[Federal Register Volume 67, Number 221 (Friday, November 15, 2002)]
[Notices]
[Pages 69186-69197]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 02-29065]


=======================================================================
-----------------------------------------------------------------------

DEPARTMENT OF COMMERCE

International Trade Administration


Antidumping Proceedings: Affiliated Party Sales in the Ordinary 
Course of Trade

AGENCY: Import Administration, International Trade Administration, 
Department of Commerce.

ACTION: Modification concerning affiliated party sales in the 
comparison market.

-----------------------------------------------------------------------

SUMMARY: The Department of Commerce is modifying its methodology in 
antidumping proceedings concerning the determination of whether sales 
to affiliated parties in the comparison market are made in the ordinary 
course of trade and thus may be considered for use in calculating 
normal value. The schedule for implementing this change is set forth in 
the ``Timetable'' section, below.

FOR FURTHER INFORMATION CONTACT: Kris Campbell (202) 482-1032, Office 
of Policy, Import Administration, International Trade Administration.

SUPPLEMENTARY INFORMATION: 

Background

    This change in methodology concerns the test used in antidumping 
proceedings to determine whether comparison market sales between 
affiliated parties are made at arm's length and thus may be considered 
to be within the ``ordinary course of trade.''
    Article 2.1 of the Agreement on Implementation of Article VI of the 
General Agreement on Tariffs and Trade 1994 (the ``AD Agreement'') 
requires that investigating authorities exclude sales not made in the 
``ordinary course of trade'' from calculations of normal value.\1\ 
Section 773(a)(1) of the Tariff Act of 1930, as amended (``the Act''), 
implements this provision by restricting comparison market sales used 
to determine normal value to those made in the ordinary course of 
trade. Under current Department practice, comparison market sales by an 
exporter or producer to an affiliated customer are treated as having 
been made at arm's length, and may be considered to be within the 
ordinary course of trade,\2\ if prices to that affiliated customer are, 
on average, at least 99.5 percent of the prices charged by that 
exporter or producer to unaffiliated comparison market customers.
---------------------------------------------------------------------------

    \1\ Article 2.1 states: ``For the purpose of this Agreement, a 
product is to be considered as being dumped, i.e., introduced into 
the commerce of another country at less than its normal value, if 
the export price of the product exported from one country to another 
is less than the comparable price, in the ordinary course of trade, 
for the like product when destined for consumption in the exporting 
county.''
    \2\ Such sales may be outside the ordinary course of trade for 
other reasons, e.g., if they are below cost.
---------------------------------------------------------------------------

    Under this 99.5 percent test, the Department determines the 
weighted-average comparison market selling price for each product for 
sales by the exporter or producer to each affiliated party. The 
Department also determines the weighted-average selling price for each 
product to the group of unaffiliated comparison market customers. For 
each affiliated customer, the Department compares the weighted-average 
price to that affiliate for each product to the weighted-average price 
of the same product to all unaffiliated customers.

[[Page 69187]]

The Department then weight averages the ratios found for all products 
sold to the affiliated customer. If the result shows sales prices to an 
individual affiliated party are, on average, at least 99.5 percent of 
the sales prices to all unaffiliated comparison market customers (i.e., 
the overall ratio is at least 99.5 percent), all of the sales to that 
affiliated party may be treated as being made in the ordinary course of 
trade and may be used in calculating normal value. Otherwise, if the 
prices to the affiliate are, on average, less than 99.5 percent of 
prices to non-affiliates, it is the Department's practice to disregard 
them. Additionally, for affiliates that pass this test (i.e., those 
whose weighted-average prices are above 99.5 percent), the exporter or 
producer may request the exclusion of individual sales to such an 
affiliate upon a showing that such sales are for other reasons outside 
the ordinary course of trade, e.g., the prices are ``aberrationally'' 
or ``artificially'' high.
    In July 2001, the WTO Appellate Body issued a report in a dispute 
involving U.S. antidumping measures on certain hot-rolled steel 
products from Japan (``Japan Hot-Rolled''),\3\ concerning, among other 
things, the Department's determination of whether sales made to 
affiliated parties in the comparison market were made in the ordinary 
course of trade and thus may be considered for use in calculating 
normal value. In its report in Japan Hot-Rolled, the Appellate Body 
found that the Department's application of its 99.5 percent arm's-
length test in the underlying proceeding was inconsistent with the 
obligations of the United States under Article 2.1 of the AD Agreement. 
In the view of the Appellate Body, ``[i]f a Member elects to adopt 
general rules to prevent distortion of normal value through sales 
between affiliates, those rules must reflect, even-handedly, the fact 
that both high and low-priced sales between affiliates might not be 
``in the ordinary course of trade'.'' \4\ Furthermore, ``the duties of 
investigating authorities, under Article 2.1 of the Anti-Dumping 
Agreement, are precisely the same, whether the sales price is higher or 
lower than the `ordinary course' price, and irrespective of the reason 
why the transaction is not in the ordinary course of trade. 
Investigating authorities must exclude, from the calculation of normal 
value, all sales which are not made in the ordinary course of 
trade.''\5\ However, investigating authorities do not need to utilize 
identical rules to scrutinize each category of sales that is 
potentially not in the ordinary course of trade.\6\ WTO Members are 
afforded discretion in this determination, but such discretion must be 
exercised in an ``even-handed'' manner.\7\
---------------------------------------------------------------------------

    \3\ Dispute Settlement Panel Report on Japan complaint 
concerning U.S. Anti-dumping Measures on Certain Hot-Rolled Steel 
Products from Japan, WT/DS184/R (Feb. 28, 2001) (``Panel Report.'' 
Appellate Body Report on Japan Complaint Concerning U.S. Anti-
dumping Measures on Certain Hot-Rolled Steel Products from Japan, 
WT/DS184/AB/R (July 24, 2002) (``AB Report'').
    \4\ AB Report, paragraph 148.
    \5\ Id., paragraph 145.
    \6\ Id., paragraph 146.
    \7\ Id., paragraph 148.
---------------------------------------------------------------------------

    The United States and Japan entered into arbitration over the 
period of time in which to implement the Appellate Body's findings in 
the Japan Hot-Rolled dispute. The arbitrator found that the United 
States has until November 23, 2002, for implementation.
    On August 15, 2002, we solicited public comment on our proposed 
modification to practice with respect to treatment of affiliated party 
sales in the comparison market.\8\ We received numerous comments and 
rebuttal comments submitted pursuant to this notice, as discussed 
below.
---------------------------------------------------------------------------

    \8\ ``Request for public comment pursuant to section 
129(g)(1)(C) of the Uruguay round Agreements Act,'' 67 FR 53339 
(August 15, 2002) (``Proposed Modification'').
---------------------------------------------------------------------------

Final Modification to Arm's-Length Methodology

    The final modification to the Department's arm's-length test is the 
same as the proposed modification, with the exception of comparing 
prices of ``similar'' products where an identical comparison product 
was not sold to unaffiliated parties, as described below. The new test 
will provide that, for sales by the exporter or producer to an 
affiliate to be included in the normal value calculation, those sales 
prices must fall, on average, within a defined range, or band, around 
sales prices of the same or comparable merchandise sold by that 
exporter or producer to all unaffiliated customers. The band applied 
for this purpose will provide that the overall ratio calculated for an 
affiliate be between 98 percent and 102 percent, inclusive, of prices 
to unaffiliated customers in order for sales to that affiliate to be 
considered ``in the ordinary course of trade'' and used in the normal 
value calculation. This new test is consistent with the view, expressed 
by the WTO Appellate Body, that rules aimed at preventing the 
distortion of normal value through sales between affiliates should 
reflect, ``even-handedly,'' that ``both high and low-priced sales 
between affiliates might not be ``in the ordinary course of trade'.''
    The single change from the proposed arm's-length methodology 
involves comparing prices of products sold to affiliates with prices of 
non-identical products sold to unaffiliated customers, with an 
adjustment for physical differences in the products, where there is no 
identical product sold to non-affiliates. This methodology corresponds 
to that used in comparing prices of products sold in the U.S. and 
comparison markets in the dumping analysis. In comparing prices across 
markets, the Department first seeks to match U.S. sales with comparison 
market sales of identical merchandise. If there are no appropriate 
sales of identical merchandise in the comparison market, the Department 
seeks the most comparable merchandise based on the relevant product 
matching characteristics. When comparing non-identical merchandise, the 
Department makes an adjustment, where appropriate, to normal value for 
differences in physical characteristics.\9\ This adjustment normally is 
based on differences in the variable costs of manufacturing 
attributable to the physical differences between the products.\10\ 
While product characteristics differ from case to case, the Department 
generally does not compare a comparison market product to a given 
product sold in the United States if the difference in variable 
manufacturing costs of the two products is greater than 20 percent.
---------------------------------------------------------------------------

    \9\ See section 773(a)(6)(C)(ii) of the Act.
    \10\ See 19 CFR 351.411.
---------------------------------------------------------------------------

