Opinion
Court No. 03-00652, Slip Op. 06-177, PUBLIC VERSION.
December 7, 2006.
Vinson Elkins LLP (Daniel L. Porter, James P. Durling, Matthew P. McCullough, Matthew R. Nicely), Washington, DC, for Plaintiffs Hynix Semiconductor Inc. and Hynix Semiconductor America Inc.
Peter D. Keisler, Assistant Attorney General; David M. Cohen, Director; Ada E. Bosque, Commercial Litigation Branch, Civil Division, U.S. Department of Justice; James M. Lyons, General Counsel, Andrea C. Casson, Assistant General Counsel, Office of the General Counsel, U.S. International Trade Commission (Marc A. Bernstein, Neal J. Reynolds), for Defendant United States.
Kelley Drye Collier Shannon (Kathleen W. Cannon), Washington, DC, for Defendant-Intervenor Qimonda North America Corp.
King Spalding LLP (Gilbert B. Kaplan, Cris R. Revaz, Daniel L. Schneiderman, Jeffrey M. Telep), Washington, DC, for Defendant-Intervenor Micron Technology, Inc.
OPINION
Hynix commenced this case in September of 2003 to challenge the final affirmative material injury determination made by the United States International Trade Commission (" ITC") with respect to dynamic random access memory semiconductors (" DRAMs") of one megabit or above from the Republic of Korea, published underDRAMs and DRAM Modules from Korea, USITC Pub. 3616, Inv. No. 701-TA-431 (Aug. 2003) (" Original Determination "). Currently before the Court are the ITC's remand results.
I. BACKGROUND
In April 2006, the Court sustained the ITC's Original Determination in all aspects but one. See Hynix Semicon., Inc. v. United States, 30 CIT ___, 431 F. Supp. 2d 1302 (2006) (" Hynix I "). The Court found that the ITC failed to support, by substantial evidence, its finding that the unprecedented drop in demand for downstream end-use products did not have such a predominant effect in producing the material injury as to prevent the subject imports from being a material cause of that injury. See id. at ___, 431 F. Supp. 2d at 1321.
The ITC had found that the drop in demand for end-use products such as personal computers, though slowing the pace of DRAMs demand growth, did not render the subject imports a merely ancillary or tangential cause. See Original Determination at 36. The ITC claimed that record evidence demonstrated a lack of any "clear correlation between growth of the DRAMs market and price movements." Id. To support that position, it cited to a table (the " McClean Report") that purported to document output growth and price changes in the global DRAMs market from the years 1990 to 2003. See id. n. 163. The ITC interpreted the McClean Report as evidence of a non-correlative relationship between demand and price in the DRAMs industry. It then concluded that since price and demand exhibited no correlation, slowing demand growth due to falling demand for underlying end-use products could not have been the sole cause for the "unprecedented severity of the price declines that occurred from 2000 to 2001 and persisted through 2002. . . ." Id. at 36. The ITC did not, however, explain why the output data from the McClean Report could be used to illustrate demand in the DRAMs industry. The Court explained in Hynix I that this finding, resting as it did on evidence consisting solely of the McClean Report, was unsupported by substantial evidence and remanded the question to the ITC for further explication. See Hynix I, 30 CIT at ___, 431 F. Supp. 2d at 1321.
In its amended remand order of May 30, 2006, the Court provided clear instructions to the ITC on how to remedy the evidentiary and explanatory deficiencies of its original determination. First, the Court directed the ITC to explain how the output growth/price movement relationship documented in the McClean Report can be used to articulate the relationship between slowing demand growth and price movement. See Am. Remand Order 1. If it was unable to provide such an explanation, the Court instructed the ITC to "point to other record evidence that shows the unprecedented drop in demand for downstream end-use products did not have such a predominant effect in producing the material injury as to prevent the subject imports from being a material factor of that injury. . . ." Id. In the event such evidence did not exist, the Court instructed the ITC to "conduct further investigations to determine the effect of the unprecedented drop in demand for downstream end-use products. . . ." Id. 2. The ITC issued its remand determination on July 11, 2006. See DRAMs and DRAM Modules from Korea, USITC Pub. 3871, Inv. No. 701-TA-431 (July 2006) (" Remand Results ").
