OPINION AND ORDER

POGUE, Chief Judge.

In prior proceedings in this matter, the Court of Appeals for the Federal Circuit ("CAFC") held that when calculating surrogate labor rates for the valuation of goods from a nonmarket economy ("NME"), the Department of Commerce ("Commerce") must use data from countries that are both economically comparable to the NME and significant producers of comparable merchandise. Dorbest Ltd. v. United States, 604 F.3d 1363, 1372-73 (Fed.Cir.2010)("Dorbest IV"). Dorbest IV thus required Commerce to redetermine, on remand, the labor rate applicable here. See Final Results of Redetermination Pursuant to Remand, Dorbest Ltd. v. United States, (Nov. 10, 2010)("Remand Results").

In its remand determination, choosing data from the record, Commerce calculated a labor wage rate by averaging industry-specific earnings and/or wages from three countries—India, Indonesia and Pakistan—that it found to be both economically comparable to China and significant producers of wooden bedroom furniture. Based on these calculations, Commerce identified an average wage rate of 0.23 USD/hour and found that using that average wage rate as a surrogate for the cost of labor in the production of Plaintiff/Respondent Dorbest's merchandise, Dorbest has a de minimis dumping margin. Remand Results at 17, 42.

Plaintiff/Petitioner American Furniture Manufacturers Committee for Legal Trade ("AFMC") now seeks review of Commerce's data choices in that redetermination on remand.1 AFMC challenges four of Commerce's specific choices: 1) Commerce's initial selection of two "bookend" countries—the Philippines and Pakistan— to limit its consideration of countries with economies comparable to China, the NME at issue; 2) Commerce's exclusion of data not available during the original investigation; 3) Commerce's use of wage rate data from India alleged to be "capped" or limited to wages of workers making 1600 Rupees ("Rs.") per month or less; and 4) Commerce's calculation of an average surrogate wage rate using only countries for which industry-specific data was available.

After a brief review of the relevant procedural history, the agency's methodology, and the applicable standard of review, the court will explain why it concludes that, given the record as a whole, the first of Commerce's choices must be remanded but the other three data choices were reasonable and therefore must be sustained.

BACKGROUND

Procedural history

This matter arises from Commerce's investigation of whether wooden bedroom furniture from China was being dumped in the United States domestic market during the time period between April 1, 2003 and September 30, 2003. Wooden Bedroom Furniture from the People's Republic of China, 68 Fed.Reg. 70,228 (Dep't Commerce Dec. 17, 2003)(Notice of Initiation of Antidumping Investigation). Commerce's final determination in the original investigation was subsequently challenged and remanded three times before it was appealed to the CAFC.

In Dorbest IV, the CAFC invalidated Commerce's wage rate calculation regulation.2 This court then remanded for further proceedings in accordance with the CAFC decision. Specifically, as noted above, the CAFC held that, contrary to Commerce's regulation, the governing statute, 19 U.S.C. § 1677b(c)(4),3 requires that when calculating surrogate labor wage rates, Commerce shall "to the extent possible," use factors of production from market economies that are both economically comparable to the non-market economy country and significant producers of the subject merchandise. Dorbest IV, 604 F.3d at 1372 (citing 19 U.S.C. § 1677b(c)(4)(A)).

After Dorbest IV, Commerce acknowledged that the data on the record was insufficient to comply with the court's remand order and re-opened the administrative record to admit new wage data. Request for Comment Regarding Wage Rate Data, A-570-890, Remand Redetermination Investigation ("RRI") 4/1/03-9/30/03 (Aug. 11, 2010), Remand Admin. R. Pub. Doc. 1. Commerce also invited interested parties to submit comments and new factual information with regards to the sole issue of labor wage valuation. Id. at 2. AFMC and Dorbest each submitted comments and wage rate data for Commerce's consideration.

