SolarWorld Americas, Inc. v. United States

Kelly, Judge: Before the court is the U.S. Department of Commerce's ("Department" or "Commerce") remand determination in the second administrative review of the antidumping duty ("ADD") order covering crystalline silicon photovoltaic cells, whether or not assembled into modules, from the People's Republic of China ("China" or "the PRC"), pursuant to the court's order in SolarWorld Americas, Inc. v. United States , 41 CIT ----, ----, 273 F.Supp.3d 1254 , 1278-79 (2017) (" SolarWorld Americas I "). See Final Results of Remand Redetermination , Jan. 18, 2018, ECF No. 123-1 (" Remand Results "); see also Crystalline Silicon Photovoltaic Cells, Whether or Not Assembled Into Modules, From the [PRC] , 81 Fed. Reg. 39,905 (Dep't Commerce June 20, 2016) (final results of ADD administrative review and final determination of no shipments; 2013-2014) and accompanying Decision Mem. for the Final Results of the 2013-2014 [ADD] Admin. Review of Crystalline Silicon Photovoltaic Cells, Whether or Not Assembled into Modules, From the [PRC], A-570-979, (June 13, 2016), ECF No. 21-5 ("Final Decision Memo"). For the reasons that follow, the court sustains Commerce's determination to include import data with reported quantities of zero in the surrogate value calculations and remands for further explanation or reconsideration consistent with this opinion Commerce's surrogate value selections for respondent Yingli Green Energy Holding Co., Ltd.'s tempered glass input and respondent Changzhou Trina Solar Energy Co., Ltd.'s scrapped solar cell and module byproduct offset. BACKGROUND The court assumes familiarity with the facts of this case as discussed in the previous opinion, see SolarWorld Americas, Inc. , 41 CIT at ----, 273 F.Supp.3d at 1259-60 , and here recounts the facts relevant to the court's review of the Remand Results . In this second administrative review of the ADD order on crystalline silicon photovoltaic cells, whether or not assembled into modules, from China, Commerce selected Yingli Green Energy Holding Co., Ltd. ("Yingli") and Changzhou Trina Solar Energy Co., Ltd. ("Trina") as mandatory respondents. See Crystalline Silicon Photovoltaic Cells, Whether or Not Assembled Into Modules, From the [PRC] , 80 Fed. Reg. 80,746 , 80,746 (Dep't Commerce Dec. 28, 2015) (preliminary results of ADD administrative review and preliminary determination of no shipments; 2013-2014) and accompanying Decision Mem. for Prelim. Results of the 2013-2014 [ADD] Administrative Review of Crystalline Silicon Photovoltaic Cells, Whether or Not Assembled into Modules, from the [PRC], A-570-979, at 2, PD 520, bar code 3427351-01 (Dec. 18, 2015) (citing 2013-2014 [ADD] Admin. Review of Crystalline Silicon Photovoltaic Cells, Whether or Not Assembled into Modules, from the [PRC]: Respondent Selection, A-570-979, at 4-5, PD 67, bar code 3264380-01 (Mar. 13, 2015) ). 1 In the final determination, Commerce valued Yingli's tempered glass input using Thai import data under Harmonized Tariff Schedule ("HTS") subheading 7007.19.9000, see Final Decision Memo at 29-34, and Trina's scrapped solar cell and module byproduct using Thai import data under HTS subheading 8548.10. 2 id="p1346" href="#p1346" data-label="1346" data-citation-index="1" class="page-label">*1346 See id. at 46-48 . Commerce included in the average unit surrogate value calculations for all factors of production import data with reported quantities of zero, finding no basis in the record to support a determination that the zero-quantity values are unreliable or incorrect, simply because quantity listed is zero. See id. at 63-64 . Plaintiff, SolarWorld Americas, Inc. ("SolarWorld"), moved for judgment on the agency record, challenging certain aspects of the final determination. See SolarWorld's Mot. J. Agency R., Jan. 26, 2017, ECF No. 44; SolarWorld Americas, Inc.'s Mem. Supp. Rule 56.2 Mot. J. Agency R. Conf. Version, Jan. 26, 2017, ECF No. 44; Summons, July 20, 2016, ECF No. 1 (commencing this action pursuant to section 516A of the Tariff Act of 1930, as amended, 19 U.S.C. § 1516a(a)(2)(B)(iii) (2012) ). 3 Relevant on remand, SolarWorld challenged Commerce's determination to value Trina's scrapped solar cell and module byproduct using Thai data for imports classified under HTS subheading 8548.10 ("Waste and scrap of primary cells, primary batteries and electric accumulators; spent primary cells, spent primary batteries and spent electric accumulators; electrical parts of machinery or apparatus, not specified or included elsewhere in this Chapter: Other"). Mandatory respondents Yingli et al. 4 and Trina et al. 5 each also commenced litigation challenging certain aspects of the final determination; both actions have been consolidated with the present action. See Mem. Points and Authorities Supp. Mot. J. Agency R., Jan. 26, 2017, ECF No. 42 ("Yingli Br."); Mem. Supp. Mot. [Trina et al.] J. Agency R., Jan. 26, 2017, ECF No. 43 ("Trina Br."); Order, Oct. 25, 2016, ECF No. 31 (order consolidating all three actions related to this administrative review). Relevant here, Yingli challenged Commerce's use of Thai import data to value Yingli's tempered glass input, contending that the Thai data is aberrational, see Yingli Br. at 9-26, and Trina challenged Commerce's inclusion, in the calculation of surrogate values, values for Thai import categories with reported quantities of zero, contending that doing so resulted in surrogate values that are not supported by substantial evidence. See Trina Br. at 16-19. In the prior decision, the court sustained in part and remanded in part Commerce's final determination in this review. 