4. Product comparisons

61. We next will consider the various characteristics of the products to assess whether there is a competitive or substitutable relationship between the imported and domestic products and draw conclusions as to whether the nature of any such relationship is direct. We will review the physical characteristics, end-uses including evidence of advertising activities, channels of distribution, price relationships including cross-price elasticities, and any other characteristics.

(i) Physical characteristics

62. Complainants argue that the defining physical characteristic of both imported and domestic products is the fact that they are distilled beverages. Other differences such as colouring or flavouring have no relevance in an analysis of whether products are directly competitive or substitutable. As summarized by the complainants:

The basic physical properties of soju and other categories of liquors concerned in this dispute are essentially the same. All distilled liquors are concentrated forms of alcohol produced by the process of distillation. At the point of distillation, all spirits are nearly identical, which means that the raw materials and methods of distillation have almost no impact on the final product. Post-distillation processes such as ageing, dilution with water or addition of flavourings, do not change the basic fact that the product sold is still a concentrated form of alcohol. ³⁷⁶

63. Korea argues that the different physical characteristics are substantial. They argue that distilled liquors can be derived from a variety of sources and that the selection of raw materials can play an important role in determining the ultimate qualities of the finished product. Korea argues that there is a distinction between brown spirits such as whisky and white spirits such as soju and gin. The brown colouring generally comes from aging in barrels whereas white spirits are not aged before bottling. Korea argues that even very minor differences in physical characteristics can be determinative if consumers perceive them as important. To put it another way, in response to a question from the panel, Korea argues that two products which are nearly physically identical can be found not directly competitive or substitutable if consumers perceive them differently. According to Korea the question's reference to nearly physically identical begs the question, because "nearly" must be defined in terms of consumer perception rather than comparison of physical characteristics by non-consumers such as chemists.
64. We do not agree with Korea's narrow interpretation. The Panel is examining the nature of the competitive relationship and determining whether there is an actual or potential relationship sufficiently direct to come within the strictures of Article III:2, second sentence. The physical characteristics themselves must be reviewed for if two products are physically identical or nearly so, then it obviously means that there is a greater potential for a direct competitive relationship. The United States argued that there can be two products of identical physical properties such as name brand and generic aspirin which are marketed somewhat differently and perceived somewhat differently by consumers. Nonetheless, they would be considered directly competitive or substitutable and the identical or nearly identical physical characteristics would be a significant factor in the analysis. We find this analogy useful.
65. The panel on Japan � Taxes on Alcoholic Beverages II referred to the usefulness of examining marketing strategies. ³⁷⁷ Marketing strategies that highlight fundamental product distinctions or, alternatively, underlying similarities may be useful tools for analysis. However, marketing strategies sometimes aim to create distinctions that are primarily perceptual between products with very similar physical characteristics. The existence of such perceptions based on marketing strategies rather than physical similarities and potential end uses does not mean that products are not at least potentially competitive. Indeed, it is natural and logical that marketers would recognize the possibility of capitalizing on the tax differentiation to create a marketing advantage.
66. As noted above, we have found it most fruitful to first examine whether the imported products are directly competitive or substitutable and only to turn to the question of whether they are like products second. The determination of whether two products are "like" has traditionally turned to a greater extent (although not exclusively), on the physical characteristics of products. It would be an incorrect reading of the law to argue that products' physical similarities were somehow less relevant for the category of directly competitive or substitutable products than for the subcategory of like products. To put it another way, if two products are physically identical or nearly so, it is highly probable that they are "like." They should not then be found to be not directly competitive or substitutable because marketing campaigns (or government tax regimes) have created a distinction in consumer perceptions. Such consumer perception distinctions are relevant but not determinative when the question is the nature of an actual or potential competitive relationship rather than merely a quantitative analysis of the current extent of competition. To find otherwise might allow allegedly discriminatory government measures to create self-justifying product distinctions between identical or nearly identical products.
67. We note that for purposes of the analysis under Article III:2, second sentence, products do not need to be identical to be directly competitive or substitutable. ³⁷⁸ However, as discussed above, physical similarities are relevant to the inquiry, particularly with respect to potential competition. All the products presented to the Panel have the essential feature of being distilled alcoholic beverages. Indeed, Korea imports ethyl alcohol for use as the base ingredient for diluted soju. Such ethyl alcohol is also used in a similar process for vodka and shochu, among other products. ³⁷⁹ All are bottled and labelled in a similar manner. ³⁸⁰ In our view, the differences due to the filtration or aging processes of the beverages described are not so important as to render the products non-substitutable. Aging in barrels will impart some flavour to the product as well as a dark colour, usually amber. But differences in color do not render products non-substitutable. We note that rum also is sold in light and dark versions, albeit not as a result of barrel aging. There have been no arguments that the two types of rums are not competitive due to this physical difference. Some beverages have flavourings added, e.g., juniper berries are added to clear spirits to make gin. However, we find that these differences in flavour or colour are relatively minor. We note that soju may also contain various sweeteners and flavourings. Indeed, the premium soju that has entered the market recently, corresponding to the entry of the imported products, has increased amounts of these additives. ³⁸¹ While there are some differences in the physical characteristics of the products, weighing the evidence presented, we find that there are fundamental physical similarities between the imported and domestic products that would support a finding that the imported and domestic products in question are directly competitive or substitutable. ³⁸²