    We plan to employ a corresponding methodology, including 
adjustments for differences in variable costs and application of the 20 
percent ``difmer cap,'' in analyzing non-identical product matches 
between sales to affiliated and unaffiliated customers for purposes of 
the arm's-length test. In many cases the information needed, including 
matching criteria and variable and total cost information, will be on 
the record pursuant to our standard information requests.\11\ Where we 
lack the necessary information we will limit our analysis to identical 
merchandise, consistent with our current methodology. That is, we will 
determine

[[Page 69188]]

the overall ratio for a given affiliate only on the basis of sales of 
those products that were also sold to non-affiliates.
---------------------------------------------------------------------------

    \11\ In determining product matches across markets, the 20 
percent difmer cap is calculated by dividing the difference in 
variable manufacturing costs between the two porducts by the total 
manufacturing costs of the U.S. product. For the arm's-length test, 
we will divide the difference invariable manufacturing costs between 
the two products by the total manufacturing costs of the product 
sold to the affiliated party. Variable manufacturing costs for home 
market sales normally are requested in all cases, while total 
manufacturing costs for home market sales currently are requested 
incases involving below-cost inquiries.
---------------------------------------------------------------------------

    The inclusion of comparisons of non-identical matches will enhance 
the reliability of the arm's-length test by increasing the pool of 
sales used to calculate the affiliate-specific ratios that are assessed 
against the 98-102 percent band. While some of the public comments 
submitted expressed concern that comparing non-identical merchandise 
will add unnecessarily to the complexity of the arm's-length test, or 
will otherwise increase the chance of error resulting from data not 
fully analyzed at the time the arm's-length test is conducted, we 
believe the benefits of bringing these matches within the ambit of the 
test outweigh these concerns.\12\
---------------------------------------------------------------------------

    \12\ See also ``Analysis of Public Comments,'' comment 5, below.
---------------------------------------------------------------------------

    Finally, as noted in the Proposed Modification and as further 
discussed in the ``Comments'' section below, we will continue our 
present practices with regard to the use of so-called ``downstream'' 
sales (sales made by an affiliated buyer to that buyer's subsequent 
customer). Specifically:
    1. If sales to all affiliates account for less than five percent of 
all comparison market sales, we normally will disregard downstream 
sales.
    2. If sales to an affiliate fail the arm's-length test, and (1) 
does not apply, we normally will request the affiliate's downstream 
sales and use those instead of the sales which failed that test.
    3. If a respondent has cooperated to the best of its ability and is 
unable to obtain downstream sales, we will not use adverse facts 
available for those sales.

Analysis of Public Comments

    Numerous comments and rebuttal comments were submitted in response 
to the Proposed Modification. We have carefully considered each of the 
comments submitted. While we have not adopted suggested alternatives to 
the proposed 98-102 percent band test, the comments were useful in 
helping to clarify the concepts underlying the ``ordinary course of 
trade'' analysis and in refining the test by allowing for comparisons 
of non-identical products. As such, we are grateful to those who took 
the time to comment on this aspect of the Department's antidumping 
methodology. Specific proposals are summarized below, along with our 
response to each. For more detail on the comments submitted, see the 
Department's web site at http://ia.ita.doc.gov, where all comments 
received have been posted in their entirety.

1. Proposals for Automatically Disregarding Comparison Market Sales 
Between Affiliates and Requesting Downstream Sales

    A number of commenters proposed that the Department should presume 
that comparison market sales between affiliates are always made outside 
the ordinary course of trade, and should automatically request 
downstream sales (sales from the affiliated purchaser to unaffiliated 
customers). These commenters maintain that such a methodology would be 
consistent with the Appellate Body report in Japan Hot-Rolled, which 
explicitly allowed for the use of downstream sales in determining 
normal value, and would also bring the normal value analysis into 
alignment with the analysis for U.S. sales, in which sales between 
affiliates are automatically disregarded. In the view of these 
commenters, such a methodology would reflect the fact that affiliated 
party sales are inherently suspect and subject to manipulation. They 
also suggest that the Act explicitly allows for use of comparison 
market downstream sales while it does not require the use of prices 
between affiliates. However, one commenter who recommends this approach 
acknowledges that it would require a change in the Department's 
regulations, in particular 19 CFR 351.403(c)-(d). This commenter 
recommends that the change in practice be accompanied by an 
announcement that the Department intends to change the regulations to 
conform to the new practice.
    Several commenters objected to this proposal. Some asserted that it 
is contrary to U.S. law, claiming that the Department must examine all 
sales in the ordinary course of trade, and citing section 773(f)(2) of 
the Act in support of the general proposition that the Department must 
make an affirmative finding that transactions between affiliates do not 
fairly reflect market value before disregarding them. Others claimed 
that it is contrary to U.S. regulations, and also is likely to give 
rise to problems of WTO consistency with respect to the obligation to 
make fair comparisons.
    Department's Position: While we disagree with the comment that U.S. 
law prohibits requesting downstream sales in lieu of upstream sales to 
affiliated parties,\13\ we are not adopting the proposal to 
automatically disregard sales to affiliates. As we stated in the 
Proposed Modification and as acknowledged by at least one proponent of 
automatically excluding sales to affiliates, this proposal conflicts 
with the assumptions underlying the Department's regulations on 
affiliated party sales (19 CFR 351.403(c)-(d)) that such sales normally 
will be used in the dumping analysis if shown to be in the ordinary 
course of trade.
---------------------------------------------------------------------------

    \13\ Discretion to request downstream sales is explicit in 
section 773(a)(5) of the Act (``If the foreign like product is sold, 
or, in the absence of sales, offered for sale through an affiliated 
party, the prices at which the foreign like product is sold (or 
offered for sale) by such affiliated party may be used in 
determining normal value.'').
---------------------------------------------------------------------------

    We do not believe it necessary or appropriate to change these 
provisions, as suggested by one commenter. The current regulations were 
developed after extensive comment, including comment on the issue of 
whether to require in all cases that respondents report downstream 
sales. In our view, the regulatory scheme for reporting and analyzing 
affiliated party sales established by 19 CFR 351.403(c) and (d) strikes 
the appropriate balance between seeking to use first-level sales from 
the respondent where such sales can be demonstrated to be within the 
ordinary course of trade, and requiring downstream sales where sales to 
affiliates do not meet this standard. While this approach does not look 
to downstream sales automatically, it places an affirmative obligation 
on respondents to report such sales where sales to an affiliate cannot 
be shown to be at arm's length. As noted in the preamble to the 
regulations, the Department ``will require a respondent to demonstrate 
in each segment of an AD proceeding that the reporting of downstream 
sales is not necessary.''\14\ This is accomplished in practice by 
maintaining a requirement that respondents report downstream sales for 
all affiliated party sales that do not pass the arm's-length test.
---------------------------------------------------------------------------

    \14\ Preamble to Dep't of Commerce Regulations, 62 FR 27296, 
27356 (May 19, 1997) (``Preamble'').
---------------------------------------------------------------------------

2. Proposals for Using Statistical Testing Methods Instead of a 
Percentage Band Approach

    Several commenters suggested that the Department incorrectly 
rejected statistically valid testing (e.g., standard deviation, 
difference in means, non-parametric tests) in the Proposed Modification 
in favor of the 98-102 percentage band approach. One commenter took 
issue with the reasons given in the Proposed Modification for not 
relying on statistical testing in determining whether sales are made in 
the ordinary course of trade, in particular the statement that ``[s]uch 
tests, properly applied, would allow

[[Page 69189]]

certain affiliated party sales to be deemed in the ordinary course of 
trade, including sales with prices below unaffiliated sales prices, 
that we believe would distort dumping calculations.''\15\ This 
statement, according to the commenter, is results-oriented reasoning 
because the Department is focusing on low-priced sales to affiliates 
and expressly rejecting statistical tests on the basis that, when 
properly applied, these tests would not exclude affiliated party 
transactions that the Department believes would result in the 
calculation of ``distorted'' margins. This commenter suggests that the 
concern over distorted margins is inappropriate in this context, since 
statistical approaches, if properly structured, by definition are 
intended to operate in a mathematically neutral manner.
---------------------------------------------------------------------------

    \15\ Proposed Methodology at 53340-53341.
---------------------------------------------------------------------------

    Another commenter proposed standard deviation testing as an example 
of a statistically valid methodology more suitable to identifying 
outlier transactions than the percentage band approach. Citing a 
proposal for such testing by one of the Japanese respondents in the 
investigation underlying the Japan Hot-Rolled report, this commenter 
suggests that, in general, respondents should be allowed on a case-by-
case basis to propose alternative testing methods that are reasonable 
and easy to administer.
    Department's Position: While we appreciate the desire for a 
statistical-testing approach to the arm's-length test, as we indicated 
in the Proposed Modification, we have been unable to identify an 
alternative test that adequately serves the purposes of a dumping 
analysis and can be readily applied in the context of the variety of 
situations we encounter, including situations that involve multiple 
products sold to an affiliate. The comment that the Department's 
reasoning is ``results oriented'' implies that the Department should be 
unconcerned that parties might manipulate pricing to affiliates for 
purposes of a dumping case. We disagree. We do not believe that the 
purpose of the types of statistical tests considered is applicable in 
this context. Moreover, the only specific proposal offered for a 
statistical test would apply the test on a CONNUM-specific basis, which 
is inconsistent with the purpose of evaluating the overall pricing 
relationship between the affiliates. (See comment 6 below.) Therefore, 
we are not persuaded that a statistical test is appropriate in this 
context.

