cannot account for observed price differences using A-A. Id. 5

Commerce takes a similar approach when it applies the A-T methodology in reviews. Id. Unlike the law governing investigations, however, the law governing reviews does not specify which comparative methodology Commerce must use to calculate margins. Instead, the statute explains only how to compute normal values when using the A-T method in reviews. 19 U.S.C. § 1677f-l(d)(2). To fill this apparent gap in the statute, federal regulations require Commerce to apply A-A in reviews unless another method is deemed more appropriate. 19 C.F.R. § 351.414(c)(1); see also Calculation of Weighted Average Dumping Margin and Assessment Rate, 77 Fed.Reg. 8101, 8102-04 (Dep’t Commerce Feb. 14, 2012) (“Final Modification ”). Commerce has used A-T instead of A-A in reviews if the Nails test reveals that an exporter engaged in targeted dumping. 6

In this case, Commerce applied the Nails test to Kelco’s sales during an administrative review. The agency concluded that some of Kelco’s sales constituted targeted dumping. Commerce also found — though obliquely — that Kelco’s targeted sales comprised more than a de minimis share of its total U.S. sales. I & D Mem. at Issue 2. After determining that the A-T methodology yielded higher margins than the A-A approach, Commerce recalculated Kelco’s margins using A-T. Commerce assigned Kelco a 12.06% margin in the Final Results, up from 5.86% in the Preliminary Results. See Public Mem. of Law in Supp. of Pis.’ 56.2 Mot. J. on Agency R. 4-5, ECF No. 28 (“Pis.’ Br.”).

STANDARD OF REVIEW

The court must review Commerce’s determinations to ensure they are supported “by substantial evidence on the record” and “in accordance with law.” 19 U.S.C. § 1516a(b)(l)(B)(i). An agency decision is based in substantial evidence if bolstered by “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Consol. Edison Co. v. NLRB, 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938); Nippon Steel Corp. v. United States, 337 F.3d 1373, 1379 (Fed.Cir.2003).

The agency’s interpretation of relevant statutes is “in accordance with law” if it passes the two-step test announced in Chevron, U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). Under Chevron, the court first determines whether a statute “directly [speaks] to the precise question at issue.” Id. at 842, 104 S.Ct. 2778. The court employs “the traditional tools of statutory construction” in this analysis, relying primarily on the statute’s plain meaning and secondarily on the “statute’s structure, canons of statutory construction, and legislative history.” Timex V.I., Inc. v. United States, 157 F.3d 879, 882 (Fed.Cir.1998) (internal quotation marks omitted). Then, if the court finds the statute’s meaning unclear, it scrutinizes the agency’s interpretation of the statute to determine whether it is permissible. The court defers to the agency if the interpretation is reasonable. See Chevron, 467 U.S. at 843-44, 104 S.Ct. 2778.

An agency action also fails to accord with law if it is arbitrary. See U.S. Steel Corp. v. United States, 37 CIT -, -, 953 F.Supp.2d 1332, 1336 (2013); Thai Plastic Bags Indus. Co. v. United States, 37 CIT -, -, 949 F.Supp.2d 1298, 1302 (2013). Under Motor Vehicle Manufacturers Ass’n v. State Farm Mutual Automobile Insurance Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983), an agency rule is arbitrary if “the agency ... relied on factors which Congress has not intended it to consider, entirely failed to consider an important aspect of the problem, ... or is so implausible that it could not be ascribed to ... the product of agency expertise.” See also Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 167-68, 83 S.Ct. 239, 9 L.Ed.2d 207 (1962) (invalidating exercise of agency discretion where agency failed to explain basis of its action).

DISCUSSION

The court evaluates Kelco’s claims in light of these standards. First, the court considers whether Commerce was authorized by statute to conduct a targeted dumping inquiry during the administrative review. See Pis.’ Br. 1-2. The law shows that Commerce was so authorized.

Second, the court assesses whether the method Commerce used to discover targeted dumping was based in substantial evidence and in accordance with law. See id. The court holds that Commerce’s targeted dumping inquiry was valid in all respects but one: Commerce’s de minimis test— which functioned either as an additional step of the Nails inquiry or as a guidepost in the agency’s discretionary analysis— was arbitrary and contrary to law.