II. STANDARD OF REVIEW
The Court will remand the ITC's determination if it is "unsupported by substantial evidence on the record, or otherwise not in accordance with law. . . ." 19 U.S.C. § 1516a(b)(1)(B) (2000); see also Nippon Steel Corp. v. ITC, 345 F.3d 1379, 1381 (Fed. Cir. 2003) (holding that 19 U.S.C. § 1516a contemplates only affirmances and remands, and never outright reversals of agency determinations). Substantial evidence "does not mean a large or considerable amount of evidence, but rather `such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.'" Pierce v. Underwood, 487 U.S. 552, 565, 108 S.Ct. 2541, 101 L.Ed.2d 490 (1988) (quoting Consol. Edison Co. v. NLRB, 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938)). It "requires `more than a mere scintilla,' but is satisfied by `something less than the weight of the evidence.'"Altx, Inc. v. United States, 370 F.3d 1108, 1116 (Fed. Cir. 2004) (citations omitted) (quoting Atl. Sugar, Ltd. v. United States, 744 F.2d 1556, 1562 (Fed. Cir. 1984); Matsushita Elec. Indus. Co. v. United States, 750 F.2d 927, 933 (Fed. Cir. 1984)).
III. DISCUSSION
Nearly all the issues in this case have already been adjudicated in Hynix I. As discussed above, that decision sustained all but one of the ITC's findings in its Original Determination. Most importantly, the Court has already determined that the subject imports contributed to the material injury to the domestic industry. The one issue remaining is whether the slowing DRAMs demand growth, precipitated by the unprecedented drop in underlying demand for downstream products such as personal computers, was so predominate as to render the subject imports a merely ancillary or tangential cause of the domestic industry's material injury.
When a complaining party raises a potential alternative cause for a domestic industry's injury, the ITC's causation inquiry must broaden to analyze the effects of that putative cause. InHynix I, the Court explained the ITC's burden:
The ITC is charged with the burden of an earnest investigation into whether other factors render the subject imports a tangential, de minimis cause of the domestic industry's material injury. An affirmative material injury determination does not rest on substantial evidence when the ITC fails to analyze compelling arguments that purport to demonstrate the comparatively marginal role of subject imports in causing that injury.
30 CIT at ___, 431 F. Supp. 2d at 1317.
In the Original Determination, the ITC posited, in essence, that a lack of a clear and discernible correlation between price and demand mooted the issue of slowing demand growth. If the ITC had pointed to evidence that indeed price is not a function of demand in the DRAMs market, then its determination would have been sustained: after all, falling demand growth cannot be said to predominate the causation analysis of a domestic industry's price woes if demand is not a primary determinant of price in the industry. It is well beyond the Court's discretion to challenge the patent reasonableness of such an argument, provided the evidence supports it.
However, in Hynix I the Court found the ITC's discussion of this factor inadequate due to the paucity of reliable evidence suggesting that price and demand were not correlated. The ITC aimed to elucidate the agency's position on remand, and the Court must now examine the ITC's Remand Results. In those Remand Results, the ITC draws on other record data in support of its characterization of the non-correlative relationship between price and demand in the DRAMs industry.
A. For Purposes of This Case, the "Output" Data from the McClean Report Is Devoid of Probative Value
As noted above, in its Original Determination, the ITC supported its characterization of the non-correlative relationship between price and demand by citing to the McClean Report. That report consists of a two column table measuring "price" and "output" over time. Because "output" refers to the quantity of goods produced, and not demand or even the quantity purchased or consumed, the citation's relevance was unclear. Though the Court issued explicit instructions to the ITC to explain the table, the ITC has failed to provide an equally clear response to the Court's concern. Instead, in the Remand Results the ITC continues to treat the "output" data as indicative of consumption trends.