Methodology

Selecting from the record data, Commerce, in its remand determination, specified five steps for calculating labor wage rates ("the 5-step methodology").4

First, Commerce created a list of economically comparable surrogate countries based on gross national income ("GNI").5 In doing so, Commerce relied on its original surrogate country memorandum,6 which provided five economically comparable countries for consideration as the primary surrogate country for this investigation.7 Remand Results at 12; Surrogate country memorandum. The countries on the list in the surrogate country memorandum are India, Pakistan, Indonesia, Sri Lanka and the Philippines. Remand Results at 12. Using, as "bookends," the high and low-income countries from that list, i.e., the Philippines and Pakistan, Commerce then added all countries with World-Bank reported per capita GNIs that fell within the "bookend" range. Remand Results at 12-13. This resulted in a list of 24 countries.

Second, Commerce proceeded to identify which of the 24 listed countries had exports of comparable merchandise between 2001 and 2003. Remand Results at 12. At this step, Commerce identified 13 countries from the list that were both economically comparable to China and significant producers of comparable merchandise.

Third, Commerce identified which of the 13 countries reported wage data between 1997 and 2002. Remand Results at 13. In doing so, Commerce relied on the International Labor Organization ("ILO") wage data from the base year and five years prior. See AFMC Br. at 8. After applying this step, six countries remained.

Commerce then added a fourth step to its methodology: It identified which countries reported an industry-specific classification within the ILO wage rate data. Remand Results at 13-14. In doing so, Commerce looked to data that was reported according to the International Standard Industrial Classification of all Economic Activities ("ISIC") code.8 Remand Results at 14. Each updated ISIC code is known as a revision, and ISIC Revision 3 was the most recent reporting period available at the time of the initial investigation. Remand Results at 14 n. 46. Commerce, however, chose to use an older revision, ISIC Revision 2, because it contained a sub-classification most specific to the production of wooden bedroom furniture.9 Remand Results at 15. After applying this step, three countries—India, Indonesia and Pakistan—remained on the list. Id.

Finally, Commerce calculated an average wage rate for these three countries by using wage rate data from a three-digit subclassification level, when that sub-classification was available.10 Remand Results at 16. Commerce used wage rate data from India and Indonesia that was reported at this additional, three-digit sub-classification level. Pakistan, however, did not report data at the three-digit sub-classification level; therefore with regards to wage rate data from Pakistan, Commerce used data which was reported at the two-digit sub-classification level. Id. Commerce then calculated a simple average using this data. See id.11

STANDARD OF REVIEW

The court will sustain a Commerce redetermination on remand "if it complies with the court's remand order, is supported by substantial evidence on the record, and is otherwise in accordance with law." Jinan Yipin Corp. v. United States, ___ CIT ___, 637 F.Supp.2d 1183, 1185 (2009)(citing 19 U.S.C. § 1516a (b)(1)(B)(i)).

An agency determination is supported by substantial evidence when the record upon which it is based contains such "relevant evidence as a reasonable mind might accept as adequate to support a conclusion." Consol. Edison Co. v. N.L.R.B., 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938). In making this evaluation of the record, the court assesses whether the agency's data choices are reasonable considering the record as a whole. See Nippon Steel Corp. v. United States, 458 F.3d 1345, 1351 (Fed.Cir. 2006). At a minimum, in making its data choices, the agency must explain the standards it applied and make a rational connection between the standards and the conclusion. See Matsushita Elec. Indus. Co. v. United States, 750 F.2d 927, 933 (Fed.Cir.1984). A rational connection is a connection that is supported by justification or evidence. See Bowman Transp., Inc. v. Arkansas-Best Freight Sys., Inc., 419 U.S. 281, 285, 95 S.Ct. 438, 42 L.Ed.2d 447 (1975) (explaining that, even under the narrower arbitrary and capricious standard of review, the agency must examine the relevant data and articulate a satisfactory explanation for its action, including a "rational connection between the facts found and the choice made" (quoting Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 168, 83 S.Ct. 239, 9 L.Ed.2d 207 (1962))).12

DISCUSSION

I. Commerce's selection of countries to act as "bookends"

The first issue before the court is Commerce's initial selection of a pair of "bookend" countries to establish a range of GNI with which to identify a list of countries that qualify as economically comparable to China.13 AFMC contends that this choice is arbitrarily skewed towards countries with a per-capita GNI that is less than that of China.14

The high-income "bookend" country selected from the list in Commerce's original surrogate country memorandum was the Philippines, with a GNI of 1,020, and the low-income bookend country was Pakistan with a GNI of 410.15 Remand Results at 12; AFMC Br. at 23. China's GNI at the time of the original investigation was 1,100.16 AFMC Br. at 23.

