LOURIE, Circuit Judge.
The United States, as Defendanh-Appel-lant, appeals from the final judgment of the United States Court of International Trade (the “trade court”), which rejected calculations advanced by the Department of Commerce (“Commerce”) regarding,
inter alia,
the profit cap applicable under 19 U.S.C. § 1677b(e)(2)(B)(iii) to merchandise sold by Italian exporter Atar S.r.l. (“Atar”) in the ninth administrative review of an antidumping duty order directed to certain Italian pasta products. In response, Commerce revised its profit cap determination, eventually including above- and below-cost sales made by profitable and unprofitable respondents in the prior administrative review to satisfy the trade court’s remand orders. The trade court thereafter sustained Commerce’s antidumping duty calculations.
Atar S.r.L. v. United States,
853 F.Supp.2d 1344 (Ct.Int’1 Trade 2012)
(“Atar TV”).
Because we conclude that Commerce’s original profit cap calculation was reasonable, we reverse.
Background
This appeal arrives with a lengthy and complex history. On June 14, 1996, Commerce determined that certain pasta products from Italy were being sold in the United States at less than fair value.
Certain Pasta from Italy,
61 Fed.Reg. 30,-326 (Dep’t Commerce June 14, 1996) (notice of final determination of sales at less than fair value). Shortly thereafter, Commerce published an antidumping duty order imposing antidumping duties against subject merchandise imported into the United States.
Certain Pasta from Italy,
61 Fed.Reg. 38,547 (Dep’t of Commerce July 24, 1996) (notice of antidumping duty order).
Some years later, Commerce conducted its ninth administrative review of that anti-dumping duty order, covering the period of July 1, 2004, through June 30, 2005. In pertinent part, Commerce arrived at an antidumping duty margin of 18.18% for Atar.
Certain Pasta from Italy,
72 Fed. Reg. 7,011, 7,012 (Dep’t of Commerce Feb. 14, 2007) (notice of final results)
(“Final Results
”). To calculate antidumping margins, Commerce ordinarily compares the export price of the subject merchandise with the “normal value,”
i.e.,
the price of like products sold in the exporter’s home market or in a representative third country.
See
19 U.S.C. §§ 1677(35), 1677b(a)(l)(A)-(C). In this case, however, Commerce determined that it could not assess normal value by reference to Atar’s proffered home-market or third-country sales data, so it approximated the normal value of Atar’s subject goods using a constructed value approach.
Atar S.r.l. v. United, States,
637 F.Supp.2d 1068, 1072-73 (Ct.Int’l Trade 2009)
(“Atar
/”); see
also
19 U.S.C. § 1677b(a)(4).
As defined by statute, the constructed value for merchandise under antidumping review ordinarily equals the sum of (1) the cost of materials and fabrication needed to produce the merchandise in the ordinary course of trade; (2) the exporter’s actual selling, general, and administrative costs incurred and actual profits realized in the production of a foreign like product in the ordinary course of trade; and (3) packing and container costs. 19 U.S.C. § 1677b(e)(l), (e)(2)(A), (e)(3). To be considered as within the “ordinary course of trade,” the sales under review generally must arise from arm’s-length, above-cost transactions.
Id.
§§ 1677(15), 1677b(b)(l), (f)(2). When the exporter under consideration lacks viable comparison market data as specified under § 1677b(e)(2)(A)—in-cluding actual profits and actual sales, general, and administrative (“SGA”) costs— the statute provides three options for deriving substitute values. Option (i) uses the actual amounts incurred and realized by the specific exporter under review from foreign sales of merchandise that falls within the same general category as the subject merchandise.
Id.
§ 1677b(e)(2)(B)(i). Option (ii) relies on the weighted average of the actual costs incurred and profits realized by the other exporters under review in selling a foreign like product in the ordinary course of trade.
Id.
§ 1677b(e)(2)(B)(ii). Option (iii) serves as a backstop that allows Commerce to derive the amounts incurred and realized “based on any other reasonable method,” provided that “the amount allowed for profit may not exceed the amount normally realized by exporters ... in connection with the sale, for consumption in the foreign country, of merchandise that is in the same general category of products as the subject merchandise.”
Id.
