Federal Circuit Judges Probe Use of Statistical Test in Unmasking Targeted Dumping

During oral argument at the U.S. Court of Appeals for the Federal Circuit, three judges questioned the use of a particular statistical test, the Cohen's d test, that is used to identify "masked" dumping in antidumping proceedings. The inquiry built off a July 2021 Federal Circuit ruling that called the use of the test into question since the Commerce Department failed to fulfill certain statistical requirements before running the test (Mid Continent Steel & Wire v. United States, Fed. Cir. #21-1747).

In antidumping duty investigations, such as the one contested by appellant PT Enterprise into steel nails from Taiwan (see 2107270042), Commerce may identify goods that are dumped into the U.S. market through "masked" dumping. Since Commerce typically conducts its investigations by comparing the average home market price of the good in question with its U.S. price, exporters may work around this by dumping the goods in certain areas and selling them at a higher price in another or at another time. To combat this, Commerce may compare the weighted average of sales in the home country to individual sales prices.

Before conducting this analysis though, Commerce must first detect the masked dumping using a differential pricing analysis. The agency breaks down the U.S. sales data into sets based on comparable product groups. Once in the product group, 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 even 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 from the comparison group. If it is, Commerce applies a "ratio test" to see if the ratio of significantly different transactions warrants using the weighted average to individual transaction comparison.

However, in Stupp Corporation v. U.S., the court issued a broad challenge to the test, questioning its overall statistical validity based on certain assumptions the test requires (see 2107150032). The creator of the test, statistician Jacob Cohen, required that his test be completed with normally distributed data of sufficient size and of roughly equal variances. At issue in the Mid Continent case is whether a simple average to individual transaction comparison or a weighting of the standard deviations by their respective quantities of sales is permissible after using the Cohen's d test to uncover masked dumping.

"Where does Cohen or any other literature say you can use a simple average to get at this when the two groups are different in size?" Judge Richard Taranto asked. "I understand that Commerce has an argument that basically says, statistics be damned, we just have a theory about equal stature of the two groups, and it doesn't matter if we're doing anything with an accepted statistical foundation, but you've also suggested that there is an acceptable statistical foundation for this, and that is what I'm simply failing to see."

Mid Continent argued that these assumptions, such as normal distribution of the data, may not even need to be met in some cases, as the simple average is the way to go. Looking beyond a simple average, Mid Continent argued that PT's proposed method of weighting the standard deviations by sales quantities is completely unsupported. "If you were to weight average at all, you have to do it by the number of observations," said Adam Gordon, counsel for Mid Continent. "The literature supports using either a simple average or a weighted average in different variations. Nothing in the literature supports an approach to weighting based on volume." Introducing volume creates distortions in the data that make it unusable, Gordon said.

Taranto questioned this claim, pointing out that weighting by volume is the only thing the statistical literature discusses when comparing two groups with a different number of observations, as is the case when comparing a test group with the comparison group. Gordon then cited a study that looks at the attitudes of three religious groups toward the United Nations to back the use of a simple average instead of weighting by volume. "The issue before the court on appeal is whether Commerce's use of a simple average to calculate the pooled standard deviation is reasonable, the respondent is saying no it's not, use our approach, which has no foundation in the literature and would introduce distortions in the calculation that would debase this goal of calculating on a reliable, predictable basis," Gordon said.