(ii) End-uses

68. The issue of end-uses for these products has drawn much attention from the parties in this case. The complainants have argued that all distilled beverages have common end-uses. They have identified these as follows:

1. all of them are drunk with the same purposes: thirst quenching, socialization, relaxation, etc.;

2. all of them may be drunk in similar ways: "straight," diluted with water or other non-alcoholic beverages or mixed with other alcoholic beverages;

3. all of them may be consumed before, after or during meals; and

4. all of them may be consumed at home or in public places such as restaurants, bars, etc.

69. These are very broad categories of end uses. In response to questions from the Panel, the complainants identified "relaxation" from the concentrated alcohol content as perhaps a defining characteristic. They also responded that such beverages as soft drinks could not be included even though they fit some of the end-uses description because they contained no alcohol.
70. Korea has structured its defense primarily around two related aspects of the case. First is price, which will be discussed below. Second is differing end-uses. The two are related because Korea argues that the overwhelming end-use for soju, particularly diluted soju, is drinking with meals in Korean-style and other traditional restaurants whereas western-style beverages allegedly are almost never utilized in such a fashion. One of the reasons, allegedly, for this distinction is the great price differences which make western-style beverages too expensive for such frequent use. There are other reasons put forward by Korea for the end-use distinction as well. For instance, soju is said by Korea to be a harsh drink particularly suitable for drinking with spicy Korean food. ³⁸³
71. Prior to the second substantive meeting of the Panel, Korea produced a study by the A.C. Nielsen company which Korea argued documented the very distinct end-uses of soju and western-style beverages. The survey concluded that while all Korean restaurants, Chinese restaurants and mobile street vendors deal in standard soju, most cafes/western-style restaurants and bars deal in whisky. The survey also found that 29.3% of the respondents consumed alcoholic beverages at home with meals, while 81% were found to have consumed such beverages with meals at restaurants. The authors of the report claimed that diluted soju was the predominantly consumed alcoholic beverage with meals. Drinking diluted soju with meals was most popular at Korean restaurants (73%), followed by Japanese restaurants (18.7%). Of the 7 beverages offered to the respondents, none of them were consumed with meals at cafes/western style restaurants, bars and hotel bars. Finally, the survey found that soju is predominantly consumed straight (98.6%), while whisky is usually consumed on the rocks (63.8%).
72. The complainants responded to this survey by pointing out that there were several categories of overlapping end-uses. For instance, all Japanese restaurants served soju and 40% of them served whisky; a further 6.7% served brandy or cognac. Of the responding Western-style restaurants and cafes, 90% served whisky and a lesser number served other types of western-style beverages. However, 21.7% served soju. Also, the complainants noted that while only 1.7% of the individual respondents drank whisky at home with meals, only 29.3% of all respondents consumed any alcoholic beverages at home with meals. Therefore, the proper comparison was of the 1.7% with the 29.3 %, thus leaving 5.8% of all respondents who consumed alcoholic beverages with meals at home as drinkers of whisky as the accompaniment. Complainants have questioned some of the findings of the Nielsen survey, but also have argued that these results are actually indications of overlapping end-uses.
73. Complainants have noted that there were almost no western-style beverages in Korea until the last five years following changes in the duty rates on imported distilled beverages. Furthermore, they argue, alcoholic beverages, like many foods and beverages, are experience based products. People tend to purchase what they are familiar with and change their tastes only over a period of time. They will only make minor substitutions for the familiar product at first and higher frequency will tend to occur over a period of time until a fairly stable rate is achieved. ³⁸⁴ The trends shown in the Nielsen survey -- as well as the substitutability shown in the EC's market survey, the Dodwell study -- constitute, according to complainants, unmistakable evidence of the beginnings of substitutability and common end-uses by imports. Trends are significant with respect to such products. We think there is merit in these arguments and further note that this is another reason why the distinction offered by Korea between potential and future competition is too stark. Reasonable projections of increasing substitutability over a period of time are relevant and valid for determinations made pursuant to Article III:2.