3. Proposals Regarding Appropriate Size of the Band

    A number of commenters proposed that, if the Department decides to 
use a ``band'' approach in determining whether comparison market sales 
to affiliates were made at arm's-length, it should alter the band size 
from the 98-102 percent range set forth in the Proposed Modification. 
Three types of proposals were made in this regard: (1) A wider band 
(e.g., 90-110); (2) a narrower band (e.g., 99.5-100.5); and an 
``asymmetrical'' band (e.g., 99.5-125).
    Those favoring a wider band argue that a 98-102 percent range does 
not sufficiently recognize natural variability within a respondent's 
pricing data, both between customers and over time. This range, 
therefore, will produce results that fail to reflect commercial 
reality, leading to the inappropriate rejection of bona fide arm's-
length sales.
    These commenters suggest that pricing differences of up to ten 
percent can occur in the normal course of business for reasons 
unconnected with affiliation, such as differences in quantities and 
relative differences in bargaining power. One commenter suggested in 
addition that some variability in POI-average prices to affiliates and 
non-affiliates can result from selling in different quantities over 
time to the two groups, e.g., a higher quantity to affiliated customers 
early in the POI and a higher quantity to unaffiliated customers later 
in the POI. Under this scenario, even where there is no variation in 
pricing to affiliates and non-affiliates at any single point in time, 
the affiliate-specific ratios calculated by the Department will show 
variance from average prices to non-affiliates.
    These commenters also contend that a restrictive band for 
determining whether sales to affiliates are within the ordinary course 
of trade is counter to the general preference in both the AD Agreement 
and U.S. law for establishing normal value based on comparison market 
sales. Further, in the event that the Department seeks to replace sales 
that fail the new arm's-length test with downstream sales (as indicated 
in the Proposed Modification), a narrow test may impose overly 
burdensome reporting requirements, in which case it may not be 
considered sufficiently ``even-handed'' as the term is used in the 
Japan Hot-Rolled report.
    Finally, certain commenters favoring a broader band suggest that, 
to the extent there is concern over manipulation of pricing (via 
clustering of sales to affiliates at the low end of the band), the 
Department could test for such pricing patterns upon receipt of a 
respondent's sales databases, and could address such problems on a 
case-by-case basis, through the fictitious markets provision \16\ as 
well as the ordinary course of trade provision.
---------------------------------------------------------------------------

    \16\ Section 773(a)(2) of the Act.
---------------------------------------------------------------------------

    Commenters arguing for a narrower band (99.5-100.5) stress that the 
change in practice under the proposed 98-102 percent band would go 
beyond the requirements of the Appellate Body report in Japan Hot-
Rolled and would enhance respondents' ability to manipulate home market 
sales to mask dumping. One commenter provides a hypothetical example of 
this potential for manipulation, highlighting perceived weaknesses both 
in the range of acceptable prices in the new standard and the fact 
that, as with the old standard, it would be applied on an affiliate-
specific, and not product-specific, basis.\17\ This combination, 
according to the commenter, would allow respondents to make sales to an 
affiliate of products matching to U.S. products at prices significantly 
below the 98 percent threshold (e.g., at 80 percent of prices to non-
affiliates) while still passing the test by selling non-matched 
products to the same affiliate at prices above the threshold (e.g., 120 
percent). This commenter maintains that, while such manipulation is 
possible under the current test, it would be ``dramatically easier'' 
under the proposed 98-102 standard.
---------------------------------------------------------------------------

    \17\ See also comment 6, below, regarding the affiliate-specific 
nature of the test.
---------------------------------------------------------------------------

    Another commenter suggests that, if the Department retains the 98-
102 standard for investigations, it should at a minimum use a 99.5-
100.5 standard for administrative reviews. This approach would place 
the arm's-length test on a consistent footing with the two percent and 
0.5 percent de minimis dumping standards used in investigations and 
reviews, respectively.
    Linking the standards used in the arm's-length test with those used 
in determining de minimis dumping would, according to this commenter, 
reduce any perceived arbitrariness over the range selected, thereby 
lowering its susceptibility to further WTO challenges. It would also 
reflect the greater potential for manipulation of pricing that can 
occur after imposition of an order than during the initial period of 
investigation.
    Commenters in favor of an ``asymmetrical'' test base their 
arguments on language from a footnote in the Japan Hot-Rolled report 
providing that, ``in finding that the application of the 99.5 percent 
test was not sufficiently even-handed, we do not suggest that the

[[Page 69190]]

methods for verifying whether high and low-priced sales to affiliates 
are `in the ordinary course of trade' must necessarily be 
identical.''\18\ Accordingly, these commenters suggest, the Department 
retains the discretion to tailor an arm's-length test for comparison 
market sales between affiliates geared toward the primary concern in a 
dumping context: namely, low-priced sales designed to reduce normal 
value. These commenters maintain that an asymmetrical test is 
consistent with the WTO report since it imposes a ``bright line'' 
standard for high-priced sales, and is otherwise appropriate because it 
would retain a broader base of profitable sales made in the normal 
course of business than the proposed 98-102 percent test. It would also 
reflect the fact that a different set of circumstances exists for high-
priced sales between affiliates, which are priced as such for internal 
company-specific reasons unrelated to the dumping analysis.
---------------------------------------------------------------------------

    \18\ AB Report, footnote 113.
---------------------------------------------------------------------------

    Department's Position: We have carefully considered each of the 
ranges proposed as alternatives to the 98-102 percent test. While some 
of these ranges (e.g., 99.5-100.5) were previously examined in the 
course of arriving at the Proposed Modification,\19\ we have 
reconsidered all options regarding upper and lower limits of the band 
in light of the arguments and hypothetical situations provided in the 
comments received.
---------------------------------------------------------------------------

    \19\ Proposed Modification at 53340. See also Premable at 27356.
---------------------------------------------------------------------------

    As indicated in the Proposed Modification, the range adopted must 
account for concerns that the band be neither overly narrow, which 
would reduce the utility of the test as few affiliates would pass, nor 
overly broad, which could increase the potential for manipulating 
normal value through clustering of sales prices to affiliates at the 
lower end of the band.\20\ Having considered the alternative 
suggestions regarding the appropriate band size, we continue to believe 
that the 98-102 range strikes the best balance in providing a 
reasonable and predictable means of assessing whether affiliated party 
sales were made at arm's-length prices. First, contrary to the argument 
of advocates for the 99.5-100.5 band, we do not believe that extending 
the lower end of the acceptable range from 99.5 percent to 98 percent 
provides a significant opportunity for manipulation of normal value, 
either in investigations or administrative reviews. The range 
established retains a standard that reasonably ensures that we only use 
sales between affiliates that are appropriate for use in the dumping 
analysis, in light of the fact that such sales are inherently suspect 
unless demonstrated to be in accord with prices negotiated by 
independent parties. While a particular concern arises regarding low-
priced sales between affiliates in an antidumping context, the 
requirement that such sales, on average, fall within two percent of 
average prices to non-affiliates will provide a reasonable means of 
continuing to ensure against such manipulation.
---------------------------------------------------------------------------

    \20\ Proposed Modification at 53340.
---------------------------------------------------------------------------

    As noted, several commenters suggest that the proposed 98-102 
standard will have largely the same effect as a 99.5-100.5 band, 
arguing that sales prices routinely diverge by more than this range in 
the normal course of business, and that the ratio can be affected by 
other factors such as the timing of sales to affiliates and non-
affiliates within the period of investigation. In response, we note 
first that the test recognizes that pricing of individual transactions 
may vary by more than two percent in the normal course of business. 
Such sales may still be found to be at arm's length and included in the 
dumping analysis as long as sales to the affiliate are, on average, 
within the band. The test in this respect is appropriately geared 
toward a recognition that, while individual sales transactions may be 
expected to vary in the normal course of business, systematic 
underpricing or overpricing between affiliates over the period examined 
in the dumping analysis is indicative of sales not made at arm's 
length.
    Second, as discussed in more detail in comment 4, below, in 
comparing prices under the arm's-length test we routinely adjust for 
many of the factors that give rise to differences in pricing, and allow 
for additional adjustments, e.g., for differences in quantities, where 
warranted.
    Third, we disagree with suggestions for a broader band (e.g., 90-
110) coupled with the proviso that, if the Department finds upon 
further analysis that sales to affiliates are clustered at the low end 
of the band, it may then consider, on a case-by-case basis, whether to 
disregard them under either the fictitious market provision or the 
ordinary course of trade provision. The fictitious market provision is 
inappropriate for this analysis; whether or not a fictitious market 
exists, prices between affiliates may not reflect arm's-length 
transactions. Applying the fictitious market standard would not 
adequately serve the purpose of identifying systematic underpricing or 
overpricing between affiliates. Furthermore, as we have stated in past 
cases, it is to be used in exceptional circumstances and not employed 
as a routine part of the Department's analysis.\21\ Such inquiries 
typically require an allegation from an interested party and call for 
analyses based on information that is quantitatively and/or 
qualitatively different from the information normally gathered by the 
Department as part of its standard antidumping analysis.\22\ In 
addition, the suggested approach is not sufficiently in accord with the 
concept that sales between affiliates are inherently suspect until 
demonstrated to be in the ordinary course of trade. In effect, it would 
reverse this concept for certain sales that in our view are suspect, 
requiring an additional finding, on a case-by-case basis, that other 
factors render such sales not at arm's length. Finally, there are 
serious concerns that any such approach would not be reasonably 
administrable within the time limits of an antidumping proceeding, 
particularly given the requirement in most instances for downstream 
sales once a determination is made that sales between affiliates are 
not at arm's length.
---------------------------------------------------------------------------