I. Commerce Acted in Accordance with Law When It Conducted a Targeted Dumping Inquiry During the Review

Kelco first argues that Commerce was not permitted to conduct a targeted dumping inquiry during the administrative review. Though not so phrased in its brief, Kelco relies almost exclusively on the interpretive maxim expressio unius est ex-clusio alterius to support its claim. See Pis.’ Br. 7-12; Pis.’ Reply Br. 1-9, ECF No. 45 (“Reply Br.”). Translated from Latin, the maxim means “to express or include one thing implies the exclusion of the other,” suggesting that if Congress grants a right or privilege in one situation, then Congress intentionally withholds that right or privilege in other situations. Black’s Law Dictionary 620 (8th ed. 2004). In this vein, Kelco asserts that while the statute expressly permits Commerce to conduct targeted dumping inquiries in investigations, the law does not authorize such inquiries in administrative reviews. Consequently, Congress must have intended to prohibit targeted dumping inquiries in reviews, and Commerce acted contrary to law by applying its targeted dumping analysis to Kelco. Pis.’ Br. 9-12; see also 19 U.S.C. § 1677f — 1(d).

A. The Targeted Dumping Statute Is Ambiguous

The court cannot agree. Under Chevron, the court must invalidate agency actions that contradict a statute’s unambiguous instructions. See 467 U.S. at 842-43, 104 S.Ct. 2778. But the statute at issue here does not clearly prohibit targeted dumping inquiries in reviews. The court notes first that 19 U.S.C. § 1675— the section that orders Commerce set dumping margins in reviews — says nothing about how those margins should be calculated. Instead, the section only requires Commerce to determine “the normal value and export price ... of each entry of the subject merchandise, and ... the dumping margin for each such entry.” 19 U.S.C. § 1675(a)(2)(A). The provision governing Commerce’s margin calculation methodology in reviews also offers little direction: 19 U.S.C. § 1677f — 1(d)(2) tells Commerce how to average normal values when using the A-T methodology in reviews, but nothing more. In sum, the statute lacks language to inform the agency’s choice between the A-A, T-T, and AT methodologies in reviews. Given this spare guidance, Commerce was free to use the targeted dumping inquiry to help it choose between A-A and A-T to calculate Keleo’s dumping margin. See Def.’s Opp. to Pis.’ M. for J. on Agency R. 8-11, ECF No. 39 (“Resp. Br.”); U.S. Steel Grp. v. United States, 96 F.3d 1352, 1362 (Fed. Cir.1996) (“So long as the [agency’s] analysis does not violate any statute and is not otherwise arbitrary and capricious, the [agency] may perform its duties in the way it believes most suitable.”); Mid Continent Nail Corp. v. United States, 34 CIT -, -, 712 F.Supp.2d 1370, 1376-77 (2010).

Even so, Kelco contends the interpretive maxim expressio unius precludes using the targeted dumping inquiry in reviews. 19 U.S.C. § 1677f-l(d)(l) explicitly permits the targeted dumping inquiry in investigations. Section 1677f — 1(d)(2), by contrast, does not mention the inquiry in the context of reviews. Kelco points to this disparity as proof that “Congress expressly withheld from [Commerce] the authority to use the targeted dumping exception in administrative reviews.... When statutory language contains no ambiguity, [Commerce] cannot create authority that has not been explicitly or implicitly granted.”' Pis.’ Br. 9.

The court does not see it that way. As the Supreme Court explained, expressio unius arguments “ha[ve] force only when the items expressed are members of an associated group or series, justifying the inference that items not mentioned were excluded by deliberate choice, not inadvertence.” Barnhart v. Peabody Coal Co., 537 U.S. 149, 168, 123 S.Ct. 748, 154 L.Ed.2d 653 (2003) (internal quotation marks omitted). Admittedly, the provisions here appear in a series. Section 1677f-l(d)(l)(B) permits targeted dumping inquiries in investigations, but § 1677f-1(d)(2) provides no such authority for reviews. Nevertheless, § 1677f-l(d)(2) mandates how Commerce must average normal values when using the A-T method to calculate margins in reviews. This implies that the legislature intended to allow A-T to be used in reviews. It thus makes little sense that Congress would prohibit targeted dumping inquiries in reviews, because the inquiry’s sole purpose is to help Commerce decide whether to apply the A-T methodology in a given case. The inference to be drawn, if any, is that Congress would allow the targeted dumping inquiry in reviews, not the opposite. See Barnhart, 537 U.S. at 168, 123 S.Ct. 748.

FAG Italia S.p.A. v. United States, 291 F.3d 806 (Fed.Cir.2002), does not mandate a different result. See Pis.’ Br. 10-12. There, the statute in question authorized Commerce to conduct duty absorption inquiries in administrative reviews, but only during the second and fourth years after an antidumping order first issued. 7 Commerce, however, attempted to conduct duty absorption inquiries in the second and fourth years following a transition order. FAG Italia, 291 F.3d at 811. 8 On appeal, the government argued the statute was silent regarding whether Commerce could conduct absorption analyses in reviews following a transition order. This silence, the government explained, served as tacit permission to carry out absorption inquiries following transition orders. Id. at 815-16. The Federal Circuit disagreed and held that the statutory silence did not authorize Commerce to conduct the inquiries following transition orders. “The fact that Commerce is empowered to take action in certain limited situations does not mean that Commerce enjoys such power in other instances.” Id. at 817.