74. Korea offers an analogy between the alcoholic beverage market and the automobile market which we do not find particularly useful. Korea argues that the imported products are Ferraris compared to soju's Renault Clio. However, the analogy is inapt. Automobiles are durable goods of great value relative to income that are only purchased periodically, generally only once every several years. It is probable that the purchaser of a Renault Clio has no option to purchase a Ferrari which might cost considerably more than the Clio purchaser's annual salary. Alcoholic beverages, on the other hand, are consumer goods which are purchased frequently, and even the Clio purchaser can afford to purchase a bottle of a more expensive beverage at least occasionally. The ratios between $10 and $100 products may be the same as between $10,000 and $100,000 products, but the purchasing decisions of ordinary consumers in the two situations are quite distinct.
75. The EC submitted a market survey conducted by Trendscope during the second substantive meeting of the Panel. It was of the same general type as Korea's Nielsen survey. It examined end-uses but did not go into specific price comparisons as did the earlier submitted Dodwell study. Korea requests that we disregard the Trendscope survey. As discussed above, we decline to disregard the survey; however, we do not, in fact, accord much weight to the submission by the EC. It adds little of probative value to the extensive prior submissions of the parties. What was of more interest was the nature of some of the substantive disagreements of the parties concerning information contained in the Trendscope survey. Among other things there was a disagreement over the correct Korean terms and whether the questioners adequately distinguished between "meals" and "food." Korea apparently puts a great deal of store in this distinction, arguing that one must look exclusively at "meals" where Korean soju predominates rather than at mere "food" which might include snacks where western-style beverages might be consumed.
76. We do not consider this a meaningful distinction between end-uses of products, certainly not enough to establish separate and non-competing product categories for purposes of Article III:2, second sentence. Neither this nor other panels should be required to draw such fine distinctions permitting significant differences in the application of the law based upon such differences as saying one beverage is used for snacks and another for meals. In reviewing the evidence of this case, we are not convinced that such a distinction, even assuming Korea's argument that it exists is correct, is sufficient. If a distilled alcoholic beverage is drunk with snacks, the nature of the competitive relationship is that it can be drunk with meals, either as marketing campaigns change or persons become more familiar with products new to the market. Indeed, we are unconvinced in general that the distinction between drinking alcoholic beverages at meals and drinking them either before or after meals was, in the context of this case, sufficient to render the products not directly competitive or substitutable. ³⁸⁵ Furthermore, we do not agree with the whole concept of basing a distinction on preferences for traditional drinks with traditional meals. ³⁸⁶ Korean food may be spicy, but it is not uniquely so in such a manner that soju and only soju is suitable for consumption with meals. It may, in fact, transpire that most Koreans will continue to prefer their traditional drink of soju with traditional food, but based on the information in the surveys and the trends in consumption patterns, it does appear that some Koreans at some times prefer other beverages and that these trends towards substitutability are likely to continue, even with respect to end-use categories we consider overly narrow.
