Opinion of the court filed PER CURIAM, dissenting opinion filed by Senior Circuit Judge ARCHER
PER CURIAM.Defendants-Appellants Dongbu Steel Co., Ltd., et al. (collectively “Korean producers”) appeal from the judgment of the Court of International Trade, British Steel P.L.C. v. United States, 941 F.Supp. 119 (Ct. Int’l Trade 1996) (British Steel II), sustaining the Department of Commerce’s decision that the government of Korea had provided the Korean steel industry with preferential and disproportionate access to long-term domestic and direct foreign loans, had constructed infrastructure at the Kwangyang Bay Industrial Estate (KBIE) for the specific benefit of the Po-hang Iron and Steel Company (POSCO), and had exempted POSCO from dockyard fees. See Final Affirmative Countervailing Duty Determinations and Final Negative Critical Circumstances Determinations: Certain Steel Products from Korea, 58 Fed.Reg. 37,338 (Dep’t Commerce 1993) (Final Determination); Final Results of Redetermination Pursuant to Court Remand in British Steel P.L.C. v. United States, Slip. Op. 95-17 (Dep’t Commerce 1995) (Redetermination).
Plaintiffs/Cross-Appellants AK Steel Corp., et al., (collectively “domestic producers”) appeal from the judgment of the Court of International Trade in British *1370Steel II upholding Commerce’s conclusion in the Final Determination that the Korean government did not provide a specific benefit to the steel industry in Korea and in particular, POSCO, by the revaluation provisions of the Tax Exemption and Reduction Control Act (TERCL) Article 56-2.
We have jurisdiction to hear these appeals pursuant to 28 U.S.C. § 1295(a)(5) (1994). We affirm the Court of International Trade’s decision with respect to the infrastructure at the KBIE, dockyard fees and the revaluation provisions of TERCL 56-2, and reverse the Court of International Trade’s decision with respect to the long-term domestic and direct foreign loans.
I. BACKGROUND
This consolidated appeal arose from proceedings before the Department of Commerce to determine whether the Korean government conferred countervailable benefits on particular members of the Korean steel industry. The proceedings are summarized in the opinion of the Court of International Trade. See British Steel II, 941 F.Supp. at 128. For purposes of this appeal, the relevant background is as follows.
On July 9, 1993, Commerce issued its Final Determination concluding, inter alia, that “the [government of Korea] ha[d] provided the steel industry with preferential access to medium- and long-term credit from government and commercial banking institutions,” 58 Fed.Reg. at 87345, and that as to both domestic and foreign loans this preferential access constituted a countervailable benefit. Commerce also determined that the Korean government’s provision of infrastructure to POSCO at the Kwangyang Bay Industrial Estate and the exemption of POSCO from dockyard fees conferred countervailable benefits. Id. at 37346^8. In addition, Commerce determined that the revaluation provisions of TERCL Article 56-2 did not provide POSCO “with a selective exemption from the 25 percent requirement in the Asset Revaluation Act.” Id. at 37351.
In the Korean producers’ appeal of Commerce’s determination, the Court of International Trade remanded the Final Determination, instructing Commerce to
explain, if it is able, what evidence on the record demonstrates that programs existed during the period of investigation to benefit the respondent steel companies by giving them preferential access to both domestic and direct foreign credit markets, and how the respondent steel companies received that access to credit.
British Steel P.L.C. v. United States, 879 F.Supp. 1254, 1331 (Ct. Int’l Trade 1995) (British Steel I).
Following Commerce’s Redetermination, on remand, the Court of International Trade upheld Commerce’s determination of the existence of a causal nexus between the Korean government’s control of the financial system and preferential access to domestic and foreign credit by the Korean steel industry, see British Steel II, 941 F.Supp. at 127-30, and that such access constituted a countervailable benefit with respect to both domestic and direct foreign loans, see id. at 133, 135-36. In addition, the Court of International Trade affirmed Commerce’s determination that the infrastructure provided to POSCO at KBIE and the exemption from dockyard fees were countervailable. See id. at 133-35. Finally, the Court of International Trade affirmed Commerce’s conclusion in the Final Determination that TERCL Article 56-2 did not constitute a countervailable benefit. See id. at 132-33.
On appeal to this court, the Korean producers challenge Commerce’s determinations. Domestic producers challenge Commerce’s finding of non-countervailability regarding TERCL Article 56-2.
II. STANDARD OF REVIEW
We review a decision of the Court of International Trade by applying *1371“anew the statutory standard of review applied by that court to the agency’s decision.” Torrington Co. v. United States, 82 F.3d 1039, 1044 (Fed.Cir.1996). If a final countervailing duty determination is “unsupported by substantial evidence on the record, or otherwise not in accordance with law,” it will be held unlawful. 19 U.S.C. § 1516a(b)(1)(B) (1988) (current version at 19 U.S.C. § 1516a(b)(1)(B)(i) (1994 & Supp. III 1997)); see also Creswell Trading Co., v. United States, 15 F.3d 1054, 1056 (Fed.Cir.1994). “[Substantial evidence is more than a mere scintilla. It means such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Aimcor, Ala. Silicon, Inc. v. United States, 154 F.3d 1375, 1378 (Fed.Cir.1998); Matsushita Elec. Indus. Co. v. United States, 750 F.2d 927, 935 (Fed.Cir.1984) (quoting Consolidated Edison Co. v. NLRB, 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938)). Thus, our analysis is not whether we agree with Commerce’s conclusions, nor whether we would have come to the same conclusions reviewing the evidence in the first instance, but only whether Commerce’s determinations were reasonable. See United States Steel Group v. United States, 96 F.3d 1352, 1357 (Fed.Cir.1996). “[T]he possibility of drawing two inconsistent conclusions from the evidence does not prevent an administrative agency’s finding from being supported by substantial evidence.” Consolo v. Federal Maritime Comm’n, 383 U.S. 607, 620, 86 S.Ct. 1018, 16 L.Ed.2d 131 (1966) (citations omitted).