    \21\ Certain Corrosion-Resistant Carbon Steel Flat Products from 
Japan, 64 FR 12951, 12956 (March 16, 1999).
    \22\ Preamble at 27357.
---------------------------------------------------------------------------

    In light of these concerns, we believe the more appropriate finding 
is that sales below the 98 percent threshold, but within the proposed 
broader band, are outside the ordinary course of trade. However, as 
discussed in comment 4, below, we will consider arguments on a case-by-
case basis that such pricing patterns were determined entirely by 
market factors not captured by the arm's-length test, such as the 
timing of sales made to affiliated and unaffiliated parties during the 
period of investigation.
    Finally, we disagree with suggestions that an ``asymmetrical'' test 
would be consistent with the WTO report in Japan Hot-Rolled or is 
otherwise appropriate as a test for sales not made at arm's length. 
While the Appellate Body provided in a footnote that the tests for 
whether low-priced and high-priced sales to affiliates are in the 
ordinary course of trade did not necessarily have to be ``identical,'' 
this was made in the context of statements that the current test was 
not sufficiently ``evenhanded'' to the extent that it ``operated 
systematically to raise normal value, through the automatic exclusion 
of marginally low-priced sales, coupled with the automatic inclusion of 
high-

[[Page 69191]]

priced sales, except those proved, upon request, to be aberrationally 
high priced.'' The Appellate Body's finding that the application of the 
99.5 percent test in the Japan Hot-Rolled case violates Article 2.1 of 
the AD Agreement was based on its assessment that the test ``focuses 
predominantly'' on the distortion that results from low-priced sales 
and does not ``take equal account of the possibility that prices `above 
the (99.5 percent) threshold' can also `distort' normal value.'' \23\ 
We believe that automatically disregarding sales to affiliates at 
prices below the 99.5 percent threshold while automatically including 
sales at prices up to a 125 percent threshold would be inconsistent 
with this reasoning.
---------------------------------------------------------------------------

    \23\ AB Report, paragraph 157.
---------------------------------------------------------------------------

4. Proposals To Take Into Account Relevant Commercial Circumstances

    Numerous commenters proposed that, in assessing whether affiliated 
party sales were made in the ordinary course of trade, the arm's-length 
test should not focus exclusively on price, but should take into 
account all relevant commercial circumstances. Referencing a statement 
in the Appellate Body report that ``price is merely one of the terms 
and conditions of a transaction,'' \24\ these commenters suggest that, 
to the extent the arm's-length test ignores the commercial 
circumstances pertaining to affiliated and unaffiliated party sales, 
the new methodology will produce distorted results. Suggestions for 
factors to examine include level of trade, customer categories, 
quantities sold, product mix, and any other terms of sale relevant to 
the transactions under examination.
---------------------------------------------------------------------------

    \24\ AB Report, paragraph 142.
---------------------------------------------------------------------------

    The suggestions vary with respect to the relationship between these 
factors and the arm's-length test as described in the Proposed 
Modification. One proposal is that affiliated party sales should be 
found within the ordinary course of trade wherever their terms of sale 
are the same as sales made at the same time to unaffiliated parties. 
Other commenters suggest that, if a price analysis is conducted, it 
needs to ensure that any differences in commercial terms and conditions 
between affiliated and unaffiliated party sales that could impact price 
are taken into account. A third proposal is that the price analysis 
should merely establish a rebuttable presumption that sales to an 
affiliated party are outside the ordinary course of trade, which could 
be countered by other information demonstrating that such sales were in 
fact made under the conditions and practices that are normal in the 
comparison market and, thus, are in the ordinary course of trade.
    Department's Position: As with the current test, the new 
methodology takes account of many of the factors suggested by 
commenters as relevant to the ordinary course of trade analysis. We 
take this opportunity to clarify those aspects of the methodology used 
to establish the affiliate-specific price ratios that relate to this 
issue.
    First, price comparisons between affiliated and unaffiliated party 
sales that are factored into the affiliate-specific price ratios (which 
are then applied against the 98-102 percent range) are made at the same 
level of trade, where appropriate. That is, the arm's-length test 
generally does not compare prices of sales made at different levels of 
trade. Any sales to affiliates for which there are no comparable sales 
to unaffiliated parties at the same level of trade are not used in 
determining the affiliate-specific price ratios. This does not mean 
that such sales are automatically disregarded from use in determining 
normal value, but simply that such sales are not used in determining 
whether, overall, sales to a given affiliate are made at arm's length. 
If, based on the sales that are used in the analysis, it is determined 
that sales to an affiliate were made at arm's length, all sales to the 
affiliate, including sales without comparable unaffiliated sales at the 
same level of trade, are included in the comparison market database 
used to establish normal value.\25\
---------------------------------------------------------------------------

    \25\ Under the current test, the same holds true regarding 
affiliated party sales that have no identical matching unaffiliated 
party sales. See comment 5 regarding the change in methodology 
allowing for non-identical comparisons. In both situations, where 
there are no sales to an affiliate that can be compared with 
unaffiliated party sales, sales to this affiliate would not be used 
in the dumping analysis.
---------------------------------------------------------------------------

    In addition to comparing sales at the same level of trade, the test 
adjusts affiliated and unaffiliated party prices for numerous 
differences relating to the sales. The adjustments account for, among 
other things, differences in packing expenses, movement expenses from 
the original place of shipment, discounts and rebates, and selling 
expenses that relate directly to the sale at issue. While the 
Department's questionnaire specifically requests information pertaining 
to a number of adjustments, it also allows for responding companies to 
claim additional adjustments for other expenses relating to the sales 
at issue. Thus, provided that a respondent has accurately reported its 
claimed differences in circumstances of sale, along with other expenses 
and price adjustments relating to the reported sales, the arm's-length 
test will account for such differences between sales to affiliates and 
non-affiliates.
    With respect to the request by numerous commenters that the test 
also take into account the price effect of any difference between sales 
to affiliates and non-affiliates in quantities sold, we note that 
adjustments for differences in quantity are addressed at Sec.  351.409 
of the Department's regulations. We do not automatically adjust for 
differences in quantities, but will do so under the conditions 
specified in this regulation. Moreover, the fact that the arm's-length 
test makes comparisons only at the same level of trade should reduce 
the number of instances in which sales of significantly different 
quantities are compared. As stated in the preamble to the Department's 
regulations, based on our experience we believe that differences in 
quantity are more likely to occur at different levels of trade.\26\
---------------------------------------------------------------------------

    \26\ Preamble at 27368.
---------------------------------------------------------------------------

    Considering these aspects of the arm's-length test in light of the 
proposals made, we believe the test adequately accounts for the factors 
alleged by the commenters to affect price comparisons between sales to 
affiliated and unaffiliated parties. Beyond this, we are not in a 
position to speculate on any case-specific circumstances that might 
warrant additional consideration. Accordingly, we have not changed the 
test in response to these comments. However, as with other aspects of 
the Department's dumping analysis, parties have a right to submit 
comments on the record of a proceeding regarding the adjustments that 
must be made under the statute in order to ensure a fair comparison. We 
will consider any comments submitted regarding case-specific 
adjustments made in the arm's-length analysis in that light. While this 
does not constitute what one commenter referred to as a rebuttable 
presumption with respect to the results of the 98-102 percent test, and 
is not a change in our practice of generally limiting the analysis to 
pricing as adjusted, as upheld by the Court of International Trade,\27\ 
it does provide a fair opportunity to ensure that all appropriate 
adjustments are made in deriving the affiliate-specific ratios to which 
the band applies.
---------------------------------------------------------------------------

    \27\ NTN Bearing Corp. v. United States, 905 F. Supp1083, 1099-
1100 (October 2, 1995).
---------------------------------------------------------------------------

