Hyundai Electronics Co., Ltd. v. United States

Slip. Op. 99 - 44

UNITED STATES COURT OF INTERNATIONAL TRADE

BEFORE: RICHARD W. GOLDBERG, JUDGE



HYUNDAI ELECTRONICS CO., LTD. AND
HYUNDAI ELECTRONICS AMERICA, INC.,

Plaintiffs,

and

LG SEMICON CO., LTD. AND
LG SEMICON AMERICA, INC.,

Plaintiffs,

v.
Cons. Ct. No. 97-08-01409

THE UNITED STATES,

Defendant,

and

MICRON TECHNOLOGY, INC.,

Defendant-Intervenor.


[Plaintiffs’ Motions for Judgment on the Agency Record, contesting
a U.S. Department of Commerce decision not to revoke an antidumping
duty order, are denied.]

Dated: May 19, 1999

Powell, Goldstein, Frazer & Murphy LLP (Lawrence R. Walders)
for plaintiffs Hyundai Electronics Co., Ltd., and Hyundai
Court No. 97-08-01409 Page 2

Electronics America, Inc.

Kaye, Scholer, Fierman, Hays & Handler, LLP (Michael P. House,
Raymond Paretzky, and R. Will Planert), for plaintiffs LG Semicon
Co., Ltd., and LG Semicon America, Inc.

David W. Ogden, Acting Assistant Attorney General; David M.
Cohen, Director, Commercial Litigation Branch, Civil Division,
United States Department of Justice (Kenneth S. Kessler); Office of
the Chief Counsel for Import Administration, United States
Department of Commerce (Duane W. Layton), of counsel, for
defendant.

Hale & Dorr, LLP (Gilbert B. Kaplan, Michael D. Esch, Paul W.
Jameson, and Cris R. Revaz), for defendant-intervenor Micron
Technology, Inc.

OPINION

GOLDBERG, Judge: In this action, the Court reviews a decision by

the U.S. Department of Commerce ("Commerce") not to revoke an

outstanding antidumping ("AD") order on dynamic random access

memory semiconductors ("DRAMs") from Korea. Plaintiffs, LG

Semicon Co., Ltd. and LG Semicon America, Inc. (collectively "LG
Semicon"), and Hyundai Electronics Co., Ltd. and Hyundai

Electronics America, Inc. (collectively "Hyundai"), are Korean

producers of the subject merchandise and seek relief from

Commerce’s action under USCIT Rule 56.2. During the underlying

administrative proceeding, plaintiffs separately asserted that

the regulatory requirements for revocation had been met, and
Court No. 97-08-01409 Page 3

requested that Commerce revoke the outstanding AD order.

Commerce rejected each invitation in its Notice of Final Results

of Antidumping Duty Administrative Review: Dynamic Random Access

Memory Semiconductors of One Megabit or Above From the Republic

of Korea, 62 Fed. Reg. 39,809 (July 24, 1997) ("Final Results").

Plaintiffs contest this determination as both contrary to law and

unsupported by substantial evidence.

The Court exercises jurisdiction to review the motions for

judgment on the agency record pursuant to 28 U.S.C. § 1581(c)

(1994). The Court sustains the Final Results.

I.
BACKGROUND

Micron Technology, Inc. ("Micron"), a U.S. manufacturer of

DRAM semiconductors, filed a petition with Commerce in April,

1992, alleging that Korean producers were selling DRAMs in the
United States at less than fair value. Following an antidumping

investigation, Commerce published an AD order on DRAMs from Korea

in May, 1993. See 58 Fed. Reg. 27,520 (May 10, 1993).

In the first anniversary month of the AD order, plaintiffs

and Micron both requested an administrative review. Commerce

found de minimis dumping margins for both plaintiffs in this
Court No. 97-08-01409 Page 4

review. See Notice of Final Results of Antidumping

Administrative Review of DRAMs from the Republic of Korea, 61

Fed. Reg. 20,216 (May 6, 1996).1 In the second anniversary

month, the parties requested another administrative review, and

Commerce again found de minimis dumping margins for both LG

Semicon and Hyundai. See Notice of Final Results of Antidumping
Administrative Review of DRAMs from the Republic of Korea, 62

Fed. Reg. 965 (Jan. 7, 1997).2

1
In January, 1998, this Court remanded to Commerce one
aspect of the first review that impacts LG Semicon. See Micron
Tech., Inc. v. United States, No. 9912 (CIT Jan. 28, 1999).
Commerce filed its remand results on March 31, 1999. Although
the Court has yet to take action on this remand determination,
the Court notes that Commerce again found LG Semicon’s margins
de minimis.
2
As with the first administrative review, this Court
remanded to Commerce one aspect of the second administrative
review pertaining to LG Semicon. See Micron Tech., Inc. v.
United States, No. 99-29 (CIT Mar. 25, 1999). As a result of
Commerce’s decision on remand, it is foreseeable that LG
Semicon’s margins for the second review period will exceed de
minimis levels. If this is the case, then LG Semicon will not
have had de minimis dumping margins for three consecutive years,
the first of three criteria for revocation, see 19 C.F.R.
§ 353.25(a)(2), and its arguments on appeal will become moot.
The remand results for the second review will not become final,
however, until the Court sustains the results and until the time
for appeal has run. In addition, no party to this action has
moved to stay these proceedings to await Commerce’s remand
findings. And, the Court notes that stays pending an appeal or
other judicial proceeding are an "extraordinary and disfavored
measure." See Gerald Metals, Inc. v. United States, 22 CIT __,
Court No. 97-08-01409 Page 5

In the third anniversary month of the order, plaintiffs

requested both a third annual review and a revocation in part of

the AD order, pursuant to 19 C.F.R. § 353.25(a)(2) of Commerce’s

regulations. Notice of Initiation of Administrative Review:

DRAMs from Korea, 61 Fed. Reg. 32,771 (June 25, 1996) (covering

the period May 1, 1995 through April 30, 1996). In pertinent

part, section 353.25(a)(2) provides that Commerce may revoke an

order if it concludes that

(i) One or more producers or resellers
covered by the order have sold the
merchandise at not less than foreign market
value for a period of at least three
consecutive years;

(ii) It is not likely that those persons will
in the future sell the merchandise at less
than foreign market value; and

(iii) ... the producers or resellers agree in
writing to their immediate reinstatement in
the order, as long as any producer or
reseller is subject to the order, if the
Secretary concludes ... that the producer or
reseller, subsequent to the revocation, sold
the merchandise at less than foreign market

__, 27 F. Supp.2d 1351, 1354 n.6 (1998) (citing Phillip Bros. v.
United States, 10 CIT 448, 453, 640 F. Supp. 261, 265 (1986)).
Thus, notwithstanding that LG Semicon’s suit potentially could
become moot, at this time it presents a genuine controversy.
Therefore, the Court issues this opinion with respect to LG
Semicon as well as Hyundai (to which the potential problem does
not apply).
Court No. 97-08-01409 Page 6

value.

19 C.F.R. § 353.25(a)(2) (1996).3

Commerce issued the preliminary results of its third review

on March 18, 1997. See Preliminary Results of Antidumping Duty

Administrative Review and Notice of Intent Not to Revoke Order:

Dynamic Random Access Memory Semiconductors of One Megabit or
Above From the Republic of Korea, 62 Fed. Reg. 12,794 (Mar. 18,

1997). Commerce again found de minimis dumping margins for both

plaintiffs during the third review period. Commerce

preliminarily determined not to revoke the AD order, however,

3
Because Commerce initiated the third review after
January 1, 1995, the agency conducted the review under the
antidumping law as amended by the Uruguay Round Agreements Act
("URAA"), Pub. L. No. 103-465, tit. II, 108 Stat. 4808, 4842
(1994). Among other things, the URAA revised certain
terminology, including substituting the term "normal value" for
the term "foreign market value." Yet, at the time Commerce
initiated the third review in June, 1996, Commerce’s revised
regulations, intended to reflect the URAA amendments, had yet to
take effect. See Notice of Final Rule: Antidumping and
Countervailing Duties, 62 Fed. Reg. 27,296 (May 19, 1997) (noting
that June 18, 1997 was the effective date of the new
regulations). Therefore, the 1996 regulations in effect during
the underlying proceeding employed the pre-URAA terminology,
i.e., "foreign market value," and that language is used
throughout this opinion. Also, all parties to the action concede
that the URAA did not alter the revocation statute nor did
Commerce alter the substantive criteria in its revocation
regulation. Currently, the regulation that governs revocation
may be found at 19 C.F.R. § 351.222(b)(2) (1998).
Court No. 97-08-01409 Page 7

because in its view plaintiffs each failed to meet the second of

the three revocation criteria. That is, plaintiffs failed to

satisfy Commerce that they were "not likely" to dump in the

future. Commerce based its preliminary determination in part on

evidence submitted by Micron regarding market trends during 1996.