The ITC lists the McClean Report data in the Remand Results and describes it as a "longer [data] series, which measures worldwide DRAM bit consumption and pricing since 1990. . . ." Remand Results at 8. Again, the ITC cites to the McClean Report data in a footnote:
We . . . observe that the longer time series data in the record also fail to show a correlation between pricing changes and changes in consumption. For example, prices declined by 60 percent in 1997, where the rate of growth of consumption was 21 percentage points higher than the previous year, but declined by only 16 percent in 1999, where the rate of growth of consumption was 11 percentage points lower than the previous year.Id. at 10 n. 40. At no point in the Remand Results does the ITC endeavor to explain why the McClean Report's output data can be utilized to illustrate a relationship between demand and prices. Defendant-Intervenor Micron Technology, Inc. (" Micron") contends that "the ITC makes clear that the reference to `output' growth actually reflects consumption, (i.e., demand) growth" by citing to footnote 40 of the Remand Results. Micron's Comments 1. No reasonable reading of that footnote could support Micron's proposition. The footnote evinces the ITC's impression that the McClean Report contains consumption data, but it does not even purport to address the key issue: why does the ITC assume that the McClean Report provides consumption data and not output data?
There is no requirement that the ITC apply the strictures of the Federal Rules of Evidence in the context of its material injury investigations, see Tarnove v. Bentsen, 17 CIT 1324, 1326 (1993), but the absence of the specific guarantees of the federal rules does not free the agency from the commonsense requirement that its findings must be based on "evidence having rational probative force." Consol. Edison, 305 U.S. at 230. Because the report tracks output data, its relevance to this proceeding is not apparent barring some explanation to the contrary. As the ITC has made no such explanation, the McClean Report's probative value as to the correlation of price and demand is nil. The citation to the McClean Report was the linchpin support of the ITC's determination that demand and price are non-correlative in the DRAMs industry, see Original Determination at 36 n. 163. Unless the ITC can point to other evidence, that determination cannot be said to rest on substantial evidence.
The ITC explains that it chose to rely on the McClean Report because of the perceived objectivity of a third-party data set.See Remand Results at 8 n. 33; cf. also DRAMs and DRAM Modules from Korea, USITC Pub. 3839 at 9, Inv. No. 701-TA-431 (Feb. 2006) (section 129 consistency determination). It appears the only other available data were provided by Hynix, which was, of course, an active participant in the investigation. While data compiled by disinterested parties remains relatively free from the shadow of suspicion, that evidence must, of course, actually have a tendency to make some material fact more or less probable,cf. Fed.R.Evid. 401.
B. The ITC Has, However, Cited Other Record Data That Bolster Its Contention That Price and Demand Are Not Directly Correlated in the DRAMs Industry
In its amended remand order, the Court instructed the ITC that, in the event the ITC is unable to elucidate the relevance of the McClean Report, it should "point to other record evidence that shows the unprecedented drop in demand for downstream end-use products did not have such a predominant effect in producing the material injury as to prevent the subject imports from being a material factor of that injury." Am. Remand Order 2.
1. Other Record Evidence in the Form of Worldwide Consumption Data Supports the ITC's Claim That Demand Is Not a Primary Determinant of DRAMs Prices
The record contains other evidence relating to world consumption of DRAMs. The ITC references another data set (" Hynix worldwide consumption data") provided by Hynix itself in Hynix's Prehearing Brief. See Remand Results at 8 n. 33 (citing Pl.'s Preh'rg Br. 62). Hynix cited the same data before the ITC during the remand proceedings. See Pl.'s Agency Remand Br., Exs. 1-2. The Hynix worldwide consumption data were not, however, cited or referred to directly in the ITC's original affirmative material injury determination. The data paint the following picture: from 1995 to 1996, DRAMs demand growth, measured by world consumption, was 77%; from 1996 to 1997, growth was 98%; from 1997 to 1998, growth was 77%; from 1999 to 2000, growth was 74%; from 2000 to 2001, growth was 60%; and from 2001 to 2002, growth was 41%.