    Finally, we have not adopted the proposal that equivalent terms of 
sale for affiliates and non-affiliates should conclusively establish 
that affiliated

[[Page 69192]]

party sales are in the ordinary course of trade. This proposal, like 
others offered, appears to be based on a sale-by-sale analysis.\28\ As 
we discuss further below, we do not believe this approach appropriately 
addresses the question of the nature of the relationship between the 
affiliates.
---------------------------------------------------------------------------

    \28\ Comment 6, below, addresses a similar proposal to retain 
all individual affiliated party sales that are priced at the level 
of any unaffiliated party sale considered to be in the ordinary 
course of trade.
---------------------------------------------------------------------------

5. Treatment of Sales to Affiliated Parties of Products Not Sold to 
Unaffiliated Parties

    Certain commenters suggested that the Department alter the manner 
in which it treats sales to affiliated parties of products not sold to 
unaffiliated parties. Currently, as with affiliated party sales that 
cannot be compared to unaffiliated party sales at the same level of 
trade (see comment 4, above), sales to affiliates with no identical 
match to an unaffiliated party sale are not used in determining the 
affiliate-specific ratios that are compared against the 99.5 percent 
threshold.
    However, such sales are not automatically disregarded for 
determining normal value; they are retained in the comparison market 
database if the affiliate passes the arm's-length test based on sales 
that could be compared with unaffiliated party sales.\29\
---------------------------------------------------------------------------

    \29\ A number of comments received on this issue assume that 
such sales are automatically considered to have ``failed'' the 
arm's-length test and, as such, are disregarded in determining 
normal value. One commenter suggests that, in a variant of the test, 
the Department makes an adverse assumption and assigns all 
affiliated party sales of products with no match to unaffiliated 
party sales a CONNUM-specific ratio of 0 percent. While the 
commenter does not cite specific cases employing different 
methodologies, we will ensure that future cases are consistent in 
their treatment of affiliated party sales with no match to 
unaffiliated sales.
---------------------------------------------------------------------------

    One commenter suggested that the new test should seek to compare 
affiliated party sales with sales of non-identical merchandise sold to 
unaffiliated parties, where there are no comparable sales of identical 
merchandise. This revision would, according to this commenter, expand 
the pool of sales used to determine whether pricing to an affiliate was 
made at arm's-length, and would also be in accord with the Department's 
regulations on affiliated party sales. These regulations provide that 
``the Secretary may calculate normal value based on [affiliated party 
sales] only if satisfied the price is comparable to the price at which 
the exporter or producer sold the foreign like product to a person who 
is not affiliated with the seller.'' \30\ The use of the term ``foreign 
like product'' in this context, according to this commenter, indicates 
that the determination of whether affiliated party sales are made at 
arm's length is to be established with reference to the price of 
identical and similar merchandise sold to unaffiliated parties.
---------------------------------------------------------------------------

    \30\ 19 CFR 351.403(c) (emphasis added).
---------------------------------------------------------------------------

    Another commenter suggests an alternative means of including sales 
to affiliates of products lacking an identical match in the arm's-
length analysis; namely, that the Department should assume that such 
sales were made at 100 percent of the price to non-affiliates, and 
factor this into the affiliate-specific ratio.
    Department's Position: As noted in the ``Final Modification to 
Arm's-Length Methodology'' section, above, we intend to match non-
identical merchandise where there are no comparable sales of identical 
merchandise. The reference in the governing regulation to comparing 
prices of affiliated party sales with sales to non-affiliates of the 
``foreign like product'' makes clear that the price of non-identical 
merchandise is appropriate for use in determining whether sales were 
made at arm's length. We expect to be able to make such comparisons 
where the respondent has provided both total and variable home market 
costs, typically in cases involving sales-below-cost inquiries. While 
we will not require total home market costs in non-cost cases solely 
for purposes of making comparisons in the arm's-length test, we will 
accept the reporting of such costs on a voluntary basis in such cases. 
While some commenters maintain that expanding the arm's-length test in 
this manner will add unnecessarily to the complexity of the analysis, 
we believe that comparisons to non-identical merchandise can be 
accommodated within the existing framework for the conduct of 
antidumping proceedings.
    We can see no reason to adopt the alternative proposal for assuming 
sales with no identical match were made at 100 percent of the price to 
unaffiliated parties. There is no claim that such an assumption is 
grounded in fact, and could lead, in effect, to an assumption that 
affiliated party sales were made at arm's length.

6. Comments Regarding Appropriate Level for Determining Whether Sales 
are at Arm's Length: by Individual Sale; by Product; by Affiliate

    As described in the Background section, above, the Department 
currently assesses whether sales were made at arm's length at the level 
of the individual affiliate. Both the methodology used in the 99.5 
percent test and the Proposed Modification weight average the product-
specific price ratios for all products sold to an affiliated customer 
to arrive at an affiliate-specific price ratio. If the result shows 
sales prices to an individual affiliated party are, on average, at 
least 99.5 percent of the sales prices to all unaffiliated comparison 
market customers (under the 99.5 percent test) or between 98-102 
percent, inclusive, of unaffiliated prices (under the 98-102 percent 
test), then all sales to that affiliated party may be treated as being 
made in the ordinary course of trade and may be used in calculating 
normal value, including any sales made at prices below the threshold. 
Otherwise, if the affiliate-specific price ratios do not meet these 
criteria, all sales to the affiliate are generally considered outside 
the ordinary course of trade, including sales at prices above the 98-
102 band.
    A variety of proposals were submitted that would allow the arm's-
length determination to be made on the basis of individual sale prices 
or weighted-average prices by product, as opposed to the affiliate-wide 
determination described above. One commenter suggests that the 
determination should be done on a sale-by-sale basis. Under this 
proposal, any individual sale to an affiliated party would be 
considered as made at arm's-length as long as it is priced at a level 
equivalent to any comparable sale to an unaffiliated party. According 
to this commenter, there is no basis to disregard such sales to 
affiliates where the comparable sale to the unaffiliated party is 
determined to be in the ordinary course of trade. Another commenter 
takes the opposite approach, recommending that all individual sales to 
an affiliate must be found to be priced at levels establishing the 
arm's-length nature of the transaction in order for any sales to the 
affiliate to be used.
    Another commenter proposes a product-specific approach for each 
customer, whereby the product-specific average price, as sold to an 
individual affiliate, must be within the band established for arm's-
length sales in order for such sales to be used in determining normal 
value. According to this commenter, a product-specific approach to 
determining sales in the ordinary course of trade is more in line with 
the rest of the statutory framework for determining normal value, which 
is centered on the price of the foreign like product, i.e., a model-
specific hierarchy of merchandise for comparison. Yet another commenter 
views ``foreign like

[[Page 69193]]

product'' broadly (akin to class or kind) and contends that the arm's-
length analysis should focus on this broad basis, since a corporation's 
pricing decisions are rarely, if ever, made on a CONNUM-specific basis.
    Department's Position: While we have carefully considered each of 
these alternative proposals for the appropriate level at which to 
determine whether affiliated party sales are made within the ordinary 
course of trade, we have decided to retain our normal practice of 
making this determination on an affiliate-wide basis. While certain 
individual sales and products that would pass the test on their own may 
be excluded under this approach, and vice-versa, an affiliate-wide 
analysis does not systematically bias the arm's-length determination in 
one direction or another. Our reasons for preferring that the 
determination of whether sales are made at arm's-length be conducted at 
the level of the individual affiliate were set forth in the 
investigation underlying the AB Report in Japan Hot-Rolled:

    With respect to NKK's concern of applying the arm's-length test 
on a customer basis, we note that the question underlying the arm's-
length test is whether affiliation between the seller and the 
customer has (in general) affected pricing. Because affiliation is 
the result of relationships between firms, the focus of the arm's-
length test is the customer, not a particular product. For this 
reason, the Department makes one up-or-down call on pricing to an 
affiliated customer: Either there is arm's-length pricing or there 
is not. However, under NKK's [product-specific] approach, 
affiliation could be found to matter for some connums, but not for 
others, even though the customer in both cases is the same.\31\

    \31\ Final Determination of Sales at Less Than Fair Value: Hot-
Rolled Flat-Rolled Carbon-Quality Steel Products from Japan, 64 FR 
24329, 24342 (May 6, 1999).
---------------------------------------------------------------------------

    This aspect of the Department's methodology was not at issue before 
either the WTO Panel or the Appellate Body in Japan Hot-Rolled, and we 
do not find sufficient reason to depart from the current approach in 
adopting the new methodology. Moreover, abandoning the focus on the 
pricing relationship with the affiliate would fundamentally alter the 
nature of the test and introduce many complicated questions about other 
aspects of the test as well as use of downstream sales.
    While the explanation cited above pertains to requests for a 
product-specific approach, its rationale applies as well to requests 
for a sale-specific approach. In particular, the proposal to retain an 
individual affiliated party sale if priced at a level equivalent to a 
comparable sale to an unaffiliated party would require that we ignore 
the potential for manipulation that results from the affiliation. Under 
this approach, affiliated party sales could be priced on average far 
below market price and still be retained for determining normal value 
as long as they are made at the price of the lowest individual sale 
price to an unaffiliated customer. The adoption of this method for 
determining arm's-length sales would, therefore, not establish that 
affiliated party sales are appropriate for use in the dumping analysis.