§ 1677b(e)(2)(B)(iii). The provision limiting the allowable profit in option (iii) is commonly referred to as the “profit cap.”
See, e.g., Atar I,
637 F.Supp.2d at 1088 n. 5.
In this case, Commerce determined that it could not proceed under § 1677(e)(2)(A) to calculate the constructed value of Atar’s products because Atar lacked data from a viable comparison market.
Certain Pasta from Italy,
Decision Memorandum for the
Final Results,
at 18 (Dep’t of Commerce Feb. 14, 2007) (“Decision Mem.”),
avail
able at
http://ia.ita.doc.gov/frn/summary/ italy/e7-2563-l.pdf. Commerce therefore turned to the three alternatives set forth in § 1677b(e)(2)(B).
Commerce disregarded option (i) because Atar did not produce any products other than the subject merchandise. Decision Mem. at 19. Option (ii) also proved unavailable because the ninth administrative review included only one other respondent, and Commerce concluded that relying on that respondent’s reported profit and SGA cost data would expose confidential business information.
Id.
at 20.
Thus finding options (i) and (ii) inappo-site, Commerce invoked option (in), which broadly authorizes the agency to derive the necessary profit and SGA values via “any other reasonable method,” subject to the statutory profit cap. In so doing, Commerce sought “a methodology that most closely simulatefd] the preferred method” of § 1677b(e)(2)(A). Decision Mem. at 19;
see also id.
at 18 (defining subsection (e)(2)(A) as “the preferred method”). Accordingly, Commerce chose to estimate Atar’s profit and SGA costs based on actual sales of a foreign like product made in the ordinary course of trade, as specified in § 1677b(e)(2)(A) as well as option (ii)—the difference being that the data Commerce used originated not from Atar or other respondents in the ninth review, but from analogous sales by the six respondents
in the prior (eighth) administrative review.
See
Decision Mem. at 14-21;
Certain Pasta from, Italy,
71 Fed.Reg. 45,017, 45,021-22 (Dep’t of Commerce Aug. 8, 2006) (notice of preliminary results). Commerce concluded that its methodology was “representative of Atar’s [SGA] and profit experience,
within the meaning of the profit cap as required by alternative (B)(iii),
and not distortive.” Decision Mem. at 19 (emphasis added). Using those data to derive the constructed value for Atar’s subject merchandise, Commerce reached its final antidumping duty margin of 18.18%.
Final Results,
72 Fed. Reg. at 7,012.
In March 2007, Atar brought suit in the trade court to challenge the
Final Results.
The trade court upheld Commerce’s decision to use a constructed value approach but concluded that the agency had not employed a “reasonable method” for calculating the constructed value of Atar’s products under § 1677b(e)(2)(B)(iii).
Atar I,
637 F.Supp.2d 1068. In particular, the court rejected Commerce’s decision to exclude data representing sales made outside the ordinary course of trade
(i.e.,
below-cost sales) from the prior administrative review period as “arbitrary.”
Id.
at 1087. According to the court, Commerce excluded below-cost sales based not on any analysis of facts and circumstances relevant to Atar but rather on the agency’s own default preferences for doing so.
Id.
at 1085-90. The court thus remanded for Commerce to reconsider its constructed value profit and SGA cost determinations.
Id.
at 1092-93.
Following
Atar I,
Commerce issued its first remand redetermination on September 3, 2009. Although it “continuefd] to believe that the methodology used in the
Final Results
constitute^] a ‘reasonable method,’ ” Commerce recalculated Atar’s constructed value using above- and below-cost sales data obtained from those respondents in the eighth administrative review that had earned a net profit during
that period of review (only two of the six total respondents). Using the weighted average profit and SGA cost values from all sales by those two respondents to gauge Atar’s constructed value, Commerce devised an amended antidumping duty margin of 14.45%.
On review, the trade court remanded again.
Atar S.r.l. v. United States,
703 F.Supp.2d 1359 (Ct.Int’1 Trade 2010)
(“Atar II”).
The court concluded that Commerce’s revised method for determining Atar’s constructed value profit was deficient for failing to separately and independently calculate the statutory profit cap.