III. DOMESTIC LOANS
A. Background
In its Final Determination, Commerce held that the Korean government effectively controlled long-term lending practices in Korea through (1) the General Bank Act and the Korean government’s Monetary Board, (2) the appointment of banking officials, (3) informal control over loan allocations, i.e., through preferential rediscounting, and (4) strict control of interest rates. 58 Fed.Reg. at 37340-41. The Korean governmental control was deemed sufficient to establish a government program. See Proposed Regulations, 54 Fed.Reg. 23366, 23379 (§ 355.2(r) defining “program” as “any government act or practice”). Although de jure preferences were terminated by 1985, Commerce found that the pattern of long-term lending to the steel industry remained largely unchanged, and that in fact lending to that industry had slightly increased. Commerce also applied a disproportionality analysis that involved the comparison of the share of long-term loans received and held by the steel industry with that industry’s share of gross domestic product (GDP). Based on data supplied by the Korean government, Commerce found that while the steel industry’s contribution to GDP had remained relatively constant at approximately 2.0-2.5% since the early- to mid-1980s, the volume of loans the steel industry received had consistently remained two to four times higher than its contribution to GDP in percentage terms. 58 Fed. Reg. at 37343, 37345.
As previously noted, the Court of International Trade held in British Steel I, that “Commerce [did] not sufficiently explain ... the connection between the government de facto program and the steel companies’ alleged preferential access to specific sources of credit” and remanded the case. 879 F.Supp. at 1325. In its Redeter-mination, following remand, Commerce concluded that the causal nexus was shown by record evidence of the Korean government’s control of the Korean financial system to create a shortage of credit and to selectively exempt the steel industry from the shortage by providing it preferential access. Commerce noted that benefits received by the steel industry resulted from “aggressive targeting” by the Korean government of scarce long-term credit to the steel industry. Redetermination, slip op. at 15. The targeting and resulting nexus, according to Commerce, “consist[ed] of (1) the de jure preferences for steel prior to *13721985; (2) direct and indirect expressions of continued support for steel after 1985; (3) a major government sponsored investment project for the steel industry, ie., the Kwangyang Bay Industrial Estate project; and (4) a volume of loans at favorable rates that exceeded the pre-1985 volume.” Id. at 21-22. Finally, Commerce considered and rejected “three economic factors identified as possible reasons why the distribution of benefits was not disproportionate ... 1) the capital intensity of the steel industry (capital needs), 2) investment cycles, and 3) the fact that the steel industry was less dependent than other industries on external financing.” Id. at 46.
The Court of International Trade thereafter held that Commerce had identified substantial evidence to support its finding of a nexus between the Korean government’s program of general control over the Korean financial system and the provision of specific benefits to the steel industry in the form of preferential long-term loans. British Steel II, 941 F.Supp. at 130. Korean producers now appeal that decision.
B. Analysis
The general rule is that a countervailing duty (in addition to any other duty) shall be imposed if
(1) the administering authority [Commerce] determines that -
(A) a country under the Agreement, or
(B) a person who is a citizen or national of such a country, or a corporation, association, or other organization organized in such a country,
is providing, directly or indirectly, a subsidy with respect to the manufacture, production, or exportation of a class or kind of merchandise imported, or sold (or likely to be sold) for importation, into the United States, and
(2) the Commission [International Trade Commission] determines that—
(A) an industry in the United States—
(i) is materially injured or
(ii) is threatened with material injury, or
(B) the establishment of an industry in the United States is materially retarded,
by reason of imports of that merchandise or by reason of sales (or the likelihood of sales) of that merchandise for importation,
then there shall be imposed upon such merchandise a countervailing duty, in addition to any other duty imposed, equal to the amount of the net subsidy. For purposes of this subsection and section 1671d(b)(1) of this title, a reference to the sale of merchandise includes the entering into of any leasing arrangement regarding the merchandise that is equivalent to the sale of the merchandise.
19 U.S.C. § 1671(a) (1988). The statute further provides that in order to countervail a governmental subsidy, Commerce must determine
whether the bounty, grant, or subsidy in law or in fact is provided to a specific enterprise or industry, or group of enterprises or industries. Nominal general availability, under the terms of the law, regulation program, or rule establishing a bounty, grant, or subsidy, of the benefits thereunder is not a basis for determining that the bounty, grant, or subsidy is not, or has not been, in fact provided to a specific enterprise or industry, or group thereof.
19 U.S.C. § 1677(5)(B) (1988). Thus, Commerce must determine whether a governmental program provided a benefit to a specific industry. Moreover, in the case of an indirect subsidy, evidence of a causal nexus between the program and the benefit is also required. See British Steel I, 879 F.Supp. at 1328.
In this case, Commerce determined that governmental control of lending institutions in Korea constituted a program, that the 'provision of preferential access to *1373loans constituted a countervailable benefit, that a causal nexus in the form of “aggressive targeting” of loans existed between the program and the benefit to the Korean steel industry, and that access to the loans was provided specifically to the steel industry.
1. The basic premise of the “aggressive targeting” theory is that the Korean government exercised its influence over Korean commercial banks to direct credit preferentially to the steel industry. In reviewing the administrative record, however, Commerce was unable to identify a single piece of direct evidence in support of its theory. In verification interviews, governmental and banking officials uniformly asserted that the government did not direct credit to the steel industry during the pertinent period. In the course of its investigation, Commerce also conducted interviews with several executives of large Korean banks, selected by the United States embassy in Korea for their experience and familiarity with the Korean financial system. None of those bankers suggested that the Korean government had sought to allocate loans preferentially to the steel industry during the period under investigation. To the contrary, the bankers stated that the government had attempted through a variety of formal and informal means to direct credit to small and medium-sized businesses, while encouraging banks to reduce their lending to large companies such as POSCO.
Third-party source materials that were introduced into the administrative record supported the bankers’ assertions that the government favored lending to small businesses during the investigation period. For example, a World Bank report on Korean economic development noted a “government policy to increase the bank lending to small/medium firms.” Similarly, a 1991 article in the Financial Times examined the government policy of encouraging banks to lend to small companies, while expressing doubts as to its effectiveness.
