Changzhou Trina Solar Energy Co. v. United States

2019 CIT 137

Opinion

Slip Op. 19-137

UNITED STATES COURT OF INTERNATIONAL TRADE

CHANGZHOU TRINA SOLAR ENERGY CO., LTD., and TRINA SOLAR (CHANGZHOU) SCIENCE & TECHNOLOGY CO., LTD.,

Plaintiffs,

CANADIAN SOLAR INC., ET AL., Before: Jane A. Restani, Judge

Plaintiff-Intervenors, Consol. Court No. 17-00198 v.

UNITED STATES, PUBLIC VERSION

Defendant,

SOLARWORLD AMERICAS, INC.,

Defendant-Intervenors.

OPINION AND ORDER

Dated: November, 2019

[Commerce’s Remand Redetermination in the Third Administrative Review of the Countervailing Duty Order pertaining to photovoltaic cells from the People’s Republic of China is partially sustained and partially remanded for reconsideration consistent with this opinion.]

Robert G. Gosselink, and Jonathan M. Freed, Trade Pacific, PLLC, of Washington, D.C., for Plaintiffs and Defendants-Intervenors Changzhou Trina Solar Energy Co., Ltd. and Trina Solar (Changzhou) Science & Technology Co., Ltd.

Craig A. Lewis, Hogan Lovells US LLP, of Washington, D.C., for Consolidated Plaintiffs Shanghai BYD Co., Ltd. and BYD (Shangluo) Industrial Co., Ltd.

Jeffrey S. Grimson, Kristin H. Mowry, Jill A. Cramer, Sara M. Wyss, James C. Beaty, and Bryan P. Cenko, Mowry & Grimson, PLLC, of Washington, D.C., for Plaintiffs-Intervenors Canadian Solar Inc., Canadian Solar International, Ltd., Canadian Solar Manufacturing (Changshu), Inc., Canadian Solar Manufacturing (Luoyang), Inc., Canadian Solar (USA) Inc., CSI Cells Co., Ltd., CSI Solar Power (China) Inc., CSI Solartronics (Changshu) Co., Ltd., CSI Solar Technologies Consol. Court No. 17-00198 Page 2 Public Version Inc., and CSI Solar Manufacture Inc.

Chad A. Readler, Jeanne E. Davidson, Tara K. Hogan, and Justin R. Miller, International Trade Field Office, U.S. Department of Justice, of New York, NY. Of counsel on the brief was Paul Keith, Office of Chief Counsel for Trade Enforcement and Compliance, U.S. Department of Commerce, of Washington, D.C.

Timothy C. Brightbill, Laura El-Sabaawi, and Usha Neelakantan, Wiley Rein, LLP, of Washington, D.C., for Defendant-Intervenor SolarWorld Americas, Inc.

Restani, Judge: This action concerns the U.S. Department of Commerce’s (“Commerce”)

remand redetermination filed pursuant to the court’s order in Changzhou Trina Solar Energy Co.

v. United States, 352 F. Supp. 3d 1316 (CIT 2018) (“Changzhou Trina I”); see Final Results of

Redetermination Pursuant to Court Remand, ECF No. 103-1 (Dep’t Commerce Apr. 25, 2019)

(“Remand Results”).

In Changzhou Trina I, the court determined that remand was necessary for Commerce to

further explain several of its decisions in the underlying review, or otherwise alter its

determination. Specifically, the court remanded for Commerce to explain and/or reconsider

whether: (1) respondents benefitted from the People’s Republic of China’s (“PRC”) Export

Buyer’s Credit Program (“EBCP”), (2) the provision of aluminum extrusions for less than adequate

remuneration (“LTAR”) was a specific subsidy, (3) the inclusion of potentially overbroad United

Nations Comtrade data in its calculation of the aluminum extrusion and solar glass benchmarks

was appropriate, (4) Commerce should have considered Canadian Solar’s data on polysilicon

imports as a tier-one metric, and (5) the provision of electricity for LTAR was a specific subsidy.

On remand, Commerce has attempted to clarify its decisions, but its decision remains largely

unaltered.

BACKGROUND

The court assumes familiarity with the facts of this case as discussed in its prior opinion, Consol. Court No. 17-00198 Page 3 Public Version Changzhou Trina I, and thus recounts relevant facts only as necessary below. This matter involves

a challenge by plaintiffs Changzhou Trina Solar Energy Co., Ltd., Trina Solar (Changzhou)

Science & Technology Co., Ltd. (collectively, “Trina”); consolidated plaintiffs BYD (Shangluo)

Industrial Co., Ltd. and Shanghai BYD Co., Ltd. (collectively, “BYD”); 1 and plaintiffs and

plaintiff-intervenors Canadian Solar Inc., Canadian Solar International, Ltd., Canadian Solar

Manufacturing (Changshu), Inc., Canadian Solar Manufacturing (Luoyang), Inc., Canadian Solar

(USA) Inc., CSI Cells Co., Ltd., CSI Solar Power (China) Inc., CSI Solartronics (Changshu) Co.,

Ltd., CSI Solar Technologies Inc., and CSI Solar Manufacture Inc. (collectively, “Canadian

Solar”) against Commerce’s remand redetermination in the Third Administrative Review of

Commerce’s Countervailing Duty Order pertaining to photovoltaic cells from the PRC.

SolarWorld Americas. Inc. (“SolarWorld”) is a defendant-intervenor. 2

JURISDICTION AND STANDARD OF REVIEW

The court has jurisdiction pursuant to 28 U.S.C. § 1581(c) and 19 U.S.C. § 1516a(a)(2)

(2012). The court upholds Commerce’s remand redetermination unless it is “unsupported by

substantial evidence on the record, or otherwise not in accordance with law[.]” 19 U.S.C. §

1516a(b)(1)(B)(i).