Id.
at 1364-67. Accordingly, the court set aside the first remand redetermi-nation and remanded again for Commerce to reconsider Atar’s constructed value profit under option (iii).
Atar II,
703 F.Supp.2d at 1370.
Commerce issued its second remand re-determination on July 19, 2010. Commerce elected to continue using data from the two profitable respondents in the prior administrative review, including their above- and below-cost sales to calculate Atar’s constructed value profit. In addition, Commerce expressly used those same data to establish a profit cap, contending that “the weighted-average profit rate of the two [profitable] respondents ... after including sales made both within and outside the ordinary course of trade, established] a reasonable profit cap.” Because its constructed value profit figure—based on the same data—did not exceed (and indeed matched) the resulting profit cap, Commerce concluded that its preexisting margin of 14.45% complied with statutory requirements.
Once again, the trade court remanded.
Atar S.r.l. v. United States,
791 F.Supp.2d 1368 (Ct.Int’1 Trade 2011)
(“Atar III”).
The court held that the second remand redetermination had yielded an unlawful profit cap because Commerce had misinterpreted the statute and misapplied the available facts. Commerce had determined that the statute required the profit cap to be a positive amount, which in the agency’s view supported its reliance on data only from profitable respondents in the prior review. The court, however, rejected Commerce’s statutory interpretation as unreasonable, holding that the statute called for a flexible, case-by-case profit cap determination that did not require Commerce to exclude representative data from unprofitable producers.
Id.
at 1376. In addition, the court concluded that Commerce’s limited focus on the two profitable respondents “ignored home-market sales data [by the unprofitable producers] that were material and probative of the general conditions in the home market of Italy affecting the profitability of domestic pasta producers operating there.”
Id.
at 1377. The court reasoned that basing the calculation on only two producers had heavily skewed Commerce’s weighted-average profit cap figure toward a single large, profitable exporter and therefore did not reflect the actual conditions affecting the “amount normally realized” by Atar and others operating in the Italian home market.
Id.
at 1377-78. The court therefore remanded again for Commerce to recalculate Atar’s constructed value profit using a lawfully determined profit cap.
Id.
at 1380-81.
Commerce then issued its third remand redetermination on December 6, 2011, once again revising its constructed value profit and profit cap calculations. To begin, Commerce returned to its original method from the
Final Results
for deriving Atar’s constructed value profit based on the above-cost sales
of
all six respondents in the prior administrative review, citing our intervening decision in
Thai I-Mei Frozen Foods Co. v. United States,
616 F.3d 1300 (Fed.Cir.2010). Under protest, Commerce also adopted a lower profit
cap to comply with the trade court’s remand instructions, using the weighted average of all data (including below-cost sales) from all six respondents in the eighth administrative review. Because the newly derived profit cap imposed a ceiling on the allowable constructed value profit figure calculated under
Thai I-Mei,
the third remand redetermination resulted in a new antidumping duty margin applicable to Atar of 11.76%.
The trade court affirmed Commerce’s third remand redetermination.
Atar TV,
858 F.Supp.2d 1344. In
Atar IV,.
the court held that Commerce had used a reasonable method for determining the profit cap— the amount of profits “normally realized” by exporters or producers in Atar’s home market. The court explained that “[t]he inclusion of all sales, both above-cost and below-cost, in the profit cap calculation produced a result that more accurately reflected the profit conditions in the home market as a whole than would one confined to sales made in the ordinary course of trade.”
Id.
at 1349. The court disposed of Atar’s complaints that Commerce should have used a simple average, rather than a weighted average, of data from the six respondents in the prior review, stating that “Commerce must be allowed a degree of discretion as to its methodological choices in determining a profit cap.”
Id.
at 1350. As a result, the final 11.76% antidumping duty margin from the third remand redetermination was sustained, and the trade court entered final judgment accordingly.
The government timely appealed. We have jurisdiction pursuant to 28 U.S.C. § 1295(a)(5).
Discussion
We review the trade court’s decision in this case
de novo,
“applying] anew the same standard used by the court, and [we] will uphold Commerce’s determination unless it is unsupported by substantial evidence on the record, or otherwise not in accordance with law.”