The absence of direct evidence to support the “aggressive targeting” theory is striking, given that successful implementation of such a program would seemingly require the knowledge and cooperation of the government and banking officials who were interviewed by Commerce. In addition, it is troubling that Commerce apparently did not seek to obtain documentary corroboration for its targeting theory. Instead, Commerce took the position that “no company or government documentation existfs] for a program such as direction of credit.” As a result, Commerce’s verification process consisted of “information gathering” rather than checking the accuracy of previously supplied information.
Commerce’s position with respect to the unavailability of documentation is difficult to accept, because Commerce encountered no difficulty in finding documentation for the Korean government’s direction of credit in other instances. The administrative record is replete with examples of preferential allocation of credit in favor of steel and other heavy industries in the late 1970s, before the period under investigation. And, as noted, there was ample evidence that the government shifted its policy and began to attempt to direct credit to smaller industries in the mid-1980s. In light of that evidence, the absence of corresponding documentation for a program of aggressive targeting to the steel industry suggests that the existence of such a program should be viewed with skepticism.
2. Because of the lack of direct evidence to support the aggressive targeting theory, Commerce had to look to circumstantial evidence to support its nexus finding. Commerce urges us to affirm based on what it terms “a series of interrelated determinations that, taken as a whole, demonstrate[ ] that a de facto program to provide steel ‘preferential access’ existfs].” The problem with Commerce’s argument, however, is that key elements of the circumstantial case for aggressive targeting are not supported by the record.
*1374The starting point for Commerce’s analysis is the Korean government’s long history of support for the steel industry, including the history of preferential direction of credit. Redetermination, slip op. at 23-29. That history is well documented. During the Heavy and Chemical Industries drive in the 1970s, the Korean government encouraged banks to lend to the steel industry and directly invested large sums of capital to help develop the industry. The Iron and Steel Promotion Act of 1970 gave further encouragement to the steel industry, promising government support in building infrastructure and securing foreign loans. Finally, the charter of the Korea Development Bank gave explicit preference to the steel industry in obtaining long-term credit. As Commerce noted, however, the entire legal framework of preferences for the steel industry was repealed by 1984. The pre-1985 system is pertinent only insofar as it provides a background against which to assess the Korean government’s subsequent conduct.
Commerce relied on what it termed “continued expressions of support” for the steel industry after the removal of legal preferences. Redetermination, slip op. at 32-33. In particular, Commerce focused on two specific references. The first was a 1988 prospectus filed by POSCO with the Securities and Exchange Commission in conjunction with a corporate bond issue. In that filing, POSCO noted that it had previously been designated as a “public company” by the Korean government and asserted that the designation indicated its strategic importance. The prospectus, however, cautioned investors that the “public company” designation carried no implication of financial support or performance. The prospectus does not seem helpful to Commerce’s argument, particularly given that the principal characteristic of a “public company” in Korea appears to be that certain restrictions are placed on foreign stock ownership.
The second reference is a reported statement by a Korean banker that even after the Korea Development Bank charter was changed to allow industries other than steel to have nominally equal access to long-term loans, the steel industry was still in a “better position” to receive loans. The probative value of that statement, however, is negated by the next sentence of the same report, which states that by 1985 “any remaining preferences” had ceased. As a result, the “expressions of support” identified by Commerce are not persuasive evidence of aggressive targeting.
The centerpiece of Commerce’s aggressive targeting theory is the observation that in the years after the removal of legal preferences in favor of the steel industry, lending to the steel industry actually increased slightly. Redetermination, slip op. at 30-31. According to Commerce, that trend indicates that the government’s policy of favoring the steel industry did not change but merely became more circumspect. Id. at 22. POSCO counters that the increase in lending from 1985 to 1991 is attributable to increased credit requirements associated with POSCO’s large steel mill project at Kwangyang Bay. Commerce rejected that argument, claiming that the Kwangyang Bay mill was in reality a project of the Korean government, not POSCO. According to Commerce, the increased borrowing associated with Kwan-gyang Bay was an example of how the Korean government manipulated both POSCO’s credit needs and the financial system to ensure that the steel industry would receive a large share of limited long-term credit. Id. at 34-37, 49-52.
The importance of Commerce’s characterization of the Kwangyang Bay project to its final conclusion of aggressive targeting cannot be overstated. If Commerce is correct in describing Kwangyang Bay as essentially a government project, Commerce can plausibly contend that a de jure preference program was replaced with a de facto system under which industry credit requirements and supplies were both managed by the government. If *1375that premise is incorrect, however, the aggressive targeting theory is clearly unsupported. Commerce itself recognized the centrality of its characterization of Kwan-gyang Bay, stating in the Redetermination that “[i]f we were convinced that POSCO, on its own, made an independent, unfettered decision, based entirely on commercial considerations, to build the Kwangyang Bay steel mill, we would not have concluded that the [government of Korea] continued to target the steel industry for preferential access to scarce long-term credit at favorable rates.” Re-determination, slip op. at 49-50.
Commerce’s evidentiary support for that central tenet of its “aggressive targeting” theory is extremely thin. The principal record support that Commerce relied on in its Redetermination is a single speech made in 1981 by Korean president Chun Doo Hwan. In that speech, President Chun stated that the government would “give special emphasis to the nation’s steel industry,” and promised to “carry on the work of constructing the second integrated iron and steel mill.” The relevance of those excerpts from President Chun’s 1981 speech is questionable. The speech took place before any construction began at Kwangyang Bay. At the time of the speech, the Korean government was in the midst of an elaborate campaign to promote the development of heavy industry. By the time of Commerce’s investigation, however, that entire framework had been dis-maiítled and government attention had shifted to emphasize aid to smaller industries. Tellingly, Commerce was unable to identify any other statements by government officials or documentary evidence indicating an active role for the government in the decision to build the steel mill at Kwangyang Bay.