AFMC asserts that the CAFC clearly intended Commerce to use countries with reported GNI's both above and below that of China in order to capture an absolute range of economically comparable countries.17 AFMC Br. at 22, 25. AFMC also points out that the surrogate country memorandum was not drafted for purposes of calculating surrogate wage rates and therefore fails to account for absolute differences in GNI by listing countries both above and below China's GNI. AFMC Br. at 23.18

AFMC also contends that Commerce has failed to provide a reasonable explanation for why it used the five countries listed in the surrogate country memorandum, given that their low GNIs would necessarily predetermine an underestimated labor wage rate. AFMC Br. at 23-24. In support of this, AFMC notes that Commerce has already recognized in an earlier proceeding that using only wage rates from countries with high GNIs will likely lead to an overestimated wage rate. AFMC Br. at 24 (citing Dorbest, Ltd. v. United States, 547 F.Supp.2d 1321, 1327 (CIT 2008)) ("Dorbest II"). For AFMC, it follows that here, when Commerce selected countries with GNIs lower than China, the data set "pre-ordained an understated wage rate." AFMC Br. at 23.

Commerce contends that it was instructed merely to base its wage value on countries that are economically comparable to China and that neither the statute nor Dorbest IV define a set range of GNI to be used when determining economic comparability. Remand Results at 32. Commerce notes that the countries on the surrogate country memo were already determined to be economically comparable to China and that the memo provided a sufficient number of economically comparable countries to act as a starting point. Remand Results at 33.

Commerce's explanation is insufficient. While Commerce has discretion to determine the countries which will act as bookends for its selection, it has not provided a reasoned explanation of its "bookend" choices. In particular, Commerce's remand decision overlooks the explicit statement in the surrogate country memo that the proposed list is non-exhaustive, allowing for the possibility of introducing a more balanced range of countries from which to draw labor wage rate data. Surrogate country memo at 1.

Here, both of the two bookend countries have GNIs that fall below China's, resulting in a range of corresponding wage rates that will likely fall below China's wage rates. Given the high correlation between per capita GNI and wage rates, a correlation that Commerce acknowledges, Commerce's selection appears arbitrarily biased towards the low end of per capita GNI. See Remand Results at 35-36 (acknowledging established global relationship between wages and GNI); also Dorbest II, 547 F.Supp.2d at 1327. Certainly Commerce does not have to achieve mathematical perfection in its choice of countries to act as bookends for its initial selection, but Commerce must explain why it selected two countries with GNIs that are lower than China's to use as bookends, and Commerce's explanation must rest "upon principles that are rational, neutral, and in accord with the agency's proper understanding of its authority." FCC v. Fox Tel. Stations, Inc., ___ U.S. ___, 129 S.Ct. 1800, 1823, 173 L.Ed.2d 738 (Kennedy, concurring) (2009); see, e.g., Matsushita Elec. Indus. Co., 750 F.2d at 933. Without such an explanation, Commerce's determination is arbitrary because it "fail[s] to consider an important aspect of the problem," and is therefore unreasonable. SKF USA v. United States, 2011 WL 73179, *6 (Fed.Cir.Jan.7, 2011)(quoting Motor Vehicle Mfrs. Ass'n. of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443(1983)).

Commerce asserts that the CAFC's explicit directive to rely on data from countries that are economically comparable to China will "necessarily result in a truncated dataset." Remand Results at 33. But this argument misses the point. While the CAFC's opinion has precluded the larger data sets that Commerce used in its invalidated regression-based methodology, that opinion did not hold that Commerce was restricted to using only countries with GNIs lower than China's. On the contrary, the CAFC noted that there were at least 16 countries from which Commerce could draw. Dorbest IV, 604 F.3d at 1372. While the CAFC explicitly declined to address exactly which countries could properly be considered economically comparable to China, it left open the possibility that countries with GNIs higher than China's could be included in the range. Id. Commerce has not provided any adequate explanation as to why these higher-income countries are necessarily excluded from the starting selection of countries.