Mittal Steel Point Lisas Ltd. v. United States,
548 F.3d 1375, 1380 (Fed.Cir.2008). Substantial evidence is “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.”
Consol. Edison Co. of N.Y. v. N.L.R.B.,
305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938). We must defer to Commerce’s reasonable construction of its governing statute where Congress “leaves a gap in the construction of the statute that the administrative agency is explicitly authorized to fill, or implicitly delegates legislative authority, as evidenced by ‘the agency’s generally conferred authority and other statutory circumstances.’ ”
Cathedral Candle Co. v. U.S. Int’l Trade Comm’n,
400 F.3d 1352, 1361 (Fed.Cir.2005) (quoting
United States v. Mead Corp.,
533 U.S. 218, 229, 121 S.Ct. 2164, 150 L.Ed.2d 292 (2001)). To review the reasonableness of agency action, “[c]ourts look for a reasoned analysis or explanation for an agency’s decision as a way to determine whether a particular decision is arbitrary, capricious, or an abuse of discretion.”
Wheatland Tube Co. v. United States,
161 F.3d 1365, 1369 (Fed. Cir.1998).
On appeal, the government contends that Commerce reasonably excluded below-cost sales from its profit cap calculations in the
Final Results.
Specifically, the government argues that Commerce followed its normal approach to constructed value profit determinations by excluding below-cost sales from its profit cap calculation. Commerce determined normal value for respondents in the eighth administrative review by excluding their below-cost sales pursuant to § 1677b(b)(l), and, according to the government, the agency’s use of an analogous approach to derive Atar’s profit cap from the same data rea
sonably reflected the “amount normally realized” from the sale of like pasta products in Italy.
The government also argues that Commerce’s profit cap calculations were consistent with a broad “statutory preference” for deriving constructed value profits from sales of a foreign like product made in the ordinary course of trade, as set forth in § 1677b(e)(2)(A) and (e)(2)(B)(ii) and reinforced by
Thai I-Mei.
Finally, the government asserts that the trade court afforded Commerce insufficient deference in interpreting and applying § 1677b (e) (2) (B) (iii) under the review standard applicable to such administrative determinations.
Atar offers no substantive response. As appellee, Atar did not file substantive briefing or participate in oral argument; instead, Atar submitted papers stating that it “has ceased commercial operations and does not possess the resources” for meaningful participation. Appellee’s Br. 2. Atar nonetheless urges affirmance of the trade court’s “well reasoned and considered decision.”
Id.
The only question presented in this appeal is whether the trade court erred in rejecting Commerce’s exclusion of below-cost sales from its profit cap calculations relating to Atar’s subject merchandise. The governing statute allows Commerce to calculate the amount allocated to profit in a constructed value determination based on any reasonable method,
except that the amount allowed for profit
may not exceed the amount normally realized
by exporters or producers (other than the [specific exporter or producer being examined]) in connection with the sale, for consumption in the foreign country, of merchandise that is in the same general category of products as the subject merchandise.
19 U.S.C. § 1677b(e)(2)(B)(iii) (emphasis added). The quoted language sets the maximum allowable profit—the profit cap—in constructed value determinations conducted under option (iii) of § 1677b(e)(2)(B) according to the “amount normally realized” from sales of similar goods in a reference market. As the trade court correctly recognized, the statutory language “does not speak directly to the question of how Commerce is to determine” the amount normally realized, nor does it “direct that data on unprofitable sales be included or excluded.”
Atar III,
791 F.Supp.2d at 1376. Because the statute in question is ambiguous, we must proceed to step two of the framework established 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), “and the deference it affords an agency’s ‘construction of a statutory scheme it is entrusted to administer.’ ”
Thai I-Me%
616 F.3d at 1305 (quoting
Chevron,
467 U.S. at 844, 104 S.Ct. 2778). The question, then, is whether Commerce acted reasonably in this case by excluding below-cost sales data from the prior administrative review in its calculation of Atar’s profit cap. We conclude that it did.
As described, subsections (i) to (iii) of § 1677b(e)(2)(B) provide Commerce with three avenues for deriving a profit figure to replicate the fraction of the constructed value that would be attributable to profit. The first option relies on sales data from the specific exporter or producer being considered, while the second uses a weighted average of profits made by
other
producers under review from sales of a foreign like product. § 1677b(e)(2)(B)(i)-(ii). Those two options set forth predefined formulas for assessing constructed value profit that are grounded in specific, objectively relevant data, and neither option includes a separate profit cap.
