Respondents Question Commerce Calculations in Remand Results on Steel Nails From Taiwan
The Commerce Department's failure to use a weighted average to calculate the denominator of the Cohen’s d test coefficient in an antidumping duty case remand was contrary to the underlying academic literature and produced unreasonable results, a group of AD respondents said in their Oct. 2 comments to the Court of International Trade. The respondents argued that Commerce failed to explain its choice in the face of its own practice and judicial precedent and have asked the court to remand the case to Commerce with instructions to calculate the denominator of the Cohen’s d equation either by weight averaging the standard deviations of the test group and the comparison group or by relying on the standard deviation of the entire population (Mid Continent Steel & Wire v. U.S., CIT Consol. # 15-00213).
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.
In September, Commerce unveiled the results of its remand redetermination of the AD investigation on steel nails from Taiwan, in which it continued to use the simple average to calculate the denominator of the Cohen’s d test coefficient (see 2309010027).
The respondents argue that Commerce failed to provide an “adequate explanation for its choice of simple averaging," as required by both the U.S. Court of Appeals for the Federal Circuit and the CIT. The literature upon which Commerce relied states that the denominator "should never be based on simple averaging of unequal sized groups," the respondents said. That literature "creates an extraordinarily high bar for Commerce to overcome in this fourth bite at the apple."
Commerce's explanation on remand that the two groups are equally reliable does not mean that the denominator can be calculated using a simple average, the respondents said. Equality in size, which is not even the case here, does mean equality in reliability, the companies said.
The equal reliability rationale fails because the two groups are not independent of each other, the respondents said. Each sale is a member of multiple groups. In a simple average methodology, a particular sale will receive more weight than other sales for certain comparisons, the respondents argued. In comparison, they say that a weighted average methodology would treat each kilogram equally relative to all others sold.
The courts have consistently agreed that weighted averaging is preferable to simple averaging and Commerce itself has stated that weighted averaging "more accurately reflects overall trade by accounting for relative import volumes," the respondents said. Commerce’s argument that the case involves “advanced statistical concepts” should be easily dismissed, the companies said. The decision to rely on a specific average is no more complex than in any of the other analyses in which the courts and Commerce have held that a weighted average is preferable.
Finally, the respondents argue that a weighted average would have led to a more reasonable result. They point to Commerce using a weighted average in calculating the d test numerator, a method that isn't in dispute. "Each kilogram affects the numerator of Cohen’s d in one way but affects the denominator in a different way," which they argue undermines the validity of the resulting ratio. The respondents submitted a statistical analysis and concluded a weighted-average would have been more reasonable to achieve Commerce's goal of determining whether prices in two groups were substantially different, the brief said.