7. Proposals for Treatment of Merchandise ``Consumed'' by Affiliates, 
as Distinguished from Merchandise Resold

    Certain commenters submitted proposals for differentiating between 
sales of the foreign like product ``consumed'' (not resold as subject 
merchandise) by an affiliate and sales to an affiliate that are resold 
as subject merchandise.\32\
---------------------------------------------------------------------------

    \32\ Sales by an affiliate of subject merchandise are referred 
to in the Preamble to the Department's regulations, and in this 
notice, as ``downstream sales.'' Preamble at 27356. Sales from the 
respondent company to the affiliated reseller are described in this 
notice as ``upstream sales.''
---------------------------------------------------------------------------

    One commenter suggested that, when sales to affiliated parties are 
not resold but are instead ``consumed,'' the standard used in the 
arm's-length test should be different. In particular, this commenter 
suggests dropping the requirement that sales, on average, be within the 
band and allowing any individual sales within the band to pass the 
arm's-length test. This commenter suggests that the broader requirement 
that pricing overall to the affiliate be within the band is less 
relevant where an affiliate consumes the merchandise by producing and 
selling a product that is outside the scope of the order.
    Another commenter, while proposing that the Department 
automatically request downstream sales in the case of resales (see 
comment 1, above), suggested applying an arm's-length test in the 
limited instance of sales of merchandise ``consumed'' by an affiliate. 
Alternatively, a third commenter, while agreeing that the Department 
should automatically request downstream resales, suggested eliminating 
sales of merchandise consumed by an affiliate from the analysis. This 
commenter suggests that the Department's concern over a methodology 
that leads to fewer comparisons based on the preferred methodology 
(home market sales) is overstated, given the U.S. Court of Appeals 
ruling in Cemex S.A. v. United States with respect to matching to 
similar merchandise \33\. Further, according to this commenter, 
disregarding all sales to affiliated consumers would not be contrary to 
the Department's regulations or Congressional intent, since the former 
must be read in light of the general suspicion of affiliated party 
sales encompassed in the Statement of Administrative Action 
accompanying the Uruguay Round Agreements Act (SAA), and the latter 
anticipates that Commerce, ``in general,'' will not rely on sales to 
affiliates in determining normal value.\34\
---------------------------------------------------------------------------

    \33\ 133 F.3d 897 (Fed. Cir. 1998) (citing also DOC Policy 
Bulletin 98.1, which specifies that, henceforth, when all sales of a 
particular home market model are below cost, instead of 
automatically resorting to constructed value to determine normal 
value, the Department will first attempt to use prices of a non-
identical model that remains above cost.).
    \34\ Citing H.R. Rep. No. 103-826, at 82 (1994).
---------------------------------------------------------------------------

    Department's Position: Consistent with our current practice and 
with Sec.  351.403(c) of the Department's regulations, we intend to 
continue using sales to affiliates, whether of merchandise consumed or 
resold, to determine normal value where such sales are shown to be at 
arm's length. The comments submitted proposing different treatment of 
sales of merchandise consumed by affiliates do not provide sufficient 
reasons to depart from this practice.
    With respect to the proposal that individual sales of merchandise 
consumed by affiliates should be found to have passed the arm's-length 
test whenever such sales prices are within the established price band, 
no underlying rationale was provided for this difference in treatment 
other than to claim that the affiliate-wide pricing requirement ``makes 
no sense'' as applied to affiliated consumers. We do not believe that 
there is sufficient reason to apply a different standard with respect 
to such sales. Whether the affiliate consumes or resells the subject 
merchandise, the question posed is the same and the test applied should 
be the same.
    With respect to the suggestions that we should automatically 
disregard sales to affiliated consumers, or that we should apply an 
arm's-length test only to such sales while automatically disregarding 
sales to affiliated resellers, our response to comment 1, above, which 
provides our reasons for applying an arm's-length test to upstream 
sales to resellers (as opposed to automatically disregarding such 
sales), applies as well to applying an arm's-length test to sales to 
affiliated consumers and using such sales to establish normal value 
when

[[Page 69194]]

they are demonstrated to be at arm's length. There is insufficient 
reason to apply different methodologies to these two groups of sales to 
affiliated parties. We also note that, to the extent there is ambiguity 
regarding reporting requirements for these two types of affiliated 
party sales, we intend in the future to make clear that sales to 
affiliates, whether consumers or resellers, will be used in the dumping 
analysis where shown to be at arm's length based on the 98-102 price 
band methodology.

8. Other Methodological Proposals for Determining Sales at Arm's Length

    Other proposals made regarding the arm's-length test include:
    [sbull] A proposal by the commenter who recommended a sale-by-sale 
approach to use, as an alternative in the event the sale-by-sale 
approach is not adopted, the quantity-based test described as an 
alternate option in the Proposed Methodology. Under this option, 
affiliated party sales would be found within the ordinary course of 
trade as long as a sufficient quantity of comparable sales to non-
affiliates were priced above and below the affiliated price. This 
commenter believes the Department's concerns over this option, 
centering on complexity, implementation, and uncertainty over the 
appropriate level of quantities needed to pass the test, are 
overstated, and provides examples of how it could be implemented 
without undue difficulty.
    [sbull] A suggestion to apply the arm's-length test only when 
common ownership between affiliates reaches a level of 50 percent or 
more. This approach, the commenter suggests, will more accurately 
reflect those situations where actual control exists sufficient to give 
rise to concerns over manipulation of pricing.
    [sbull] A request for clarification of the methodology with respect 
to a single affiliate with multiple customer codes in the reported home 
market database, due to, for instance multiple billing addresses. This 
commenter requests that Commerce adopt in all cases the methodology 
used in Certain Cold-Rolled Carbon Steel Flat Products from France,\35\ 
where it ``collapsed'' multiple customer codes and performed the arm's-
length test on an aggregate basis.
---------------------------------------------------------------------------

    \35\ 67 FR 31204 (May 9, 2002).
---------------------------------------------------------------------------

    [sbull] A request that the Department explain how a band approach, 
containing an upper-level ceiling on affiliated party prices, is 
consistent with the test applied for valuing inputs sold between 
affiliates, as prescribed at sections 773(f)(2) and (3) of the Act. The 
commenter believes any differences could be interpreted as reflecting 
inconsistent definitions of the term ``foreign like product,'' one 
relating to price-based normal value (arm's-length test) and one 
relating to constructed value (the provisions of the Act cited above). 
This commenter requests that this explanation be made with reference to 
a recent remand by the Court of International Trade (as directed by the 
Court of Appeals for the Federal Circuit), in which the Department was 
asked to clarify why it uses different definitions of the term 
``foreign like product'' for price-based and cost-based 
calculations.\36\ The commenter also references the recent 
determination in Certain Cold-Rolled Carbon Steel Flat Products from 
France,\37\ where, the commenter maintains, the arm's-length and cost 
valuation issues were joined, since a transaction that failed the 
current arm's-length test could be evaluated under the major input rule 
for use in determining input costs.
---------------------------------------------------------------------------

    \36\ FAG Italia v. United States, 291 F.3d 806 (CIT 2002).
    \37\ 67 FR 31204 (May 9, 2002).
---------------------------------------------------------------------------

    [sbull] A request for clarification that, when the Department finds 
an insufficient volume of sales to unaffiliated purchasers, it will 
continue its practice, as noted in the preamble to the Department's 
regulations,\38\ of disregarding affiliated party sales.
---------------------------------------------------------------------------

    \38\ Preamble at 27355.
---------------------------------------------------------------------------

    [sbull] A request that the Department explicitly place on 
respondents the burden of proof for establishing that affiliated party 
sales are in the ordinary course of trade, and clarify that all such 
sales will be disregarded until this burden of proof is met.
    [sbull] A request for clarification regarding whether all 
affiliated party sales that fail the arm's-length test will continue to 
be excluded from the CEP profit calculation. This commenter notes that 
the current practice is centered on low-priced sales falling below the 
99.5 percent threshold,\39\ and asks whether high-priced sales above 
the 98-102 band would also be excluded. This commenter suggests that 
``capping'' the CEP profit calculation by excluding high-priced sales 
that fail the arm's-length test would conflict with the preamble to the 
Department's regulations and with its statutory obligations.\40\
    Department's Position: We respond to each item, in turn. With 
respect to the suggestion favoring the use of a quantity-based test, 
our concerns with this test, as set forth in the Proposed Modification, 
remain despite the suggestions by the commenter. These include, in 
addition to the general complexity and implementation concerns cited by 
the commenter, concerns over whether to apply the test by affiliate or 
for all affiliates combined by product, and questions as whether this 
might not be an overly narrow definition of the ``normal'' price range 
of sales to affiliated parties. We continue to believe the 98-102 
percent band provides a more reasonable, predictable, and administrable 
test.
---------------------------------------------------------------------------