The parties have devoted substantial briefing to a phantom issue: whether domestic DRAMs consumption is a permissible proxy for DRAMs demand. The Court did not remand that issue, and has always assumed that the ITC's practice of gauging demand by measuring domestic consumption is permissible. At issue is the demand for DRAMs, so the court will look to consumption of DRAMs, and not consumption of end-use products, as being indicative of DRAMs demand. Plaintiff's contention that end-use products are a more accurate proxy for demand is belied by its own summary judgment brief, where it observed categorically: "The key is that DRAM demand never actually drops; rather, the rate of growth falls." Pl.'s J. Agency R. Br. 39. Because end-use product demanddid drop, Hynix therefore was referring to DRAMs consumption — and not end-use product consumption — as a proxy for demand. The confusion is perhaps the product of the Court's instructions to the ITC to execute the causation analysis in light of the "underlying drop in demand" for "downstream end-use products."Hynix I, 30 CIT at ___, ___, 431 F. Supp. 2d at 1319-21; Am. Remand Order 2. In choosing such language, the Court was not inviting the parties to re-brief the question of what is an appropriate measure of DRAMs demand; instead, the Court simply referred to the fact that both parties agreed that the drop in demand for end-use products occasioned the decline in DRAMs demand growth. See Pl.'s J. Agency R. Br. ("The demand for DRAMs is a derived demand; that is, it is based solely on the downstream demand for products that use DRAMs, such as personal computers"); Pl.'s Agency Remand Br. 7; Def.'s Mem. Opp. J. Agency R. 43.
The data was excluded because the McClean Report data was perceived to be more objective. See supra note 1.
The Hynix worldwide consumption data measure global demand, but both parties have indicated their acceptance of worldwide demand data as indicative of general trends in U.S consumption. See Remand Results at 8; Pl.'s Agency Remand Br. 7.
The Hynix worldwide consumption data track trends in global demand growth from 1996 to 2002, a shorter window than the McClean Report, which reports "output" growth from 1990 to 2003. However, the data largely mirror the McClean Report output data for the years 1996 to 2002. The only significant deviation between the two sets occurs in 2002, for which the Hynix worldwide consumption data reports a demand growth number (41%) that is 16% lower than the output number from the McClean Report (49%).
The Hynix worldwide consumption data evince the same lack of correlation between price and global demand that the ITC claimed was evident in the McClean Report. For example, DRAMs demand growth in both 1996 and 1999 was 77%. However, the price changes for those periods, as indicated in the McClean Report, were wildly divergent (negative 65.3% in 1996 and a mere negative 12.5% in 1999). The data relating to the period of investigation show that the deceleration of demand growth was less severe in 2001 (74% growth in 2000 to 60% growth in 2001) than in 2002 (60% growth in 2001 to 41% growth in 2002). However, the price effects were much more pronounced in the former period (price declined 72.7% in 2001, and merely 33.3% in 2002). The price decline in 2001 was the worst in DRAMs history. The fact that price rebounded significantly in 2002 despite the aggravated decline in demand growth undercuts significantly Hynix's contention that demand trends predominate any analysis of price movements in the DRAMs industry. The ITC reasoned that if demand trends were a cardinal determinant of DRAMs prices, the 1999 price movements would have more closely tracked those in 1996.
The domestic consumption data during the period of investigation tell a similar story. The rate of demand growth measured by apparent domestic consumption in 2000 was 67.2%. See Remand Results at 7. In 2001, domestic demand growth was 48.6%, and in 2002 it fell further still to 27.4%. See id. Thus, the 2002 price rebound occurred during a time when demand growth was still plummeting. This data similarly testifies to a lack of any direct and discernible relationship between demand and price in the DRAMs industry.
2. The ITC Properly Explained Why Hynix's Proffered Variance Analysis Was Unreliable
Hynix prepared and presented to the ITC a variance analysis purporting to measure the effects of the decreasing demand growth on the DRAMs selling price, given the price elasticity estimated by the ITC. This variance analysis draws on the record data, and claims to individuate the relative importance of the decline in demand growth from the loss of market share due to increased imports. Subject imports are distinguished from nonsubject imports.
According to Hynix, the ITC routinely utilizes a variance analysis to isolate the effects of changes in price, volume, and unit cost on U.S. producers' profitability. The current proposed variance analysis supposedly borrows from that methodology.
