AD Respondent Rails Against Use of 'd' Test, Relies on Past CAFC Opinion
The Commerce Department's use of the Cohen's d statistical test to carry out its differential pricing analysis in rooting out "masked" dumping violates "well-recognized statistical principles," plaintiff HiSteel Co. argued in an Oct. 17 motion for judgment at the Court of International Trade. Commerce's assertions that certain statistical assumptions typically required of the d test are not relevant since it using the entire population of data and not just a sample "is mathematically dishonest," the brief said (HiSteel Co. v. United States, CIT #22-00142).
Sign up for a free preview to unlock the rest of this article
Timely, relevant coverage of court proceedings and agency rulings involving tariffs, classification, valuation, origin and antidumping and countervailing duties. Each day, Trade Law Daily subscribers receive a daily headline email, in-depth PDF edition and access to all relevant documents via our trade law source document library and website.
The case concerns the administrative review of the antidumping duty order on heavy walled rectangular welded carbon steel pipes and tubes from South Korea. In the review, to detect masked dumping, Commerce used Cohen's d test as part of its differential pricing analysis. The agency found that masked dumping was occurring, then decided to use the average-to-transaction method to calculate the dumping margin. HiSteel took to the trade court to argue this was illegal (see 2206080062).
The U.S. Court of Appeals for the Federal Circuit in the Stupp Corp. v. U.S. July 2021 decision called into question use of the Cohen's d test, finding that it failed to abide by certain assumptions required of the test, including a normal distribution of the data and roughly equal variances (see 2107150032). The U.S., though, has argued that it does need to abide by these assumptions since it is using the total population of the data in question instead of just a sample.
HiSteel said in its brief that this "assertion is simply wrong as a matter of mathematics. Any analysis based on the standard deviation of data -- as Cohen’s d is -- depends on the shape of the distribution being analyzed, whether that distribution consists of the entire population or just a sample. Commerce’s attempt to assume away this fundamental issue is mathematically dishonest."
Commerce's position that it can ignore the limitations on the test because it is not using the test in the manner that statisticians do "is fatal to its position," the brief argued. As part of the test, the agency used a 0.8% cutoff laid out by statisticians to see whether HiSteel's sales passed the Cohen's d test, meaning masked dumping was occurring. "Commerce cannot reasonably claim that its reliance on the 0.8 cut-off that statisticians use as a rule-of-thumb for assessing whether Cohen’s dis 'large' represents the application of a “widely adopted' statistical method," HiSteel said. "Instead, if, as Commerce admits, it is not following normal statistical practices, then it must justify its use of a 0.8 cut-off without reference to those practice."
HiSteel further argued that Commerce here is bound by the Stupp decision, even though Commerce says it is not bound by this opinion since it is not "final and conclusive." The respondent said that "while the remand proceeding related to the Stupp appeal may be ongoing, Commerce must still address the CAFC’s judgment as it applies to the use of the Cohen’s d test in this case. In that regard, the CAFC has made clear that its decision in Stupp applies to all cases where Commerce’s use of Cohen’s d in its DPA presents 'identical concerns' to those raised in Stupp."