Finally, Commerce claims that the range of economically comparable countries is not unfair—just because that range is not centered around China's GNI—and points out that there is no statutory requirement that Commerce select the "most comparable country." Remand Results at 33-34. AFMC responds that at a minimum, Commerce should achieve "substantial balance" in its data set by selecting bookend countries that are roughly equally above and below China's per capita GNI.19 Commerce replies that focusing on the ranking of each country will create an "illusion of precision." Remand Results at 34. In making this argument, Commerce relies on Fujian Lianfu Forestry Co., Ltd. v. United States, 638 F.Supp.2d 1325, 1348-49 (CIT 2009), which held that India was economically comparable to China despite a wide difference between their respective GNIs (India 620, China 1290).

Fujian Lianfu Forestry is distinguishable, however, because it involved the choice of a single country to act as a primary surrogate country. See Fujian Lianfu Forestry, 638 F.Supp.2d at 1347. Here, Commerce is selecting a range of countries. Moreover, there is no indication here that the methodology applied in Fujian Lianfu Forestry to select a primary surrogate country is similar to the methodology for determining surrogate wage rates. See id. at 1348-49. On the contrary, in the context of wage rate calculation, Commerce has stated that there is a high correlation between wage rates and GNI. Remand Results at 35-36. Given this statement, Commerce has not explained, beyond conclusory reasoning, how relying on broader GNI rankings of countries could produce an "illusion of precision." See Amanda Foods (Vietnam), Ltd. v. United States, 647 F.Supp.2d 1368, 1377 (CIT 2009)(holding that Commerce must provide more than conclusory reasoning for treating all countries on surrogate country memorandum as identical).

For the above reasons, the court remands this issue to Commerce so that it may 1) explain why it is justified in selecting this particular pair of countries to act as bookends for the selection process, in light of their low GNIs and the high correlation between GNI and wage rates, or 2) otherwise reconsider its determination in accordance with this opinion.

II. Commerce's decision to use data available at the time of the original investigation

AFMC next contends that Commerce's reliance on 2002 GNI data and 2002 ILO wage data does not constitute use of the "best available data" under 19 U.S.C. § 1677b(c)(1). AFMC Br. at 11.20 We disagree.

This court held in Dorbest, Ltd. v. United States, 462 F.Supp.2d 1262 (CIT 2006), that given that administrative law defines "available" in terms of the underlying investigation, "available" may reasonably mean "available during the investigation." Dorbest, 462 F.Supp.2d at 1299 ("Dorbest I")(citing Vermont Yankee Nuclear Power Corp. v. Natural Resources Def. Council, Inc., 435 U.S. 519, 555, 98 S.Ct. 1197, 55 L.Ed.2d 460 (1978)).

AFMC contends that once an agency has reopened the record, it must consider all evidence properly before it and therefore, the best available information currently consists of the 2003 ILO wage data. AFMC Br. at 16. AFMC's argument, however, disregards the procedural posture of this case. Remand proceedings do not grant the parties the right to a new antidumping investigation from the current date. See, e.g., 2 Am.Jur.2d Administrative Law § 575. Rather, in remand proceedings, an administrative agency must modify its original determination in accordance with the remand order. See id.

Here Commerce reopened the record to admit new data because it needed a new type of data to comply with our remand order; that order, however, did not require data from a different time period. See Request for Comment Regarding Wage Rate Data, A-570-890, RRI 4/1/03-9/30/03 (Aug. 11, 2010), Remand Admin. R. Pub. Doc. 1 at 2. The error in Commerce's original determination arose not from the time period for which the ILO wage data were selected, but rather from the methodology applied to select the data.21 See Dorbest IV, 604 F.3d at 1372-73. Because we are to treat Commerce's calculations on remand as if they were made at the time of the original investigation, it is reasonable for Commerce to consider only data that was available to it during the original investigation, namely, the 2002 ILO wage data. See Vermont Yankee Nuclear Power Corp., 435 U.S. at 554-55, 98 S.Ct. 1197 ("[a]dministrative consideration of evidence . . . always creates a gap between the time the record is closed and the time the administrative decision is promulgated") (citation omitted).