In this vein, Kelco argues Commerce misinterpreted statutory silence in § 1677(d)(2) to permit targeted dumping inquiries in reviews. The comparison to FAG Italia, however, is inapt. In FAG Italia, the same provision that authorized duty absorption inquiries also limited those inquiries to the second and fourth years following an order. See 19 U.S.C. § 1675(a)(4). Congress endowed Commerce with an investigative power and cabined it in the same penstroke. Here, by contrast, the provisions authorizing Commerce to calculate dumping margins in investigations and reviews are separate from provisions describing how to perform those calculations. 19 U.S.C. § 1673 empowers Commerce to calculate margins in investigations, and § 1677f-l(d)(l) channels Commerce’s exercise of that power. 19 U.S.C. § 1675(a) charges Commerce to compute margins in reviews, and § 1677f-1(d)(2) provides limited mathematical guidance regarding those computations. Paragraphs 1677f — 1(d)(1) and (2), in short, were designed to be read in light of their parent provisions and not as a unit. Given this statutory scheme, one cannot easily infer that language authorizing targeted dumping inquiries in investigations bars the agency from conducting similar inquiries in reviews. See NTN Bearing Corp. v. United States, 368 F.3d 1369, 1373 (Fed. Cir.2004) (holding expressio wains does not apply when “its application would thwart the legislative intent made apparent by the entire act”) (internal quotation marks omitted).

Kelco’s case is more akin to NTN Bearing than to FAG Italia. In NTN Bearing, Commerce used cost data from the respondents’ affiliates to calculate respondents’ inventory carrying costs and difference in merchandise (“difmer”) adjustment. Id. at 1371-72. The statute, however, expressly authorized Commerce to use affiliate data for other purposes, including to calculate a respondent’s production costs and constructed normal value. 19 U.S.C. § 1677b(f). The Federal Circuit nevertheless approved Commerce’s use of the affiliate data. The court found that statutory language governing carrying costs and dif mer adjustments did not prohibit Commerce from using affiliate data to compute those values. 368 F.3d at 1373. And although the statute expressly referenced affiliate data only in the context of production costs and constructed normal value, the law did not preclude using affiliate data in other contexts. Id. Likewise, 19 U.S.C. § 1677f — 1 (d)(1)(B) — the provision authorizing targeted dumping inquiries in investigations — does not bar Commerce from using targeted dumping inquiries in reviews. Indeed, the statute did not preclude, whether expressly or implicitly, the targeted dumping analysis in the review below.

B. Commerce’s Decision to Conduct a Targeted Dumping Inquiry in the Review Was Reasonable

Nor was the agency’s decision to conduct a targeted dumping inquiry an unreasonable interpretation of the statute. See Chevron, 467 U.S. at 843-44, 104 S.Ct. 2778. As discussed above, the statute permits Commerce to use A-T in reviews. See 19 U.S.C. § 1677f-l(d)(2). The regulations regarding reviews also allow Commerce, when appropriate, to apply A-T instead of A-A. See 19 C.F.R. § 351.414(c)(1). Because neither the statute nor the regulations dictate when using A-T would be “appropriate” in reviews, it was reasonable for Commerce to use the targeted dumping inquiry as a principled way of choosing between A-A and A-T to calculate Kelco’s margins.

Kelco nevertheless argues that the agency’s interpretation was invalid because antidumping investigations and reviews serve different purposes. See Reply Br. 7-8 (citing Union Steel v. United States, 713 F.3d 1101, 1109 (Fed.Cir.2013)). In investigations, Commerce examines overall pricing patterns to decide whether to impose anti-dumping duties. Reviews, by contrast, exist to determine the amount of those duties. Because the targeted dumping inquiry focuses on “overall pricing patterns,” Kelco argues the inquiries are appropriate only in investigations. Id. at 8.

Kelco’s own citation refutes this argument. In Union Steel — a case Kelco offers to illustrate the difference between investigations and reviews — the Federal Circuit upheld Commerce’s decision to use the A-A methodology in investigations and the A-T methodology without offsets in reviews. 713 F.3d at 1108-09. But as discussed previously, the targeted dumping inquiry is simply a threshold analysis Commerce conducts before applying A-T to calculate dumping margins. It does not follow that Congress would prohibit targeted dumping inquiries in reviews when the inquiry’s purpose is to help Commerce decide whether to apply A-T in a given case. The policy differences between investigations and reviews do not render the agency’s interpretation of the statute unreasonable.

