ITA Proposes to Normally Use Random Sampling to Select AD Respondents

Toward this end, the ITA is requesting comments by January 18, 2011 on a proposed random sampling methodology for respondent selection and related issues (e.g. voluntary respondents, adverse facts available rates, etc.) in antidumping proceedings.

(In support of the President's National Export Initiative (NEI), the Commerce Department recently announced 14 upcoming proposed changes to AD/CV duty practices, including random respondent sampling, on which it would seek input on and introduce in practice. See ITT's Online Archives or 08/27/10 news, 10082711, for BP summary.)

ITA Often Selects Largest Volume Companies as Respondents

When the number of producers/exporters (companies) involved in an AD investigation or review is so large that the ITA finds it impracticable to examine each company individually, the ITA has statutory authority to limit its examination to (1) a sample of exporters, producers, or types of products that is statistically valid based on the information available to the administering authority at the time of selection, or (2) exporters and producers accounting for the largest volume of subject merchandise from the exporting country that can be reasonably examined.

Current Process Overlooks Companies Selling Lower Volumes

The ITA has, to date, used the second option in virtually every one of its proceedings. A consequence of this practice is that companies under investigation or review with relatively smaller import volumes have typically not been selected by the ITA for individual examination.

ITA Proposes Random, Stratified Sampling Methodology

To ensure the statistical validity of the samples, the ITA proposes to employ a sampling technique that (1) is random, (2) is stratified, and (3) uses probability-proportional-to-size (PPS) samples. The ITA proposes the following methodology:

Determining When to Sample

Where possible, the ITA proposes to use sampling to select respondents rather than limiting its examination to companies accounting for the largest import volume that can be reasonably examined. However, the ITA would, in general, forgo sampling under the following circumstances:

1. If, due to resource constraints, the ITA is unable to examine at least three companies,
2. When the largest companies by import volume account for at least 75 percent of total imports, or
3. When characteristics of the underlying population make it highly likely that results obtained from the largest possible sample, given resource constraints, would be unreasonable to represent the population.

To make a determination under (3) above the ITA proposes to announce a ten-day period for interested parties to comment on the existence of significant variation in company characteristics that are likely to have a substantial effect on the variation in dumping margins of the companies in the population in question. If the ITA receives any comment, there would be a five-day rebuttal period before the ITA announces its decision on the respondent selection method for that segment of the proceeding.

Definition of Sample Population

Currently, the ITA does not require any evidence of shipment from a non-selected company before making its respondent-selection decision. However, in the sampling context, the existence of shipments would be required in order to both define the population, and if the company is selected, establish a dumping margin for the company.

Therefore, the ITA proposes to normally use CBP data as evidence of shipment, and to define the relevant population from which to sample as (1) all companies subject to investigation with shipments of subject merchandise, and (2) all companies named in a review with shipments of subject merchandise.

In a non-market economy (NME) case, the relevant population should not include companies that are a part of the NME entity. Since the ITA may not be able to determine a company’s eligibility for a separate rate before respondent selection, the ITA proposes to exclude from the relevant population companies that have not submitted separate rate applications.

Sampling Technique

The ITA proposes to use stratified PPS samples. The first step in the proposed sampling technique is to sort all companies in the relevant population from largest to smallest, based on import volumes. Second, companies would be segregated into a number of strata equal to the sample size, with each stratum accounting for approximately the same share of import volume. Third, one respondent from each stratum would be selected using PPS.

If a single company accounts for more than 33 percent or more of imports, that company would be assigned its own stratum and the remaining companies would be divided into two strata accounting for an equal share of the remaining imports. Then, one respondent would be selected from each of these strata. If two companies each account for more than 33 percent or more of imports, each of the two companies would be assigned its own stratum and one respondent would be randomly selected from the remaining companies.

Calculating and Assigning Rates

After examination of selected respondents by the sampling method, the ITA proposes to calculate a “sample rate,” which would be an average of all selected respondent rates, weighted by the import share of their corresponding strata. In a market economy case, all companies in the relevant population who were not selected for individual examination would receive the sample rate.

In NME cases, only companies in the relevant population that qualify for separate rates would receive the sample rate; those that do not qualify for separate rates would receive the NME country-wide rate.

Comments also Sought on Replacement/Voluntary Respondents, AFA/Zero Rates, Etc.

In addition to comments on the methodology, the ITA requests comments on how it should (1) address the case in which a selected respondent needs to be replaced, due to withdrawal or disqualification; (2) treat voluntary respondents in the sampling context; and (3) treat adverse-facts-available, de minimis, and zero AD duty rates in its calculation of the sample rate.

(See ITT's Online Archives or 12/16/10 news, 10121635, for BP summary of ITA's request for comments on the NME separate AD rate test.)

ITA contact - Albert Hsu (202) 482-4491

(FR Pub 12/16/10)