DISCUSSION

I. Export Buyer’s Credit Program

1 As in Changzhou Trina I, here BYD does not present its own arguments, but rather adopts the arguments made by Trina and Canadian Solar. See BYD’s Comments on the Final Remand Redetermination, ECF No. 115 (June 19, 2019). 2 Although SolarWorld filed a response to plaintiff and plaintiff-intervenors’ comments objecting to the remand results, its response is simply a statement of agreement with Commerce’s decision on the EBCP and its specificity findings regarding aluminum extrusions and electricity. See SolarWorld’s Response to Comments on Final Results of Redetermination Pursuant to Court Remand, ECF No. 122 (Aug. 14, 2019). Consol. Court No. 17-00198 Page 4 Public Version Lately, the Export Buyer’s Credit Program (“EBCP’) has been the subject of frequent

litigation in the court. See Clearon Corp. v. United States, 359 F. Supp. 3d 1344, 1358–60 (CIT

2019) (collecting cases). The EBCP promotes PRC exports by providing preferential loan rates to

foreign purchasers of PRC goods. See GOC Initial CVD Questionnaire Response, at 147–51, P.R. 3

100–102, C.R. 16–18, 20 (May 3, 2016). Commerce found that following 2013 revisions to the

EBCP, that the prior $2 million-dollar contract minimum to qualify for the program had been

repealed and that EBCP loans may be routed through third-party banks and not simply issued from

the Export-Import Bank of China (“EX-IM Bank”) as previously understood. See I & D Memo at

13; Prelim I & D Memo at 31. The Government of China (“GOC”) refused to provide information

on the 2013 revisions, including internal guidelines. See I & D Memo at 13. Because of the GOC’s

non-cooperation, Commerce found that it was unable to verify respondent’s certifications of non-

use. I & D Memo, at 13. Accordingly, Commerce found that respondents, through the application

of AFA, 4 had used the program despite their cooperation in the review. Id. 5

The court remanded this issue concluding that Commerce did not demonstrate that

respondent’s certifications were unverifiable. Changzhou Trina I, 352 F. Supp. 3d at 1327. The

3 “P.R.” refers to a document contained in the public administrative record. “C.R.” refers to a document contained in the confidential administrative record. “Rem.” refers to documents submitted following Commerce’s Remand Redetermination. 4 When a party fails to cooperate to the best of its ability, Commerce may “use an inference that is adverse to the interest of that party in selecting from among the facts otherwise available.” See 19 U.S.C. § 1677e(b). Commerce refers to this process as “AFA” or “adverse facts available.” 5 In its third administrative review Commerce found that cooperating parties used the EBCP despite respondents’ certifications of non-use, whereas in the second administrative review Commerce found certifications to be sufficient to support non-use of the program. See Changzhou Trina I, 352 F. Supp. 3d at 1324. Commerce claims the 2013 revisions called into question the verifiability of non-use certifications such that they would no longer be considered sufficient to establish non-use of the EBCP. Id. Consol. Court No. 17-00198 Page 5 Public Version court held that although Commerce may apply AFA in a way that collaterally affects a cooperating

party, Commerce had not tried to avoid that undesirable consequence. See id. at 1325–27.

Additionally, Commerce did not explain “how an adverse inference regarding the operation of the

EBCP logically leads to a finding that respondents used the program.” Id. at 1326.

On remand, Commerce continues to find the certifications unverifiable and imputes usage

of the EBCP based on the application of AFA. Remand Results at 12–24. Commerce refers to a

discussion with an EX-IM Bank official who apparently indicated that the 2013 revisions

eliminated the contract minimum. See id.; see also Administrative Review of Countervailing Duty

Order on Citric and Certain Citrate Salts: Verification of the Questionnaire Resp. Submitted by

the GOC, at 2, P.R. 150 (Oct. 7, 2014) (“EX-IM Discussion”). In addition, Commerce cites a

questionnaire submitted by the GOC in a different investigation indicating that an EBCP “borrower

must be an importer or a bank approved by the China EX-IM Bank.” See GOC’s 7th Supp. Resp.,

Certain Amorphous Silica Fabric from China CVD Investigations (C-570-039), P. R. 150 (Sep. 6,

2016) (“GOC Silica Questionnaire Resp.”). Commerce states that it cannot conduct verification

using its normal practices given these uncertainties about the EBCP’s potential use of third-party

banks to distribute EBCP funds. Remand Results at 19–20. Commerce claims it requires the

GOC’s disclosure of the 2013 internal guidelines and other information, because without this

information, effective verification is stymied, if not completely impeded, as Commerce would be

unable to effectively sort through and identify potentially-suspect transactions given the size of the

respondent companies. 6 Id. at 21–23. Finally, Commerce finds that respondents benefitted from

the program after applying an adverse inference to evidence that the EX-IM Bank provided loans

6 Commerce infers that given the size of the respondent companies and their “substantial amount of business activity,” there would be too much financial data to sort through. Id. at 21–22. Consol. Court No. 17-00198 Page 6 Public Version to “new and high-tech projects” and because “energy projects are eligible for this financing.” Id.

at 24.

Canadian Solar and Trina argue that Commerce has failed to show that it is missing any

information regarding the usage of the EBCP and rather, that Commerce identifies a potential gap

in information concerning the operation of the EBCP. Canadian Solar Comments on Final remand

Redetermination, ECF NO. 113 at 2–6 (June 19, 2019) (“Canadian Solar Br.”); Trina Comments

on Final Results of Redetermination Pursuant to Court Remand, ECF No. 111 at 5–6, 9–12 (June

19, 2019) (“Trina Br.”). They also argue that the record demonstrates that obtaining loans through

the EBCP actively involves both the U.S. importer and Chinese exporter such that either can verify

usage. Canadian Solar Br. at 8–9; Trina Br. at 9 (citing GOC Initial CVD Questionnaire Response

at 151). Most notably, Canadian Solar cites evidence showing that the exporter receives funds

directly from the EX-IM Bank. Canadian Solar Br. at 9–10. Respondents claim that Commerce

overstates the difficulty in verifying whether respondents or their customers used the program. See

Canadian Solar Br. at 13–14; Trina Br. at 15–18. 7 They assert because Commerce has not

requested relevant records, Commerce’s claim about the difficulty of verification is speculative.