Because Commerce’s interpretation passes both steps of Chevron, the court holds that the decision to conduct a targeted dumping inquiry during the review accorded with law.

II. Commerce’s Method for Discovering Targeted Dumping Was Not in Accordance with Law

Kelco next claims that Commerce’s method for discovering targeted dumping was neither grounded in substantial evidence nor in accordance with law. This argument breaks into three subparts. First, Kelco argues the Nails test is itself arbitrary and unsupported by substantial evidence. Second, Kelco alleges Commerce did not base in substantial evidence its decision to apply the Nails test to Kelco. Third, Kelco claims Commerce acted arbitrarily by refusing to excuse Kelco’s targeted sales from A-T treatment under the de minimis test. Of these arguments, only the last persuades.

A. The Nails Test Itself Is Neither Arbitrary Nor Unsupported in Evidence

Kelco argues the Nails test as applied in administrative reviews is arbitrary and unsubstantiated in evidence. In particular, Kelco alleges Commerce failed to explain how the Nails test’s standard deviation metric and thirty-three and five percent thresholds unmask targeted dumping. Pis.’ Br. 13-14.

These arguments do not advance Kelco’s case. Under State Farm, 463 U.S. at 43, 103 S.Ct. 2856, an agency rule is arbitrary if “the agency ... relied on factors which Congress has not intended it to consider, entirely failed to consider an important aspect of the problem, ... or is so implausible that it could not be ascribed to ... the product of agency expertise.” The Nails test avoids these pitfalls because it identifies targeted dumping as described by statute. Under § 1677f-l(d)(l)(B)(i), targeted dumping exists if “there is a pattern of export prices ... for comparable merchandise that differ significantly among purchasers, regions, or periods of time.” As explained in Mid Continent Nail in the context of investigations, the Nails test’s standard deviation analysis pinpoints “pattem[s] of export prices” by measuring “the dispersion of values in an exporter’s price data, to aid in identifying which of the exporter’s sales were relatively low compared to others.” 34 CIT at -, 712 F.Supp.2d at 1377. The court has upheld Commerce’s standard deviation test as a statistically valid means of determining price dispersion. See id. at -, 712 F.Supp.2d at 1377-78. Kelco’s arguments give no occasion to abandon this holding in the context of reviews.

Commerce’s thirty-three and five percent thresholds are also valid. If thirty-three percent of allegedly targeted prices fall one standard deviation below average prices by CONNUM, then the agency concludes that a “pattern of export prices” existed. See I & D Mem. at Issue 2; Mid Continent Nail, 34 CIT at -, 712 F.Supp.2d at 1378. Next, Commerce takes the sales that pass the standard deviation test and determines whether they pass the gap test. I & D Mem. at Issue 2. If sales that pass the gap test comprise five percent of the exporter’s sales to the alleged target, Commerce concludes that prices “differed] significantly” among purchasers, regions, or periods of time. Mid Continent Nail, 34 CIT at -, 712 F.Supp.2d at 1378-79. The court has held that these thirty-three and five percent thresholds together identify targeted dumping as defined in 19 U.S.C. § 1677f — l(d)(l)(B)(i). Id. Kelco furnished no arguments or record evidence demonstrating the contrary. 9 The court thus holds that the Nails test as applied in reviews is based in substantial evidence and not arbitrary.

B. Commerce’s Application of the Nails Test to Kelco Was Neither Unsupported in Evidence nor Contrary to Law

Next, Kelco argues that Commerce’s choice to deploy the Nails test in the review below was contrary to law and unsupported in evidence. It notes that Commerce returned to “case-by-case adjudication” to find targeted dumping after the agency withdrew a targeted dumping regulation in 2008. Pis.’ Br. 14; see Withdrawal of Regulatory Provisions Governing Targeted Dumping in Antidumping Duty Investigations, 73 Fed.Reg. 74,930, 74,931 (Dep’t Commerce Dec. 10, 2008) (interim final rule) (the “Withdrawal”). 10 Consequently, Commerce needed to conduct “a specific analysis as to the facts of the present case in order to determine whether the Nails test was the most appropriate way to unmask any alleged ... targeted dumping.” Pis.’ Br. 14. Kelco says Commerce never undertook this “specific analysis.”