    \39\ Citing Import Administration Policy Bulletin 97.1: 
Calculation of Profit for Constructed Export Price Transactions, at 
pages 3-5 (September 4, 1997).
    \40\ Citing Preamble at 27354 (``the statute does not authorize 
a cap on the amount of profit deducted from CEP'').
---------------------------------------------------------------------------

    With respect to the suggestion that we only apply the arm's-length 
test in situations involving 50 percent or greater cross-ownership 
between affiliates, as we stated in the preamble to the Department's 
regulations, we believe an arm's-length analysis is appropriate 
``whenever there are transactions between parties within the meaning of 
section 771(33) of the Act. Therefore, if two parties are affiliated, 
any transactions between them are subject to paragraphs (c) and (d) of 
19 CFR 351.403, allowing use of transactions between affiliated party 
sales only if found to be made at arm's length.''\41\ We have not 
changed our view in this regard.
---------------------------------------------------------------------------

    \41\ Preamble at 27356.
---------------------------------------------------------------------------

    With respect to the issue of multiple customer codes for a single 
affiliate, we confirm that we intend to aggregate sales to a single 
affiliate for purposes of the arm's-length test.
    With respect to the comment regarding a perceived inconsistency 
between the arm's-length standard as set forth in the Proposed 
Modification and the statutory requirements for valuing affiliated 
party inputs (sections 773(f)(2) and (3)), we disagree that the arm's-
length test must apply the standard or test used for valuing affiliated 
party inputs. These tests are employed for different purposes in 
analytically distinct areas of the dumping analysis. As for the CIT 
remand cited by the commenter, we note that this remand concerned a 
separate issue relating to the statutory definition of ``foreign like 
product'' as the term is used in various parts of the antidumping 
statute. The commenter did not explain the relevance of this court 
decision, nor do we believe that the modification of the arm's-length 
test depends on or implies any application of different definitions of 
the term ``foreign like product.''

[[Page 69195]]

    With respect to the request for clarification on our intended 
practice regarding insufficient unaffiliated party sales, we confirm 
that, consistent with the preamble to our regulations, affiliated party 
sales will not be used where there are insufficient unaffiliated party 
sales for use in the arm's-length test.
    With respect to the comment on burden of proof, we believe the 
Department's regulations speak for themselves, namely that affiliated 
party sales will be used only where the Department is satisfied that 
the price to an affiliate is comparable to unaffiliated prices.\42\
---------------------------------------------------------------------------

    \42\ 19 CFR 351.403(c).
---------------------------------------------------------------------------

    With respect to the request for clarification regarding affiliated 
party sales used in determining CEP profit, the Department's current 
practice is to exclude non-arm's-length sales and include downstream 
sales of the same merchandise where such sales are reported. We have 
not changed that policy.

9. Treatment of Downstream Sales

    Aside from the methodology used to determine whether sales to 
affiliates are made in the ordinary course of trade, numerous 
commenters submitted proposals regarding the use of downstream sales by 
affiliated parties where upstream sales fail the arm's-length test.\43\
---------------------------------------------------------------------------

    \43\ See page 7, above, and Proposed Modification, 67 FR at 
53340, for a summary of the Department's practice concerning 
downstream sales.
---------------------------------------------------------------------------

    Several commenters maintain that the 98-102 percent test, if 
adopted, will increase reliance on downstream sales and will, as a 
result, create greater potential for facts available given the frequent 
reluctance on the part of affiliated resellers to provide information 
regarding downstream sales. One commenter suggests that, in order to 
balance this likely effect, the current ``five percent'' exemption for 
reporting downstream sales \44\ should be broadened to a ``20 percent'' 
exemption, analogous to the rule for determining whether ``substantial 
quantities'' of sales were made below cost. Under this approach, the 
Department would not request downstream sales for any respondent whose 
comparison market sales to affiliates comprise less than 20 percent of 
the value (or quantity) of all comparison market sales of the foreign 
like product. Alternatively, this commenter suggests applying the five 
percent test on a different basis than that currently used. 
Specifically, instead of determining whether sales to all affiliates 
are less than 5 percent of total sales of the foreign like product, the 
Department would under this proposal determine whether only those sales 
of merchandise to affiliates that (1) failed the arm's-length test and 
(2) are resold (not consumed) are less than five percent of all sales 
of the foreign like product, and would not request any downstream sales 
if this standard was met.
---------------------------------------------------------------------------

    \44\ See Sec.  351.403(d) of the Department's regulations, 
specifying that the Department generally will not calculate normal 
value based on downstream sales where sales of the foreign like 
product to affiliated parties constitute less than five percent of 
the total value (or quantity) of the respondent's sales of the 
foreign like product in the market in question.
---------------------------------------------------------------------------

    Another commenter suggests that the Department should not request 
downstream sales under the following circumstances: (1) Where sales to 
an individual affiliate constitute less than one percent of all 
comparison market sales of the foreign like product, regardless of 
whether the five percent exemption is met in the aggregate; (2) where 
respondents demonstrate that downstream sales prices are lower than 
upstream sales prices, provided they agree that upstream prices would 
be used in determining normal value; and (3) where resales are made in 
small quantities or at different levels of trade than the other 
comparison market and U.S. sales.
    Other commenters propose stricter reporting requirements and 
expanded coverage of downstream sales. One suggestion is to eliminate 
or lower (to 0.5 percent) the five percent exemption for reporting 
downstream sales in order to counteract what is likely to be a larger 
amount of sales disregarded--particularly high-priced sales--under the 
revised test compared with the 99.5 percent test.
    Another commenter recommends a different standard be applied in 
investigations and reviews regarding the respondent's obligations to 
report downstream sales. This proposal would allow for downstream sales 
to be disregarded in investigations when a respondent demonstrates to 
the Department that it cannot obtain such sales, but would require 
respondents to include, as a condition of sale to affiliates, a 
requirement that such affiliates provide information on their sales in 
antidumping reviews. This proposal would have the Department issue a 
statement of practice pertaining to administrative reviews providing, 
among other things, that ``[i]f a respondent claims that it is 
otherwise unable to submit the downstream sales data of an affiliated 
seller, the Department will apply adverse facts available.''
    Finally, another commenter asks that the Department make clear that 
it will apply an arm's-length test to downstream sales, where such 
sales are sold to a second-level affiliate.
    Department's Position: We have not changed our requirements 
regarding downstream sales based on these suggestions. With respect to 
the five-percent threshold for reporting downstream sales by affiliates 
set forth at Sec.  351.403(d) of the Department's regulations, the 
proposals to raise or to lower this standard do not address the 
proposed change in the arm's-length test itself. In any event, we do 
not believe that a change in the regulations is warranted by these 
suggestions.
    The adoption of the five-percent threshold was based on the premise 
``that imposing the burden of reporting small numbers of downstream 
sales often is not warranted, and that the accuracy of determinations 
generally is not compromised by the absence of such sales.''\45\ We 
continue to believe that a five-percent standard normally balances 
these considerations appropriately. The proposed 20 percent standard is 
too high to warrant confidence that exceptions to reporting downstream 
sales based on this threshold would not compromise the accuracy of our 
determinations. On the other hand, the proposed 0.5 percent threshold 
is based on a misplaced analogy to the de minimis dumping standard in 
administrative reviews. We do not believe that exempting downstream 
reporting where a respondent sells less than five percent of the 
foreign like product to affiliates, and basing normal value on other 
sales or on constructed value, gives rise to concerns about the 
accuracy of our determinations.
---------------------------------------------------------------------------

    \45\ Preamble at 27356.
---------------------------------------------------------------------------

    With respect to the proposal that the sales of the foreign like 
product used to determine whether the five-percent threshold is met 
should be narrowed to only those that fail the arm's-length test and 
are not consumed by the reseller, we continue to believe that the five-
percent standard, as stated in the regulation, is appropriate. The 
assessment by the Department, in the preamble to the regulations, that 
excusing reporting of downstream sales would not compromise the 
accuracy of its determinations was predicated on a finding that the 
respondent's total sales of the foreign like product to affiliates were 
less than five percent of all sales of the foreign like product. While 
we may determine in certain cases that it is appropriate to excuse 
downstream reporting along the lines suggested by this commenter, we do 
not believe the