Importantly, Micron’s data included information for the period

after April 30, 1996, i.e., the last day covered under the third

administrative review. In the Final Results, Commerce affirmed
its decision not to revoke the AD order. See 62 Fed. Reg. at

39,811.

Plaintiffs challenge Commerce’s decision not to revoke the

order, alleging that it was neither in accordance with law nor

supported by substantial evidence. More specifically, both

plaintiffs assert that, as a matter of law, Commerce did not

apply the proper standard to determine the likelihood of future
dumping nor did it use data from an appropriate time period when

it applied its faulty standard. Hyundai further challenges

Commerce’s characterization of the DRAM market and of Hyundai’s

behavior in that market as inconsistent with the record evidence.

LG Semicon also challenges various conclusions regarding LG

Semicon’s current and future activities in the DRAM market as
Court No. 97-08-01409 Page 8

unsupported by record evidence.

Commerce opposes all of plaintiffs’ challenges, as does

Micron.

II.
STANDARD OF REVIEW
Commerce’s determination will be sustained if it is

supported by substantial evidence on the record and is otherwise

in accordance with law. See 19 U.S.C. § 1516a(b)(1)(B) (1994).
To determine whether Commerce’s interpretation of the

statute is in accordance with law, the court applies the two-

prong test set forth in Chevron U.S.A., Inc. v. Natural Resources

Defense Council, Inc., 467 U.S. 837 (1984). Chevron first

directs the court to determine "whether Congress has directly

spoken to the precise question at issue." Id. at 842-43

(internal quotations omitted). In doing so, the court must

inquire "whether Congress’s purpose and intent on the question at
issue is judicially ascertainable." Timex V.I., Inc. v. United

States, __ Fed. Cir. (T) __, __, 157 F.3d 879, 881 (1998) (citing

Chevron, 467 U.S. at 842-43 & n.9); see also Micron Tech., Inc.

v. United States, No. 99-12, slip op. at 4-5 (CIT Jan. 28, 1999).

Congress’s purpose and intent must be divined using the

traditional tools of statutory construction. Timex, 157 F.3d at
Court No. 97-08-01409 Page 9

882 (citation omitted). Of course, the "first and foremost tool

to be used is the statute’s text," and "if the text answers the

question, that is the end of the matter." Id. (citations and

internal quotation omitted). In addition to the plain language

of the statute, the other tools include the statute’s structure,

canons of statutory interpretation, and legislative history. See
id. (citing Dunn v. Commodity Futures Trading Comm’n, 117 S.Ct.

913, 916-20 (1997)); Chevron, 467 U.S. at 859-63; Oshkosh Truck

Corp. v. United States, 123 F.3d 1477, 1481 (Fed. Cir. 1997)).

If, using these tools, Congress’s intent is unambiguous as to the

issue at hand, the court must give effect to that intent.

On the other hand, if Congress’s intent is "silent or

ambiguous with respect to the specific issue, the question for

the court is whether the agency’s answer is based on a

permissible construction of the statute." Chevron, 467 U.S. at
843. Thus, the second prong of the Chevron test directs the

court to consider the reasonableness of an agency’s

interpretation.

If asked to review Commerce’s factual findings, the court

will uphold the agency if its findings are supported by

substantial evidence. "Substantial evidence is something more
Court No. 97-08-01409 Page 10

than a ‘mere scintilla,’ and must be enough reasonably to support

a conclusion." Ceramica Regiomontana, S.A. v. United States, 10

CIT 399, 405, 636 F. Supp. 961, 966 (1986), aff’d, 5 Fed. Cir.

(T) 77, 810 F.2d 1137 (1987). In applying this standard, the

court sustains Commerce’s factual determinations so long as they

are reasonable and supported by the record as a whole, even if

there is some evidence that detracts from the agency’s

conclusions. See Atlantic Sugar, Ltd. v. United States, 2 Fed.
Cir. (T) 130, 138, 744 F.2d 1556, 1563 (1984).

III.
DISCUSSION
In the discussion below, the Court first reviews plaintiffs’

argument that Commerce’s application of the "not likely" standard

was not in accordance with law. The Court then considers

plaintiffs’ claim that Commerce unlawfully examined an
inappropriate time period to make its "not likely" determination.

In both instances, the Court rejects plaintiffs’ arguments. The

Court concludes its discussion by considering plaintiffs’ claims

that Commerce’s decision not to revoke the order was unsupported

by substantial evidence. Here, too, the Court sustains

Commerce’s determination.
Court No. 97-08-01409 Page 11

A. "Not Likely" Standard

Plaintiffs first argue that Commerce applied an improper

standard to determine whether plaintiffs were "not likely" to

dump in the future. Each plaintiff advances a different theory,

however. Hyundai contends that since Commerce adopted 19 C.F.R.

§ 353.25(a)(2)(ii) in 1989, the agency has granted revocation in

virtually every case where a respondent has established three

consecutive years of no dumping and has furnished the required

agreements. See Pl.’s (Hyundai’s) Br. in Supp. of Mot. for J. on
Agency R. at 9 n.5 ("Hyundai’s Br.") (citing fourteen

determinations between 1990 and mid-1997 where Commerce granted

revocation). Hyundai thus maintains that respondents who meet

these criteria should not be denied revocation barring extremely

unusual circumstances, which it asserts do not exist in this

case.
In addition, Hyundai argues that Commerce’s final

determination in this case imposes a nearly impossible standard

for revocation of AD orders on merchandise sold in cyclical

markets. Hyundai asserts that Commerce presumes market downturns

in cyclical industries always lead to dumping. Specifically,

Hyundai points to the Final Results, where Commerce stated that
Court No. 97-08-01409 Page 12

"[t]he DRAM industry is highly cyclical in nature with periods of

sharp upturn and downturn in market prices," and then noted that

"because DRAMs are currently a commodity product, DRAM

producers/resellers must price aggressively during a downturn

period in order to stay competitive and maintain their customer

base." Final Results, 62 Fed. Reg. at 39,810. Commerce then

went on to find that "it is reasonable to conclude that

information regarding the selling activities and pricing patterns

of the respondents, as well as other market conditions, during

periods of significant downturn are relevant to whether dumping

is not likely to occur in the future." Id. According to

Hyundai, if the assumption that market downturns automatically

lead to dumping is accepted, it will effectively make it

impossible to revoke AD orders on products sold in cyclical

markets because, by definition, there will be market downturns.
LG Semicon also challenges Commerce’s use of the "not

likely" standard in the instant case, although on different

grounds. LG Semicon argues Commerce ignored the plain meaning of

the regulation when it found LG Semicon did not meet the "not

likely" criterion. LG Semicon asserts that future dumping is

"not likely" if the chance that respondent will not dump in the
Court No. 97-08-01409 Page 13

future is greater than the chance that respondent will dump in

the future. Therefore, LG Semicon contends Commerce should

revoke an AD order if it finds at least a 51% chance that

respondent will not dump in the future. And LG Semicon argues

that, contrary to this proposed standard, Commerce applied a more

rigorous test to ascertain whether it was "not likely"

respondents would dump in the future. In doing so, LG Semicon

maintains Commerce’s decision was contrary to law.