The variance analysis purportedly isolates four factors that impact "the demand for U.S. producers' shipments": low domestic growth relative to historic averages; low export growth relative to historic averages; decreased market share from nonsubject imports; and decreased market share from subject imports. Pl.'s Agency Remand Br. 13. Assuming that growth in demand for both domestic and export DRAMs would be around 75% absent the decreased demand growth, the "direct" decline in demand growth accounted for 65.8% of the decline in demand for domestically produced DRAMs products. See id. 15. The loss of market share to subject imports amounted to merely [____]% of the decline in demand for U.S. producers' products. See id. The remaining decline in demand for U.S. producers' products was attributed to the loss in market share to nonsubject imports. Thus, "the direct decline in demand for all DRAMs alone had [____________________] more effect on U.S. producer shipments than did subject imports." Id. 15 (emphasis omitted). Hynix also compared the [____]% effect of subject imports to the [_____]% combined effect of lost market share to nonsubject imports and decline in direct demand growth. See id.
It appears that the variance analysis indicates that the decline in demand growth (65.8%) was [____________________], and not [____________________], as important as the loss in market share to subject importers ([____]%).
Furthermore, the variance analysis maps the demand growth data onto the ITC's price elasticity estimates to yield information on the price effects of the slowing demand growth. Positing a price elasticity of 0.4, Hynix concludes that the price effects of the decrease in demand growth alone should have amounted to a 74% decrease in price. Other record evidence suggests that such a decline in fact occurred, with price declines in individual products ranging from 68% to 84%. See Pl.'s Agency Remand Br. 16. The implication is obvious: since the predicted price effect of the decline in demand growth is roughly equal to the actual price declines witnessed during the period of investigation, the true cause of the price declines was the decrease in demand growth. The final step in Hynix's analysis is stated succinctly at the end of this section of its brief: "using the [ITC's] own price elasticity estimate, virtually all the price decline from 2000 to 2001 can be directly attributable to the decline in demand for U.S. produced DRAMs." Id. 16 (emphasis omitted).
The ITC estimated that the U.S. producers' supply elasticity fell in the range of 0.3 to 0.5. See Pl.'s Agency Remand Br. 16 (citing Original Determination at II-9).
In the Remand Results, the ITC acknowledged the potential probative value of Hynix's variance analysis, noting that the variance analysis "is pertinent to the scope of the proceedings. . . ." Remand Results at 3 n. 13. However, the ITC ultimately disregarded the analysis on account of what it perceived to be two fundamental flaws that undercut its probative value. First, the ITC characterized the assumption that domestic DRAMs producers could have sustained a 75% demand growth rate as unrealistic. See id. at 14 n. 54. Second, the ITC alleged certain "computational difficulties with Hynix's analysis." Id.
The first claimed deficiency is supported by a citation to the McClean Report, which the ITC insists is evidence of "annual DRAM demand increases fluctuat[ing] considerably." Id. Moreover, "material collected for the [ITC's] preliminary determination . . . indicated that the DRAM demand increased by substantially less than 75 percent in 2000. . . ." Id. The McClean Report, for the reasons stated above in Part III.A of the Court's decision, is devoid of probative value. However, the 2000 demand growth data collected during the preliminary determination creates problems for Hynix's variance analysis. The Remand Results cite the ITC's estimate that domestic demand grew by 67.2% during 2000. See id. at 7. Hynix ripostes by citing the Hynix worldwide consumption data discussed at length above. Looking exclusively at those data, Hynix is almost correct that "for each of the . . . five years [before 2001] the DRAM demand growth rate was at [sic] least 75 percent." Pl.'s Comments 12 (emphasis omitted). Since the apparent U.S. domestic consumption data relate to the domestic market, they are more reliable than the Hynix worldwide consumption data. Therefore, the citation to the Hynix worldwide consumption data does not rebut the ITC's position that the 75% growth assumption — on which the entire variance analysis hinges — was unrealistic in light of the significantly lower apparent U.S. consumption growth (67.2%) in 2000.
Hynix is only "almost correct" because the Hynix worldwide consumption data document a growth rate of 74% in 2000.
Hynix also cites a comment made by Steve Appleton, CEO of Micron, during a 2003 conference call with analysts. Mr. Appleton, discussing the demand profile of the DRAMs industry, said that "`We have had a fundamental shift I think in the demand profile. As you know, it historically ran at 75 percent. . . .'" Pl.'s Comments 12 (quoting Micron Q3 Financial Release Conf. Call, Pl.'s Post-Hr'g Br., Ex. 13) (emphasis omitted). Again, the ITC's own investigation of the period immediately preceding the 2001 downturn revealed that apparent domestic consumption grew at 67.2%. It is hardly unreasonable for the ITC to have relied on data from its own investigation rather than the transcript of a conference call between a CEO and Wall Street analysts covering his company.