77. We also are of the view that the presence of ad-mixtures in the Korean market lends credence to the conclusion of the substitutability of the imported and domestic products. Korea argued that the domestic ad-mixtures are not soju, but as soju is defined in the Liquor Tax Law as essentially diluted ethyl alcohol with some flavorings and additives, it is unclear what point Korea is making with this alleged distinction. If alcoholic beverages can be and often are drunk mixed, either as pre-mixes or mixed after purchase, it shows the potential for substitution between the base drinks and the lack of importance of the distinction which Korea attempts to draw based on alcohol strength. ³⁸⁷
78. End-uses constitute one factor which is particularly relevant to the issue of potential competition or substitutability. If there are common end-uses, then two products may very well be competitive, either immediately or in the near and reasonably predictable future. In this regard, we do find it relevant, albeit of less relative evidentiary weight, to consider the nature of the competitive relationship in other markets. If two products compete in a market that is relatively less affected by government tax policies, it might shed light on whether those same two products are potentially competitive in the market in question. Such a comparison is not dispositive by any means; neither should it be ignored. Its relevance consists primarily in whether it tends to corroborate the trends seen in the market in question or whether it reveals inconsistencies with complainants' case which deserve further consideration. In this regard, we note that in Japan there was increasing end-use substitutability between western-style beverages and Japanese shochu as consumers became increasingly familiar with the new product. Both soju and shochu are traditional drinks in their respective countries. Both markets involved small but growing import penetration following partial liberalization. The trends that the panel and Appellate Body observed in Japan appear to be beginning in Korea.
79. Article III cases deal with markets ³⁸⁸ and the response of Korean producers to changes in the markets provides significant evidence of at least a competitive relationship between soju and the imports. The trend towards increasingly overlapping end-uses are supported by the marketing strategies of the domestic Korean companies. These companies met the potential threat of imports of western-style beverages by creating and selling premium diluted soju. This beverage had more flavourings and was marketed in a manner more similar to western-style beverages than standard diluted soju. However, it remained within the definition of diluted soju in the Korean tax law. The physical characteristics were changed enough to be more similar to such imports while still enjoying the price advantages provided by lower tax rates. The complainants also produced evidence that these products were being advertised as competitive with western-style beverages. ³⁸⁹ Indeed, one advertisement referred to soju as a vodka-like product and also showed a new product called barley soju which clearly is intended to be comparable to imported products such as whiskies. ³⁹⁰ Evidence was also produced from various sources including Korean Air's in-flight magazine showing very similar advertising strategies for distilled soju and western-style beverages. ³⁹¹
80. Korea argued that advertisements in Korean Air's in-flight magazine should not be considered as aimed at the broad domestic market. Similarly, according to Korea, information from the website of the largest Korean soju producer, Jinro, in English or other advertisements in Japanese would be aimed at the export market not at the Korean domestic market, which is the only relevant one here. We take note of Korea's criticisms of these materials. However, we continue to disagree that the only relevant market for collecting data is the Korean domestic market. ³⁹² Rather the Korean market is the one that is the subject of our decision. In assessing the potential for products to be directly competitive with or substitutable for the domestic products it is relevant to look at how the domestic Korean companies produce, advertise and distribute their products in other markets as well as in Korea. Such evidence may be valuable for confirming or challenging trends and identifying important characteristics of the market which is the subject of the determination. In this case, the trends in the Japanese market where shochu and imported western-style beverages became increasingly used for the same purposes and the behaviour of Korean firms that met the challenge of imports with versions of soju increasingly similar to such imported beverages are relevant confirmation of what exists, albeit in a somewhat nascent form, in the Korean market.