Asserting that Commerce's decision is not in accord with the statute, AFMC incorrectly cites Port of Seattle v. Fed. Energy Regulatory Comm'n, 499 F.3d 1016, 1035 (9th Cir.2007). AFMC's reliance misses the point. In Port of Seattle, the Ninth Circuit reviewed a decision made by the Federal Energy Regulatory Commission ("FERC") which disregarded evidence added to the record after a preliminary evidentiary proceeding but before FERC rendered its final decision. Port of Seattle, 499 F.3d at 1025. Port of Seattle does not involve or address new data that was not available at the time of the original determination or investigation.22

In the alternative, AFMC contends that the decision to use only data available during the time of the original investigation is arbitrary and not supported by substantial evidence. AFMC Br. at 17. AFMC makes three arguments in support of this assertion.

First, AFMC contests Commerce's finding that the interests of administrative finality and efficiency overcome an interest in conducting accurate fact finding and that allowing later-discovered evidence sets an undesirable precedent. AFMC Br. at 17-18. AFMC argues that it is not attempting to circumvent the finality of Commerce's determination with new evidence and that Commerce has strayed from its own precedent in choosing not to use the updated data.

The case AFMC relies on to make this argument, Shakeproof Assembly Components Div. Of Illinois Tool Works, Inc. v. United States, 30 CIT 1173, 2006 WL 2457626 (Aug. 25, 2006), does not support AFMC's claim. The court in Shakeproof upheld Commerce's decision to reject evidence that was not contemporaneous with the period of investigation and noted in dicta that Commerce has traditionally used "valuation information contemporaneous with a period of investigation or review." Shakeproof, 30 CIT at 1177. Here the data used by Commerce is contemporaneous with the period of investigation in that it represents the data available at the time of the original investigation. Thus Shakeproof is not contrary authority.

Second, AFMC asserts that 2003 ILO wage data was available at the time of the original investigation because this court acknowledged in Dorbest I that the 2004 download, which happened to include 2003 data, was materially the same as the data available during the original investigation. AFMC mis-states our finding in Dorbest I. In Dorbest I, this court discussed the availability of 2002 ILO wage data shortly after the original investigation was completed and found that the 2004 download of that data was materially the same. Dorbest I, 462 F.Supp.2d at 1299. The availability of the 2003 ILO wage rate data during the original investigation was not before us in Dorbest I and therefore AFMC may not rely on Dorbest I to establish the existence of the 2003 ILO wage data and its concomitant availability during the original investigation.

Finally, AFMC asserts that Commerce "cherry picked" from the data, using both 2002 and 2003 ILO wage data in its calculations. AFMC Br. at 20. Commerce acknowledged in the remand results that it needed to extract 2002 ILO wage data that had not been retained at the time of the original investigation. In the remand results, Commerce stated that it relied on "a current download of 2001-2003 export data" to determine which countries were significant producers of comparable merchandise. Remand Results at 22. AFMC asserts that there is no reasonable basis for Commerce to conclude that the 2003 ILO wage data was not available during the time of the investigation while at the same time concluding that it was available. AFMC Br. at 21. Commerce responds that, by necessity, certain portions of the data it used were newly extracted, but that it relied upon data that would have been available during the original investigation. Defendant's Response to AFMC's Remand Comments, Dorbest Ltd. v. United States, (Dec. 22, 2010) at 14 ("Commerce Reply Br."). Commerce acknowledges that this is not a perfect procedure but the best it can make of the available data sources. Commerce Reply Br. at 15. This is reasonable. Accordingly, Commerce's decision to rely solely on data that would have been available during the original investigation is affirmed.

III. Indian wage rate

The AFMC next contends that Commerce's calculations are unsupported by substantial evidence because they rely in part on Indian wage data that appear to exclude workers making more than Rs.1600 per month, and thus appears "capped" or limited to wages under that amount. AFMC Br. at 37. We disagree.