The Court first addresses the standard under which these

arguments are reviewed. In Section 751(c) of the Trade

Agreements Act of 1979, Pub. L. No. 96-39, tit. I, § 101, 93

Stat. 144, 176 (1979), Congress provided that Commerce, as the

administering authority, "may revoke, in whole or in part, a

countervailing duty order or an antidumping duty order or

finding, or terminate a suspended investigation, after [an

administrative review]". See 19 U.S.C. § 1675(d)(1). Although

Congress offered no further guidance to assess when revocation

might be appropriate, Commerce has promulgated its own series of

regulations to fulfill the statutory mandate.4 In this case,

4
Commerce published its initial regulation to guide
operation of the revocation statute in 1980. See Final Rule:
Antidumping Duties, 45 Fed. Reg. 4932 (Feb. 6, 1980). Then, as
Court No. 97-08-01409 Page 14

neither plaintiff contends that section 353.25(a)(2) itself is at

odds with the statute’s intent. Rather, plaintiffs contest

Commerce’s application of the regulation. Thus, the Court must

consider whether Commerce’s actions were reasonable.

i. Commerce is Not Automatically Required to Revoke an
Antidumping Order Where Respondent Has Not Dumped for
Three Consecutive Years and Has Furnished the Requisite
Agreements.

Hyundai maintains that except in extraordinary cases,

Commerce should automatically revoke an AD order when respondent

can show three years of no dumping and has furnished the required

no-dumping agreements. This argument lacks merit. Hyundai’s

assertion ignores both the language of the statute and the

regulation. First, the statute expressly states that Commerce

"may revoke, in whole or in part, . . . an antidumping duty

order." 19 U.S.C. § 1675(d)(1) (emphasis added). Similarly, the

regulation provides that Commerce "may revoke an order in part"
if the three criteria for revocation have been met. 19 C.F.R. §

353.25(a)(2) (emphasis added). The use of the word "may" affords

Commerce the discretion not to revoke an order even if all three

noted earlier, Commerce issued revised regulations in 1989. See
supra text accompanying note 3. The current regulation that
governs revocation can be found at 19 C.F.R. § 351.222(b)(2).
Court No. 97-08-01409 Page 15

criteria are satisfied. Indeed, the court has held that the pre-

1989 incarnation of the regulation, 19 C.F.R. § 353.54 (1988),

which similarly stated that Commerce "may act to revoke" when

certain conditions were met, indicates that "Commerce is not

compelled to grant revocation" even where plaintiffs satisfy the

requirements. Toshiba Corp. v. United States, 12 CIT 455, 463,

688 F. Supp. 617, 623 (1988), aff’d, 7 Fed. Cir. (T) 13, 861 F.2d
257 (1988); see also Tatung Co. v. United States, 18 CIT 1137,

1144 (1994) (finding that the "second requirement for revocation,

that the respondent is not likely to resume dumping, necessarily

involves an exercise of discretion and judgment").

Second, the regulation plainly establishes a three-part test

for revocation, not a two-part test. Contrary to Hyundai’s

argument, the second prong of the regulation, i.e., the "not

likely" exercise, is an independent criterion that must be
established to Commerce’s satisfaction to attain revocation.

Thus, the plain language of the statute and the regulation

refutes Hyundai’s argument that Commerce acted contrary to law

when it denied revocation even though respondent established

three consecutive years of no dumping and furnished the requisite

agreements.
Court No. 97-08-01409 Page 16

ii. Commerce’s Application of the "Not Likely" Standard
Allows for the Possibility That an Outstanding Order on
a Product Sold in a Cyclical Market May Be Revoked.

Hyundai next asserts that Commerce’s application of the "not

likely" standard makes it impossible for an AD order to be

revoked in a cyclical industry. The record and Commerce’s past

practice belie this argument. First, Commerce did not simply

decide that because the DRAM industry is cyclical, Hyundai and LG

Semicon are likely to dump in the future; rather, in its effort

to predict whether resumption of dumping is not likely, Commerce

examined the history of companies in the DRAM industry  and this

industry only  to determine how those companies react to market

downturns. It noted that

[i]n the past, the DRAM industry has been characterized
by dumping during periods of significant downturn. For
instance, various foreign producers were found to have
dumped in the mid-1980s, and the Korean respondents in
this proceeding were found to have dumped in the less
than fair value investigation during 1991-1992, the
last period when there was a significant downturn in
the DRAM industry.

Final Results, at 39,810 (citation omitted). Commerce then

stated that "it is reasonable to conclude that information

regarding the selling activities and pricing practices of

respondents, as well as other market conditions, during periods

of significant downturn are relevant to whether dumping is not
Court No. 97-08-01409 Page 17

likely to occur in the future." Final Results, at 39,810

(emphasis added). Importantly, some of the other market

conditions Commerce considered were (1) spot pricing and company-

specific pricing data for 1996 and 1997, (2) production and

inventory data for 1996 and 1997, and (3) industry reports on the

general health of the DRAM industry for both 1996 and 1997. See
Final Results, at 39,817-19; see also discussion infra Section

III.C (finding that Commerce’s decision on this matter was based

on substantial evidence). Thus, summary review of the Final

Results demonstrates that Commerce did not base its "not likely"

decision on the mere fact that DRAMs are sold in a cyclical

market. Rather, Commerce considered respondents’ historical

selling and pricing behavior in addition to other market

conditions. Consequently, the record here does not support the

conclusion that Commerce will automatically find future dumping
is likely where there is a cyclical market.

Second, Commerce has, in at least one instance, revoked an

AD order on products sold in a cyclical market. In Frozen

Concentrated Orange Juice from Brazil, 56 Fed. Reg. 52,510 (Oct.

21, 1991), Commerce revoked an order even though the merchandise

at issue was commodity based and sold in a highly cyclical
Court No. 97-08-01409 Page 18

market. In that case, Commerce identified three years of price

fluctuations in a cyclical market but discerned no correlative

trend between dumping and market downturns. Id. at 52,511.

Thus, Commerce has revoked an AD order involving a cyclical

industry in the past, and nothing in the instant case indicates

that Commerce is precluded from doing so in the future.

iii. The Plain Language of the Regulation Does Not Require
Commerce to Use a Set Numerical Threshold to Assess
Whether Future Dumping is "Not Likely."
Turning to LG Semicon’s argument, the Court finds that

Commerce’s application of the "not likely" standard did not

contravene the regulation’s plain language. LG Semicon maintains

the phrase "not likely" requires Commerce to find a greater-than-

fifty-percent chance that dumping will occur before it may deny a

request for revocation. The Court does not agree. The "not

likely" calculus is not a mathematical formula, but rather a
fact-intensive, case-by-case determination. Within reason,

Commerce has the discretion to apply the "not likely" standard as

it deems fit. And, as Commerce points out, the proposed greater-

than-fifty-percent standard would be difficult if not impossible

to administer. In addition, the court has held that Commerce

"need not affirmatively find that LTFV sales are likely[,] to be
Court No. 97-08-01409 Page 19

unsatisfied that there is no likelihood of LFTV sales." Toshiba

Corp. v. United States, 15 CIT 597, 600 (1991).5 In other words,

even if Commerce cannot or does not find that LTFV sales are

likely to occur, it can still, within its discretion, reject a

respondent’s claim that it is "not likely" to make LTFV sales in

the future. Therefore, Commerce reasonably exercised its

discretion when it applied the "not likely" standard in this

case.

iv. Commerce’s "Not Likely" Requirement is Consistent With
International Obligations.
In ancillary fashion, LG Semicon contends that Commerce’s

"not likely" standard is at odds with U.S. international

5
In 1989, Commerce substituted the phrase "not likely" for
"no likelihood." 19 C.F.R. § 353.25(a)(2)(ii) (1989). Yet, the
Toshiba case involved a pre-1989 revocation review and, hence,
the phrase "no likelihood" instead of "not likely" was
dispositive there. See 15 CIT at 600. When Commerce revised its
regulation in 1989, Commerce gave no explanation for the change
in phraseology; indeed, the agency has continued to use the two
phrases interchangeably in its rulings. See, e.g., Final Results,
62 Fed. Reg. at 39,81213; see also Wieland-Werke v. United
States, 22 CIT __, __, 4 F. Supp.2d 1207, 1211 (1998) (holding
that the phrases "no likelihood" and "not likely" have the same
meaning in determining whether exporters will resume dumping).
While the Court here has no occasion to revisit whether
distinctions between the two phrases result in variant standards,
to avoid confusion, the Court urges Commerce to refrain from
using the "no likelihood" phrase in future revocation decisions
when it means "not likely."
Court No. 97-08-01409 Page 20

obligations.6 Although not adequately addressed by the parties,

this issue merits more than cursory analysis due in large measure

to a World Trade Organization ("WTO") panel report that addressed

the same question.7

When asked to review the same underlying administrative

decision, a WTO dispute settlement panel recently found that

Commerce’s "not likely" requirement violates WTO rules. See WTO
Dispute Panel Report: United States  Anti-Dumping Duty on DRAMs

of One Megabit or Above from Korea, 1999 WL 38403 (WTO Jan. 29,

1999, adopted March 19, 1999) ("Korean DRAMs WTO Report").