These arguments fail to persuade. Under the substantial evidence standard, the court must uphold Commerce’s choice if it was based on “relevant evidence [that] a reasonable mind might accept as adequate to support a conclusion.” Consol. Edison, 305 U.S. at 229, 59 S.Ct. 206. Commerce’s decision to apply the Nails test to Kelco clears this hurdle. In the Preliminary Results, Commerce noted that petitioner Aqualon based its allegation of targeted dumping on record evidence. See Preliminary Results, 77 Fed.Reg. at 47,-037-38; Pet’r’s Allegation 3-7. 11 Even so, Commerce initially refused to administer the targeted dumping test to afford the “parties an opportunity to meaningfully comment on the Department’s implementation of this recently adopted methodology in the context of this administrative review.” Preliminary Results, 77 Fed.Reg. at 47,038.

After receiving comments, Commerce decided to run the targeted dumping analysis as Aqualon had petitioned. In the Post-Preliminary Analysis, the agency acknowledged its authority to use A-T under 19 C.F.R. § 351.414(c)(1) and explained how it applied the Nails test to Kelco’s sales. Posh-Preliminary Analysis at 2-3. The Post-Preliminary Analysis further concluded that Kelco engaged in targeted sales by time period. See id.

Commerce again explained how it applied the Nails test in the I & D Memo, adding that the Court of International Trade had upheld the Nails test as reasonable. I & D Mem. at Issue 2. The agency also noted that it used the Nails test in reviews of other antidumping orders. Id.; see also Ball Bearings and Parts Thereof from France, Germany, and Italy, 77 Fed.Reg. 73,415 (Dep’t Commerce Dec. 10, 2012) and accompanying I & D Mem. at cmt. 1 {“Ball Bearings ”); Tapered Roller Bearings and Parts Thereof, Finished and Unfinished, from the People’s Republic of China, 78 Fed.Reg. 3396 (Dep’t Commerce Jan. 16, 2013) and accompanying I & D Mem. at cmt. 1. The court finds that this evidence — garnered from the review below and other reviews where the Nails test was used — sufficed to support Commerce’s decision to apply the Nails test to Kelco.

Kelco, by contrast, does not identify any specific conclusions Commerce made that were unsupported by evidence or logic. In its brief, Kelco faults Commerce for failing “to state what facts it looked at and how those facts supported” the agency’s choice to employ the Nails test during the review. Pis.’ Br. 15. Yet Kelco does not suggest what more Commerce could have done to shore up its decision. It does not, for example, offer any factors independent of the Nails inquiry that would corroborate whether Kelco had perpetrated targeted dumping. Nor did Kelco propose anything better at oral argument, where it made a vague pitch for Commerce to issue supplemental questionnaires before undertaking a Nails inquiry. Oral Argument at 17:43. Having considered these arguments and the record, the court finds Commerce’s decision to use the Nails test below was based in substantial evidence and in accordance with law.

C. Commerce’s Use of the De Minim-is Test Was Arbitrary

Finally, Kelco argues Commerce acted arbitrarily by applying an ill-defined de minimis test to Kelco’s targeted sales. In reviews of other antidumping orders, Commerce declined to apply the A-T methodology where targeted sales constituted a small fraction — or a de minimis portion — of an exporter’s total U.S. sales. See Pis.’ Br. 15-16. Here, Commerce held Kelco’s targeted sales were more than de minimis but did not explain what “de minimis sales” means. The agency consequently applied the A-T methodology to compute Kelco’s margins. Kelco says Commerce should have offered some definition of de minimis sales before finding that its sales exceeded the de minimis threshold. See id. at 17-18.

The court agrees with Kelco. 12 As discussed above, agency decisions are arbitrary if they cannot “be ascribed to ... the product of agency expertise.” State Farm, 463 U.S. at 43, 103 S.Ct. 2856. Administrative decisions are similarly invalid if they fail to state “the basis on which the [agency] exercised its expert discretion.” Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 167, 83 S.Ct. 239, 9 L.Ed.2d 207 (1962); see also Nat’l Org. of Veterans’ Advocates, Inc. v. Sec’y of Veterans Affairs, 314 F.3d 1373, 1380-81 (Fed.Cir.2003) (remanding where agency did not articulate rationale for stat utory interpretation). Commerce’s de minimis test founders under either of these standards.

