495 F.Supp.3d 1298
Ct. Intl. Trade2021Background
- Commerce investigated antidumping of certain steel nails from Taiwan and applied a differential-pricing analysis using Cohen’s d to detect pricing "patterns" among purchaser/region/time subgroups.
- In calculating the pooled standard deviation for Cohen’s d, Commerce used a simple (unweighted) average of the two groups’ variances, rather than a weighted average by transactions, volume, or value.
- Commerce’s preliminary and final determinations applied the mixed (A-to-T for passing sales / A-to-A for others) methodology in the final determination for respondent PT, producing a positive dumping margin.
- PT challenged Commerce’s use of a simple average; the CIT initially sustained Commerce, but the Federal Circuit (Mid Continent III) vacated in part and remanded, directing Commerce to better explain why a simple average (rather than a weighted average) was appropriate.
- On second remand Commerce defended the simple-average approach as better reflecting two distinct, equally-weighted pricing behaviors and avoiding dilution by a larger subgroup; the CIT sustained Commerce’s Second Remand Results and entered judgment for the government.
Issues
| Issue | Plaintiff's Argument | Defendant's Argument | Held |
|---|---|---|---|
| Whether Commerce reasonably used a simple (unweighted) average for the pooled standard deviation in Cohen’s d. | Simple averaging is inconsistent with the statistical literature and the statute; weighted average (by observations/volume/value) is correct. | Simple average better captures two distinct, equally valid pricing behaviors and prevents dilution of a smaller group’s variance; statute allows Commerce’s method. | Court sustained Commerce: simple average is reasonable in this context. |
| Whether PT’s numerical examples show simple averaging distorts Cohen’s d and produces unreasonable results. | Examples show simple average yields a dumping margin while weighted average gives de minimis—proving distortion. | Examples are insufficient; differing outputs alone do not prove unreasonableness; Commerce’s statutory objective and rationale control. | Court found PT’s examples unpersuasive and upheld Commerce’s explanation. |
| Whether ANOVA/statistical principles require weighted averaging. | ANOVA and cited statistical principles establish that pooled variance should be weighted. | PT largely failed to exhaust this argument administratively; Commerce considered the literature and explained why simple averaging fits the non-sampling, context-specific task. | Court rejected PT’s ANOVA claim as not overturning Commerce’s reasoned choice. |
| Whether PT’s additional examples/arguments on remand are barred by exhaustion. | PT sought to submit further examples as clarification and rebuttal to Commerce’s draft; should be considered. | Many new arguments/evidence were not raised to Commerce and are barred by exhaustion; to the extent they merely illustrate earlier points, they are admissible. | Court applied exhaustion: barred truly new matters but considered illustrative material; no reversible error. |
Key Cases Cited
- Mid Continent Steel & Wire, Inc. v. United States, 940 F.3d 662 (Fed. Cir. 2019) (vacating in part and remanding for further explanation of averaging choice in Cohen’s d analysis)
- Mid Continent Steel & Wire, Inc. v. United States, 219 F. Supp. 3d 1326 (Ct. Int’l Trade 2017) (CIT sustained Commerce’s differential-pricing determination)
- Fujitsu General Ltd. v. United States, 88 F.3d 1034 (Fed. Cir. 1996) (agency interpretations reasonable if not the only plausible view; deference to technical economic decisions)
- Consol. Bearings Co. v. United States, 348 F.3d 997 (Fed. Cir. 2003) (exhaustion of administrative remedies doctrine applies in CIT review)
- Apex Frozen Foods Private Ltd. v. United States, 208 F. Supp. 3d 1398 (Ct. Int’l Trade 2017) (addressing when issues are considered exhausted/previously developed before the agency)