6 SolarWorld Americas, Inc. , 41 CIT at ----, 273 F.Supp.3d at 1278-79 . Specifically, the court remanded three issues. The court remanded Commerce's selection of a surrogate value for Yingli's tempered glass input to explain why the selection is reasonable in light of evidence of the disproportionate impact of Hong Kong input data and the allegation of aberrational benchmarks. See id. , 41 CIT at ----, 273 F.Supp.3d at 1261-65, 1278-79 . The court remanded Commerce's determination to value Trina's scrapped solar cells and modules byproduct offset using import data for Thai HTS category 8548.10, determining that Commerce had not sufficiently explained why the selection is reasonable given that the category is not specific to the solar cells and modules and in light of SolarWorld's evidence that the selection results in a surrogate value for the byproduct that is higher than the value of the input itself. See id. , 41 CIT at ----, 273 F.Supp.3d at 1267-68, 1278-79 . Finally, the court remanded Commerce's use of surrogate values for factors of production with reported quantities of zero for Commerce to explain why the inputs are reliable in light of the evidence on the record that the values are not within range of the values for other low-quantity imports on the record. See id. , 41 CIT at ----, 273 F.Supp.3d at 1273-75, 1278-79 . Commerce filed the Remand Results on January 18, 2018. Plaintiff SolarWorld continues to challenge Commerce's selection of Thai data for imports classified under HTS subheading 8548.10 as a surrogate to value Trina's scrapped solar cell and module byproduct. See Pl. [SolarWorld]'s Comments on Final Results of Redetermination Pursuant to Remand at 5-8, Mar. 7, 2018, ECF No. 133 ("SolarWorld Remand Comments"). SolarWorld contends that Commerce on remand continues to insufficiently explain its selection of an HTS category specific to scrapped battery cells, a product with which the scrapped solar cells and modules share no components, making it an unreasonable surrogate value. Id. Consolidated Plaintiff Yingli continues to challenge Commerce's selection of Thai import data for valuing Yingli's tempered glass input, contending that Commerce has insufficiently explained its selection and failed to address the court's request to explain why the selection of the Thai data is reasonable in light of the disproportionate impact that the import data from Hong Kong has on the overall Thai data value. See Comments of Pls., [Yingli] et al., on [Commerce]'s Final Results of Remand Redetermination, Mar. 6, 2018, ECF No. 129 ("Yingli Remand Comments"). Consolidated Plaintiff Trina continues to challenge Commerce's inclusion of import data with reported quantities of zero in its surrogate value calculations, contending on remand that Commerce has relied upon erroneous calculations to support its analysis of the data sets. See [Trina]'s Comments on [Commerce]'s Final Results of Redetermination Pursuant to Remand, Mar. 5, 2018, ECF No. 127 ("Trina Remand Comments"). JURISDICTION AND STANDARD OF REVIEW The court has jurisdiction pursuant to 19 U.S.C. § 1516a(a)(2)(B)(iii) and 28 U.S.C. § 1581 (c) (2012), which grant the court authority to review actions contesting the final determination in an administrative review of an antidumping duty order. "The court shall hold unlawful any determination, finding, or conclusion found ... to be unsupported by substantial evidence on the record, or otherwise not in accordance with law." 19 U.S.C. § 1516a(b)(1)(B)(i). "The results of a redetermination pursuant to court remand are also reviewed 'for compliance with the court's remand order.' " Xinjiamei Furniture (Zhangzhou) Co. v. United States , 38 CIT ----, ----, 968 F.Supp.2d 1255 , 1259 (2014) (quoting Nakornthai Strip Mill Public Co. v. United States , 32 CIT 1272 , 1274, 587 F.Supp.2d 1303 , 1306 (2008) ). DISCUSSION I. The Use of Import Data with Reported Quantities of Zero In the final determination, Commerce included values for import data with reported quantities of zero in the surrogate value calculations. See Final Decision Memo at 63-64. Trina argued that the values with zero quantities were not reliable, such that their inclusion results in distorted surrogate values. Trina Br. at 16-19. The court remanded on this issue, determining that Commerce's conclusory explanation that the values are reliable because there was no reason to conclude the zero-quantity values were errors was insufficient given Commerce's acknowledgment that the majority of the zero-quantity values are not within range of other low-quantity import values on the record. SolarWorld Americas, Inc. , 41 CIT at ----, 273 F.Supp.3d at 1273-75, 1278-79 . The court requested that, on remand, Commerce explain how the inclusion of these values is reasonable. Id. , 41 CIT at ----, 273 F.Supp.3d at 1275 . On remand, Commerce provided additional analysis of the zero-quantity values and further explanation of its determination that the values were reliable and not in error. See Remand Results at 33-52. For the reasons that follow, the court sustains Commerce's determination on this issue. In the final determination, Commerce stated that, in this review, "in most instances, the values for zero quantity import data points are not within the range of other lower quantity and value import data points ...." Final Decision Memo at 64. The court requested that Commerce explain, on remand, how its explanation that there is no indication that the values are unreliable is reasonable in light of this acknowledgment that the values are out of range of other low-quantity values on the record. See SolarWorld Americas, Inc. , 41 CIT at ----, 273 F.Supp.3d at 1275 . On remand, Commerce "acknowledge[s] that [the agency] did not directly address the data Trina placed on the record" regarding the zero-quantity values in the final determination. Remand Results at 34. In its remand determination, Commerce analyzes the import data to determine whether the zero-quantity values were in fact reliable. See id. at 33-52. Commerce determined, after examining the data on the record, "that the values of the zero-quantity imports are, in fact, relatively low, especially when the overall trend of the import data reflects that low quantity imports have a higher [average unit value ("AUV") ] than high quantity imports." Id. at 34-35. Based on this determination, Commerce "continue[s] to find that the zero-quantity data are attributable to rounding small quantities down to zero, rather than random errors that might result in unreliable data." Id. at 35. Commerce summarized and explained the results of the analyses and comparisons it ran on remand to determine whether the zero-quantity values are within range of other low-quantity import values on the record. See Remand Results at 34-38. Commerce concluded that the "comparisons and analysis do indicate a strong correlation between low value and zero quantities," which it determined demonstrates that the data is reliable, reasoning that, if the zero values were the results of random errors, the zero-quantity imports would not be "consistently in the low value range of all imports" as the comparisons demonstrate that they are. Id. at 38. Specifically, one analysis demonstrated that, "of any imported quantity with more than one import ( i.e. , data point), the average value of zero-quantity imports is lower than the AUV of any other quantity of imports." Id. at 35. In