First, Commerce never explained what purpose the de minimis test serves in the statutory scheme. Under 19 U.S.C. § 1677f-l(d)(l)(B)(i), Commerce may apply A-T in investigations if an exporter’s sales constitute “a pattern of export prices” differing significantly among purchasers, regions, or time periods. In this vein, the de minimis test could serve as part of the Nails inquiry in investigations and reviews, signaling whether targeted sales are voluminous enough to form a “pattern” of significantly differing prices as described in statute. Post-Preliminary Analysis at 3 (suggesting pattern existed “based on the percentage of U.S. sales found to have been targeted”); Oral Argument at 25:41 (arguing for Kelco that de minimis test is third prong of Nails analysis). Alternatively, the de minimis test could guide the agency’s discretion when deciding whether to apply A-T to an exporter’s targeted sales. The statute does not compel Commerce to apply A-T to exporters who made targeted sales. Instead, it states that Commerce “may determine” to use A-T when deciding whether such exporters made sales at less-than-fair value. 19 U.S.C. § 1677f-l(d)(l)(B); see also Oral Argument at 27:27 (arguing for government that de minimis test guides agency’s discretion to apply A-A or A-T). The agency could use the de minimis test to serve this discretionary function, to identify a pattern of prices, or both. Yet nothing in the record establishes which role the test played in the review below.

Second, Commerce never explained the quantum of an exporter’s sales that must be targeted to fall above or below the de minimis threshold. In the I & D Memo, Commerce said only that “the percentage of sales by quantity which was found to be targeted in this case is far too high to be considered de minimis, and so CP Kelco’s argument [regarding the de minimis threshold] is not relevant in the context of this case.” I & D Mem. at Issue 2. 13 Commerce’s Post-Preliminary Analysis also mentioned that Commerce considered applying the A-T methodology only after finding “sufficient sales ... passed the Nails test.” Post-Preliminary Analysis at 3. But the Analysis furnished neither a qualitative nor a quantitative explanation of what a “sufficient” number of sales is.

Nor do administrative reviews under other antidumping orders define de minimis sales. See Pis.’ Br. 16. Commerce applied the de minimis analysis in a number of proceedings other than the review contested here. See Certain Frozen Warmwater Shrimp From India, 78 Fed.Reg. 42,492 (Dep’t Commerce July 16, 2013) and accompanying I & D Mem. at cmt. 1 (applying A-T to exporter with “sufficient volume of targeted sales” but A-A to exporter with “insufficient volume of targeted sales”); Ball Bearings, 77 Fed.Reg. 73,415 (Dep’t Commerce Dec. 10, 2012) (final admin, reviews) and accompanying I & D Mem. at cmt. 1; Circular Welded Carbon Steel Pipes and Tubes from Turkey, 77 Fed.Reg. 72,818 (Dep’t Commerce Dec. 6, 2012) and accompanying I & D Mem. at cmt. 1. None of these proceedings, however, furnished a useful definition of de minimis sales.

In an effort to reduce this ambiguity, the government tries to sketch the contours of the de minimis threshold on appeal. In its brief, the government cites a de minimis provision from a dumping margin regulation to show that Kelco’s sales were more than de minimis. Resp. Br. 18-19; see also 19 C.F.R. § 351.106(c) (treating dumping margins under 0.5% as de minimis). As a general rule, however, agencies cannot rely on post hoc rationalizations to justify actions taken during administrative proceedings: “[A]n administrative order cannot be upheld unless the grounds upon which the agency acted in exercising its powers were those upon which its action can be sustained.” SEC v. Chenery Corp., 318 U.S. 80, 95, 63 S.Ct. 454, 87 L.Ed. 626 (1943). Here, although the government cites a potentially relevant de minimis provision in its brief, Commerce did not do so anywhere in the record. The court is thus left without a basis to evaluate the de minimis test’s substance and place in the statutory scheme. See Burlington, 371 U.S. at 167-68, 83 S.Ct. 239.

And so the court must return the undercooked de minimis dish to the administrative kitchen. On remand, Commerce must define the de minimis test’s function (i.e., does the test identify a “pattern” of differing prices, does it guide the agency’s discretion to apply A-T, or both?). Commerce must then outline the quantitative data, qualitative variables, or other information it considers when determining whether an exporter’s targeted sales fall above or below the de minimis threshold. Finally, Commerce must use this definition to find whether Kelco’s targeted sales pass or fail the de minimis test. Conclusory explanations of the variety found in the I & D Memo below will not be accepted. 14

CONCLUSION AND ORDER

The law allowed Commerce to scrutinize Kelco’s sales for targeted dumping during the review. The law also supported the agency’s choice to use the Nails test to discover targeted sales. But Commerce could not deny Kelco a de minimis exception to its A-T margin calculation methodology without saying what de minimis means. The agency must correct this error on remand.