The other pieces of evidence used by Commerce to support its conclusions regarding Kwangyang Bay are even weaker. For example, in the Redetermination Commerce pointed to an Industry Basic Placement Plan published by the Korean government that described the Kwan-gyang Bay industrial estate as centered around the iron and steel industry. See Redetermination, slip op. at 36-37. That document does not discuss how the decision to build Kwangyang Bay was made. More importantly, the document cited by Commerce dates from 1990, after the steel mill was completed. The only other evidence that Commerce cited is a single newspaper article regarding the completion of the Kwangyang Bay steel mill, which describes the mill as “implemented by the Fifth Republic.” That statement could be plausibly interpreted as referring either to Kwangyang Bay’s relative position in the National Industrial Estate system implemented by the Korean government to disperse economic development, or to more- direct involvement by the government in the specific project. In any case, it is hardly a sufficient base from which to draw a conclusion that the Korean government directed POSCO to build the plant.
Conversely, there was considerable evidence in the record that indicated that POSCO had every reason to decide on its own to-build the steel mill at Kwangyang Bay. During the construction of the Kwan-gyang Bay facility, the demand for steel in Korea almost doubled. Completion of the plant moved POSCO into a position as one of the world’s largest steel manufacturers, and the plant has proved to be a profitable investment. Finally, the record shows that POSCO internally financed a high percentage of the development costs. Although not conclusive, each of those facts suggests that the Kwangyang Bay development was what it appears to be: a rational decision by a private actor to expand facilities to service a growing market.
Apart from the materials relating specifically to Kwangyang Bay, Commerce relied on a number of third-party materials allegedly showing the continued direction of credit by the Korean government to the steel industry. A review of each of those documents demonstrates that, singly or collectively, they do not provide support *1376for Commerce’s aggressive targeting theory.
First, Commerce points to statements made by certain Korean bankers that the government influenced lending to favor manufacturing sectors of the economy over service sectors. Inferentially, the steel industry could be expected to benefit from that intervention. The same bankers, however, also made clear that lending to small businesses was favored over large ones, that the government had instructed the bankers to reduce lending to the largest corporations in Korea, of which POSCO was one, and that there were no specific preferences in favor of the steel industry.
Several sources referred to the ability of the Korean government to allocate credit in commercial banking. A report of the Organization for Economic Co-operation and Development, for example, stated that supervision by the Korean Ministry of Finance “goes beyond the mere monitoring and supervisory functions usually exercised by official bodies and has included allocation of financial resources.” Similarly, articles in the Far Eastern Economic Review and the Economist made general references to official intervention in credit allocation. None of those sources, however, suggested that government officials intervened in lending decisions on behalf of the Korean steel industry during the investigation period.
The only sources identified by Commerce in support of credit allocation in favor of steel related to the period in which the Korean government was overtly favoring heavy industry. For example, a World Bank report assessed the distortive effect of preferences to steel and other favored industries before noting that in the mid-1980s government policy changed to favor smaller industries. An academic paper presented at the Korea Development Institute in 1989 also described credit policies undertaken by the government to favor steel, but did not suggest that those policies continued after the 1970s.
In sum, Commerce’s conclusion that there is a causal nexus between the Korean government’s control of the financial system and the domestic loans received by the steel industry from private sources is unsupported by substantial evidence. While it is clear that the Korean government exercised control to benefit the steel industry at one time, Commerce has not pointed to evidence from which it is reasonable to infer that the government’s control continued into the period of investigation. The portion of the Court of International Trade’s judgment sustaining the imposition of countervailing duties based on domestic credit provided to the Korean steel industry by private Korean lenders therefore must be reversed.
IV. FOREIGN LOANS
A. Background
In its Final Determination, Commerce found that the Korean government’s control of long-term lending in Korea constituted a government act or practice, and that this control led to preferential access to direct foreign loans for the steel industry, which- was a specific benefit not accorded other industries. Commerce based its benefit determination “on the disproportionate share of foreign loans obtained by the steel industry in relation to all other industries.” 58 Fed.Reg. at 37345. In particular, Commerce noted that “the iron and steel industry has received approximately 60 percent of all direct foreign loans to the manufacturing sector since 1985, and over 40 percent of all direct foreign loans to all industries since this time.” Id.
In British Steel I, the Court of International Trade held that “Commerce has failed to direct the [Court of International Trade’s] attention to other reasons that may explain why respondent steel companies have such a large share of direct foreign loans.” 879 F.Supp. at 1327. “Commerce has failed to point out or to invite the [Court of International Trade’s] *1377attention to evidence on the record to permit the [Court of International Trade] to conclude there is substantial evidence in the record to support the conclusion of Commerce that this program bestowed industry-specific benefits on respondent steel companies.” Id. As a result, the Court of International Trade remanded the Final Determination to Commerce for further explanation.
In its Redetermination, Commerce concluded that record evidence demonstrates the Korean government’s targeting of the steel industry for a disproportionate share of foreign credit through preferential access to foreign loans. Redetermination, slip op. at 38. Commerce pointed to record evidence of the Korean government’s involvement in the allocation of foreign loans. In particular, the Foreign Capital Inducement Act and the companion regulations require governmental approval of all direct foreign loans, establish eligibility criteria, and require initial consultations with the Korean government before negotiating for a direct foreign loan. Id. at 37-38.
The Court of International Trade thereafter upheld Commerce’s Redetermination, finding substantial evidence supporting its finding of a nexus between the government program of control of the financial system and the preferential access to direct foreign loans provided to the steel industry. British Steel II, 941 F.Supp. at 130, 135-36. Korean producers now appeal that decision.
B. Analysis
Countervailing duties are authorized in response to government programs only when those programs constitute subsidies under the statute. See Kajaria Iron Castings Pvt. Ltd. v. United States, 156 F.3d 1163, 1174 (Fed.Cir.1998). Commerce characterized the foreign loans as subsidies on the ground that preferential access to those loans benefited the Korean steel industry. See Final Determination, 58 Fed.Reg. at 37344. That is an inadequate basis under the governing statute for determining that the foreign loans constituted subsidies.
The statute in effect during the period pertinent to this case, 19 U.S.C. § 1677(5)(A) (1988), defined the term “subsidy” as including the following:
(ii) The following domestic subsidies, if provided or required by government action to a specific enterprise or industry, or group of enterprises or industries, whether publicly or privately owned and whether paid or bestowed directly or indirectly on the manufacture, production, or export of any class or kind of merchandise:
(I) The provision of capital, loans, or loan guarantees on terms inconsistent with commercial considerations....