Id. at 14–15; Trina Br. at 15–20. Finally, Trina claims that Commerce improperly relies on

“uncorroborated statements from the petition” in its AFA analysis. Trina Br. at 11–12. The

government claims Commerce sufficiently explained its need to understand the operation of the

EBCP in order to conduct verification and that the use of AFA was appropriate. See Defendant’s

7 For instance, respondents argue that at verification Commerce often performs a “spot check” rather than full analysis, and that such a sampling could be done here. Canadian Solar Br. at 15; Trina Br. at 18–20. The government rejects that a “spot-checking verification procedure” would be possible given what it expects will be a “substantial amount of business activity” to search through. Gov. Br. at 12–13. Consol. Court No. 17-00198 Page 7 Public Version Reply to Comments on Remand Redetermination, ECF No. 120 at 10–11 (Sep. 14, 2019) (“Gov.

Br.”).

The court must determine whether substantial evidence exists by reviewing the record as

a whole. See e.g., Nippon Steel Corp. v. United States, 337 F.3d 1373, 1379 (Fed. Cir. 2003).

From the documents submitted by the government, it appears that the Commerce became

concerned about the verifiability of customer certifications of non-use following a discussion

with an EX-IM Bank official in a different administrative review. See Remand Results at 17–18.

During that discussion, the official apparently informed Commerce that in 2013 the $2 million-

dollar contract minimum was eliminated. See EX-IM Discussion at 2. This prompted Commerce

to review EX-IM Bank documents including “The Implementing Rules for the Export-Buyer’s

Credit of the Export-Import Bank of China” which Commerce claims appears to indicate the

involvement of “intermediary Chinese bank[s].” See Remand Results 18. When asked to clarify,

the GOC failed to do so. Id. at 19. The record indicates, however, that the GOC had in another

investigation a month earlier explained that:

According to the Ex-Im Bank, in order to make a disbursement, the Ex-Im Bank lending contract requires the buyer (importer) and seller (exporter) to open accounts with either the Ex-Im Bank or one of its partner banks. While these accounts are typically opened at the Ex-Im Bank, sometimes a customer prefers another bank (e.g., the Bank of China) which is more accessible than an account with the Ex-Im Bank. The loan agreement also stipulates that the borrower (generally the importer/customer) must grant the Ex-Im Bank authorization to conduct transactions in the account opened specifically for this financing. After all conditions for disbursement are met, the Ex-Im Bank will disburse the funds according to the lending agreement. The funds are first sent from the Ex-Im Bank to the borrower’s (importer) account at the Ex-Im Bank (or other approved partner bank). The Ex-Im Bank then sends the funds from the borrower’s (importer) account to the seller’s (exporter) bank account.

GOC Silica Questionnaire Resp. at 4–5. Thus, it appears that “other approved partner bank[s]”

may be involved in some capacity in the disbursement of EBCP funds. The discussion with the

EX-IM official indicates that after the importer’s application for the EBCP is approved, “[t]he Consol. Court No. 17-00198 Page 8 Public Version foreign importer will then instruct EXIM bank to pay the Chinese exporter by assigning payment

to the Chinese exporter’s bank account.” See EX-IM Discussion at 2. Considering the evidence

as a whole, it may be that even if funds are temporarily routed to banks outside the EX-IM bank,

funds are sent back to the EX-IM bank and that it disburses those funds to the exporter. If this is

indeed the situation, then Commerce would apparently need to verify only whether the exporter

had received any funds from the EX-IM bank and then, if so, ask them to provide documentation

showing the purpose of those funds. In this situation, verification seems relatively

straightforward.

If, however, the funds are not routed back through the EX-IM bank prior to reaching the

exporter, verification would admittedly be more difficult. But, so long as Commerce were able to

access the importer’s and exporter’s records, it appears that Commerce could cross-reference the

records to see if any funds appeared to originate from the EX-IM bank, even if the funds went

through an intermediary bank at some point. This seems especially doable with Trina and its

affiliated U.S. importer, given that it has only one U.S. customer. See Trina Br. at 3. The court

suspects that doing so will either confirm non-use or at least help clarify how the EBCP operates.

The court cannot sustain Commerce’s determination that verification would be impossible

or unduly onerous. Although Commerce has shown that the GOC failed to answer certain questions

regarding the EBCP’s operation, it is still not entirely clear to the court that the missing information

is required to effectively verify respondent’s non-use of the program. In order to avoid

unnecessarily impacting cooperating parties because of the GOC’s failure to cooperate, Commerce

needs to at least attempt to verify the certifications of non-use in this case. See Archer Daniels

Midland Co. v. United States, 917 F. Supp. 2d 1331, 1342 (CIT 2013) (noting that Commerce

should “seek to avoid” adversely impacting a cooperating party). There appears to be enough Consol. Court No. 17-00198 Page 9 Public Version information on the record for Commerce to identify potential suspect financial entries. As

respondents indicate, this may require Commerce to deviate from its standard verification

procedures. The court suggests ways in which Commerce might attempt verification, but

respondents have suggested others that may be preferable. On remand, the parties should discuss

potential ways forward and Commerce should request records that may answer the question of

EBCP use from respondents, and, if necessary, their importers. Commerce should detail its process

in its remand redetermination.