Upon consideration of all papers and proceedings herein, it is hereby:

ORDERED that the final determination of the International Trade Administration, United States Department of Commerce (“Commerce”), published as Purified Carboxymethylcellulose from Finland, 78 Fed.Reg. 11,817 (Dep’t Commerce Feb. 20, 2013) (final results), be, and hereby is, REMANDED to Commerce for redetermination; it is further

ORDERED that Plaintiffs Rule 56.2 Motion for Judgment on the Agency Record be, and hereby is, GRANTED as provided in this Opinion and Order; it is further

ORDERED that Commerce must issue a redetermination (“Remand Redetermination”) in accordance with this Opinion and Order that is in all respects supported by substantial evidence, in accordance with law, and supported by adequate reasoning; it is further

ORDERED that Commerce must fully explain the purpose of the de minimis test and provide a reasoned definition of the quantum of total sales of subject merchandise that must be targeted for Kelco to fall above or below the de minimis threshold discussed in this Opinion and Order; it is further

ORDERED that Commerce must apply the de minimis test as defined in the Remand Redetermination to Kelco’s targeted sales and recalculate Kelco’s dumping margins in accordance with the results of that test; it is further

ORDERED that Commerce shall have ninety (90) days from the date of this Opinion and Order in which to file its Remand Redetermination, which shall comply with all directives in this Opinion and Order; that the Plaintiff and Defendanfr-Intervenor shall have thirty (30) days from the filing of the Remand Redetermination in which to file comments thereon; and that the Defendant shall have thirty (30) days from the filing of Plaintiff and Defendant-Intervenor’s comments to file comments.

1 . Carboxymethylcellulose is "an acid ether derivative of cellulose that in the form of its sodium salt is used as a thickening, emulsifying, and stabilizing agent and as a bulk laxative in medicine." Merriam-Webster’s Collegiate Dictionary 172 (10th ed. 1993).

2 . Defendant-Intervenor Ashland Specialty Ingredients, G.P. ("Ashland”) was previously known as Aqualon Company. See Order Granting Consent Mot. to Amend Caption, ECF No. 36 (Sept. 10, 2013).

3 . Commerce rarely uses T-T to compute dumping margins. See Calculation of Weighted Average Dumping Margin and Assessment Rate, 77 Fed.Reg. 8101, 8102 (Dep’t Commerce Feb. 14, 2012) (final modification) (discussing investigations).

4 . Commerce does not use the terms "target gap” and “non-target gap” in its analysis. These terms were coined as shorthand for ease of explanation.

5 . There is rulemaking underway concerning whether Commerce may apply A-T to all sales, both targeted and untargeted, when Commerce finds targeted sales in an investigation. See Gold East Paper (Jiangsu) Co. v. United States, 37 CIT -, -, 918 F.Supp.2d 1317, 1325-28 (2013); Non-Application of Previously Withdrawn Regulatory Provisions Governing Targeted Dumping in Antidumping Duty Investigations, 78 Fed.Reg. 60,240 (Dep't Commerce Oct. 1, 2013). The court is unaware of any similar rulemaking for reviews.

6 . Before publishing the Pinal Modification in February 2012, Commerce used A-T without offsets as its default comparative methodology in reviews. 77 Fed.Reg. at 8101. For a history of Commerce’s evolving method for calculating margins in administrative reviews, see Timken Co. v. United States, Slip Op. 14-24, 2014 WL 763124, at *1-2 (CIT Feb. 27, 2014).

7 . "Duty absorption” occurs when an exporter pays the cost of antidumping duties without passing those costs to U.S. consumers. FAG Italia, 291 F.3d at 809. If Commerce finds an exporter has absorbed duties, it may pass those findings to the International Trade Commission for use in assessing material injury. 19 U.S.C. § 1675(a)(4).

8 . A transition order is an "antidumping duty order ... which is in effect on the date the WTO Agreement enters into force with respect to the United States.” FAG Italia, 291 F.3d at 811. Transition orders were deemed issued on January 1, 1995, for the purposes of subsequent sunset reviews. Id.; 19 U.S.C. § 1675(c)(6)(D).

9 . At oral argument, Kelco called the Nails test arbitrary in reviews because Commerce applies A-A using month-to-month comparisons in reviews and twelve-month comparisons in investigations. Oral Argument at 16:18, 38:35. This distinction does not render the Nails test arbitrary in reviews, though. First, Kelco never made this argument in its brief. See Pis.’ Br. 13-14. Second, even if month-to-month A-A comparisons reveal targeted dumping better than annual A-A comparisons, this does not bar Commerce from using the targeted dumping inquiry and the A-T methodology in reviews. See 19 C.F.R. § 351.414(c). As recognized in Union Steel, 713 F.3d at 1109, Commerce does not average export prices at all under A-T, yielding transaction-specific margins that are likely more accurate even than month-to-month A-A comparisons. Third, and most important of all, A-T is explicitly permitted in reviews under the statute. See 19 U.S.C. § 1677f-1(d)(2).