another analysis, Commerce "divide[d] the value of each zero-quantity import by the AUV of the HTS category under which the zero-quantity import was classified" and averaged those quantities, which resulted in a quantity for the values that was "nearly 50,000 times less than the average of the quantities of all data points." Id. at 36. Commerce uses this figure to demonstrate that the value of zero-quantity imports is lower than the AUV for other quantities for each import category. Commerce further discovered that more than one-third of the zero-quantity import values "have a value that is less than the AUV of the HTS category under which the zero-quantity import was classified," which Commerce suggests implies a quantity of less than one for those values because "the AUV is the average value of one unit imported under that HTS category[.]" Id. at 36-37. In an additional comparison methodology, Commerce replaced the zero with 0.49, the highest quantity that would be rounded down to zero, and discovered that, "overall, the AUVs for zero-quantity imports are consistent with the AUVs of other low quantity imports." Id. at 37. Based on all of these methods of comparison, Commerce concluded that the zero-quantity import values "exhibit characteristics consistent with other low-quantity imports." Id. at 37. Accordingly, Commerce "continue[s] to find that the zero-quantity data are attributable to rounding small quantities down to zero, rather than random errors that might result in unreliable data." Id. at 35. Commerce has responded to the court's order. Commerce explained that it had not previously addressed the data thoroughly, and accordingly ran a thorough analysis of the data on remand. The results of that analysis demonstrate that the zero-quantity values are in fact consistent with other low-quantity values on the record. Having confirmed that the values are consistent with other low-quantity values, it is reasonable for Commerce to conclude that the zero-quantity values are not the result of error but are the result of rounding quantities between 0.01 and 0.49 down to zero, and to determine that the data is reliable. Trina contends that, on remand "Commerce has ignored substantial information that demonstrates that the values of zero quantity Thai entries vary tremendously even within the same HTS classification and therefore cannot reasonably be considered to reflect only small quantity imports," such that these values are unreliable and should not be included. Trina Remand Comments at 5. Commerce's analysis of the data emphasizes that the overall trend of the zero-quantity values suggests that the zero-quantity values are consistent with other low-quantity values on the record. This explanation does not require that the value within each HTS category be unvaried. The determination that the values are generally consistent with other low-quantity values for each HTS category is reasonable despite some variation within each HTS category. Trina also argues that Commerce's analysis is misleading and inaccurate because the agency has presented an analysis of the figures which relies upon grouping "together completely different imported products with widely divergent import average unit values." Trina Remand Comments at 5. However, in explaining its analysis, Commerce stated that its methodology specifically accounted for the fact that there are differences in the AUVs of individual HTS categories. See Remand Results at 35-36, 46. To account for these differences, Commerce "divid[ed] the total value of each import by the AUV of the HTS category under which the import was classified to normalize for differences in the AUVs of different HTS categories," and then compared the averages that resulted for each distinct import quantity to assess each individual HTS category. Id. at 46. This methodology reasonably assesses the reliability of each individual HTS category, and Trina's argument that the agency erroneously assessed all values together without taking those differences into account is therefore unpersuasive. II. Tempered Glass The court remanded Commerce's valuation of Yingli's tempered glass input using Thai import data, requesting that Commerce explain how the selection is reasonable in light of its past practice, record evidence of the disproportionate impact of the Hong Kong input values, and the claim of aberrational benchmarks. See SolarWorld Americas, Inc. , 41 CIT at ----, 273 F.Supp.3d at 1261-65, 1278-79 . Yingli reported tempered glass as a factor of production in this review. In proceedings involving imports from a nonmarket economy country, 7 such as the PRC, Commerce obtains a normal value by adding the value of the factors of production used to produce the subject merchandise with other costs, expenses, and profits. See 19 U.S.C. § 1677b(c)(1). Pursuant to the statute, Commerce selects "the best available information regarding the values of such factors in a market economy country or countries" as a surrogate with which to value each factor of production. 8 Id. Commerce has broad discretion in deciding what constitutes the best available information. See QVD Food Co. v. United States , 658 F.3d 1318 , 1323 (Fed. Cir. 2011) (noting the absence of a definition for "best available information" in the ADD statute). The agency has developed a methodology to determine which data source is the best available information, which is to select a source that is (1) specific to the input; (2) tax and import duty exclusive; (3) contemporaneous with the period of review; (4) representative of a broad market average; and (5) publically available. See Import Admin., U.S. Dep't Commerce, Non-Market Economy Surrogate Country Selection Process , Policy Bulletin 04.1 (2004), available at http://enforcement.trade.gov/policy/bull04-1.html (last visited May 15, 2018) (" Policy Bulletin 04.1 "). Despite its discretion, Commerce's determination of what constitutes the best available information must be based in the objective of the ADD statute, to calculate accurate dumping margins. See Rhone Poulenc, Inc. v. United States , 899 F.2d 1185 , 1191 (Fed. Cir. 1990) ; see also Parkdale Int'l. v. United States , 475 F.3d 1375 , 1380 (Fed. Cir. 2007). As explained in SolarWorld Americas I , it is Commerce's practice not to use aberrational values as surrogate values. Antidumping Duties; Countervailing Duties , 62 Fed. Reg. 27,296 , 27,366 (Dep't Commerce May 19, 1997); SolarWorld Americas, Inc. , 41 CIT at ----, 273 F.Supp.3d at 1262 . It is the agency's practice, "[w]hen