6
In a footnote to its principal brief, LG Semicon argued
that Commerce violated U.S. international obligations through its
application of the "not likely" standard. See Pl.’s (LG
Semicon’s) Br. in Supp. of Mot. for J. on Agency R. at 14 n.4
("LG Semicon’s Br."). For its part, Hyundai only alludes to U.S.
international obligations in its principal brief when it argues
that Commerce’s alleged policy of automatic revocation after
three years of no dumping is consistent with its international
obligations. See Hyundai’s Br. at 9. Only in its reply brief
does Hyundai make the express claim that Commerce’s "not likely"
standard is inconsistent with international obligations -- and
even then it does so in just one sentence. See Pl.’s (Hyundai’s)
Reply Br. in Supp. of Mot. for J. on Agency R. at 6.
7
Although the WTO issued its report after the briefing
period concluded in the case at bar, no party moved to file
supplemental briefs. Nevertheless, the Court takes judicial
notice of the fact that the WTO panel published a report
examining the same issue because it is particularly relevant to
the scope of U.S. international obligations.
Court No. 97-08-01409 Page 21

Although the WTO panel rejected the "not likely" approach, it

declined to suggest that the United States should act to revoke

the Korean DRAMs order. Rather, the panel concluded that the

United States has a "range of possible" options to implement its

recommendation. Korean DRAMs WTO Report, 1999 WL 38403, at *151.

For purposes here, however, the salient point is that the

WTO panel found the "not likely" standard inconsistent with

Article 11.2 of the WTO’s Agreement on Implementation of Article

VI of the 1994 General Agreement on Tariffs and Trade (the

"Antidumping Agreement"). Article 11.2 provides as follows:

The authorities shall review the need for continued
imposition of the duty, where warranted, on their own
initiative or, provided that a reasonable period of
time has elapsed since the imposition of the definitive
antidumping duty, upon request by any interested party
which submits positive information substantiating the
need for a review. Interested parties shall have the
right to request the authorities to examine whether the
continued imposition of the duty is necessary to offset
dumping, whether the injury would be likely to continue
or recur if the duty were removed or varied, or both.
If, as a result of the review under this paragraph, the
authorities determine that the antidumping duty is no
longer warranted, it shall be terminated immediately.

Antidumping Agreement, art. 11.2 (emphasis added). In essence,

the WTO panel concluded that the "not likely" approach results in

a more rigorous standard for respondents than the "likely"

approach embodied in Article 11.2. Korean DRAMs WTO Report, 1999
Court No. 97-08-01409 Page 22

WL 38403, at *141. More precisely, the panel reasoned as

follows:

6.45 We consider that a failure to find that an event
is "not likely" is not equivalent to a finding that the
event is "likely." We see a clear conceptual
difference between establishing something as a positive
finding, and failing to establish something as a
negative finding. It is perfectly possible that one
could not determine that someone was unlikely to dump
and find that they were also likely to dump. But the
former determination does not, in and of itself, amount
to a demonstrable basis for concluding the latter.
This is evident from the fact that the former finding
is manifestly compatible also with the reverse of the
latter situation, i.e., it is perfectly logical to find
that you cannot determine that someone is unlikely to
dump, yet also be unable to determine that they were
actually likely to dump. In other words, determining
something is not "not likely" is entailed by, but does
not itself entail, that something is likely.

6.46 . . . .

6.47 Given this reality, it is a priori possible that
situations could arise where the not "not likely"
criterion is satisfied but where the likelihood
criterion is not satisfied. Reliance on the not likely
criterion clearly fails to provide any reliable means
to avoid or preclude this flaw. Given such a
fundamental flaw, it cannot constitute a demonstrable
basis for consistently and reliably determining that
the likelihood criterion is satisfied.

Korean DRAMs WTO Report, 1999 WL 38403, at *141-42.

As an initial matter, the WTO report itself has no binding

effect on the court. In Footwear Distributors and Retailers of

America v. United States, 18 CIT 391, 852 F. Supp. 1078 (1994),
Court No. 97-08-01409 Page 23

the court was confronted with a claim that an adopted GATT panel

decision should govern the outcome of the case. Upon thorough

review, the Footwear Distributors court reasoned that the

response to a panel report is the prerogative of the executive

branch, not the judiciary, because it implicates political

decisions. Id. at 414, 852 F. Supp. at 1096. "[T]he courts

traditionally refrain from disturbing the ‘very delicate, plenary

and exclusive power of the [executive] as the sole organ of the

federal government in the field of foreign relations.’" Id.
(quoting United States v. Curtiss-Wright Export Corp., 299 U.S.

304, 320 (1936)). The Footwear Distributors court therefore

concluded that "[h]owever cogent the reasoning of the GATT

panel[]" judicial relief from this court does not attach. Id.

While Footwear Distributors concerned an adopted GATT panel

report, the same principles apply to the WTO report at issue
here. Most importantly, Congress made this clear when it

codified the principles espoused in Footwear Distributors as part

of the URAA. Specifically, Congress provided that the response

to an adverse WTO panel report is the province of the executive

branch and, more particularly, the Office of the U.S. Trade

Representative. See URAA § 129 (codified as 19 U.S.C. § 3538).
Court No. 97-08-01409 Page 24

Thus, the WTO panel report does not constitute binding

precedential authority for the court. Of course, this is not to

imply that a panel report serves no purpose in litigation before

the court. To the contrary, a panel’s reasoning, if sound, may

be used to inform the court’s decision.

The Antidumping Agreement, on the other hand, is a different

matter. When it enacted the URAA, the Senate noted that the

Uruguay Round Agreements, including the Antidumping Agreement,

"are not self-executing and thus their legal effect in the United

States is governed by implementing legislation." S. Rep. No.

103-412, at 13 (1994); accord H.R. Rep. No. 103-826, pt. I, at 25
(1994). Nevertheless, as a signatory to the Uruguay Round

agreements, the United States has obligations under these

agreements irrespective of whether the agreements are self-

executing. See Federal-Mogul Corp. v. United States, __ Fed.
Cir. (T) __, __, 63 F.3d 1572, 1581 (1995) (noting that GATT

agreements, including the Uruguay Round agreements, are

international obligations); Footwear Distribs., 18 CIT at 410,

852 F. Supp. at 1093 (same). Indeed, the Statement of

Administrative Action to the URAA, H.R. Doc. No. 103-316 (1994),

at 669, provides that the URAA was "intended to bring U.S. law
Court No. 97-08-01409 Page 25

fully into compliance with U.S. obligations under [the Uruguay

Round] agreements." Accordingly, the Antidumping Agreement is

properly construed as an international obligation of the United

States.

When confronted with a conflict between an international

obligation and U.S. law, it is of course true that an unambiguous

statute will prevail over the international concern. See, e.g.,
Federal-Mogul, __ Fed. Cir. (T) at __, 63 F.3d at 1581;

Suramerica de Aleaciones Laminadas, C.A. v. United States, 10

Fed. Cir. (T) 74, 83, 966 F.2d 660, 668 (1992). Yet, absent

express language to the contrary, a statute should not be

interpreted to conflict with international obligations. This

time-honored canon of statutory construction was first applied by

Chief Justice Marshall in Murray v. Schooner Charming Betsy, 6

U.S. (2 Cranch) 64 (1804).
It has also been observed that an act of Congress ought
never to be construed to violate the law of nations, if
any other possible construction remains, and
consequently can never be construed to violate neutral
rights, or to affect neutral commerce, further than is
warranted by the law of nations as understood in this
country.

These principles are believed to be correct, and they
ought to be kept in view in construing the act now
under consideration.
Court No. 97-08-01409 Page 26

Id. at 118; see also Federal-Mogul, __ Fed. Cir. (T) at __, 63

F.3d at 1581; Footwear Distribs., 18 CIT at 408, 852 F. Supp. at

1091. And, Chevron must be applied in concert with the Charming

Betsy doctrine when the latter doctrine is implicated. See

DeBartolo Corp. v. Florida Gulf Coast Bldg. and Constr. Trades

Council, 485 U.S. 568, 574-75 (1988).