The Korean steel companies argue (1) that the terms of the foreign loans were negotiated with commercial entities, e.g., foreign banks and equipment suppliers, none of which were affiliated with the Korean government, and (2) that neither the government nor the domestic producers have identified evidence in the record indicating that the foreign lenders would have agreed to terms inconsistent with commercial considerations. In fact, Commerce found that the foreign loans were provided “from foreign banks at world market rates” suggesting that the terms of those loans were consistent with the terms to which those entities normally agreed. See Final Determination, 58 Fed.Reg. at 37345.
The Korean steel companies’ statutory argument is persuasive, especially in light of the government and domestic producers’ failure to confront the statutory language. Neither the government nor the domestic producers suggest that the Korean government induced or otherwise arranged for foreign lenders to provide below-market rate loans to POSCO, and there is no evidence that the Korean government took any other steps that affected the rates of those foreign loans. Com*1378merce’s entire case on the foreign loans rests on its assertion that the Korean government granted a subsidy to POSCO by preventing other Korean industries from borrowing on the international market, and that the steel industry thus obtained a benefit.
It may be that the steel industry obtained preferential treatment with respect to foreign loans vis-a-vis other Korean industries, but that is not enough to satisfy the statute. Section 1677(5)(A)(ii)(I) required Commerce to show that POSCO obtained loans “on terms inconsistent with commercial considerations.” There was no evidence that the loans to POSCO resulted from anything but arm’s-length commercial transactions governed by the market in international loans. Absent some evidence that the Korean government’s conduct in preventing some Korean companies from obtaining foreign loans had some impact on the terms of the loans obtained by POSCO - and neither the government nor the domestic producers point to any such evidence - there is no basis on which to conclude that POSCO was provided “loans ... on terms inconsistent with commercial considerations.” 19 U.S.C. § 1677(5)(A)(ii)(I).
In its brief, Commerce characterizes the Korean government’s conduct as the “allocation” of foreign loans. Underlying that characterization of the governmental prohibition of foreign loans to some Korean companies as an “allocation” of loans to the steel industry is the unarticulated assumption that foreign lenders had some irrational predisposition to provide a fixed number of loans to Korean entities regardless of their credit-worthiness and, barred from negotiating with a broad range of companies, they chose to provide all the earmarked Korean loans to the Korean steel industry. While such a scenario is not impossible, neither Commerce nor the domestic producers have identified record evidence of any such distortion of the international credit market. Nor has the government pointed to any record evidence of more direct intervention, e.g., loan guarantees by the government.
The domestic producers offer an analogy to support Commerce’s characterization of the foreign loans as a countervailable subsidy. That analogy supposes that natural gas is available on the world market at $2 per cubic meter but available in Korea’s internal market only at $5 per cubic meter. The domestic producers contend that if only one Korean company is selected to have access to the world market that company receives a benefit from the Korean government.
Whether or not a benefit is received in the natural gas case depends upon the baseline adopted: If the baseline is a system of no government regulation, then the selected company is not receiving a benefit, but all the other Korean companies are being harmed; if the baseline is a system of complete government prohibition then the selected company is receiving a benefit. While the natural gas analogy may present difficult questions in future cases, it is irrelevant to the issue at hand, because the section of the countervailing duty statute applicable to this case sets forth special and specific conditions for loans. Under the statute, Commerce was required to review the evidence in the record to determine whether the Korean government’s actions somehow resulted in the foreign lenders agreeing to terms inconsistent with commercial considerations. Because Commerce did not conduct such a review and did not conclude that the terms on which foreign credit was offered to the Korean steel companies were inconsistent with commercial considerations, the loans cannot be regarded under the terms of the statute as the countervailable products of a subsidy from the Korean government. We therefore reverse the portion of the Court of International Trade’s judgment affirming the assessment of countervailing duties based on preferential access to foreign credit.
*1379V. INFRASTRUCTURE
A. Background
The Kwangyang Bay Industrial Estate (KBIE) was established under a national industrial estate program. In April 1983, the Ministry of Construction (MOC) issued a Broad Kwangyang Development Base Plan describing the proposals for future development of the area for industrial purposes, including a harbor facility. The KBIE consisted of fifty-two operating companies in the steel, cement, petrochemical, assembled metal, and services industries. POSCO was one of the operating companies.
According to the Korean producers, the major items of infrastructure provided by the Korean government at the KBIE included: (1) an expansion of existing waterworks; (2) dredging of the harbor and construction of harbor structures; (3) construction of a road up to the boundary of the industrial estate; (4) construction of a railroad spur up to the boundary of the estate; and (5) construction of eighteen port berths, fifteen built by POSCO and three built by the Korean government.
In its Final Determination, Commerce determined that the Korean government limited access to the KBIE, that POSCO was the primary user of the infrastructure at the KBIE and that the provision of infrastructure conferred a specific benefit that was countervailable. Commerce rejected the Korean producers’ argument that the provision of infrastructure at the KBIE should be analyzed in the context of the national industrial estate system as a whole. Commerce relied upon Carbon Steel Wire Rod From Saudi Arabia, 51 Fed.Reg. 4206 (Dep’t Commerce 1986), as support for analyzing individual industrial estates rather than the entire national system when determining specificity.
Commerce also determined that the free use by POSCO of the fifteen port berths that had been built by POSCO was a coun-tervailable benefit because POSCO received a benefit not accorded any other company at the KBIE. The exemption from dockyard fees was deemed separate from the infrastructure benefit.1 Commerce rejected the Korean producers’ argument that the exemption was merely a reimbursement by the Korean government for the costs POSCO incurred in building the berths and deeding them to the Korean government.
On appeal, the Court of International Trade found substantial evidence on the record to support all of the determinations made by Commerce regarding infrastructure and dockyard fees in the Final Determination. See British Steel II, 941 F.Supp. at 133-35. The Court of International Trade specifically addressed the issue of the construction of the fifteen port berths at the KBIE, namely the domestic producers’ argument that POSCO did not build the berths themselves. See id. at 130-32. In rejecting their argument, the Court of International Trade concluded that substantial evidence existed that POS-CO built and paid for the fifteen port berths.