See id.
In contrast, the catch-all third option permits Commerce to calculate construct
ed value profit using “any other reasonable method,” subject to a profit cap that serves to prevent the various possible calculation methods from yielding anomalous results that stray beyond the “amount normally realized” from sales of merchandise in the same general category.
See
§ 1677b(e)(2)(B)(iii). Like option (ii), however, the profit cap provision expressly requires the use of data from different exporters or producers than the one under consideration.
See id.
(“[T]he amount allowed for profit may not exceed the amount normally realized by exporters or producers)(oiher
than the exporter or producer described in clause (i)
.... ” (emphasis added)).
In this case, Commerce had no choice but to proceed under option (iii). Atar had no comparison sales suitable for use under option (i), and Commerce rejected option (ii) because the ninth administrative review included only one other respondent, and using that respondent’s data alone would have revealed business-proprietary information.
Atar I,
637 F.Supp.2d at 1082. If the ninth administrative review had included sufficient data from one or more additional respondents, however, under option (ii) Commerce would have been
required
to exclude data from those respondents’ below-cost sales in calculating Atar’s constructed value profit.
See
19 U.S.C. § 1677b(e)(2)(B)(ii) (“[T]he weighted average of the actual amounts ... realized by exporters or producers that are subject to the investigation ... in connection with the production and sale of a foreign like product,
in the ordinary course of
trade----” (emphasis added)).
While such data were not available here, Commerce did have access to analogous data from the six respondents in the previous administrative review. Using those surrogate data, Commerce used methods that otherwise mirrored option (ii) to calculate Atar’s constructed value profit as the weighted average of the actual profits realized from above-cost sales of a foreign like product by the respondents in the eighth review. Decision Mem. 19. The resulting profit figure undeniably would have qualified as a permissible measure of constructed value profit if it had been derived pursuant to option (ii) during the eighth administrative review, and the record contains no indication that the relevant market underwent any substantial intervening change that would meaningfully distinguish the period of the eighth review (July 1, 2003 to June 30, 2004) from the timeframe now at issue (July 1, 2004 to June 30, 2005). Once Commerce decided to mirror option (ii) by using data from other respondents, the exclusion of below-cost sales made sense. As the Statement of Administrative Action (“SAA”) that accompanied the Uruguay Round Agreements Act
explains with regard to option (ii), the consideration of below-cost sales of a foreign producer’s competitors could allow that producer to “benefit perversely from its own unfair pricing.” SAA, H.R.Rep. No. 103-316, at 840 (1994),
reprinted in
1994 U.S.C.C.A.N. 3,773, 4,176. That is, a foreign producer might have undercut the market, thereby forcing its competitors to make unprofitable sales. Because the approach Commerce selected under option (iii) raised those same concerns, it decided, consistent with the SAA’s guidance, to avoid the risk of lowering Atar’s profit cap based on any such unprofitable sales. We therefore cannot conclude that Commerce acted unreasonably in deciding that the same weighted-aver
age profit rate would not exceed the “amount normally realized” under option (iii).
We recognize that the enumerated reference products differ between option (ii) and the profit cap provision of option (iii). Specifically, option (ii) contemplates profit data from other respondents’ sales of a “foreign like product,” while the profit cap provision references normal returns from sales of “merchandise that is in the same general category” as the subject products.
Compare
§ 1677b(e)(2)(B)(ii),
with
(e)(2)(B)(iii). As explained in the SAA, however, “[t]he term ‘general category of merchandise’ encompasses a category of merchandise broader than the ‘foreign like product.’ ” SAA at 840. In other words, a “foreign like product” also necessarily qualifies as merchandise within the “same general category” for purposes of § 1677b(e). As such, the use of data from sales of a foreign like product to address the profit cap requirement fits within the scope of § 1677b(e)(2)(B)(iii).
Furthermore, the “same general category” language of option (iii) does not prohibit Commerce from excluding below-cost sales when calculating a profit cap. The trade court concluded that “Congress did not intend for Commerce to exclude data on below-cost sales from its calculation when determining a profit cap.”