Should verification fail to clarify whether respondents benefited from the EBCP, and

Commerce continue to apply adverse facts, the court would consider Commerce’s reliance on the

verification checklist in this analysis problematic. Although Commerce cites the results of the

investigation and the accompanying issues and decision memorandum as facts available

supporting the notion that EX-IM Bank and EBCP loans are made to “new and high-tech projects”

such as energy projects, that finding appears to be based on the petitioner’s allegations in the

petition. 8 See Crystalline Silicon Photovoltaic Cells, Whether or Not Assembled Into Modules,

From the People’s Republic of China, 77 Fed. Reg. 63,788 (Dep’t Commerce Oct. 17, 2012), and

accompanying Issues and Decision Memorandum at 59 (Oct. 9, 2012). This does not logically lead

to Commerce’s remand results for two reasons. First, after review of the Initiation Checklist, it

appears that this allegation may relate to the EX-IM Bank’s seller’s credit and not the buyer’s

credit program at issue here. See Import Administration Office of AD/CVD Operations

Countervailing Duty Investigation Initiation Checklist, C-570-980 at 24 (Nov. 8, 2011) (“Initiation

8 In the Remand Results, Commerce cites the underlying investigation and issues and decision memorandum to support its claim that energy projects use the EBCP. See Remand Results at 24. In turn, the issues and decision memorandum cites the Initiation Checklist, which lists documents supporting this claim that have not been placed on the record in this case. Consol. Court No. 17-00198 Page 10 Public Version Checklist’”). This potential discrepancy relates to the court’s other problem with Commerce’s

remand redetermination on this issue—Commerce relies on the Initiation Checklist, and ostensibly

the underlying supporting documentation, but does not submit the latter to the court.

As noted in both the relevant statutory and regulatory provisions, a petition can serve as a

source of information for the selection of adverse facts. 19 U.S.C. § 1677e(b)(2)(A); 19 C.F.R.

§ 351.308(c)(1)(i) (stating that information derived from the petition is “secondary information”).

These provisions also state, however, that when relying on secondary information, Commerce shall

“to the extent practicable, corroborate that information from independent sources that are

reasonably at [Commerce’s] disposal.” 19 U.S.C. § 1677e(c)(1); 19 C.F.R. § 351.308(d). In this

case, it is unclear what, besides the allegations made in the petition, supported Commerce’s finding

regarding solar industry usage of the EBCP. Although there may be underlying documents that

corroborate this finding, those documents have not been submitted to the court. See Deacero

S.A.P.I. de C.V. v. United States, 393 F. Supp. 3d 1280, 1286 (CIT 2019) (noting that the Initiation

Checklist is part of Commerce’s pre-initiation analysis and that “[t]he independent sources may

be embedded in the pre-initiation analysis; however, the pre-initiation analysis itself is not an

independent source”) (footnote omitted). If Commerce continues to apply AFA in determining that

respondents’ buyers benefitted from the EBCP, Commerce should explain what evidence beyond

petitioner’s allegations in the investigation supports Commerce’s finding.

II. Provision of Aluminum Extrusions for LTAR

The court remanded Commerce’s finding that the subsidization of aluminum extrusions

represented a countervailable subsidy as Commerce failed to adequately explain whether “the

actual recipients of the subsidy, whether considered on an enterprise or industry basis, [were]

limited in number.” See 19 U.S.C. § 1677(5A)(D)(iii)(I). The record appeared to indicate that Consol. Court No. 17-00198 Page 11 Public Version aluminum extrusion use was widespread, which would render the subsidy non-specific, and thus

non-countervailable. See Changzhou Trina I, 352 F. Supp. 3d at 1330–31 (noting that aluminum

extrusions were utilized for “building and construction; transportation; electrical; machinery and

equipment; consumer durables; and other industries”).

On remand, Commerce continued to find the subsidization of aluminum extrusions de facto

specific, but further explained its finding that “those sectors that actually consume aluminum are

limited in number.” Remand Results at 25–27. Commerce noted that within the six broad industries

mentioned, the actual users within those industries are also limited in number. Id. Commerce

explained that “usage is limited to certain enterprises involved in the production” of a narrow set

of applications, 9 even though those enterprises may fall into larger industrial categories. Id. at 26–

27. Thus, Commerce found that although six broad sectors use aluminum extrusions, upon further

examination the enterprises within those industries are limited. Id. Finally, Commerce considered

whether the makeup of users of aluminum extrusions “could be considered something akin to the

whole of the Chinese economy,” and found, based on record evidence, that comparatively

numerous industries do not benefit from its subsidization.

Canadian Solar argues that Commerce misunderstands the court’s order and still fails to

properly assess specificity. Canadian Solar Br. at 18–22. It argues that as most industries in the

GOC use aluminum, a specificity finding is unwarranted. Id. For support it cites a GOC submission

9 Specifically, Commerce points to evidence that demonstrates that even though aluminum extrusions are used in “building and construction,” within that potentially broad category, the major applications of aluminum extrusions are “’frames of doors and windows,’ ‘curtain wall,’ ‘structural frames,’ ‘bridges,’ and ‘guard bars.’” See Remand Results at 26. Similarly, although finding that aluminum extrusions are used by “machinery and equipment,” the GOC only listed as major applications “‘elevator and escalator,’ ‘shield, handrail and terrace,’ ‘agricultural machinery,’ ‘radiator,’ and ‘shape-setting equipment and assembly-line equipment.’” Id. Commerce asserts that this shows that within these broad sectors, usage is limited to a narrow range of applications and thus only certain enterprises. Id. at 26–27. Consol. Court No. 17-00198 Page 12 Public Version claiming that “as many as 113 industries out of 124 industries in China consume aluminum.” GOC

Initial CVD Questionnaire Resp. re Canadian Solar, P.R. 123 at 44 (June 10, 2016). The

government responds that Commerce’s finding of specificity should nonetheless be sustained

because although aluminum may be used by many industries, the usage within those industries is

limited on an enterprise basis. See Gov. Br. at 14–18.

Commerce has adequately explained its aluminum extrusions specificity determination.