10 . The court invalidated the Withdrawal on procedural grounds in Gold East, 37 CIT at -, 918 F.Supp.2d at 1325.

11 . At oral argument, Kelco claimed Aqualon's targeted dumping allegation was inadequate to trigger Commerce’s Nails inquiry. Oral Argument at 16:58. The court disagrees. As an initial matter, Kelco never made this argument in its brief. See Pis.’ Br. 14-15. Furthermore, neither the statute nor the regulations specify when Commerce must investigate for targeted dumping or what form a petitioner’s targeted dumping allegations must take. See 19 U.S.C. § 1677f-1(d)(1)(B); 19 C.F.R. § 351.414(c)(1). In this case, Aqualon based its targeted dumping allegation on its preliminary application of the Nails test to Kelco's sales. See Pet'r's Allegation 3-7; Comments of Pet’r Aqualon Company on Post-Prelim. Targeted Dumping Analysis Mem. at 1-2, PD 71 at bar-code 3112811— 01 (Jan. 2, 2013), ECF No. 30 (July 2, 2013). Though it seems circular that Aqualon based its claim on the results of the very test Commerce deploys to find targeted dumping, it was not unreasonable for Commerce to launch its own, independent Nails inquiry in response to the allegation, however derived.

12 . In its brief, Defendant-Intervenor Ashland appeared to challenge whether Commerce is permitted to conduct de minimis inquiries in reviews at all. The de minimis test, Ashland alleged, "would undermine Commerce’s targeted dumping analysis by eliminating its ability to take action at the very first sign of the targeted dumping behavior.” Def.-Intervenor’s Resp. Br. 14, ECF No. 37. Thus "a de minimis principle should not and cannot be applied to targeted dumping.” Id. At oral argument, however, Ashland stated that it did not contest Commerce’s decision to conduct a de minimis analysis during the review. Oral Argument at 24:12. The court thus declines to consider whether the statute permitted the agency to conduct a de minimis inquiry.

13 . With respect to alleged targeted dumping by customer, Commerce found [[Confidential Data Deleted ]]% of sales by quantity and [[Confidential Data Deleted ]]% of sales by value were targeted. Respecting targeted dumping by time period, (footnote continued) Commerce found [[Confidential Data Deleted ]]% of sales by quantity and [[Confidential Data Deleted]]% of sales by value were targeted. Confidential Mem. of Law in Supp. of Pis.' 56.2 Mot. J. on Agency R. 17, ECF No. 27 (“Pis.’ Confid. Br.”).

14 . The court’s decision does not conflict with another recent targeted dumping case, Timken Co. v. United States, 2014 WL 763124. There, Commerce found that respondents' targeted sales were de minimis and hence insufficient to justify using the A-T methodology to calculate dumping margins. Petitioner Timken alleged Commerce erred by failing to define what de minimis targeted sales were. Id. at *8.

Rejecting Timken's claim, Judge Restani gave three reasons why Commerce did not need to define de minimis sales: (1) Commerce was never presented with and did not consider petitions to specify and justify a de minimis threshold; (2) Timken never argued that the targeted sales found were more than de minimis; and (3) respondents’ targeted sales were small. Id. at *9. Judge Restani also noted the government’s argument that “Commerce is not obligated to justify relying on” the default A-A methodology to calculate margins in reviews. Id. at *8.

Yet unlike the petitioner in Timken, Kelco clearly asked Commerce to find its sales were "minimal and insufficient to meet the [de minimis ] standard.” I & D Mem. at Issue 2. To make this finding, the agency needed to have some principled definition of what de minimis sales are. See Pis.' Br. 18. The record, however, sported no such explanation. Furthermore, although Kelco’s targeted sales by time period were appreciable, they were not obviously more than de minimis. See .Pis.’ Confid. Br. 17. Finally, Commerce applied A-T to calculate Kelco’s margins, departing from the methodology (A-A) the agency normally applies in reviews. See 19 C.F.R. § 351.414(c)(1) (mandating A-A in reviews unless "another method is appropriate in a particular case”).

In short, because Commerce used an exceptional methodology to generate Kelco's margins, because Kelco’s sales were not clearly more or less than de minimis, and because Kelco specifically asked the agency to find its targeted sales were de minimis, Kelco’s case differs from Timken’s. Commerce must provide a principled definition of de minimis targeted sales on remand.