The U.S. Department of Commerce has issued a formal request for public comment on potential alternatives to its long-used Cohen’s d test in detecting patterns of differential pricing in antidumping investigations. This action follows a recent Federal Circuit ruling that found Commerce’s application of the d-test unreasonable when used on datasets that fail to meet critical statistical assumptions.
The notice, published by the Enforcement and Compliance division of the International Trade Administration, seeks responses by May 30, 2025, specifically addressing how Commerce should determine whether “there is a pattern of export prices (or constructed export prices) for comparable merchandise that differ significantly among purchasers, regions, or periods of time” under section 777A(d)(1)(B)(i) of the Tariff Act of 1930.
“Commerce is considering possible alternatives to the current approach… to identify if there is a pattern of prices for comparable merchandise that differ significantly among purchasers, regions, or periods of time,” the agency wrote.
Commerce uses various price comparison methodologies in antidumping investigations. Under statute, it may depart from standard comparison methods—average-to-average or transaction-to-transaction—when it finds a significant pattern of price differences that are not otherwise captured by those methods. In such cases, the agency may use an average-to-transaction approach to account for “masked” dumping—low-priced transactions offset by high-priced ones.
For over a decade, Commerce relied on Cohen’s d statistic within its differential pricing analysis framework to identify significant price variation. However, in Marmen Inc. v. United States, the Federal Circuit ruled on April 22, 2025 that applying Cohen’s d is statistically unsound when the underlying data violate assumptions of normal distribution, equal variances, or adequate sample size.
Although the court’s mandate has not yet issued and the decision is not final, the ruling signals significant scrutiny of Commerce’s methodology. According to the decision:
“It is unreasonable to use the Cohen’s d test… when applied to data that do not satisfy the statistical assumptions of normal distribution, equal variances, and sufficiently numerous data.” — Marmen Inc. v. United States, 2025 U.S. App. LEXIS 9506
Legal experts say the Marmen decision could have wide-reaching implications. In commentary published by trade law firms following the decision, attorneys noted that many past determinations may now be vulnerable to challenge unless Commerce adopts a more robust or flexible statistical approach.
“The Federal Circuit’s decision casts doubt on years of administrative practice,” said a recent bulletin from Cassidy Levy Kent, warning that “the Department must find a test that meets both statistical rigor and legal clarity.”
“This is not merely a technical tweak. It opens a broader conversation about transparency, predictability, and fairness in the administration of trade remedy laws,” noted King & Spalding in its client advisory.
Commerce emphasized that the Marmen ruling does not yet compel an immediate change but justifies proactive consultation:
“There is no requirement for specific agency action in response to this decision at this time… however, we are not precluded from seeking public comment regarding potential alternatives.”
Submission Guidelines
Comments must reference Docket No. ITA-2025-0004
Contact: Melissa Porpotage, Enforcement and Compliance Communications Office, (202) 482-1413, ECCommunications@trade.gov
Document ID: 2025-08914
Federal Register Publication Date: May 15, 2025
Comment Deadline: May 30, 2025
⸻
Sources:
• Federal Register Notice, Docket No. ITA-2025-0004
• Marmen Inc. v. United States, 2025 U.S. App. LEXIS 9506
• Section 777A(d)(1)(B)(i) of the Tariff Act of 1930, as amended
• Commentary from King & Spalding and Cassidy Levy Kent client alerts
Comments
No comments on this item Please log in to comment by clicking here