Although some evidence indicates that aluminum extrusions are used by many Chinese industries,

the record also supports Commerce’s contention that in practice these subsidies are primarily used

in a narrow range of applications. Although the evidence cited by Commerce is not definitive,

Commerce’s decision is nonetheless sufficiently supported by the record. See Nippon Steel, 458

F.3d at 1351. Although aluminum extrusions may be used by broad sectors of the PRC’s economy,

the actual applications of extrusions are primarily limited to a relatively narrow range of

applications. Accordingly, the court sustains Commerce’s aluminum extrusions specificity

determination.

III. Use of Comtrade/IHS data in Computing a Benchmark for Aluminum Extrusions

In computing the benchmark calculation for aluminum extrusions, Commerce averaged

datasets from UN Comtrade (“Comtrade”) and IHS Technology/Markit (“IHS"). 10 The court

remanded and ordered Commerce to consider whether the Comtrade dataset was overinclusive of

irrelevant aluminum products such that it was “too flawed to be probative of the world market

price” for aluminum extrusions. Changzhou Trina I, 352 F. Supp. 3d at 1332–33.

10 As detailed throughout the court’s previous opinion, the Comtrade data is based on various HTS subheadings but is computed monthly whereas the IHS data is based on the exact product at issue but is a weighted annual average. Changzhou Trina I, 352 F. Supp. 3d at 1331. Consol. Court No. 17-00198 Page 13 Public Version On remand, Commerce continues to average both data sets. 11 It distinguishes a similar issue

involving the benchmark price for solar glass, detailed below, and states that no record evidence

distinguishes solar frames from other types of aluminum extrusions contained in the Comtrade

data. Remand Results at 27–28. Commerce cites the IHS Report as evidence to support that the

price of solar frames fluctuates [[ ]]. Id. at 30 (IHS PV

Materials Report at 303 (Nov. 15, 2016) (“IHS Report”)). Because the Comtrade data is the only

data on record that captures monthly-price fluctuations, Commerce continues to find its inclusion

necessary. Id.

Canadian Solar and Trina continue to argue that the Comtrade data is overinclusive. See

Canadian Solar Br. at 22–26; Trina Br. at 29–39. Canadian Solar argues that the use of a six-digit

heading is a “basket category” that is “overly general regardless of whether they individually

reflect the inputs used by Canadian Solar.” Canadian Solar Br. at 24–25. Trina argues that the

differences between solar frames and the products contained within the subheadings used by the

Comtrade data contains physical differences and applications. Trina Br. at 30. Finally, Canadian

Solar and Trina argue that Commerce misreads the IHS Report that it claims supports a finding of

monthly price fluctuations and that the Comtrade data alone shows monthly fluctuations. Canadian

Solar at 25–26; Trina Br. at 24–28. The government defends Commerce’s position and cites a

recent decision by the Court of Appeals for the Federal Circuit that recently upheld the use of HTS

heading 7604.29 for valuing solar frames in the surrogate value context. Gov. Br at 18–29; see

also SolarWorld Americas, Inc. v. United States, 910 F.3d 1216, 1223 (Fed. Cir. 2018).

11 Commerce ceased using HTS subheading 7610.10 after finding that it was not a subheading under which aluminum solar frames are imported. See Remand Results, at 30–31. It now solely relies on HTS subheadings 7604.21 and 7604.29 from the Comtrade data. Id. at 31. Consol. Court No. 17-00198 Page 14 Public Version Commerce has failed to address the court’s concerns that the monthly fluctuations evinced

by the Comtrade data might be caused by fluctuations in the price other products encompassed in

the Comtrade headings unrelated to solar frames. Although Commerce’s inference from the IHS

data that monthly variations in price exist for solar frames is not unreasonable, its decision that the

Comtrade data is reflective of this fluctuation is unreasonable.

First, the IHS Report appears to indicate [[

]] while the Comtrade data shows the opposite trend. See IHS Report; Revised

Benchmarks and Final Rates Calculations for Trina Solar, Rem. P.R. 19, Rem. C.R. 6–9 (Dep’t

Commerce Apr. 26, 2019) (“Calculations”). While the IHS Report states that the price of

aluminum frames is [[

]], it does not necessarily follow that the HTS headings used in the Comtrade data are

reflective of this correlation. See id. at 13.

Second, the record indicates that solar frames [[

]] and while this does not necessarily mean that the aluminum extrusions in HTS

headings 7604.21 and 7604.29 are not comparable, it undercuts Commerce’s finding that they are

comparable. IHS Report. Rather than address evidence that contradicts Commerce’s ultimate

conclusion, Commerce simply states that there “is no evidence on the record that such differences

are significant enough to warrant abandoning the sole source of a monthly benchmark on the

record.” Remand Results at 71. Although the evidence that solar frames differ from aluminum

extrusions more generally is not definitive, it is enough to render unsupported by substantial

evidence Commerce’s dismissal of this concern without further analyzing the merchandise

captured by the HTS headings used by Comtrade. See Zhejiang DunAn Hetian Metal Co. v. United

States, 652 F.3d 1333, 1340 (Fed. Cir. 2011) (the court looks “to the record as a whole, including Consol. Court No. 17-00198 Page 15 Public Version evidence that supports as well as evidence that fairly detracts from the substantiality of the

evidence”). Commerce has not adequately accounted for “factors affecting comparability.” 19

C.F.R. § 351.511(a)(2)(ii). 12

A preference for monthly values cannot overcome data that does not reasonably relate to

the product at issue. Accordingly, the court again remands Commerce’s decision to use the

Comtrade data in computing a benchmark. Commerce must use the IHS data alone in computing

the benchmark. Unless Commerce can demonstrate, however, that the HTS subheadings used by

Comtrade are not grossly overinclusive and determines that the merchandise is sufficiently

comparable to solar frames.

IV. Use of Comtrade/IHS data in Computing a Benchmark for Solar Glass

After finding the provision of solar glass to be a countervailable subsidy, Commerce

constructed a benchmark based on an average of Comtrade and IHS data.13 See Changzhou Trina

I, 352 F. Supp. 3d at 1333. As with the aluminum extrusions benchmark, the court remanded after

finding the Comtrade data potentially overinclusive of non-subject merchandise. Id. at 1334–35.