81. The issue was raised whether the Panel should use the same criteria for defining markets under Article III:2 as under competition law. Korea was generally supportive of utilizing competition law market definitions for purposes of Article III and even went further and queried whether competition law market definitions might be too broad for purposes of Article III. Complainants argued, on the other hand, that Article III has a different purpose from competition law. Article III, they argue, is an anti-discrimination provision aimed at ensuring that government measures do not result in competitive conditions which favor the domestic industry. Therefore, the interpretation should be broad. According to the complainants, antitrust law has a different purpose of addressing the actions of individual firms or persons that threaten competition and such laws generally do not recognize any distinction between foreign or domestic persons. While the specifics of the interaction between trade and competition law are still being developed, we concur that the market definitions need not be the same. Trade law generally, and Article III in particular, focuses on the promotion of economic opportunities for importers through the elimination of discriminatory governmental measures which impair fair international trade. Thus, trade law addresses the issue of the potentiality to compete. Antitrust law generally focuses on firms' practices or structural modifications which may prevent or restrain or eliminate competition. It is not illogical that markets be defined more broadly when implementing laws primarily designed to protect competitive opportunities than when implementing laws designed to protect the actual mechanisms of competition. In our view, it can thus be appropriate to utilize a broader concept of markets with respect to Article III:2, second sentence, than is used in antitrust law. We also take note of the developments under European Community law in this regard. For instance, under Article 95 of the Treaty of Rome, which is based on the language of Article III, distilled alcoholic beverages have been considered similar or competitive in a series of rulings by the European Court of Justice ("ECJ"). ³⁹³ On the other hand, in examining a merger under the European Merger Regulation, ³⁹⁴ the Commission of the European Communities found that whisky constituted a separate market. ³⁹⁵ Similarly, in an Article 95 case, bananas were considered in competition with other fruits. ³⁹⁶ However, under EC competition law, bananas constituted a distinct product market. ³⁹⁷ We are mindful that the Treaty of Rome is different in scope and purpose from the General Agreement, the similarity of Article 95 and Article III, notwithstanding. Nonetheless, we observe that there is relevance in examining how the ECJ has defined markets in similar situations to assist in understanding the relationship between the analysis of non-discrimination provisions and competition law. ³⁹⁸
82. In making our assessment with respect to degree and nature of overlapping end-uses, we wish to make clear that we are not putting a burden of proving the negative on Korea. Rather, we think that the complainants submitted adequate evidence, inter alia, in the form of the Dodwell study, anecdotal evidence, and evidence of trends and results in other markets to establish this portion of their case. We also have taken note of the information in the Nielsen and Trendscope studies. All the beverages described are utilized for socialization purposes in situations where the effect of drinks containing relatively high concentrations of alcohol is desired. They may be used in a variety of social settings, including with food, either meals or otherwise. Korea's attempts to rebut this argument ultimately were unpersuasive. The distinctions that Korea would have us draw are too narrow and transitory. We decline to base a decision on whether a particular type of food is a meal or merely a snack. Indeed, as discussed above, we are sceptical of the whole meal-based rationale which is an important part of the Korean case. In balancing the evidence in this regard, we are mindful of the examples offered by the drafters of Article III and Ad Article III who considered apples and oranges directly competitive or substitutable products. Thus, we conclude that, on balance, the evidence is that there are current and potential overlapping end-uses sufficient to be supportive of a finding that the domestic and imported products are directly competitive or substitutable. ³⁹⁹

