In late April, the Federal Circuit issued its opinion in Stupp Corporation v. United States, a case originally decided by the Court of International Trade that we have been tracking because it attracted an amicus brief. In this case, the Federal Circuit considered whether the use of a statistical test by the Department of Commerce was reasonable. Judge Stark authored the Federal Circuit’s opinion, which described why the court vacated and remanded the judgment of the Court of International Trade. Judges Lourie and Bryson joined Judge Stark’s opinion. This is our opinion summary.
Judge Stark’s opinion provided the relevant background:
The facts of this case are set out in detail in our earlier decision, Stupp Corporation v. United States, 5 F.4th 1341, 1344-45, 1348 (Fed. Cir. 2021). A short summary suffices for present purposes.
In the course of its LTFV investigation, Commerce determined that SeAH engaged in targeted dumping; that is, “a pattern of export prices (or constructed export prices) for comparable merchandise that differ significantly among purchasers, regions, or periods of time.” Such targeted dumping can be “masked,” and go undetected – and, therefore, unaddressed by imposition of an anti-dumping duty – “because a respondent’s sales of low-priced dumped merchandise would be averaged with (and offset by) sales of higher-priced masking merchandise, giving the impression that no dumping was taking place.” “To address the problem of targeted dumping, Congress created an exception to the use of the average-to-average [A-to-A] method” Commerce ordinarily uses “for calculating a dumping margin.” “Commerce refers to the alternative method of calculating a weighted average dumping margin as the ‘average-to-transaction’ [A-to-T] method.” Commerce sometimes also employs “some hybrid of the two,” combining the average-to-average and average-totransaction methods.
To “implement[] Congress’s directive” to uncover targeted dumping, and determine whether to use the A-to-A, A-to-T, or hybrid comparison methodology, Commerce uses a “differential pricing analysis.” “The differential pricing analysis involves three tests . . . : (1) Cohen’s d test, (2) the ratio test, and (3) the meaningful difference test.”
This appeal is focused on the first of these steps, Cohen’s d test, which is “named after statistician Jacob Cohen” and is used “to evaluate whether the test group differs significantly from the comparison group.” “If the Cohen’s d value is equal to or greater than 0.8 for any test group, the observations within that group are said to have ‘passed’ the Cohen’s d test, i.e., Commerce deems the sales prices in the test group to be significantly different from the sales prices in the comparison group.”
As we set out when this case was before us in 2021, “Commerce applied its differential pricing analysis to SeAH’s sales of welded line pipe and selected the hybrid approach for calculating SeAH’s weighted average dumping margin. That approach resulted in a weighted average dumping margin of 2.53%.” After the Trade Court affirmed Commerce, SeAH appealed to us and contended (as relevant here) that “Commerce misused the Cohen’s d test in its differential pricing analysis,” because “the data in this case did not satisfy the conditions required to achieve meaningful results from the Cohen’s d test.” We “agree[d] that there are significant concerns relating to Commerce’s application of the Cohen’s d test in this case and, more generally, in adjudications in which the data groups being compared are small, are not normally distributed, and have disparate variances.” We expressed concern, in particular, that “Commerce’s application of the Cohen’s d test to data that do not satisfy the assumptions on which the test is based may undermine the usefulness of the interpretive cutoffs.” Therefore, we remanded “to give Commerce an opportunity to explain whether the limits on the use of the Cohen’s d test prescribed by Professor Cohen and other authorities were satisfied in this case or whether those limits need not be observed when Commerce uses the Cohen’s d test in less-than-fair-value adjudications.”
On remand, Commerce again applied its differential pricing analysis, including Cohen’s d test, and concluded that the “statistical criteria do not serve as a basis for Dr. Cohen’s thresholds.” Accordingly, in Commerce’s view, Cohen’s d test can reasonably be used even when the data being analyzed does not satisfy the three statistical criteria about which we had raised concern in the earlier appeal. SeAH again appealed to the Trade Court, which determined that “Commerce has adequately explained how its methodology,” including its use of Cohen’s d, “is reasonable.” SeAH timely appealed to us. We have jurisdiction pursuant to 28 U.S.C. § 1295(a)(5).
On the merits, Judge Stark rejected the government’s argument that “it was reasonable to use Cohen’s d test under circumstances in which the statistical requirements for use of Cohen’s d are not met.” Judge Stark explained how in a prior opinion the court “held it is ‘unreasonable for Commerce to use Cohen’s d as part of its differential pricing analysis when the test is applied to data sets that do not satisfy the statistical assumptions’ on which Cohen’s d is predicated.” According to Judge Stark, here the Federal Circuit was faced with “the very same situation,” and as a result the panel was “compelled to reach the same result.” As a result of this analysis, the Federal Circuit vacated the Court of International Trade’s judgment and remanded this case “for proceedings consistent with this opinion.”
Notably, Judge Stark wrote a separate opinion to provide additional views. In it, he clarified that, if the Federal Circuit was “writing on a blank slate,” he “might well be persuaded by Commerce that its use of Cohen’s d test, as one step in its three-step differential pricing analysis, is reasonable.” But, he noted, the panel was “not addressing a question of first impression” because in a prior precedential opinion the court held “it is ‘unreasonable’ for Commerce to use Cohen’s d when the three assumptions underlying it” are not satisfied. Judge Stark additionally noted that the Federal Circuit had “remanded this very case nearly four years ago to give Commerce an opportunity to explain why meeting these assumptions is not necessary,” and that Commerce “did not do so in a particularly persuasive manner.”