The court stressed the importance of accounting “for factors affecting comparability.” See id. at

1335 (citing 19 C.F.R. § 351.511(a)(2)(ii)). Finally, the court expressed concern that the Comtrade

data did not include information from major solar glass producing countries. Id.

12 The decision in Solarworld Americas, does not control here as it involved surrogate values, which are inherently less reliable and involve differing procedural and factual considerations. See 910 F.3d at 1222–25. 13 The Comtrade data was not specific to solar glass, which Commerce recognizes is a type of glass with unique properties. See Remand Results at 27, 32–33. Instead the Comtrade data included glass under the U.S. Harmonized Tariff Schedule (“HTS”) headings 7007.19 and 7007.29, which includes other types of non-solar tempered or laminated glass. Id. at 32–34. In contrast, the IHS dataset is specific to solar glass. Id. at 33. Consol. Court No. 17-00198 Page 16 Public Version On remand, Commerce continues to average the Comtrade and IHS datasets, finding that

neither is sufficient on its own to compute a benchmark price. Commerce did, however, remove

HTSUS heading 7007.29 from the Comtrade benchmark, recognizing that it is “an overbroad

representation of the price of solar glass on the record of this case.” Id. at 34. 14 To support its

continued use of the Comtrade data, Commerce points to record evidence showing that the price

for solar glass fluctuates year-to-year. Id. at 34. It extrapolates that as prices fluctuate on a year-

to-year basis, fluctuation is “likely to occur on a month-to-month basis.” Id. Usage of the Comtrade

data is suitable, Commerce claims, as it is the only data on record that can “provide any evidence

of how such fluctuations might have occurred on a month-to-month basis.” Id.

Canadian Solar and Trina continue to find the use of the Comtrade data problematically

overinclusive. Canadian Solar Br. at 31–35; Trina Br. at 34–40. Canadian Solar notes that the

Comtrade data is also underinclusive in that it fails to contain data from at least two major solar

glass producing countries. Canadian Solar at 32–33. The government responds that use of the

Comtrade data is appropriate, especially now that Commerce removed HTS 7007.29 data. Gov.

Br. at 29–34. It argues that solar glass is similar in “in design, core function, and purpose” to other

glass categorized under HTS heading 7007.19, such that the Comtrade data is probative of the

price of solar glass. Id. at 32–33.

Commerce has not adequately addressed the court’s concern that the Comtrade data’s

monthly fluctuations may be caused by non-solar glass merchandise. See Changzhou Trina I, 352

F. Supp. 3d at 1333–34. In Changzhou Trina I, the court faulted Commerce for using the Comtrade

data based on purported price fluctuations of solar glass, when the only evidence of price

14 Commerce also removed Comtrade data from countries that were shown to have not produced solar glass. See Remand Results at 35–36. Consol. Court No. 17-00198 Page 17 Public Version fluctuations appeared to be the challenged Comtrade dataset. Id. at 1335. Although Commerce

provides evidence that the price of solar glass fluctuates on a year-to-year basis, this does not

necessarily mean that the monthly fluctuations in the Comtrade data set are caused by solar glass

rather than non-solar glass. In fact, the IHS Report that Commerce cites to support the notion that

the price for solar glass fluctuates year-to-year [[

]], but the Comtrade data shows the opposite trend. Compare

IHS Report with Calculations (listing the monthly Comtrade data points). Thus, the data cited by

Commerce as evidence that monthly data needs to be included undermines the inclusion of the

Comtrade data and supports respondents’ contention that non-solar glass accounts for the price

variation.

Further, Commerce failed to explain whether the inclusion of non-solar glass in the

Comtrade data set made it unusable. The Remand Results do not take into account the factors of

comparability required of its regulations. See 19 C.F.R. § 351.511(a)(2)(ii). Although the glass

covered by HTS heading 7007.19 may have similarities to solar glass, this does not adequately

address the court’s concern that the fluctuations in the Comtrade data may be due to fluctuations

in the price of the non-solar glass products in that subheading. Finally, Commerce made no effort

to address concerns that the Comtrade data failed to include data from major solar glass-producing

countries. See Changzhou Trina I, 352 F. Supp. 3d at 1335. Commerce’s contention that the

countries included in the Comtrade data set “are not on the ‘non-producer’” country list provided

by Canadian Solar does not sufficiently address this deficiency. See Remand Results at 35–36.

Even if every country included in the Comtrade data did in fact produce solar glass, the lack of

data from major producers further undermines its use in constructing a solar glass benchmark.

Because the Comtrade data is fatally overinclusive of non-solar glass and underinclusive Consol. Court No. 17-00198 Page 18 Public Version of data from countries with major solar glass producers, its usage in deriving a benchmark is

unsupported by substantial evidence. See 19 U.S.C. § 1516a(b)(1)(B)(i). Thus, the court remands

this issue and directs Commerce to use the IHS data alone in computing a solar glass benchmark.

In the alternative, if Commerce chooses to reopen the record because it has identified a dataset that

is both specific to solar glass and computed on a monthly basis, it should use that dataset in

computing the benchmark.

V. Commerce’s Rejection of Canadian Solar’s Import Pricing Data in Computing a Benchmark Price for Polysilicon

After finding that the provision of polysilicon for LTAR was a countervailable subsidy,

Commerce estimated adequate remuneration using a tier-two metric claiming that the GOC’s

involvement in the polysilicon market distorted domestic prices, thus rendering unusable tier-one

prices based on a “market-determined price for the good or service resulting from actual

transactions in the country in question.” See 19 C.F.R. § 351.511(a)(2)(i)–(ii); I& D Memo at 31.

Canadian Solar contested Commerce’s decision to resort to tier two data, claiming that its “arms-

length imports with market-economy suppliers” should have been considered as tier-one data.