B. Analysis
1. Infrastructure
Whether the Korean government’s provision of infrastructure constitutes a countervailable subsidy requires a determination that it is provided to a specific industry, and is not merely a program of general availability. See 19 U.S.C. § 1677(5)(B) (1988). Because infrastructure may nominally be viewed as a program of general availability, Commerce has employed a three-prong specificity test (Infrastructure Specificity Test) to determine whether infrastructure is provided specifically to a certain industry:
The Secretary will not regard the provision of infrastructure by a government *1380as specific ... provided the Secretary determines that:
(i) The government does not limit who can move into the area where the infrastructure has been built;
(ii) The infrastructure that has been built is in fact used by more than a specific enterprise or industry, or group thereof; and
(iii) Those that locate in the area have equal access to, or receive the benefit of the infrastructure on the basis of neutral and objective criteria.
Proposed, Regulations, 54 Fed.Reg. at 23379-80. We agree with the domestic producers that substantial evidence on the record supports Commerce’s determination that the Korean government’s provision of infrastructure at the KBIE was a countervailable benefit provided specifically to POSCO.
The first prong of the Infrastructure Specificity Test was deemed by Commerce not to be satisfied because: (1) the Korean government admitted that certain industries are considered inappropriate for Kwangyang Bay, (2) the Korean government admitted that an industry is not free to locate in any industrial estate, (3) the Korean government selected priority industries, in particular the steel and iron industry, pursuant to the Industry Basic Placement Plan, (4) the Korean government’s description of the KBIE as centering on the “core mother enterprises” of steel and iron industry and its related and subsidiary industries pursuant-to the Industry Basic Placement Plan, id. at 37347; and (5) the Korean government’s listing of “Steel, Iron Works” as the “Industries of Specialization” at the KBIE.
As to the second prong, Commerce determined that POSCO was the dominant user of the KBIE because: (1) the Korean government admitted that the vast majority of the goods shipped into and out of the port are by POSCO, (2) the Korean government admitted that POSCO is the primary user of the port facilities at Kwan-gyang Bay, and (3) only eight of the fifty-two other companies located at the KBIE use the Kwangyang port facilities.
The information on the record as to the third prong, i.e., whether other companies located in the area have equal access to the infrastructure, was found by Commerce to be “not conclusive.” The Infrastructure Specificity Test, however, does not require a definitive answer to all three prongs. It is sufficient to deem the provision by the Korean government of infrastructure at the KBIE as a specific subsidy if two of the prongs have not been satisfied. Thus, the benefit to POSCO was properly countervailed.
The Korean producers argue that Commerce erred by not treating the infrastructure at the KBIE as part of a national industrial estate system. In particular, the Korean producers argue that Commerce’s conclusion in the Final Determination that the Korean government had not submitted sufficient information regarding the KBIE’s relationship to a national system is contrary to law-because Commerce was required to notify the Korean government of any deficiencies in information. The Korean producers assert that sufficient information is of record with respect to the national industrial estate system to satisfy their burden of production, thus shifting that burden onto Commerce to request any additional information. Commerce insists that the burden never shifted because the Korean producers failed to establish a prima facie demonstration that the KBIE was integrally linked to the national industrial estate system. The domestic producers also argue that the Korean producers failed to carry their burden regarding integral linkage because the information the Korean producers rely upon provided only limited detail on the national industrial estate system.
The Korean producers are incorrect that Commerce was required to notify the Korean producers of a deficiency in their information' as to integral linkage. The Korean producers were clearly aware, on *1381the basis of the Preliminary Determination, that Commerce was analyzing specificity based solely on the KBIE. The burden of showing that KBIE was integrally linked to the national industrial estate system was squarely on the Korean producers if they intended to use the integral linkage exception. See Creswell Trading Co. v. United States, 15 F.3d 1054, 1060 (Fed.Cir.1994) (although the burden of production initially was on Commerce to establish a prima facie case for imposing a countervailing duty, the burden of production then shifted to the exporter to come forward with sufficient evidence to rebut the prima facie case).
The Korean producers supplied some general information regarding the nature of the national industrial estate system in Korea, and more specific information as to three of the forty-seven national industrial estates. Other information cited by the Korean producers concerns future plans for the industrial estates, not actual data of Korean government investment or industry use of existing estates. We are not convinced that the information supplied was sufficient to carry the Korean producers’ burden.
The Korean producers also argue that Commerce erred by treating the infrastructure at the KBIE as a grant to POS-CO when the Korean government built the infrastructure, owned it, and charged POS-CO standard user fees. In particular, the Korean producers assert that absent preferential user fees, no benefit accrued to POSCO and thus there is nothing to be countervailed. Commerce and the domestic producers argue that the provision of preferential user fees is s separate issue and not properly part of the determination of whether a specific benefit was conferred on POSCO through the Korean government’s construction of infrastructure at the KBIE.
We agree with Commerce and the domestic producers that the proper measure of the benefit to POSCO was what POSCO would have had to spend to build the infrastructure if the Korean government had not built it. Under the three-pronged Infrastructure Specificity Test, there was substantial evidence on the record to show that the Korean government provided a specific benefit to POSCO by constructing infrastructure at the KBIE and that POSCO was the dominant user of that infrastructure. The Korean government’s construction of the infrastructure is an independent countervailable program separate from the provision of preferential rates. Thus, while preferred user fee rates is one way in which a subsidy can be determined, it is not the only way. See 19 U.S.C. § 1677(5)(A)(ii)(II) (1988). Commerce’s use of the Infrastructure Specificity Test here to determine that the Korean government’s construction of infrastructure at the KBIE was a grant or subsidy to POSCO was proper.