presented with sufficient evidence to demonstrate that a particular surrogate value is aberrational, and therefore unreliable," to "examine relevant price information on the record, including any appropriate benchmark data, in order to accurately value the input in question." Remand Results at 6 (citations to past practice omitted). Commerce explains that its practice is to assess aberrationality by examining HTS data both across potential surrogate countries and within the surrogate country over multiple years. Id. Commerce considers import data to be aberrationally high if that data is "many times higher than the import values from other countries." Id. (quoting Final Decision Memo at 33). On remand, Commerce continues to value the tempered glass input using Thai import data, again determining that the import data is not aberrational based on a revised explanation of its practice for determining aberration. See Remand Results at 4-9, 12-33. Commerce clarifies that, in the final determination, the agency erred by citing two cases as current practice which no longer reflect its current practice. Id. at 13 (citing Final Results of Redetermination Pursuant to Catfish Farmers of America v. United States, Consol. Court No. 08-00111 (Sept. 14, 2009), ECF No. 100-1 (" Catfish Farmers Remand Results "); Issues and Decision Mem. for the Investigation of Steel Wire Rope from the [PRC], A-570-859 (Feb. 28, 2001), available at http://ia.ita.doc.gov/frn/summary/prc/014895-1.txt (last visited May 15, 2018) (" Steel Wire Rope from the PRC ") ) ). In those cases, Commerce had explained its practice as evaluating individual inputs within the overall import data for a certain HTS category and, if found to be aberrational, removing those component inputs from the import data before calculating the surrogate value. See id. ; Catfish Farmers Remand Results at 4-7; Steel Wire Rope from the PRC at Comment 1. The court expressed concern that Commerce's practice in fact runs counter to the methodology that the agency employed in the present case and requested that, on remand, the agency reconcile its methods in this case with that practice. SolarWorld Americas, Inc. , 41 CIT at ----, 273 F.Supp.3d at 1264-65 . On remand Commerce explains that its current practice is in fact to require interested parties to demonstrate that the import data is aberrational in the aggregate, rather than to evaluate each individual input that forms the overall value for aberrationality. Remand Results at 14-15. Commerce states: The underlying rationale is that "[w]hen determining whether data are aberrational, [Commerce] has found that evidence of a high or low AUV does not necessarily establish that GTA data for the suspect countries are unreliable, distorted or misrepresentative. Rather, interested parties must provide specific evidence showing whether the value is aberrational." Commerce's current practice considers whether the AUV, in the aggregate , is aberrational for the economically comparable surrogate countries or as compared to historical AUVs of the surrogate country at issue. Id. at 25-26 (emphasis in original) (citing and quoting Issues and Decision Mem. for the Final Results of the 2012-2013 [ADD] Admin. Review of Multilayered Wood Flooring from the PRC at Comment 11.D, A-570-970 (July 8, 2015) (" Wood Flooring Decision Memo ") ). Commerce emphasizes that this current practice does not require the agency to evaluate whether certain imports with a high or low value have a disproportionate impact on the overall import value, but instead to assess whether the overall AUV "is consistent with surrogate values for the input from other economically comparable countries identified as potential surrogate countries." 9 Id. at 15. Commerce states that this practice is reasonable because it would be administratively burdensome to require the agency to assess the potential aberration of each data point on the record, id. at 16, and would invite interested parties to request "distortive cherry picking of data" to suit their objectives. Id. at 15 (quoting Polyethylene Terephthalate Film, Sheet, and Strip from the [PRC]: Issues and Decision Mem. for the Final Results of the 2010-2011 Admin. Review at 12, A-570-924, (June 5, 2013), available at http://ia.ita.doc.gov/frn/summary/prc/2013-13985-1.pdf (last visited May 15, 2018) (" PET Film Decision Memo ") ). 10 Commerce emphasizes that this practice is grounded in its "judicially-affirmed preference to base surrogate values on broad data that reflect the surrogate country's market as a whole." Id. Commerce contends that, because its practice is to determine whether the overall AUV of the HTS category is aberrational in the aggregate, it is unnecessary to analyze individual data points for aberration. Id. at 28. Commerce explains that its actions in the present case were consistent with this practice, as the agency compared the Thai AUV for tempered glass with AUVs for tempered glass from other potential surrogate countries, which resulted in the determination that the Thai AUV is not aberrational. See id. at 15. On remand Commerce emphasizes that it followed its current practice here as reflected in the Wood Flooring proceeding. See Remand Results at 13-15, 25-27. In Wood Flooring , although Commerce asserted that it is only concerned with aberrationality in the aggregate, it nevertheless explained why the allegedly aberrational inputs were in fact representative of market-driven prices by assessing the share each input represented of the aggregate data. Wood Flooring Decision Memo at 43 (finding "imports from Taiwan and the United States represent the vast majority of imports into Thailand (77.1%) and, therefore, are a true representation of market-driven prices."). Thus, it is not clear from Wood Flooring whether Commerce's practice is to assess what percentage of the market the allegedly aberrational input data constitutes, to determine whether that data is representative of the market. If that is the case, Commerce must clarify how its practice is relevant here, where the allegedly aberrational Hong Kong data comprises just 1.6% of the overall import data into Thailand and yet constitutes more than 75% of the overall value of the Thai import data. 11 See Yingli Remand Comments at 9; Reply of Pls. [Yingli et al.] Supp. Mot. J. Agency R. at 4-5, June 2, 2017, ECF No. 67 ("Yingli Reply Br."). If Commerce does not have a practice of considering what percentage of market share is made up by the input data in question, Commerce should explain why it focused on market representation in the Wood Flooring Decision Memo . Because Commerce has invoked Wood Flooring here to explain its practice, the agency must clarify what exactly that practice is and why, in light of that practice, its selection of the Thai import data constitutes a reasonable surrogate value for the tempered glass input. 