Here, Congress was silent on the mechanics of the revocation

procedure. Commerce acted to fill the void and, in doing so,

promulgated the "not likely" requirement. Meanwhile, as

described earlier, Article 11.2 of the Antidumping Agreement also

outlines procedures for revocation and provides that, in deciding

whether to revoke an outstanding order, a signatory may consider

whether dumping is "likely to continue or recur if the duty were

removed." Because Congress declined to enact procedures for

revocation, under the Charming Betsy doctrine, the Court must
consider whether Commerce formulated its regulation consistent

with Article 11.2 of the Antidumping Agreement.

Commerce fulfilled its statutory mandate in consonance with

U.S. international obligations. As a general matter, Article

11.2 of the Antidumping Agreement provides the administering

authority discretion to determine whether revocation is
Court No. 97-08-01409 Page 27

appropriate. See Antidumping Agreement, art. 11.2 (stating that

"[i]f, . . ., the authorities determine that the antidumping duty

is no longer warranted, it shall be terminated immediately"

(emphasis added)). It follows that the administering authority

also has discretion to determine whether injurious dumping would

be "likely" to occur in the future. And, as the WTO panel aptly

noted,

the certainty inherent to such a prospective analysis
could be conceivably somewhat less than that attached
to purely retrospective analysis . . . . this reflects
the fact that the necessity involved in Article 11.2 is
not to be construed in some absolute and abstract
sense, but as that appropriate to circumstances of
practical reasoning intrinsic to a review process.
Mathematical certainty is not required, but the
conclusions should be demonstrable on the basis of the
evidence adduced.

Korean DRAMs WTO Report, 1999 WL 38403, at *140-41. In short, an

analysis of whether dumping is "likely" or "not likely" to occur

in the future is inherently predictive. As a result,

operationally, Article 11.2 provides that an administering

authority has considerable discretion to make an inherently

predictive analysis.

The Court concedes that "it is a priori possible that

situations could arise where the not ‘not likely’ criterion is
Court No. 97-08-01409 Page 28

satisfied but where the likelihood criterion is not satisfied."

Korean DRAMs WTO Report, 1999 WL 38403, at *142. And, therefore,

there is not perfect overlap between the two standards. Yet, the

discretion afforded to predict the state of future dumping erases

any clear conflict between the two approaches. Indeed, the Court

is satisfied that, as applied, the discretionary authority to

make such a predictive analysis must result in general overlap

between the two approaches. So viewed, unless the conflict

between an international obligation and Commerce’s interpretation

of a statute is abundantly clear, a court should take special

care before it upsets Commerce’s regulatory authority under the

Charming Betsy doctrine. In sum, Commerce’s "not likely"

requirement is consistent with U.S. international obligations

and, more specifically, its obligation under Article 11.2 of the

Antidumping Agreement.

B. Gap Period

Plaintiffs next argue that Commerce based its "not likely"

decision on data culled from an inappropriate time period.

Specifically, plaintiffs contend that, in making its "not likely"

determination, Commerce’s review of data from the period after
Court No. 97-08-01409 Page 29

the last day of the review period until the publication date of

the Final Results, i.e., April 30, 1996 until July, 1997, was not

in accordance with law. Plaintiffs maintain that Commerce’s

action in this respect resurrects the so-called "gap-period"

review, a procedure expressly eliminated by Commerce when it

issued revised regulations in 1989.

Before 1989, an AD order might have been revoked after only

two consecutive years of zero or de minimis dumping margins. See

19 C.F.R. § 353.54(b) (1988). But, before a final decision on

revocation could be issued, the regulations also required an

additional administrative review of sales made during the "gap

period"  the day after the second review period until the date

the preliminary results for the revocation decision were

published. Id. Because an indeterminate publication date,

rather than a regulatory deadline, governed the length of the
"gap period," a review of sales made during the "gap period"

might cover a period ranging from nine months up to several

years.8 Also, a "gap-period" review proceeded much like a

8
See generally L. Shambon, Revocation Under the Antidumping
and Countervailing Duty Law?: You Should Live So Long!, The
Practicing Law Institute, Vol. 2, 241, 293 & n.52 (1987) (noting
that while the "gap period" lasted at least nine months,
Commerce’s "gap-period" reviews covered on average fifteen
Court No. 97-08-01409 Page 30

traditional administrative review in that questionnaires were

issued, responses were submitted, and even verifications were

conducted. If, after the review, Commerce found no dumping in

the "gap period" and all other criteria were satisfied, the

agency then could exercise its discretion to revoke the order.

In 1989, Commerce revised its revocation regulation and

specifically acted to excise the "gap-period" review. Rather

than forcing a respondent to demonstrate no dumping for an

indeterminate "gap period," Commerce increased the number of

consecutive years of zero or de minimis dumping from two to

three. See 19 C.F.R. § 353.25(a)(2)(i) (1989). In doing so,

Commerce explained that

the adoption of a three-year period for
revocation or termination based on the absence of
dumping does not substantially modify the period
of time that must be examined under the current
regulations. . . . The adoption in §353.25(c)(3)
of the day after the end of the three-year period
as the effective revocation date eliminates the
need for an examination of the gap period.

Notice of Final Rule: Antidumping Duties, 54 Fed. Reg. 12,742,

12,75758 (Mar. 28, 1989) (emphasis added).

months).
Court No. 97-08-01409 Page 31

Plaintiffs generally argue that in its review of May to

December, 1996 data, Commerce did more than simply assess whether

the future occurrence of dumping was "not likely." Instead,

plaintiffs claim Commerce effectively required the companies to

demonstrate the absence of dumping after the third period of

review and, in so doing, resurrected the "gap-period" review.

Also, plaintiffs contend the phrase "not likely that

[respondents] will in the future" make LTFV sales, 19 C.F.R.

§ 353.25(a)(2)(ii), only allows for review of those events that

might occur after a revocation notice is published, not those

events that occur immediately after the close of the third review

period. Plaintiffs thus maintain Commerce erred when it

considered the May to December, 1996 data because "in the future"

mandates a prospective analysis of events that might occur after

the date of publication, which in this case was July, 1997.

Plaintiffs’ arguments are unpersuasive. Under the current

regulation, it is true that respondents need to establish only

three years of no dumping to satisfy the first criteria. By

eliminating the "gap-period" review, Commerce effected this

change. Yet, it does not follow that because Commerce reviewed

the May to December, 1996 data in making its "not likely"
Court No. 97-08-01409 Page 32

determination, Commerce required plaintiffs to prove an absence

of dumping. Commerce did not issue questionnaires, require

responses, conduct a verification, or calculate a dumping margin

for sales made during the May to December period. Thus, Commerce

simply did not conduct an administrative review or a quasi-review

of May to December, 1996 sales and, hence, did not resurrect the

"gap-period" review.

In addition, while Commerce may not conduct a typical full

review of post-POR sales as part of its "not likely" analysis,

nothing in section 353.25(a)(2)(ii) proscribes its analysis to a

review of projected trends in the period afer publication of the

revocation notice. Commerce may use post-POR data, as well as

analysis of post-revocation models, to assess whether future

dumping is "not likely." Commerce has discretion to decide which

data it will use to assess whether future dumping is "not

likely." In sum, Commerce’s decision to review post-POR and pre-

revocation notice data was in accordance with law.
Court No. 97-08-01409 Page 33

C. Substantial Evidence Supports Commerce’s Decision.

Plaintiffs point to a host of evidentiary flaws in

Commerce’s decision and claim that, on the whole, the decision is

not supported by substantial evidence. More specifically,

plaintiffs identify eight alleged defects in support of their

claim: (1) both plaintiffs contend Commerce’s reliance on below-

cost sales for the May and June, 1996 period to support its "not

likely" decision is unsupported by the record evidence; (2) both

plaintiffs contend Commerce’s reliance on spot-pricing data as

evidence of future dumping is misplaced and runs counter to

company-specific pricing and cost data; (3) Hyundai contends

Commerce ignored crucial evidence showing that the DRAM industry

improved in 1997; (4) Hyundai contends the record does not show

that production levels and inventory levels were increasing; (5)

Hyundai contends Commerce ignored evidence that it had no

incentive to dump in the U.S. market because it was building a

DRAM facility in the United States; (6) LG Semicon contends it

had no incentive to dump in the future because its U.S. market

share is so small; (7) LG Semicon contends Commerce failed to

consider crucial exchange rate data that indicates LG Semicon was

"not likely" to dump in the future; and, finally, (8) both
Court No. 97-08-01409 Page 34

plaintiffs contend Commerce ignored the fact that the companies

were willing to participate in a data collection program to

dispel the notion that future dumping might occur. The Court

considers each argument posed by plaintiffs in turn below and

holds that, while other conclusions might easily be drawn from

the record, Commerce’s determination is nevertheless supported by

substantial evidence.

i. Commerce’s Reliance on May and June 1996 Data Indicating
Sales Below Cost is Supported By Substantial Evidence.
In the Final Results, Commerce examined cost data submitted

by the parties as part of the third administrative review.