The Korean producers’ contend that Certain Steel Products From Brazil, 58 Fed.Reg. 37295 (Dep’t of Commerce 1993), Phosphoric Acid From Israel, 52 Fed.Reg. 25447 (Dep’t of Commerce 1987), and Carbon Steel Wire Rod From Trinidad & Tobago, 49 Fed.Reg. 480, 482 (Dep’t of Commerce 1984) show that a preferential rate methodology should have been used. In Certain Steel Products From Brazil, an issue was whether the construction of certain port facilities for the benefit of the steel industry constituted a provision of subsidized infrastructure. 58 Fed.Reg. at 37300. However, Commerce did not countervail the construction of the port facilities, but relied on a preferential rate analysis to determine whether a subsidy had been granted. In Carbon Steel Wire Rod From Trinidad & Tobago, Commerce actually employed the grant methodology, and not the user fee methodology, to countervail governmental construction of a marine terminal specifically for the Iron and Steel Company of Trinidad and Tobago. 49 Fed.Reg. at 482. Finally, in Phosphoric Acid From Israel, Commerce employed the Infrastructure Specificity Test and found that certain rail facilities failed the *1382second prong of the test (i.e., the infrastructure was used by a limited number of chemical companies). 52 Fed.Reg. at 25450. Nevertheless, Commerce did not classify the rail facilities as a subsidy because the user rates charged were not preferential. Id.
Commerce and the domestic producers both argue that the analysis in Phosphoric Acid is flawed because it did not take into consideration the separate issue of the construction cost to build the infrastructure.
The Korean producers’ arguments regarding the user fee methodology are not persuasive. None of the cases relied upon limits Commerce to a specific methodology in determining whether a grant or subsidy was provided to POSCO through the construction of the KBIE infrastructure. The preferential user fee methodology is not an exclusive method that must be used. See 19 U.S.C. § 1677(5)(A)(ii)(II) (“subsidy ... includes, but is not limited to ... (II) The provision of goods or services at preferential rates.”) (emphasis added). Thus, Commerce properly applied the Infrastructure Specificity Test in determining whether the Korean government’s construction of the KBIE infrastructure conferred a eountervailable benefit to POSCO.
2. Exemption From Dockyard Fees
Commerce also determined that POSCO received a benefit from the exemption from dockyard fees. POSCO built and paid for fifteen of the eighteen port berths at the KBIE because the Korean government was experiencing budget-related delays. The parties agree that Korean law required POSCO to cede ownership of the fifteen berths to the Korean government upon completion. The Korean government sought to reimburse POSCO by exempting POSCO from dockyard fees that all other companies using the port were required to pay. The parties contest the characterization of that exemption and whether a benefit was conferred on POSCO by the exemption. The Korean producers contend that the exemption is not a benefit because POSCO relieved the Korean government of costs it would have normally assumed.
In the Final Determination, Commerce determined that a countervailable benefit was conferred on POSCO by the exemption from dockyard fees regardless of whether the fifteen port berths were built directly by the Korean government or by POSCO. Commerce explained that if the Korean government had built the port berths, instead of having them ceded by POSCO, Commerce would have “countervailed the construction funding as a specific infrastructure benefit,” as in the case of the infrastructure of the KBIE. We are convinced that Commerce’s determination is supported by substantial evidence, and that the exemption provided POSCO a eountervailable benefit.
Commerce’s determination that (1) the Korean government provided a specific eountervailable benefit to the steel industry, in particular POSCO, through the construction of infrastructure at the KBIE and that (2) POSCO’s exemption from dockyard user fees constituted a counter-vailable benefit are supported by substantial evidence. The judgment of the Court of International Trade in these respects is affirmed.
VI. TERCL ARTICLE 56-2
A. Background
Korea adopted Article 56-2 of the Tax Exemption and Reduction Control Act (TERCL) on November 28, 1987. TERCL Article 56-2 allowed a company making an initial public offering after January 1, 1987 to revalue its assets without meeting the requirement in the Asset Revaluation Act2 (ARA) of a 25% change in the wholesale price index since the company’s last reval*1383uation. Any company newly listed on the Korean Stock Exchange from January 1, 1987 to December 31, 1988 could revalue its assets until December 31, 1989. Pursuant to TERCL Article 56-2, 207 out of a total of 316 newly listed companies revalued their assets.
POSCO had previously revalued its assets, in 1982. After its shares were offered to the public and listed on the Korean Stock Exchange on June 10, 1988, POSCO revalued its assets in 1989, as permitted by TERCL Article 56-2, even though wholesale prices had increased by only 5.09% from 1982 to 1989. As a result of the revaluation, the recorded value of POS-CO’s assets increased by 2.614 trillion won (approximately $3.3 billion).
In its Preliminary Determination, Commerce determined that the revaluation was not granted to POSCO on a selective basis, as the domestic producers contended, due to the fact that over 200 other companies also revalued under TERCL Article 56-2. See Certain Steel Products from Korea, 57 Fed.Reg. 57761, 57770 (Dep’t Commerce 1992). Nevertheless, Commerce requested additional information concerning the magnitude of the revaluation and whether the methodology used in the revaluation was performed in accordance with standard Korean accounting principles.
The Korean government provided Commerce with data from a database maintained by the Korea Listed Companies Association (KLCA) on the 207 companies that had revalued under TERCL Article 56-2. POSCO also provided Commerce with information concerning the circumstances surrounding its revaluation. Thereafter, domestic producers submitted data to Commerce, contending that it showed that POSCO’s revaluation may have been significantly more than that of other companies. The data, however, related to 45 of the 207 companies that revalued pursuant to Article 56-2. Because the information was untimely submitted, it could not be verified.
In its Final Determination, Commerce found that the evidence in the record indicated that POSCO did not revalue more of its assets than allowed under Korean law and that the Korean government did not intervene to allow POSCO to revalue more assets than other companies. Regarding the information submitted by the domestic producers, Commerce determined not only that the submissions were untimely and thus unable to be verified, but also that the information only contained data concerning 45 companies. Moreover, Commerce said the information was unclear as to which of the 45 companies had revalued under the general provisions of the ARA and which had revalued under TERCL Article 56-2. Commerce therefore drew no conclusions from this information regarding the relative benefit POSCO derived from the program compared to other companies. Because it found no de jure or de facto selectivity regarding the timing or the method of revaluation, Commerce determined that the revaluation program was not countervailable.