Changzhou Trina I, 352 F. Supp. 3d at 1336–37. The court concluded that Commerce failed to

explain how the GOC’s market participation resulted in import distortion and remanded for

Commerce to either explain or else use the import data as a tier-one metric. Id. at 1336. The court

noted that while market interference could result in import price depression, without “sufficient

information about polysilicon’s fungibility or the dynamics of the market,” the court could not

conclude that Commerce’s decision was supported by substantial evidence. Id. at 1337.

On remand, Commerce continues to resort to tier-two metrics, finding that the GOC’s

participation in the market forced importers to lower their prices to compete in the domestic

market. Remand Results at 36–39. Commerce calls this a “basic economic inference drawn from Consol. Court No. 17-00198 Page 19 Public Version the logic of a competitive marketplace.” Id. at 38. It claims that this inference is reasonable when,

as here, domestic producers supply most of the polysilicon domestically consumed. 15 Commerce

further noted that there was “no information on the record indicating that imports and domestic

purchases are not fungible.” Id.

Canadian Solar claims that Commerce confuses the analysis. In its estimation, what matters

is not whether the domestic production accounts for the majority of domestically consumed

polysilicon, but whether the government involvement is so great as to manipulate the market.

Canadian Solar Br. at 29–30. It notes that this occurs when the “‘government provider’ constitutes

a majority of the market.” Id. at 29. Canadian Solar claims that record evidence shows that this is

not the situation in the GOC. Id. at 29–30. The government responded that Canadian Solar is

attempting to re-litigate Commerce’s finding that the market in the PRC is distorted by government

participation. Gov. Br. at 41–42.

After reviewing the record, the court finds Commerce’s rationale in rejecting Canadian

Solar’s import data unsupported. The GOC submitted information about its involvement in the

polysilicon industry, but noted that it did not track the solar-grade polysilicon specifically.

Commerce found that this made the GOC’s responses unreliable such that an adverse inference

was warranted. See I & D Memo, at 31; Prelim. I & D Memo, at 25–27. 16 Accordingly, Commerce

15 Specifically, Commerce cites record evidence supplied by the GOC that shows domestic producers supply 66 percent of domestic consumption and imports account for the remaining 34 percent. Remand Results at 38 (citing GOC Initial CVD Questionnaire Resp. at 73–74). 16 In the preliminary issues and decisions memorandum, Commerce listed other documents relevant in making its determination regarding the polysilicon industry including a WTO Dispute Settlement Panel Determination, as cited by the petition in the underlying antidumping and countervailing duty investigation on solar cells from the PRC; a New York Times article; and an article on the production of polysilicon. See Prelim. I & D Memo at 26. The parties have not made any argument regarding these documents and it is unclear whether and how these materials influence Commerce’s finding that the solar-grade polysilicon market is distorted. Consol. Court No. 17-00198 Page 20 Public Version stated that “the GOC’s involvement in the PRC’s solar grade polysilicon market leads to

significantly distorted solar grade polysilicon prices in the PRC.” I & D Memo at 31. The record

shows that the GOC has an ownership or management interest in roughly [[ ]] of the

domestic polysilicon produced which equates to just [[ ]] of the total amount of

polysilicon domestically consumed, obviously a small percentage. See GOC Initial CVD

Questionnaire Resp. at 73–74. Although the court must accept Commerce’s findings so long as “a

reasonable mind might accept the evidence as sufficient to support the finding,” Maverick Tube

Corp. v. United States, 857 F.3d 1353, 1359 (Fed. Cir. 2017) (citations omitted), the court cannot

conclude that the evidence reasonably supports Commerce’s finding that the price of solar-grade

polysilicon imports is distorted by the GOC’s participation in the market.

Commerce “must explain the evidence which is available, and must offer a rational

connection between the facts found and the choice made.” See Motor Vehicle Mfrs. Ass'n of U.S.,

Inc. v. State Farm Mut. Auto. Ins., 463 U.S. 29, 42 (1983) (quotation and citation omitted). It has

not done so here. As noted in the Preamble; Countervailing Duties; Final Rule, unless a

“government provider constitutes a majority, or in certain circumstances, a substantial portion of

the market,” the effect on the market will normally be minimal. 63 Fed. Reg. 65,348, 65,377 (Dep’t

Commerce Nov. 25, 1998) (“Preamble”); see also, Maverick, 857 F.3d at 1362.

The government neither provides evidence to support that either of these circumstances

exist nor provides an explanation for why, even without these circumstances, the solar-grade

polysilicon market is so distorted as to depress the price of arm’s-length imports. Although it is

theoretically possible that the GOC’s influence over a small percentage of the general polysilicon

industry results in a majority control of the production of solar-grade polysilicon, the court

concludes that this possibility is too remote, without more, to serve as substantial evidence that Consol. Court No. 17-00198 Page 21 Public Version this influence disrupts import pricing. The court has, in different circumstances, given credence to

Commerce’s finding that market distortion makes import prices unreliable as a tier-one price. See

Archer Daniels, 917 F. Supp. 2d at 1343 (CIT 2013). In Archer Daniels, however, the finding was

based on the state’s controlling over half of domestic production of the input at issue and the

GOC’s imposition of an export tax on that input during the relevant period. Id.; see also Guangdong

Wireking Housewares & Hardware Co., v. United States, 900 F. Supp. 2d 1362, 1380–82 (CIT

2013) (ruling that where 47.97 percent of domestic production was state-controlled, imports only

comprised 1.53 percent of the domestic market, and export tariffs were in place, a finding of market

distortion was reasonable). Here, Commerce’s decision to resort to tier-two price information was

not reasonable. See Preamble at 65,377 (noting that “[w]here it is reasonable to conclude that

actual transaction prices are significantly distorted as a result of the government's involvement in

the market, we will resort to the next alternative in the hierarchy”).