12 Additionally, in the Remand Results , Commerce did not respond to the court's request that the agency explain why its selection of the surrogate value for tempered glass is reasonable given the evidence of the Hong Kong data's disproportionate impact on the overall value of the Thai import data. See SolarWorld Americas, Inc. , 41 CIT at ----, 273 F.Supp.3d at 1263-65 . The inclusion of the Hong Kong data, which has such a disproportionate effect on the overall Thai import value, appears to contradict Commerce's stated "preference to base surrogate values on broad data that reflect the surrogate country's market as a whole." Remand Results at 15. Commerce's only response to the court's request is its contention that the Hong Kong data does not distort the Thai value for tempered glass because there is no evidence to support "Yingli's claim that the Thai AUV for tempered glass does not reflect a range of prices that is representative of the Thai import market." Id. at 17. The Hong Kong data represents 1.6% of the import volume but increases the value of the imports by more than 75%, see Yingli Reply Br. at 4-5; it is not clear how the Thai AUV as reflected in this record represents the Thai market as a whole. Commerce's response is inadequate and does not speak to the underlying issue: that the Hong Kong data skews the Thai AUV in a way that renders the Thai AUV unrepresentative of the Thai market. On remand Commerce must reconsider or explain the issue of the disproportionate impact of the Hong Kong data. The court acknowledges Commerce's concern regarding the potential administrative burden involved with assessing individual inputs for aberration. Commerce notes that [i]dentifying and defining what is and what is not aberrational among these thousands of data points spread along a vast spectrum of relatively high and low values is an impossible task. An argument that an import value that is ten times the average value is clearly aberrational may on its face appear plausible, and yet the record in this case contains thousands of such values. Our current practice of only examining whether an entire country's AUV is aberrational prevents what could be a never-ending process of removing allegedly relatively high and low individual data points from our calculations. The second concern with using the proposed analysis is that it undermines the use of broad-market average prices for a particular input. Both concerns relate to Commerce's statutory obligation to rely on the "best information available" to value [factors of production], which the parties recognize. Remand Results at 26-27 (citations omitted). Commerce claims that "the proposed analysis ... undermines the use of broad-market average prices," id. at 27, but it is unclear how retaining the Hong Kong input, which so disproportionately effects the overall value, serves the objective of using broad market average prices. Further, Commerce asserts that disaggregating inputs that are ten times higher (or lower) than the average value would create an insurmountable administrative burden, as there could be "thousands of such values." Id. at 26. However, here, there is a claim that the Hong Kong input is close to two-hundred times higher than the average unit values from the rest of the import data. See Yingli Remand Comments at 9. Although the administrative burden concern is significant, it does not outweigh the accuracy concerns raised in a case such as this where the Thai data includes unit values from Hong Kong which make up only 1.6% of the import volume yet, at 191 times higher than the average unit values from other countries, quadruple the Thai AUV. See id. ; Yingli Br. at 17-18. Commerce of course may change its practice; however, its practice must still be reasonable. Here, Commerce has supported its claimed change in practice, stating it will no longer disaggregate data and exclude aberrational values, see Remand Results at 13-15, but Commerce has not explained how its practice supports its stated preference to "base surrogate values on broad data that reflect the surrogate country's market as a whole," id. at 15, where unit values representing 1.6% of the import volume account for more than 75% of the total value of Thai imports of tempered glass. A practice that considers values in the aggregate to avoid administrative burdens may be reasonable in other cases but, without further explanation, does not appear reasonable on this record. III. Scrapped Solar Cells and Modules The court remanded Commerce's selection of Thai HTS category 8548.10 to value Trina's scrapped solar cell and module byproduct. See SolarWorld Americas, Inc. , 41 CIT at ----, 273 F.Supp.3d at 1267-68, 1278-79 . The court concluded that Commerce had not explained its decision to value the solar cell and module byproduct using an HTS category which is specific to scrapped electric battery cells. Id. The court requested that Commerce explain on remand why its selection is reasonable given the fact that Thai HTS category 8548.10 is not specific to solar cells or modules and results in a value for the scrapped cell and module byproduct that is higher than the value of the input itself. Id. On remand, Commerce has continued to value the byproduct using Thai HTS category 8548.10. See Remand Results at 53-64. SolarWorld continues to challenge the selection and argues that Commerce has still not explained why the selection is reasonable given that the category is not specific to the input. SolarWorld Remand Comments at 5-8. For the reasons that follow, the issue is remanded for further explanation or reconsideration. To calculate normal value for a nonmarket economy country, Commerce removes the value of reported byproducts from the values calculated for the factors of production, expenses, and profits. See 19 U.S.C. § 1677b(c)(1). As with factors of production, Commerce selects a surrogate with which to value each byproduct using "the best available information regarding the values of such factors in a market economy country or countries." Id. Commerce's methodology for selecting the best available information evaluates data sources based upon their: (1) specificity to the input; (2) tax and import duty