On appeal to the Court of International Trade, the domestic producers challenged Commerce’s determination of non-specificity regarding the 1989 revaluation by POS-CO. The Court of International Trade, however, agreed with Commerce, finding substantial evidence to support Commerce’s determination that POSCO was not provided with a selective exemption from the 25% requirement in the ARA, and that the Korean government did not provide preferential treatment to POSCO concerning the revaluation.
B. Analysis
The domestic producers do not contest that POSCO complied with the requirements of TERCL Article 56-2 or that POSCO was ineligible to revalue its assets pursuant to that provision. In this court, the domestic producers make three arguments: (1) Commerce erred when it determined the dominant or disproportionate *1384share of the benefit conferred based on a percentage basis rather than on an absolute (or total benefit) basis; (2) Commerce erred by not relying on the information supplied by domestic producers as best information available (BIA), particularly given the Korean producers’ earlier lack of cooperation; and (3) Commerce erred by relying on the information contained in the KLCA Report, belatedly supplied by the Korean government, which was not part of the record.
The domestic producers criticize Commerce’s specificity analysis because the Final Determination failed to discuss whether the benefit received by the steel industry was dominant and disproportionate. According to the domestic producers, when analyzing disproportionality, Commerce must compare the industry’s benefit to some reasonable benchmark of a nonspecific distribution of governmental benefits, e.g., comparing an industry’s share of benefits to its contribution to the gross domestic product (GDP) of the economy as a whole. In this regard, the domestic producers assert that the Korean producers contributed only 2% to GDP and received 86.6% of the benefit (POSCO itself received 75.7%), demonstrating the dispro-portionality of the benefit.
The Korean producers argue that the program did not confer a disproportionate or dominant benefit because 81 of the 207 companies that participated in the program had a greater percentage increase than POSCO in the revaluation of their assets and because POSCO’s percentage increase fell below the average percentage increase for the companies that revalued their assets in the year in which POSCO revalued its assets.
In addition, the Korean producers argue that POSCO was not a dominant user of the program because Commerce determined that the program was available to over 200 companies in a wide variety of industries. Thus, Korean producers argue that Commerce’s conclusion in the Final
Determination that “POSCO did not revalue more of its assets than is generally allowed under Korean law” implies that Commerce did not consider POSCO to be a dominant user or to have received a disproportionate benefit.
Whether the benefit provided to the Korean steel industry, POSCO in particular, by the revaluation provisions of TERCL Article 56-2 is countervailable depends on whether the benefit was de facto specific.3 See 19 U.S.C. §§ 1671, 1677(1988). At the time of Commerce’s investigation, de facto specificity was analyzed by considering four factors:
(1) The extent to which a government acts to limit the availability of a program;
(2) The number of enterprises, industries, or groups thereof that actually use a program;
(3) Whether there are dominant users of a program, or whether certain enterprises, industries, or groups thereof receive disproportionately large benefits under a program; and
(4) The extent to which a government exercises discretion in conferring benefits under a program.
Proposed Regulations, 54 Fed.Reg. at 23379 (emphasis added) (codified at 19 U.S.C. § 1677(5A)(D)(iii) (1994 & Supp. Ill 1997)). At issue here is factor (3), whether the Korean steel industry was a dominant user of or received a disproportionate benefit under TERCL Article 56-2.
Although not specifically discussed in the Final Determination, we conclude that Commerce relied on record evidence, specifically the Verification Report of the Government of Korea, as demonstrating that POSCO was not a dominant user of and did not receive a disproportionate benefit from the revaluation program. POSCO was not a dominant user because over 200 companies from a wide variety of industries revalued *1385their assets pursuant to TERCL Article 56-2. Also, POSCO did not receive a disproportionate benefit because in the same year when POSCO revalued its assets upward by 94.0%, the average increase in asset value was 94.2%. Moreover, for the companies for which complete data was available, 81 out of 181 companies revalued their assets by a greater percentage than POSCO. Thus, Commerce’s conclusion that the asset revaluation pursuant to TERCL Article 56-2 was not de facto specific is supported by substantial evidence.
The domestic producers’ argument that Commerce “must analyze (and always has analyzed)” disproportionality by looking at the percentage of the total benefit of a subsidy program accruing to a particular company or industry is not correct. Determinations of disproportionality and dominant use are not subject to rigid rules, but rather must be determined on a case-by-case basis taking into account all the facts and circumstances of a particular case. See Proposed Regulations, 54 Fed.Reg. at 23368 (“[T]he specificity test cannot be reduced to a precise mathematical formula. Instead,.the Department must exercise judgment and balance various factors in analyzing the facts of a particular case.”); see also PPG Indus. v. United States, 928 F.2d 1568, 1576 (Fed.Cir.1991). Thus, it was not error for Commerce to rely on record evidence demonstrating no disproportionality based on the relative percentage benefit rather than on the absolute benefit conferred on POSCO. The domestic producers’ methodology could produce an untenable result, i.e., that a benefit conferred on a large company might be disproportionate merely because of the size of the company. Because Commerce’s methodology in determining dis-proportionality was reasonable, we will not disturb it on appeal. See Saarstahl, 78 F.3d at 1544.
We have considered the domestic producers’ other arguments regarding dispro-portionality of the TERCL Article 56-2 benefits, including their contentions that best information available should have been used by Commerce and that reliance on the KLCA Report was improper, but find them unpersuasive. Because Commerce’s determination that TERCL Article 56-2 did not provide a specific benefit to the Korean steel industry or to POSCO is supported by substantial evidence, the judgment of' the Court of International Trade in this respect is affirmed.
VII. CONCLUSION
The judgment of the Court of International Trade is affirmed-in-part and reversed-in-part.
AFFIRMED-IN-PART, REVERSED-IN-PART
. The other three port berths were built by the Korean government and thus were coun-tervailable as part of the infrastructure at the KBIE.
. In general the Asset Revaluation Act allowed assets to be revalued if they were (1) actually used for carrying out the business purposes of the company; and (2) capable of being depreciated/amortized under Korea’s tax laws.
. None of the parties argue that TERCL Article 56-2 is de jure specific.