Commerce Again Defends Use of Simple Average in 'd' Test Denominator After CIT Remand

The Commerce Department released the results of its fourth remand redetermination in an antidumping duty investigation on steel nails from Taiwan, sticking with its use of a simple average to calculate the denominator of the Cohen’s d test coefficient. The department said it complied with the remand order by providing reasonable justification for its methodology in its test to identify "masked dumping" (Mid Continent Steel & Wire v. U.S., CIT # 15-00213).

Agencies are not duty-bound to follow published literature when it is inapplicable to the specific problem being addressed, Commerce said. The case has long centered around whether Commerce can and should depart from academic literature in its calculation methodology. In the results of its previous, and third, remand, Commerce stuck by its use of a simple average, saying that the literature supported its use in this specific analysis. Commerce said it "reexamined the academic literature" and found it supported the use of a simple average when the underlying data encompassed full populations of sale prices rather than samples "drawn from a larger universe of sales" (see 2211140064).

In the current remand results, Commerce acknowledged that the Federal Circuit "twice found Commerce’s explanations to be inadequate to justify reliance on a simple average" but nonetheless continued with its use, offering the explanation that its use of the d test was based on "the full universe of sale prices" in the test group, and in the comparison group, those groups were equally reliable. The use of a simple average was reasonable for purposes of calculating the denominator when comparing samples of equal reliability, Commerce said.

In antidumping duty investigations, Commerce can detect masked dumping using differential pricing analyses. The agency breaks down U.S. sales data into sets based on comparable product groups. Commerce then breaks that data into various subsets, including the region where the U.S. sales took place, the purchasers involved in the sales and the time periods of the sales. Commerce will then pick one subset as the "test group" while aggregating the remaining subset into the "comparison group." The agency then employs the Cohen's d test to find whether the test group is significantly different than the comparison group.

Commerce argued that both groups were "100% reliable full populations” and it is the reliability of the calculated-from-populations that make the use of a simple average reasonable. "Even though an estimated value is not 100% reliable, it does not follow that two equally reliable samples, which are equally reliable because of equal sample size, should not be equally weighted."

The department said that this time, it provided reasonable justification to support departing from the U.S. Court of Appeals for the Federal Circuit’s understanding of the academic literature. Commerce's justifications were “unsupported but not unsupportable,” Judge Claire Kelly said when she sent the results of the third remand back in April (see 2304040009).

Mid Continent and Commerce didn't immediately respond to requests for comment.