[[Page 69196]]

proposal could be applied generally without compromising accuracy. For 
similar reasons, we also disagree with the proposal to exempt 
individual affiliates from reporting downstream sales based on the 
proposed ``one-percent'' standard, though we may exempt reporting of 
such sales in individual cases. In our view, the five-percent standard, 
based on a company's aggregate sales to all affiliates, provides a 
reasonable test for whether to exempt a respondent from downstream 
reporting.
    Regarding the proposal that we exempt respondents from downstream 
sales reporting where they can show such sales were made at prices 
below the relevant upstream sale and agree to use the upstream sale in 
its place, we do not believe it would be appropriate to address such 
hypothetical situations. We will do so if and when such issues are 
raised in a case.
    Regarding the proposal that we exempt downstream sales made at 
different levels of trade than other comparison market sales or U.S. 
sales, such an exemption could conflict with our practice of matching 
U.S. and comparison market sales at different levels of trade in the 
absence of comparable sales at the same level of trade. As such, it 
could inappropriately reduce the number of price-based comparisons in 
the dumping analysis. However, as stated in the Preamble to the 
Department's regulations, the Department does not believe it necessary 
or appropriate to require the reporting of downstream sales in all 
instances, though the Department will require a respondent to 
demonstrate in each segment of a proceeding that the reporting of 
downstream sales is not necessary.
    Regarding the proposal that we exempt downstream sales made in 
small quantities, as noted above, we believe that, as a general matter, 
the correct level at which to determine whether sales are so small as 
to warrant not reporting is at the level of the upstream sale between 
affiliates. This is the level at which the five-percent threshold is 
applied. Any other requests for exemptions from reporting based on a 
small quantity of sales would need to be considered on a case-by-case 
basis.
    Regarding the proposal that we apply different standards in 
investigations and administrative reviews regarding a respondent's 
claim that it cannot submit downstream sales data, we disagree with the 
suggestion that we automatically resort to adverse facts available in 
administrative reviews. We will continue to determine, based on the 
facts of each case, the extent to which an individual respondent has 
failed to cooperate by not providing requested information. This 
approach is consistent with our statutory and WTO obligations regarding 
the use of adverse facts available. While we do not disagree in 
principle with the suggestion that a respondent who has participated in 
an initial investigation may be expected in subsequent administrative 
reviews to have gone to greater lengths to secure such data, any 
finding of uncooperativeness must be made with reference to the 
particular facts of each segment of the proceeding.
    Finally, we intend to continue our practice of applying the arm's-
length test to any sales made to affiliated parties, including 
downstream sales to second-level affiliates.

10. Proceedings/Entries Governed by Revised Arm's-Length Test

    One commenter argued that the Department's proposed timetable for 
applying the new methodology with respect to other proceedings and 
segments of the Japan hot-rolled proceeding other than the 
investigation (i.e., reviews initiated on the basis of requests 
received on or after the first day of the month following the date of 
publication of the Department's final notice of that new methodology) 
would contravene section 129(c) of the URAA (19 U.S.C. 3538(c)). That 
section, the commenter claimed, requires that such changes be 
implemented only with respect to entries made, not proceedings 
requested or initiated, on or after the implementation date.
    Department's Position: The Department's timetable for applying its 
new methodology beyond the Japan hot-rolled investigation is legally 
permissible and appropriate. Specifically, contrary to the commenter's 
assertions, section 129 of the URAA applies only to changes implemented 
with respect to the segment of the proceeding that gave rise to the WTO 
challenge. That is, section 129 of the URAA applies only to changes 
made as a result of ``an action by the administering authority in a 
proceeding under title VII * * * [that] is not in conformity with the 
obligations of the United States under the Antidumping Agreement * * 
*.'' Section 129(b)(1) (emphasis added). Therefore, the timing 
provisions of section 129(c) (which deal with implementation under 
section 129) also apply only to changes to measures ``as implemented'' 
with respect to the segment of the proceeding which served as the basis 
for the WTO challenge.
    In contrast, changes in agency practice (such as this change with 
respect to the arm's-length test) made in connection with an adverse 
WTO panel or Appellate Body are governed by a different provision of 
the URAA. See section 123(g) of the URAA. Section 123 has its own 
``effective date of modification'' provision (section 123(g)(2)). This 
provides for a single limitation on the effective date: ``the final 
rule or other modification may not go into effect before the end of the 
60-day period beginning on the date on which consultations [with the 
appropriate congressional committees on the proposed content of the 
modification] begin [unless the President determines that an earlier 
effective date is in the national interest].'' Because this new 
methodology will ``go into effect,'' for other proceedings and other 
segments of the Japan hot-rolled proceeding, after the 60-day period 
will have ended, the timetable for implementation is lawful. Thus, 
Commerce's decision to apply its new methodology prospectively, 
beginning with segments of proceedings initiated on or after November 
23, 2002,\46\ is proper.
---------------------------------------------------------------------------

    \46\ See ``Timetable'' section, below.
---------------------------------------------------------------------------

    The fact that, under the proposed implementation timetable, the new 
arm's-length methodology ``would affect'' margins on imports which 
entered prior to the implementation date, but for which the margins 
would be calculated in a review initiated after the implementation 
date, does not compel the result urged by the commenter. The 
commenter's broad reading of the legislative history of section 129 
does not provide authority for extending the effective date provision 
of that section to areas covered instead by section 123, especially 
given that section 123 has its own, different, provision that controls 
such a new methodology.
    It is significant that section 123 uses the term ``go into effect'' 
(which refers to the beginning of use of a methodology), rather than 
language of section 129, which refers to which entries will be 
affected. There is no legislative inconsistency with the use of a new 
methodology ``affecting'' entries made prior to the date on which the 
methodology changed. Indeed, except where otherwise specified (as in 
section 129 with respect to the actions of the Department in the 
contested segment of the proceeding), the Department's practice has 
normally been to begin application of a new methodology with respect to 
segments of proceedings requested or initiated after a given date, 
rather than applying different methodologies within the same segment

[[Page 69197]]

of the proceeding. See, e.g., section 291(a)(2) of the URAA (the URAA 
amendments shall ``take effect'' on the date the WTO Agreements enter 
into force and ``shall apply with respect to'' reviews initiated 
pursuant to a request filed after such date); 19 CFR 351.701 
(regulations implementing the changes made by the URAA ``apply to all 
administrative reviews initiated on the basis of requests made after 
June 18, 1997'' (the ``effective date'' provided in the notice of final 
rule published in the Federal Register on May 19, 1997).

11. Applicability of Administrative Procedures Act To Revised Arm's-
Length Test

    One commenter contended that the change to the arm's-length test is 
tantamount to creating a rule as set forth in 5 U.S.C. 553 of the 
Administrative Procedures Act (APA). More specifically, citing Carlisle 
Tire & Rubber Co. v. United States, 10 C.I.T. 301, 305-06 (1986) 
(Carlisle), the commenter suggests that the Department's notice and 
comment procedures should comply with those set forth under APA. In 
this commenter's view, the 15 day notice and comment period provided by 
the Department falls short of the 60 day period required under the APA.
    Department's Position: As discussed above, the revised arm's-length 
methodology has been developed taking into account the finding in the 
AB Report that the application of the 99.5 percent arm's-length test in 
the underlying investigation was inconsistent with the obligations of 
the United States under Article 2.1 of the AD Agreement. As a result, 
the revised arm's-length test represents a methodology consistent with 
section 2.1 of the AD Agreement in accordance with the AB Report. 
Unlike the methodologies contested in Carlisle, our arm's-length 
methodology does not create an inflexible rule. In short, the 
Department's arm's-length methodology is not subject to the APA 
because, unlike the methodology underlying Carlisle, it only interprets 
the law.
    The Department also notes that section 123(g) does not provide for 
application of the APA within the context of the remediation of the 
Department's practice. Section 123(g) only requires, in relevant part, 
that the Department provide the public with the proposed change, an 
explanation of how that change would implement the panel or Appellate 
Body report, and an opportunity for comment. Consequently, under a 
plain language reading of section 123(g), the Department's announced 
change in practice would not be subject to the notice and comment 
procedures of the APA.

Timetable

    This methodology will be used in implementing the Japan Hot-Rolled 
findings pursuant to section 129 of the URAA. In accordance with 
section 129(c)(1) of the URAA, the section 129 determination in Japan 
Hot-Rolled will establish new cash deposit rates for all producers for 
whom the investigation rates are still applicable and will apply with 
respect to unliquidated entries of the subject merchandise which are 
entered, or withdrawn from warehouse, for consumption on or after the 
date on which the United States Trade Representative directs the 
Department to implement that determination. With respect to other 
proceedings and other segments of the Japan hot-rolled proceeding, the 
new methodology will be applied in all investigations and reviews 
initiated on or after November 23, 2002.\47\
---------------------------------------------------------------------------

    \47\ This is a slight modification of the Timetable as set forth 
in the Proposed Modification. Previously, the Timetable anticipated 
that the implementation of this practice would go into effect with 
respect to investigations initiated on the basis of requests 
received after the publication date of this notice, and for reviews 
initiated on the basis of requests received in the month following 
publication of this notice. Upon further consideration, we have 
determined that it is appropriate to employ this methodology in all 
investigations and reviews initiated on or after November 23, 2002.

    Dated: November 8, 2002.
Faryar Shirzad.
Assistant Secretary for Import Administration.
[FR Doc. 02-29065 Filed 11-14-02; 8:45 am]
BILLING CODE 3510-DS-P