Accordingly, the court must one again remand this issue. Commerce must either use

Canadian Solar’s proffered import data as a tier-one metric or else provide sufficient evidence

supporting Commerce’s contention that the GOC’s participation in the solar-grade polysilicon

industry renders this data unreliable.

VI. Electricity Subsidy Specificity

After the GOC failed to provide information required to assess whether electricity prices

were set in accordance with market principles, Commerce continued to find, resorting to facts

available with an adverse inference, that the provision of electricity in the PRC is a countervailable

subsidy. See I & D Memo at 40–41; Prelim. I & D Memo at 27–28. The court remanded for

Commerce to fully “explain how adverse inferences lead to the conclusion that the provision of

electricity in China is a countervailable subsidy.” Changzhou Trina I, 352 F. Supp. 3d at 1342. Consol. Court No. 17-00198 Page 22 Public Version Commerce now explains that despite the GOC’s claim that electricity prices are based on

market principles, the GOC refused to provide “key information” necessary to verify these claims.

Remand Results at 39–40. Commerce claims that the GOC refused to provide documents to clarify

why prices vary among provinces. Id. at 40. Commerce placed on the record, as evidence of

specificity, the Initiation Checklist from the investigation, which asserts that preferential electricity

rates are “limited to priority industries, such as the solar power industry.” Id. at 40–41. Thus,

Commerce finds that the program is de facto specific under 19 U.S.C. § 1677 (5A)(D)(iii)(i). 17 See

id. Commerce explains that the GOC’s failure to provide sufficient information regarding the

variation among provincial electricity pricing and how price adjustments occur between provinces

and the National Development and Reform Commission (“NDRC”), creates gaps in the record

justifying its use of AFA. Remand Results at 39–40; see also 19 U.S.C. § 1677e(b).

Canadian Solar claims that “[t]he only gap in the record has to do with variation among

provinces.” Canadian Solar Br. at 39. It argues that Commerce can confirm that electricity prices

do not vary on the basis of industry and that Commerce has insufficient support for the notion that

the solar industry disproportionately benefits from the subsidy. Id. at 38–42. Canadian Solar also

takes issue with Commerce’s citation to the Initiation Checklist for support. Id. at 41–42.

The court concludes that Commerce has identified potentially-material gaps in the record

that could allow Commerce to rely on facts otherwise available and draw adverse inferences based

on non-cooperation. Submissions by the GOC appear to indicate that the NDRC maintains some

input over provincial electricity pricing and it is unclear whether adjustments made by the NDRC

17 To impose countervailing duties, Commerce must show that an authority is providing a subsidy, that confers a benefit, and that is specific. See 19 U.S.C. § 1677(5). Canadian Solar challenges specificity only. See Canadian Solar Br. at 39. Consol. Court No. 17-00198 Page 23 Public Version are made in accordance with market principles as the GOC claims. See GOC Initial CVD

Questionnaire Response, at 95–102. When asked directly about how adjustments are made, the

GOC provided some documents, including price schedules, but Commerce maintains that these

documents do not clarify “whether preferential prices might be limited to certain industries or

enterprises,” Remand Results at 40. 18

For the first time in this administrative review, 19 Commerce states that the provision of

electricity is specific because it is limited to certain industries, rather than geographical regions.

See 19 U.S.C. § 1677 (5A)(D)(iii)(i). Commerce relies on information in the Initiation Checklist

as “facts available,” as Commerce is permitted to do. See 19 U.S.C. § 1677e(b)(2); 19 C.F.R. §

351.308(c). Commerce, however, has not added to the record the supposedly supportive

documentation it specifically relied on in the Initiation Checklist. 20 As noted above, in this case,

without the underlying exhibits relied upon as support, the court cannot evaluate Commerce’s

18 Commerce claims that the GOC failed to provide “price proposals for each of the relevant provinces that might demonstrate the provinces are setting prices and that they are setting prices in accordance with supply, demand, and cost; a detailed description of the cost elements and price adjustments that were discussed between the provinces and the NDRC; and, province- specific explanations linking particular costs to retail prices.” See Remand Results at 39–40. 19 The court has recently encountered the question of whether the provision of electricity in the PRC is a specific subsidy, in a case involving a challenge to the final determination and countervailing duty order on certain aluminum foil from the PRC. See Jiangsu Zhongji Lamination Materials Co. v. United States, Slip Op. 19-122, 2019 WL 4467099 (CIT Sept. 18, 2019). In that case, the court held that Commerce relied in part on submissions from the GOC that indicated that price variations were at least in part due to type of use. Id. at *11–12. The court upheld Commerce’s finding in that case given the substantial evidence supporting its decision. Although sufficient evidence may exist in this case, Commerce has not yet put forth such evidence.

20 The Remand Results cite to the Initiation Checklist, which in turn lists documents supporting the allegation that solar cell producers benefit from subsidized electricity rates. See Remand Results at 41 (citing Initiation Checklist at 12–13). Those supporting documents have not been placed on the record. Consol. Court No. 17-00198 Page 24 Public Version specificity finding. See Deacero, 393 F. Supp. 3d at 1286. Further, as indicated Commerce does

not state how its finding that the GOC is subsidizing specific industries relates to the gap created

by the GOC’s non-cooperation, so that it may draw the adverse inference that specific industries

are subsidized. The court surmises that Commerce understands that the subsidization of these

specific industries is, at least in part, a reason for the price variation among provinces and because

of the gaps in the record it cannot determine exactly why or how that occurs. On remand,

Commerce should expressly set forth its reasoning under the statutory steps for drawing adverse

inferences to fill record gaps.

CONCLUSION

For the foregoing reasons this matter is remanded for proceedings consistent with this

opinion. Commerce may reopen the record and supplement it as necessary. Remand results should

be filed by January 7, 2020. Objections are due February 6, 2020 and Responses to Objections are

due February 20, 2020.

____/s/Jane A. Restani_____ Jane A. Restani, Judge

Dated: November 8, 2019 New York, New York