exclusivity; (3) contemporaneity with the period of review; (4) representativeness of a broad market average; and (5) public availability. See Policy Bulletin 04.1 . Although Commerce has discretion to determine what constitutes the best available information, see QVD Food Co. v. United States , 658 F.3d 1318 , 1323 (Fed. Cir. 2011), the agency must ground its selection of the best available information in the overall purpose of the statute, which is to calculate accurate dumping margins. See Rhone Poulenc, Inc. , 899 F.2d at 1191 ; see also Parkdale Int'l. , 475 F.3d at 1380 . There were two potential surrogate values placed on the record of this review with which Commerce could value Trina's scrapped solar cell and module byproduct: the Thai import value for HTS category 8548.10, covering "waste and scrap of primary cells, primary batteries and accumulators; spent primary cells, spent primary batteries, and spent electrical accumulators," and the Thai import value for HTS category 2804.69, covering silicon of less than 99.99 percent purity. Remand Results at 53. On remand, Commerce continues to value the scrapped solar cell and module byproduct using Thai HTS category 8548.10 for scrapped battery cells. See id. at 53-64. Commerce explains that this category constitutes an appropriate surrogate value because "the manufacturing processes and the raw materials used to produce primary cells and batteries are more similar to the processes and inputs used in producing solar cells than those used to extract or produce silicon." Id. at 54. Commerce elaborates upon this similarity by noting that scrapped solar cells and modules, like scrapped battery cells, "are scrap products that were initially assembled together from many different inputs to create negative and positive electronic charges capable of conveying electricity." Id. Commerce asserts that, in contrast, the silicon covered under HTS category 2804.69, SolarWorld's preferred category, is minimally processed and "is not a product that is now scrap or one that was originally manufactured/assembled together from many different inputs for the purpose of producing or conveying electricity." Id. Commerce also emphasizes that it found HTS category 2804.69 not specific to Trina's scrapped solar cells and modules because the polysilicon in a solar cell is of a higher purity than the silicon covered by HTS category 2804.69. 13 Id. This explanation does not sufficiently explain why, on this record, the category is a reasonable choice for the best available information. Products that are assembled from multiple inputs, convey electricity, undergo certain unspecified manufacturing processes, and are ultimately scrapped do not inherently share a similar value. In emphasizing these similarities, Commerce misses the point of a surrogate value for a byproduct. The surrogate value should be a product that is similarly valued in order to achieve an accurate valuation for the respondent's byproduct and, ultimately, for the respondent's normal value. See Nation Ford Chem. Co. v. United States , 21 CIT 1371 , 1375-76, 985 F.Supp. 133 , 137 (1997), aff'd , 166 F.3d 1373 (Fed. Cir. 1999). 14 Commerce has not provided an explanation as to why the selection of a category covering scrapped electrical batteries accurately values the respondent's scrapped solar cells and modules byproduct. 15 Without an explanation which demonstrates that the selected import data, for HTS category 8548.10, provides a representative value for the scrapped solar cells and modules in a market-economy PRC, the court cannot say that the selection of this category is reasonable. 16 See id. The issue is remanded to Commerce to reconsider or further explain its selection of HTS category 8548.10 in light of this opinion. CONCLUSION For the foregoing reasons, the court sustains Commerce's determination to include import data with reported quantities of zero in the surrogate value calculations. This matter is remanded to Commerce for reconsideration or further explanation consistent with this opinion Commerce's surrogate value selection for Yingli's tempered glass input and Trina's scrapped solar cells and modules byproduct offset. In accordance with the foregoing, it is ORDERED that Commerce's surrogate value selections for the respondents' tempered glass input and scrapped solar cells and modules byproduct offset are remanded for further explanation or reconsideration consistent with this opinion. Commerce shall file its remand determination with the court within 60 days of this date; and it is further ORDERED that the parties shall have 30 days thereafter to file comments on the remand determination; and it is further ORDERED that the parties shall have 30 days thereafter to file a reply to comments on the remand determination. Footnotes [1] On September 14, 2016, Defendant submitted indices to the public and confidential administrative records for this review. These indices are located on the docket at ECF Nos. 21-2 and 21-3. All further references to documents from the administrative records are identified by the numbers assigned by Commerce in these indices. [2] In the final determination, Commerce valued Yingli's, but not Trina's, scrapped solar cells using Thai import values for HTS 2804.69, explaining: Yingli reported that it removes the polysilicon from its scrap solar cells and reintroduces it into production. Thus, the value of these scrap solar cells is in the silicon content. Hence, consistent with Solar ARI , we valued Yingli's scrap cells based on HTS 2804.69, which is the HTS category applicable to silicon. Final Decision Memo at 47. Commerce noted that, "[i]n contrast," because Trina reported that its scrap is composed of broken cells and modules that could not be reintroduced into production, the agency "determined that Trina's cell scrap consisted of every component of the cell, not simply polysilicon, and its modules scrap consisted of every component of the module." Id. [3] Further citations to the Tariff Act of 1930, as amended, are to the relevant provisions of Title 19 of the U.S. Code, 2012 edition. [4] The following parties are plaintiffs in the action Yingli Green Energy Holding Co., Ltd. v. United States, Ct. No. 16-00135, which has been consolidated with the present action: Yingli Green Energy Holding Company Limited; Yingli Green Energy Americas, Inc.; Yingli Energy (China) Co., Ltd.; Baoding Tianwei Yingli New Energy Resources Co., Ltd.; Tianjin Yingli New Energy Resources Co., Ltd.; Hengshui Yingli New Energy Resources Co., Ltd.; Lixian