In support of its claim, AFMC cites the Annual Survey of Industries (ASI), which can be read to indicate that a "cap" limits the Indian wage data to the bottom 2% of wage earners. Petitioners' Comments Concerning Draft Results of Redetermination Pursuant to Remand in Dorbest Limited v. United States, A-570-890, RRI 4/1/03-9/30/03 (Oct. 22, 2010), Remand Admin. R. Pub. Doc. 13 at 35-37. Commerce, however, declines to use this information because it was not made available until 2006. Remand Results at 40. As discussed supra, Commerce's decision to exclude data that would not have been available at the time of the original investigation is reasonable. See Dorbest I, 462 F.Supp.2d at 1299.

In addition, Commerce further supports its stance by explaining that the ASI data do not represent industry-specific 2002 wages.23 See Remand Results at 40 n. 99. Commerce notes that even though the 2003 ASI report on the record contains a "trends" column which shows a FY 2002 country-wide rate, no source data is on the record for this column, nor does the report include industry-specific data, which is what Commerce used in its calculations. Id. Thus Commerce decided that the ASI submission is not an appropriate benchmark because, even if it were available at the time of the original investigation, it does not contain industry-specific 2002 ILO wage data and thus would not be relevant to Commerce's calculations.24 This determination is reasonable.

IV. Commerce's calculation of the average wage rate

Finally, AFMC challenges Commerce's data choices at the fourth and fifth steps of the wage rate calculation, asserting that Commerce's choices arbitrarily reduced the number of countries from which Commerce could calculate a labor wage rate. We disagree.

AFMC argues that the use of industry-specific data from ISIC Revision 2 is arbitrary and capricious because of Commerce's stated preference for a large "basket" of countries from which to choose. AFMC claims that requiring industry-specific wage data unnecessarily reduces the number of available countries from which to draw data when country-wide wage data is available from more countries. AFMC Br. at 32.

Commerce, in response, asserts correctly that the governing statute is silent on this issue, leaving the determination to Commerce's reasonable discretion. See 19 U.S.C. § 1677b(c)(1). Commerce explains that using industry-specific data is preferable because it comports with Commerce's long-standing practice of valuing the most specific data for production factors, and, at the very least, such data is more specific to the subject merchandise than country-wide data. Remand Results at 28-29.

While Commerce acknowledges AFMC's concerns, it notes, and we agree, that AFMC has failed to provide evidence to show that industry-specific data are unsuitable for calculating wage rates. Remand Results at 28. Furthermore, AFMC's argument misses the point. Our inquiry here is not whether Commerce used a certain number of countries in its calculations. Rather it is whether Commerce reasonably adhered to the remand order and the statutory requirements set forth in 19 U.S.C. § 1677b(c)(4), and whether Commerce's determination is supported by a reasonable reading of the record evidence as a whole.

Nonetheless, with regards to whether Commerce used the best available information from the record, Commerce states that it believes industry-specific data will yield the most accurate results, and explains that it used ISIC Revision 2 because it contains a two-digit sub-classification of industry-specific wages which Commerce feels to be most relevant to the production of wooden furniture. Remand Results at 15 ("[Commerce] identified the two-digit series most specific to wooden bedroom furniture as Sub-Classification 33", which is described as "Manufacture of Wood and Wood Products, Including Furniture"). Commerce also stated that it chose to use ISIC Revision 2 rather than the updated Revision 3 because Revision 2 contained the specific sub-classification which was more specific to, and thus a better match for, the subject merchandise.25 Remand Results at 15. Here, Commerce has explained the standards it applied and made a rational connection between this standard and its decision to use ISIC Revision 2 data. See Matsushita Elec. Indus. Co., 750 F.2d at 933.

For the reasons given above, Commerce's decision to use industry-specific data is reasonable and in compliance with the statutory requirements set forth in 19 U.S.C. § 1677b(c)(4).

Conclusion

Accordingly, Commerce's initial selection of two "bookend" countries—the Philippines and Pakistan—to limit its consideration of countries with economies comparable to China, is remanded for further consideration in accordance with this opinion. Commerce shall have until March 28, 2011 to complete and file its remand determination. Plaintiffs shall have until April 11, 2011 to file comments. Defendant and Defendant-Intervenors shall have until April 25, 2011 to file any reply. Commerce's other data choices are affirmed.

It is SO ORDERED.