Included in this data was information for the two shoulder months

immediately following the close of the third review period, i.e.,

May and June, 1996. Commerce decided to use this data for its

revocation analysis. In doing so, Commerce found that both

respondents made "a substantial number of home market sales [] at
prices below [cost of production] during [the] two months

immediately following the close of the third administrative

review." Final Results, at 39,817. Commerce stated that the

existence of below-cost home market sales in May and June, 1996

"is suggestive of deteriorating market conditions that often give

rise to dumping." Id.
Court No. 97-08-01409 Page 35

Plaintiffs point out that the statute only allows Commerce

to disregard below-cost sales when calculating the dumping margin

if they (1) have been made within an extended period of time in

substantial quantities, and (2) were not at prices which permit

recovery of all costs within a reasonable period of time. See 19

U.S.C. § 1677b(b)(1). And, the statute defines an extended

period of time as "a period that is normally 1 year, but not less

than 6 months." 19 U.S.C. § 1677b(b)(2)(B). Because the sales

at issue plainly were not made over an extended period of time,

plaintiffs maintain that the May and June, 1996 data cannot be

used to support Commerce’s revocation decision.

The Court agrees in part. Plaintiffs are correct in noting

that for purposes of calculating a dumping margin, more than two

months of below-cost sales are needed before such sales may be

disregarded. Commerce even concedes as much in the Final
Results:

We note that, according to the Department’s cost test
methodology, these below cost sales were not
sufficiently numerous for the Department to reject as
a basis for determining normal value in the third
review. We also agree with LGS that whether it made
home market sales at prices below COP during the two
months immediately following the close of the third
review period in and of itself does not demonstrate
that dumping occurred.
Court No. 97-08-01409 Page 36

Final Results, at 39,817. Thus, standing alone, below-cost sales

made over a limited time period cannot amount to substantial

evidence that a respondent will engage in future dumping. See,

e.g., Steel Wire Rope from Korea, 62 Fed. Reg. 17,171, 17,174

(Apr. 9, 1997) (finding that evidence of below-cost sales alone

did not support petitioner’s claim that LTFV sales were likely to

be made in the future). Yet, as discussed earlier, Commerce has

discretion to decide which data it will use to make its "not

likely" determination. While the Court agrees that below-cost

sales over a limited period alone are not indicative of future

dumping, Commerce may, in its discretion, use such data as one

factor among several to guide its "not likely" decision.

Consequently, the Court finds that the May and June, 1996 below-

cost sales, in conjunction with other factors, constitutes

substantial evidence to support a "not likely" determination.

ii. Commerce’s Finding that the Submitted Pricing and Cost
Data is Indicative of Future Dumping is Supported by
Substantial Evidence.

Plaintiffs next contend that Commerce misconstrued and

ignored the submitted pricing and cost data to such an extent

that the "not likely" determination is not based on substantial

evidence. Both plaintiffs critique Commerce’s analysis of spot
Court No. 97-08-01409 Page 37

and actual prices and contend the data show that future dumping

is not likely. In addition, plaintiffs generally claim that

Commerce improperly rejected some cost data, then inexplicably

accepted other cost figures such that its price-to-cost

comparisons were fundamentally skewed.

More specifically, plaintiffs claim Commerce’s reliance on

spot-pricing data is misplaced because these prices did not

reflect contract prices to large U.S. original equipment

manufacturers ("OEMs"), which represented a significant portion

of respondents’ U.S. sales. According to Hyundai, Commerce also

failed to give adequate weight to its submitted cost data. In

particular, Hyundai points to an economist’s report it presented

to Commerce during the review, illustrating that in all scenarios

16M DRAM prices would exceed Hyundai’s costs by substantial

margins in 1997 and 1998. See Hyundai’s Letter to DOC
Transmitting Case Brief (Apr. 21, 1997), Ex. 2, P.R. Doc. No. 121

("Hyundai’s DOC Br.") (attaching an April 1997 report of Dr.

Kenneth Flamm entitled "Economic Analysis of 16 Megabit DRAM Cost

and Pricing: Projections for 1997 and 1998"). Hyundai argues

that the econometric forecasts noted in Dr. Flamm’s report were

based on conservative assumptions, and Commerce gave insufficient
Court No. 97-08-01409 Page 38

credence to the report. Hyundai also notes that Commerce’s

price-to-cost comparison was flawed in part because it compared

the average price for all types of one DRAM model with the cost

of only one type within that model. Hyundai maintains these

combined defects in Commerce’s analysis render its decision

unsupported by substantial evidence.

Similarly, LG Semicon notes that, as part of the third

review, Commerce verified that LG Semicon’s actual prices during

May and June, 1996 were higher than U.S. spot prices for the same

period. Furthermore, LG Semicon argues the record shows that its

actual prices  typically based on contract prices that lagged

far behind the declining spot market  remained above its own

declining costs of production throughout the downturn. See LG
Semicon’s Br. at 29. And, contrary to Commerce’s finding, LG

Semicon insists that its submitted data, showing declining costs
in the second half of 1996, were reliable. Finally, LG Semicon

points out that even though DRAM prices were declining rapidly

from January to April, 1996, Commerce found plaintiffs did not

engage in dumping during this period.

Plaintiffs claims have merit. Indeed, it is fair to say

that one could reasonably find respondents were not likely to
Court No. 97-08-01409 Page 39

dump in the future based on the record evidence. Yet, it is not

for the Court to reweigh the evidence; rather, the Court must

simply review the record to ascertain if Commerce’s determination

is supported by substantial evidence. See Matsushita Elec.

Indus. v. United States, 3 Fed. Cir. (T) 44, 54, 750 F.2d 927,

936 (1984) (noting that it is not the function of the court to

reweigh evidence).

Commerce dedicated a significant portion of the Final
Results to analysis of the pricing and cost data and its effect

on the "not likely" determination. Commerce summarized its

analysis in the following passage:

(1) The respondents’ own sales and cost data indicate
that there were a substantial number of home market
sales made at prices below COP during the two months
immediately following the close of the third
administrative review; (2) the lowest point of the
downturn, in terms of DRAM pricing and other market
conditions, did not occur until after mid-1996 (well
after the end of the third administrative review
period); (3) publicly available spot market pricing
data, when viewed in conjunction with the respondent’s
[sic] cost data, extrapolated to a future point in
time, indicate that LGS and Hyundai may have made U.S.
sales at prices below COP during 1996; (4) respondent’s
[sic] own pricing data indicate that contract prices
generally follow the same pricing patterns as spot
market prices; and (5) many of the respondents’
arguments concerning the alleged distortions and
inaccuracies in the petitioner’s analysis lack merit.
In addition, we find that the respondents made several
changes to their costs immediately following the third
Court No. 97-08-01409 Page 40

review period, including changes in depreciation and
foreign exchange loss write-offs.

Final Results, at 39,817. In addition, Commerce noted that

industry revenues worldwide plunged during the 1996 downturn and,

in particular, respondents both reported "dramatic decreases" in

their 1996 financial statements. Id. at 39,816.

Turning to the record, the evidence shows that while there

typically is a lag, contract prices to OEMs did in fact trend in

the direction of spot-market prices. See Mem. from Program
Manager/IA to Principal Deputy Asst. Sec’y/IA (July 16, 1997),

Charts C-E, C.R. Doc. No. 42 ("Final Analysis Mem.") (showing

comparison of respondents’ actual U.S. prices and spot prices).

As Commerce found, the record also contains press reports

indicating that contract prices were below spot prices in 1997.