Yingli New Energy Resources Co., Ltd.; Baoding Jiasheng Photovoltaic Technology Co., Ltd.; Beijing Tianneng Yingli New Energy Resources Co., Ltd.; Hainan Yingli New Energy Resources Co., Ltd.; and Shenzhen Yingli New Energy Resources Co., Ltd. [5] The following parties are plaintiffs in the action Changzhou Trina Solar Energy Co., Ltd. v. United States, Ct. No. 16-00132, which has been consolidated with the present action: Changzhou Trina Solar Energy Co., Ltd.; Trina Solar (Changzhou) Science and Technology Co., Ltd.; Yancheng Trina Solar Energy Technology Co., Ltd.; Changzhou Trina Solar Yabang Energy Co., Ltd.; Turpan Trina Solar Energy Co., Ltd.; and Hubei Trina Solar Energy Co., Ltd. [6] Specifically, the court sustained: Commerce's surrogate value selections for valuing respondents' aluminum frames, semi-finished polysilicon ingots and blocks, solar backsheet, and nitrogen inputs; Commerce's selection of financial statements for calculating financial ratios for the respondents' overhead, selling, general, and administrative expenses, and profit; and Commerce's application of adverse facts available to Trina's unreported, purchased solar cells. See SolarWorld Americas, Inc. , 41 CIT at ----, 273 F.Supp.3d at 1278 . [7] The term "nonmarket economy country" means any foreign country that Commerce determines "does not operate on market principles of cost or pricing structures, so that sales of merchandise in such country do not reflect the fair value of the merchandise." 19 U.S.C. § 1677 (18)(A). In such cases, Commerce must "determine the normal value of the subject merchandise on the basis of the value of the factors of production utilized in producing the merchandise ... [together with other costs and expenses]." Id. § 1677b(c)(1). [8] To the extent possible, Commerce uses "the prices or costs of factors of production in one or more market economy countries that are-(A) at a level of economic development comparable to that of the nonmarket economy country, and (B) significant producers of comparable merchandise." 19 U.S.C. §§ 1677b(c)(4)(A)-(B). Commerce also has a regulatory preference for valuing all factors of production using surrogate value data from a single surrogate country where practicable. 19 C.F.R. § 351.408 (c)(2) (2015). [9] The court notes that Commerce, after dismissing the practice demonstrated in Catfish Farmers Remand Results and Steel Wire Rope from the PRC (in which the agency disaggregated import data and excluded component data that it found aberrational) as contrary to its current practice, Remand Results at 13, later cites those cases as support of its methodology for determining aberration. Id. at 32 . Commerce invokes Catfish Farmers Remand Results and Steel Wire Rope from China to support its finding in this case that the Thai AUV is not aberrational: [W]hile the POR Thai AUV for tempered glass is approximately four and a half times the average of Thai AUVs for tempered glass from the first administrative review and the investigation in this proceeding, in Steel Wire Rope , Commerce stated that it would determine whether unit values are aberrational if they are many times higher than the import values from other countries. Similarly, in Fish from Vietnam , the Department found the surrogate values for labels to be aberrational where the AUVs varied between 30 and 79 times greater than the average of the rest of the import data. Hence, our comparison to historical data does not demonstrate that the POR Thai AUV for tempered glass is aberrational, particularly because it is within the POR AUVs of tempered glass from the other potential surrogate countries. This failure to refute the POR Thai AUV for tempered glass, in the aggregate, with credible benchmarks supports the reasonableness of this value. Id. at 32 . Although Commerce invoked these cases for its practice of only finding values aberrational when they are many times higher than other benchmarks, in the sections cited in both cases the agency was in fact discussing excluding certain aberrational components within the overall data, see Steel Wire Rope from the PRC at Comment 1; Catfish Farmers Remand Results at 4-7, which is exactly the approach the plaintiffs are seeking here. Commerce cites Catfish Farmers Remand Results and Steel Wire Rope from the PRC for the proposition that a value has to be many times higher to be considered aberrational while ignoring the fact that at issue in those cases were allegedly aberrational component input data. Commerce also cites a section in Steel Wire Rope from the PRC in which the agency excluded an aggregate surrogate value, rather than one component data input, that it determined was aberrational overall. See Remand Results at 32 (citing Steel Wire Rope from the PRC at Comment 6). [10] Although Commerce cites to the PET Film Decision Memo as well as the Wood Flooring Decision Memo as evidence of its current practice, see Remand Results at 13-14, there is no discussion in the PET Film Decision Memo regarding the share of the market represented by the data that was allegedly aberrational for which exclusion was sought. See PET Film Decision Memo at 17-18. [11] Yingli presented calculations to Commerce and this court which demonstrate the extreme impact of the Hong Kong input data on the overall Thai AUV: if the Hong Kong data was excluded, the overall Thai value would drop from $4.14 USD per kilogram to $1.00 USD per kilogram, which constitutes a 76% decrease in value. See Yingli Br. at 22-23; Reply of Pls. [Yingli et al.] Supp. Mot. J. Agency R. at 4-5, June 2, 2017, ECF No. 67 ("Yingli Reply Br."); Yingli's Case Br. at 22-27, PD 563, bar code 3438258-01 (Feb. 2, 2016). Yingli's data demonstrates that the value of the imports for this HTS category into Thailand from Hong Kong was $7,351,945 USD, which is 76.16% of the total value, $9,652,802 USD, of imports into Thailand for this HTS category, while the quantity of the same Hong Kong imports was 38,398 kilograms, which is 1.6% of the total quantity, 2,331,015 kilograms, of imports into Thailand for this HTS category. See Yingli Reply Br. at 4-5. [12] Commerce's explanation on remand on the issue of aberrational benchmark data from Ecuador and Ukraine is sufficient. Regarding the contention that the Ecuador and Ukraine import values are aberrational benchmarks, Commerce emphasizes that the fact that an AUV is of a lower import quantity does not in and of itself render the AUV aberrational; additional indication that the import data is aberrational is needed, and no such indication