To continue with Channels of distribution and points of sale

³⁷⁶ EC First Submission at para. 97; US First Submission at para. 68.

³⁷⁷ Report of the Panel on Japan -- Taxes on Alcoholic Beverages II, supra., at para. 6.28.

³⁷⁸ Appellate Body Report on Canada -- Certain Measures Concerning Periodicals, adopted on 30 July 1997, WT/DS31/AB/R, at p. 28.

³⁷⁹ See para. 6.153 and para.6.161. Korea has argued that this statement takes in too much. It would also imply that certain industrial products might also be like soju or directly competitive or substitutable for it. We agree that this commonality of source material is not, in and of itself, sufficient for our analysis in this case. However, it is a factor which we take into consideration for it does go to show that there is a fundamental similarity in the basic materials used in the manufacturing process.

³⁸⁰ Korea attempted to place a great deal of emphasis on the differences in bottle sizes and labelling. Korea argued that bottles of soju were very different from bottles of shochu which were, according to Korea, made to look more like whisky. We find these differences relatively minor compared to the similarities in presentation.

³⁸¹ See para. 5.55 and para. 7.18.

³⁸² We note that these findings with respect to physical characteristics support our conclusion in section 3, above, that the identified imported products should be considered as a single category.

³⁸³ We note that Korea elsewhere emphasizes the sweetness of soju for purposes of distinguishing it from shochu. (See para.5.55.) Also, Korea submitted a copy of an advertisement for a standard diluted soju which emphasized its mildness. (Attachment 6 to Korean First Submission) This would seem to imply that Koreans would be willing to substitute allegedly less harsh western-style beverages were they to experience them. Furthermore, it is unclear how Korea's emphasis on the lower alcohol level of diluted soju relative to western-style beverages accords with the assertion of its singular harshness. Finally, it also is unclear why food should be seen as necessary to cushion the effect on the stomach of a lower alcohol drink compared to a higher alcohol drink.

³⁸⁴ See para. 6.120

³⁸⁵ Korea attempts to draw a number of product distinctions based on quite narrow differences. For example, Korea drew distinctions between some products based on whether they were given as gifts. Some of these assertions appeared to be contradictory. For instance, Korea stated that a distinction between diluted soju and brandy was that cognac and brandy were generally used as gifts, unlike diluted soju. However, Korea later stated that distilled soju was distinguishable from cognac and brandy because it was used for gifts while cognac and brandy were sold in "high-brow restaurants." (See para. 5.259 and para.5.295) When asked to explain, Korea responded with an even finer distinction as to what occasions particular gift beverages are used for. (See para. 7.23) Korea attempts to draw too fine a line between products for purposes of analysis under Article III:2, second sentence.

³⁸⁶ Korea has argued that the notion that distilled alcoholic beverages are not to be consumed with food is a peculiarly western notion. It is not clear that this assertion concerning "western notions" is necessarily true. Such drinks as vodka and whisky may very well be associated with traditional meals in some of the countries of origin. Furthermore, even if it is true that westerners do not generally drink distilled beverages with meals, it begs the question as to whether Koreans would like to, or sometimes do now, drink western style beverages with their meals. Also, Korea's assertion that "soju is a volume drink; vodka is not" (See para.5.269) could be questionable based on drinking behaviour in other markets. Again, while our decision is in regard to the Korean market, consumption patterns elsewhere are relevant, at least for purposes of assessing potential competitiveness.

³⁸⁷ The question of mixes highlights another of the inconsistencies that emerge from Korea's narrow product-by-product comparisons. For example, Korea argues that an important distinction between soju and vodka is that soju is almost always drunk straight whereas vodka is a mixing drink. (See paras. 5.268-5.269) If this is an important distinction, it would seem an important similarity then between soju and whisky, for example, that the two are often drunk straight. Or if whisky is mixed, it generally is with ice or water which also would seem to highlight the similarity with diluted soju on the basis of alcohol strength of the consumed product. It is also unclear what the basis is for Korea's assertion that vodka is a mixing drink. While it frequently is served that way, it also is served straight.

³⁸⁸ Appellate Body report on Japan -- Taxes on Alcoholic Beverages II, supra., at p. 25.

³⁸⁹ See US Exhibits E and F and EC First Submission, Annex 12. See also the Sofres Report, supra., at pp. 23-24.

³⁹⁰ US Exhibit Q.

³⁹¹ US Exhibit D.

³⁹² It is not clear that all the advertisements were aimed completely outside the Korean market as Korea claims. The advertisements in U.S. Exhibit D appear to be in Korean as well as English.

³⁹³ See Commission v. France, Case 168/78, 1980 ECR 347; Commission v. Kingdom of Denmark, Case 171/78, 1980 ECR 447; Commission v. Italian Republic, Case 319/81, 1983 ECR 601; Commission v Hellenic Republic, Case 230/89, 1991 ECR 1909.

³⁹⁴ Council Regulation No. 4064/89 of 21 December 1989 on the control of concentrations between undertakings.

³⁹⁵ Case No. IV/M 938 � Guinness/Grand Metropolitan.

³⁹⁶ Commission v. Italy, Case 184/85, 1987 ECR 2013.

³⁹⁷ United Brands v. Commission, Case 27/76, 1978 ECR 207.

³⁹⁸ In finding the relationship of the provisions to each other relevant, we do not intend to imply that we have adopted the market definitions defined in these or other ECJ cases for purposes of this decision.

³⁹⁹ We note that the conclusions we reach in this section regarding end-uses supports our conclusion in section 3, above, that the identified imports should be considered a single category.