See Hyundai DOC Br., Ex. 9 (Computer Reseller News, About-Face:

DRAM Reverses Course  Memory Prices on the Rise, Mar. 17, 1997,
noting that "spot market prices are higher than OEM pricing by as

much as $4"), and Ex. 11 (Electronic Buyer’s News, Koreans Reduce

DRAM Flow, Apr. 14, 1997, noting that spot market prices were

about fifty cents higher than contract prices). The confidential

record also illustrates this point, at least with respect to one

respondent. And, review of the confidential record pertaining to
Court No. 97-08-01409 Page 41

Dr. Flamm’s study shows that report even supports some of

Commerce’s price-to-cost conclusions for the 1997 period.

Similarly, the record affirms that Commerce had reason to

express skepticism about LG Semicon’s submitted cost data.

Specifically, LG Semicon did not report its change in the

depreciation schedule and its corresponding effect on cost until

after the verification and after Commerce issued its preliminary

results. Because LG Semicon decided not to disclose this

significant change until late in the proceedings, Commerce had

reason to question the effect of the change on LG Semicon’s cost

data. It was therefore reasonable for Commerce to accord little

weight to LG Semicon’s unverified cost data and the corresponding

cost projections and for it to rely instead on Micron’s submitted

projections. Finally, LG Semicon is correct that Commerce found

de minimis margins for LG Semicon for the January to April, 1996
period, even though DRAM prices were declining rapidly during

this period. Yet, the record also shows that DRAM prices

continued to decline between May and December, 1996, and in some

cases by as much as sixty percent.

Thus, while the Court again notes that in its view, Commerce

could have decided that the voluminous record showed respondents
Court No. 97-08-01409 Page 42

were not likely to dump in the future, this is neither surprising

nor persuasive. "[T]he possibility of drawing two inconsistent

conclusions from the evidence does not prevent an administrative

agency’s finding from being supported by substantial evidence."

Consolo v. Federal Maritime Comm’n, 383 U.S. 607, 620 (1966); see

also Matsushita, 3 Fed. Cir. (T) at 54, 750 F.2d at 936 (noting

that it is "neither surprising nor persuasive" that a plaintiff

can point to record evidence that detracts from Commerce’s

determination). Upon careful review of the record evidence here,

the Court is satisfied that Commerce’s assessment of the pricing

and cost data and its impact on the "not likely" decision is

supported by substantial evidence.

iii. Commerce’s Characterization of the DRAM Market’s Health
is Supported by Substantial Evidence.
Hyundai argues that Commerce’s decision is unsupported by

substantial evidence because it focused on the condition of the
DRAM market in 1996, rather than 1997. Hyundai maintains that

evidence on the record illustrated that the DRAM market had

stabilized in 1997 and, hence, the record demonstrates that

respondents were not likely to dump in the future. In

particular, Hyundai points to record evidence indicating that

spot prices for two high volume DRAM models increased between
Court No. 97-08-01409 Page 43

October, 1996 and April, 1997. See Hyundai DOC Br., Ex. 4.

Hyundai also references two reports from industry analysts in

1997 -- one stating "[w]e believe that the momentum of the

current drive to raise prices will carry on at least through May,

[1997]" id. at Ex. 5 (De Dios & Associates, The DRAM Market

Advisor, Feb. 5, 1997, at 11), and the other noting that "[t]he

long awaited DRAM cycle has turned up. . . Memory bit growth is

believed to be strong, faster than supply at the margin and less

excess capacity exists than many have estimated." Id. at Ex. 15
(Merrill Lynch, Semiconductors Update, Mar. 5, 1997). Finally,

Hyundai notes the record establishes that Micron itself even

acknowledged the improved market conditions in 1997.

It is true that in the Final Results, Commerce devoted

significant attention to the declining state of the DRAM industry

in 1996. See Final Results, at 39,816-17 (noting repeatedly the
various indicators illustrating a market downturn in 1996). In

addition, however, Commerce considered the health of the industry

in 1997. Indeed, consistent with Hyundai’s claim, Commerce

remarked that "market conditions in the DRAM industry have

recovered somewhat in 1997." Id. at 39,814. Yet, Commerce later

was careful to temper this statement and make clear that, in its
Court No. 97-08-01409 Page 44

view, the stability of the market was still uncertain in 1997.

More specifically, Commerce found that

Wholly apart from the data concerning the 1996
downturn, . . . our analysis indicates that market
conditions in the DRAM industry remain volatile. As
stated previously, while the plunge in prices began to
stabilize somewhat in early 1997, recent data indicate
that prices are headed downward again. For example,
according to publicly available data, the average U.S.
price for a 16M DRAM fell from approximately $18.00 in
May 1996 to approximately $7.00 in December 1996.
According to Dataquest, the price for the 16M as of
June 30, 1997, is approximately $6.50. This represents
a 64 percent decline in prices between the end of the
third period of review (April 30, 1996) and June 1997.

Final Results, at 39,818. The record supports Commerce’s

analysis. For instance, just prior to Commerce’s decision, spot

prices for one DRAM model dropped nearly fifteen percent in two

weeks. See Letter from Micron to DOC (July 3, 1997), Attach.,

P.R. Doc. No. 153 (Dataquest, The Semiconductor DQ Monday Report,

June 23, 1997, and June 30, 1997). Also, the record indicates
that some industry analysts expressed skepticism about the

market’s rebound. In particular, one of the reports cited by

Hyundai commented that "[t]he momentum and market psychology can

still shift back in the opposite direction," Hyundai DOC Br.,

Ex. 5 (De Dios & Associates, The DRAM Market Advisor, Feb. 5,

1997, at 2), while another report indicated that "[t]he mainstay
Court No. 97-08-01409 Page 45

16-Mbit market last week continued to be highly unstable.

Analysts and independent distributors all agreed that average

selling prices slipped about 10% in the spot market." Letter

from Micron to Commerce (May 2, 1997), Ex. 1, P.R. Doc. No. 133

(Electronic Buyer’s News, DRAM Price Skid Reaches 64-M Parts,

Apr. 28, 1997). In sum, the Court is satisfied that Commerce

reached its "not likely" determination after considering the

state of the DRAM industry in both 1996 and 1997. And, the Court

is satisfied that substantial evidence supports Commerce’s

analysis in this respect.

iv. Commerce’s Assessment of Supply and Demand Trends in
1997 is Supported by Substantial Evidence.
During the underlying administrative proceeding, respondents

publicly announced that they planned to reduce production on

lower priced DRAMs. Hyundai contends the production cutbacks

were intended to bring supply and demand into balance, thereby
stimulating recovery in the DRAM market. As support, Hyundai

points to an article on the record, reporting that Korean

producers actually were reducing production in early 1997. "The

longer the flow of South Korean DRAMs into the spot market

remains low, the more credible are claims of the Big Three chip

makers in that country that they aren’t building up excessive
Court No. 97-08-01409 Page 46

stocks that must be dumped later." Hyundai’s DOC Br., Ex. 11

(Electronic Buyer’s News, Koreans Reduce DRAM Flow, Apr. 14,

1997). Hyundai also identifies record evidence forecasting

continued growth in personal computers for 1997 and claims

corresponding demand for DRAMs was likely to result. Hyundai

maintains Commerce gave insufficient weight to this record

evidence, which, according to Hyundai, tends to show that

respondents would not be likely to dump in the future.

In the Final Results, Commerce acknowledged that respondents
announced cutbacks in DRAM production yet concluded "it is

unclear how much of an effect this will have on the overall

supply of DRAMs." Final Results, at 39,817. Commerce’s

conclusion is supported by substantial evidence. First, some

reports on the record indicated that the announced production

cuts were actually production shifts from the 16M DRAM to the 64M
DRAM, see Hyundai DOC Br., Ex. 7 (Electronic News, Korean Big 3

in Partial Shift to 64M, Feb. 3, 1997), and that "excess supply"

of 64M DRAMs was expected to persist throughout 1997. See Letter

from Micron to DOC (May 2, 1997), Ex. 1, P.R. Doc. No. 133

(Electronic Buyer’s News, DRAM Price Skid Reaches 64-M Parts,

Apr. 28, 1997). In addition, one industry report issued in
Court No. 97-08-01409 Page 47

April, 1997 concluded that "Korean DRAM makers have not reduced

their production of DRAMs in the last 6 weeks. However they have

started holding product off the market and are constraining

supply to prop [up] prices, while building inventories in die

banks that will be unleashed on the market later." Letter from

Micron to DOC Transmitting Resp. Br. (Apr. 30, 1997), Ex. 1, P.R.