was presented by interested parties in this case. Id. at 17-18 . [13] It is not clear to the court that this difference in purity on which Commerce relies to reject HTS category 2804.69 is a rational reason to reject that category, given that Commerce values Yingli's scrapped solar cells using HTS category 2804.69, which Commerce describes as "the HTS category applicable to silicon," because Yingli reported extracting the polysilicon from its solar cell byproduct. See Final Decision Memo at 47. Even though Commerce believes that SolarWorld's position "calls for speculation that parties are purchasing scrapped solar cells and modules only for their polysilicon," Remand Results at 61, Commerce's determination with respect to Yingli makes clear that, if purchasers were doing so, HTS 2804.69 would be the appropriate category for valuation of the byproduct, despite the difference in purity levels between the silicon covered by the category and the respondent's polysilicon. See Final Decision Memo at 47 (finding "the nature of the process, and the additional chemicals and additives used during cell production, introduce impurities ... suggest[ing] that the recycled polysilicon is not at the purity level required for solar grade polysilicon (99.9999 percent silicon)."). [14] In the final determination, Commerce explained that it "has a long-standing practice of rejecting or capping byproduct surrogate values in instances where the byproduct surrogate value exceeds the surrogate value of the input from which it was derived," but that this practice did not apply in this case because the argument was presented by SolarWorld in relation to an alternate HTS category that was ultimately not selected. Final Decision Memo at 48. The court requested that, on remand, Commerce explain why its selection of import data for HTS category 8548.10 is reasonable, in light of the concerns raised by SolarWorld regarding selecting a surrogate value for a byproduct that is higher than the value for the input itself. SolarWorld Americas I , 41 CIT at ----, 273 F.Supp.3d at 1268 . In the Remand Results , Commerce states that the value of the HTS category should be compared to the cost of finished solar cells rather than to the cost of polysilicon. Remand Results at 55 ("SolarWorld improperly compared the scrap surrogate value to the value of polysilicon when the proper comparison is to the cost of solar cells, which is approximately $200 per kilogram (over four times greater than the scrap surrogate value)."). Commerce's position that the value of the scrap stems from all of the components is at odds with SolarWorld's position that it is the polysilicon, the primary input into the cells and modules, that gives value to the scrap as it is "the raw material that is reclaimed when cells and modules are scrapped." SolarWorld Remand Comments at 6. Record information supports SolarWorld's position. Commerce determined that Yingli's scrap solar cells and modules derive their value from the reclaimed polysilicon. Final Decision Memo at 47. Commerce noted that Yingli reported that it extracts the polysilicon from the scrapped solar cells and that Trina reported that it does not. Id. Commerce notes that SolarWorld speculates that Trina sells the scrap for others to extract the polysilicon. Remand Results at 61. However, Commerce speculates that the scrap is sold not for the polysilicon but for all the components. Id. ("... [I]t is not appropriate to assume that purchasers valued these defective products, particularly scrapped modules that continued to function, solely for polysilicon.") Although there is record evidence to support SolarWorld's speculation, Commerce does not point to anything in the record which supports its own speculation that Trina's scrapped solar modules are resold for components other than their polysilicon. In light of the record evidence that scrapped solar cells and modules are valuable to some for the polysilicon, Defendant has not explained why it is reasonable for Commerce to assume that scrapped solar cells and modules are valuable to others for more than the polysilicon. [15] Commerce attempts to explain why its position makes sense by comparing it to an example: Commerce's practice is to select surrogate values as specific to the input (or byproduct/scrap) being valued as possible. There are no surrogate values specific to scrapped solar cells or modules on the record. Therefore, we looked for surrogates covering products of the same nature as completed solar cells and modules. We would not use the price of an automobile to value a suitcase. Rather, we would rely on suitcase prices to value suitcases and if there were no such prices on the record, then we would rely on the value of a broader category of products consisting of containers used to convey or transport items to value a suitcase. Because solar cells and modules are electrical products manufactured using a multiple array of inputs, including chemicals and metals, we find that the potential surrogate covering scrapped manufactured electrical products comprising various inputs is the better surrogate compared to a potential surrogate covering silicon rocks. Remand Results at 62. It is unclear how this comparison demonstrates that the selection of a category specific to scrapped electrical batteries provides a reasonable surrogate value for scrapped solar cell and module byproduct. This analogy demonstrates that a certain different inapposite valuation would be unreasonable and it implies that the scrapped batteries are broader category of solar modules. Commerce does not explain why its implication is reasonable other than to say that solar modules and batteries contain metals and chemicals and produce electricity. This explanation does not support Commerce's implication. [16] Commerce and Plaintiff-Intervenor each allude to the fact that HTS category 8548.10 for scrapped battery cells contains an "other" subcategory, suggesting that scrapped modules could possibly be represented within this other category. See Remand Results at 59; Pl.-Intervenors BYD (Shangluo) Industrial Co., Ltd. and Shanghai BYD Co., Ltd. Comments on Remand Results at 5, Mar. 21, 2018, ECF No. 136. It is not clear that it is reasonable to suggest that this "Other" category would cover scrapped solar modules. However, this suggestion, even if reasonable, would not address the lack of a rationale that focuses on the representativeness of the selected value, as would be necessary to make the selection of this HTS reasonable on this record.