Atar II,
703 F.Supp.2d at 1365. In so doing, the court recited the following language from the SAA: “[T]he Administration does not intend that Commerce would engage in an analysis of whether sales in the same general category are above-cost or otherwise in the ordinary course of trade.”
Id.
(quoting SAA at 841).
Despite this statement of intention, however, the SAA does not prohibit Commerce from excluding below-cost sales when deriving a profit cap. Read in context, the cited language instead addresses the ordinarily limited availability of data concerning third-party sales of merchandise that does not qualify as a foreign like product in the review but does fall within the same general category of merchandise. The associated full paragraph from the SAA reads as follows:
The administration does not intend Commerce to require companies to submit all data necessary to apply each alternative. For example, Commerce will not require a company which has provided profit information on its own sales of the particular foreign like product also to submit profit information on its sales of the same general category of products solely to enable Commerce to use the latter information to calculate profit for a different company. Likewise, the Administration does not intend that Commerce would engage in an analysis of whether sales in the same general category are above-cost or otherwise in the ordinary course of trade.
SAA at 841.
The SAA thus recognizes that Commerce faces practical limitations on its
ability to obtain comprehensive data regarding the foreign sales of various related products in a typical investigation. In general, Commerce “is unlikely to have sale-specific data on merchandise in the same general category because such merchandise is not subject to the investigation or review.”
Thai I-Mei,
616 F.3d at 1308 (internal quotation marks omitted). Accordingly, the SAA makes clear that, in contrast to sales covering the specific foreign like product, Commerce need not gather the detailed and potentially voluminous data necessary to differentiate sales made within and outside the ordinary course of trade across an entire general category of merchandise. But relieving Commerce of the requirement to collect such detailed information in every instance does not prohibit its use when available. Here, data collected during the eighth administrative review allowed Commerce to distinguish between above- and below-cost sales of relevant comparison products, and, having such information readily available, Commerce reasonably elected to exclude below-cost sales in its profit cap calculations in this case. The SAA passage on which the trade court relied does not state an “unambiguously expressed intent” of Congress to the contrary.
Chevron,
467 U.S. at 843, 104 S.Ct. 2778.
Finally, the trade court’s decision was in large part founded on its perception that “below-cost sales were a significant feature of the home-market conditions affecting the marketing of pasta in Italy” during the relevant periods of review, such that Commerce’s attempts to minimize or exclude below-cost sales distorted its calculation of the “amount normally realized” from sales in that market.
Atar III,
791 F.Supp.2d at 1380;
see also Atar I,
637 F.Supp.2d at 1088 (faulting the
Final Results
for “a failure to ground the decision to exclude those [below-cost] sales in findings of fact ... that are pertinent to Atar’s specific situation”). Nevertheless, we have long recognized that antidumping determinations “involve complex economic and accounting decisions of a technical nature, for which agencies possess far greater expertise than courts.”
Fujitsu Gen. Ltd. v. United States,
88 F.3d 1034,1039 (Fed.Cir. 1996). Moreover, the statutory language at issue here is ambiguous. Accordingly, Commerce is entitled to substantial deference in its choice of accounting methodology, and, as a reviewing court, we may not substitute one reasonable approach for another according to our own preferences. In addition, we note that Commerce has not advocated a rigid requirement that below-cost sales data must be excluded from
all
profit cap determinations.
See
Oral Arg. at 14:37,
available at
http://www. cafc.uscourts.gov/oral-argument-record-ings/2013-1001/all (“Commerce ... must consider each situation on a case-by-case basis, so, in this case, Commerce did not automatically exclude below-cost sales.”). That approach is consistent with the SAlA’s directive that Commerce should “determine on a case-by-case basis the profits ‘normally realized’ by other companies on merchandise of the same general category.” SAA at 841.
Conclusion
In view of the foregoing, we conclude that Commerce acted reasonably in excluding below-cost sales data from the pri- or administrative review when calculating the constructed value profit cap applicable to Atar’s subject merchandise under § 1677b(e)(2)(B)(iii). Accordingly, we re
verse and remand for further action consistent with this opinion.
REVERSED AND REMANDED