Doc. No. 132 (Goldman, Sachs & Co. Investment Report, Apr. 14,

1997). Upon review of the record evidence, the Court finds that

Commerce’s analysis of the effect of respondents’ announced

production cuts on overall supply and demand is based on

substantial evidence.

v. Commerce’s Decision is Based on Substantial Evidence,
Notwithstanding Hyundai’s Construction of a DRAM
Facility in the United States.
Hyundai next argues that it has no economic incentive to

dump in the future. More concretely, Hyundai informed Commerce
that the company was "in the final stages of building" a $1.4

billion DRAM wafer fabrication facility in Eugene, Oregon. See

Hyundai DOC Br., at 26-27. Hyundai noted that when operational,

the Eugene facility will be one of the largest DRAM plants in the

United States and will be capable of producing the most advanced

DRAM models. Id. Commerce made no mention of this evidence in
Court No. 97-08-01409 Page 48

its Final Results; because Hyundai maintains this evidence

plainly indicates it has no incentive to dump in the future, it

argues Commerce’s "not likely" decision is not based on

substantial evidence.

The Court does not agree. Hyundai is correct in that

Commerce’s Final Results did not probe the effect of Hyundai’s

new facility on future dumping. And, the Court is puzzled by

government counsel’s following defense at oral argument: "What

[the building of the plant] tells you at a minimum is that

[Hyundai] may not need to ship to the United States. It doesn’t

tell you a [] thing about dumping, price discrimination." Tr. of

Oral Arg., at 57. It is unclear to the Court how future dumping

is possible if there are no shipments to the United States.

Notwithstanding the infirmity in counsel’s presentation at oral

argument, Commerce correctly points out that there is no evidence

on the record to indicate that Hyundai will stop, or even

decrease, shipments to the United States after the facility comes

on line. While one surely might reach this conclusion by virtue

of the plant’s existence, one also might not. And, no record

evidence exists to resolve the question. So viewed, it was

reasonable for Commerce to ignore the existence of the Eugene
Court No. 97-08-01409 Page 49

facility, and its decision to do so does not detract from the

whole of the record evidence that supports the "not likely"

determination.

vi. Commerce’s Finding that LG Semicon In Fact Had Economic
Incentive To Dump, Notwithstanding Its Small Market
Share, is Supported by Substantial Evidence.

LG Semicon contends that Commerce’s "not likely" decision is

unsupported by substantial evidence because the company is not a

major participant in the U.S. market and, thus, has no economic

incentive to dump. In particular, LG Semicon highlights record

evidence, illustrating that the U.S. market made up only a small

percentage of its total global DRAM sales in 1995 and 1996. And,

LG Semicon notes that it ranked 12th among 17 U.S. DRAM suppliers
in 1996 by volume, holding only two to three percent of the

market throughout the period of review. On this evidence, LG

Semicon asserts it had no rational business incentive to dump in
the future.

In the Final Results, Commerce responded to LG Semicon’s

argument as follows:

the United States is part of the world’s largest
regional market for DRAMs, with considerable growth
potential. Given the importance of the U.S. market, as
a general matter, even a producer with a relatively
small market share would have an incentive to ride out
industry downturns. The fact that DRAM producers,
Court No. 97-08-01409 Page 50

including Korean respondents, have historically been
found to have dumped during downturns supports this
conclusion.

Final Results, at 39,819. Record evidence shows that the United

States is the world’s largest market for DRAMs. See Final

Analysis Mem., Point 4. Moreover, record evidence shows that in

absolute terms, the value of LG Semicon’s DRAM sales in the U.S.

market was significant through the POR. Id. Based on this

evidence alone, it was entirely reasonable for Commerce to

conclude that the U.S. market would remain significant for LG

Semicon despite its small market share.

vii. Commerce Considered Exchange Rate Data and Its
Conclusions are Based on Substantial Evidence.

LG Semicon next argues that Commerce failed to consider

significant exchange rate data. The record demonstrates that the

Korean won appreciated against the U.S. dollar over the course of

the POR, thereby raising the cost of producing DRAMs in Korea.
LG Semicon concedes this point yet notes that the record also

shows the won depreciated against the dollar after the POR,

thereby decreasing Korean production costs and the potential

incidence of sales below cost. LG Semicon also offered data,

showing that Korean producers purchased much of their equipment
Court No. 97-08-01409 Page 51

from Japan, which it considers significant because the yen

depreciated against the won after the POR. Thus, LG Semicon

asserts that its purchases of Japanese equipment after the POR

were less expensive, thereby further reducing production costs.

LG Semicon maintains that Commerce failed to consider these

currency movements and their effect on Korean DRAM prices and

costs.

Commerce considered the exchange rate data. First, Commerce

acknowledged this claim in the Final Results, noting that LG
argued "the won is currently depreciating against the dollar,

negating the possibility of exchange rate dumping." Final

Results, at 39,816. Commerce then responded to this argument

when it stated,

we note that Korean DRAM producers import machinery and
equipment and many raw materials. In fact, both
respondents recorded large foreign exchange losses for
fiscal year 1996. Therefore, the depreciation of the
won may have actually tended to increase the
respondent’s COP, making dumping more likely in the
United States. At the very least, we find no basis in
the record to conclude that this exchange rate
depreciation entirely favored the respondents.

Id., at 39,818. It is true that at first blush the depreciating

won should lead one to conclude that production costs would

decrease, thus making sales below cost less likely. Yet, upon
Court No. 97-08-01409 Page 52

careful review of the record, the Court is satisfied that

Commerce’s conclusion was reasonably drawn. Two factors

particular to this case support this finding. Specifically, the

effect of the won’s depreciation on (1) servicing foreign-

currency denominated loans, and (2) depreciation costs for

machinery and equipment, an important cost factor, made it

conceivable that LG Semicon’s costs might have increased as a

result of the won’s depreciation. Thus, Commerce considered LG

Semicon’s argument and its conclusion is reasonable based on the

record evidence.

viii. Commerce Correctly Decided that the Proposed Data
Collection Program Was Not Relevant to the "Not
Likely" Determination.
Finally, both plaintiffs argue that Commerce ignored a

crucial piece of evidence that had direct bearing on whether

respondents were "not likely" to dump in the future -- a proposed
data collection program between a U.S. importer of Korean DRAMs,

Compaq Computer Corporation, and the government of Korea that had

been accepted by both LG Semicon and Hyundai. The parties

intended the proposal as additional assurance that Korean

producers would not dump in the future. Plaintiffs maintain
Court No. 97-08-01409 Page 53

Commerce disregarded the import of this program and, hence, its

decision is not based on substantial evidence.

This argument is without merit. For purposes of the "not

likely" assessment, the Court fails to see how the proposal at

issue is to be distinguished from the requisite agreements that

respondents must enter pursuant to 19 C.F.R. § 353.25(a)(2)(iii).

That is, both forms of agreement appear to provide that

respondents’ activity in the DRAMs market will be monitored to

assure that dumping does not recur. It is unclear then why the

proposal at issue has some purported relevance for the "not

likely" determination, though, as the framing of the regulation

makes clear, the requisite agreements do not. Thus, Commerce was

correct when it stated that "while we have considered this

proposed data program, we find that this program has no bearing

on the likelihood issue." Final Results, at 39,811.
* * *

In sum, plaintiffs have identified record evidence that

detracts from Commerce’s "not likely" determination. Indeed,

after combing the record, the Court finds that the agency very

easily might have reached the opposite conclusion. Yet, this is

unsurprising given the extensive record in this case. Commerce
Court No. 97-08-01409 Page 54

was charged with the very difficult task of predicting future

behavior. And, while certain record evidence detracts from its

decision, the Court is also satisfied that, on the whole,

substantial evidence supports its decision.

IV.
CONCLUSION
For the foregoing reasons, the Court sustains Commerce’s

decision not to revoke the antidumping order on DRAMs from Korea.

A separate Order will be entered accordingly.

_________________________________
Richard W. Goldberg
JUDGE

Dated: May 19, 1999
New York, New York.