The ITC also questions the computational accuracy of the variance analysis. Specifically, the ITC claims that Hynix's analysis "assigns to demand declines reductions in shipments that Hynix's own calculations acknowledge were attributable to increases in market penetration of subject and nonsubject imports." Remand Results at 14 n. 54. The variance analysis apportions the causation of the price declines to the various factors by dividing the estimated number of total shipments lost by the effect of a given factor (e.g., loss of domestic shipments due to demand growth decline). Hynix derives the estimate of total lost shipments by adding the number of shipments lost on account of the four separate factors that the variance analysis purports to individuate: (1) lost shipments due to increased market share for subject imports, (2) lost shipments due to increased market share for nonsubject imports, (3) lost export shipments due to demand growth decline, and (4) lost domestic shipments due to demand growth decline. The ITC correctly points out that the figure representing lost domestic shipments due to demand growth decline also includes lost shipments due to the loss in market share. As a result, the variance analysis exaggerates the relative effect of the demand growth decrease and in the process deflates the effect of increased import market share. A close examination of the variance analysis confirms that this computational objection is valid, and further undermines the foundation of the variance analysis.
In its brief before the ITC, Hynix claimed that "U.S. producers' U.S. shipments [in 2001] would have been 11,342,832 higher had consumption grown at its historical rate." Pl.'s Agency Remand Br. 14 (emphasis added). The 11,342,832 figure was then added to the estimates for (1) lost export shipments due to demand growth decline, (2) lost shipments due to increased market share for nonsubject imports, and (3) lost shipments due to increased market share for subject imports. As noted above, in sum these quantities yielded an estimate of the "total" loss of shipments for domestically-produced DRAMs in 2001, which amounted to 39,175,541 shipments. See id. 15.
The consumption figures are in billions of bits.
The 11,342,832 figure itself was derived by measuring the difference between actual domestic shipments for 2001 and likely domestic shipments assuming continued growth of 75%. Ignoring for the moment the problems associated with this growth assumption, this computation holds constant the domestic producers' market share. Therefore, the figure represents (1) the amount of shipments lost on account of decreased demand growth; and (2) the amount of shipments lost on account of increased market share for import products. Hynix incorrectly summarized the analysis' import, claiming that "U.S. shipments would have been 11,342,832 higher had consumption grown at its historical rate." Id. 14. Rather, that assertion is subject to a further qualification that touches on the core issue in this case: the increased market share of imports, including subject imports. The domestic industry could only have achieved that number of shipments if the market grew at 75% and if the U.S. producers' preserved their competitiveness. Therefore, to treat the 11,342,832 figure as indicative of only slowing demand growth, and to then add separately the number of shipments lost due to loss in market share, is to double-count the latter.
The 11,342,832 figure, because it fails to discount the likely shipments in light of the decreased market share and U.S. producer competitiveness, is thus a wholly unrealistic estimation of what the 2001 market would look like absent the decline in demand growth for DRAMs products. The resultant price calculations derived from the variance analysis data are similarly devoid of persuasiveness.
IV. CONCLUSIONS
Because the ITC has failed to establish the foundational relevance of the McClean Report, the Court must disregard it. On the other hand, pursuant to the Court's earlier remand order, the ITC has pointed to other record evidence that suggests that demand and price are not directly correlative in the DRAMs industry. The Court therefore finds that the ITC has supported by substantial evidence its finding that the slowing growth of demand does not so predominate the causation analysis as to render the subject imports a mere ancillary cause of the domestic industry's woes.
Moreover, the Court finds that the ITC properly disregarded Hynix's proposed variance analysis. The computational inaccuracies, as well as the justifiable preference for record data over the CEO's growth predictions, place the ITC's decision to disregard the proposed variance analysis well within the bounds of reasonableness.
The Court sustains the ITC's Remand Results. Recalling that the Court, in Hynix I, sustained all other challenged aspects of theOriginal Determination, the Court must now enter judgment in the ITC's favor.