(iii) Channels of distribution and points of sale

83. There is a considerable degree of overlap between the questions of common end-uses and common channels of distribution. Often, consumer products will be distributed in a manner that reflects their intended end-uses. Channels of distribution tend to reveal present market structure while end-uses deals with both the current overlap, if any, and potential for future overlap. In the present case, it is evident that soju and western-style beverages are currently sold through similar retail outlets in a quite similar manner for off-premise consumption. ⁴⁰⁰ Korea has argued that when taken from such outlets soju is consumed differently; this argument is addressed in the preceding section. Similarly, the complainants have shown that there are some similarities in other presumably more minor outlets such as duty free sales.
84. The primary area of disagreement is with respect to the channels of distribution for on-premise consumption. Korea argues that soju is sold primarily for use with Korean food in Korean-style restaurants. This was broadened and further explained by Korea through the Nielsen survey, which Korea argues provides evidence that most soju for on-premise consumption was sold to traditional Korean-style restaurants, as well as Japanese and Chinese restaurants and mobile vendors. Conversely, western-style beverages were sold for on-premise consumption primarily to cafes/western-style restaurants and bars.
85. As discussed above, the complainants have noted that there was overlapping distribution in the Japanese-style restaurants and cafes/western-style restaurants in Korea's Nielsen survey. We also noted from the Nielsen survey that, with respect to sales to cafes/western-style restaurants, while only 13 of the 60 survey respondents said they sold soju compared with 54 of the 60 saying they sold whisky, they sold 22,710 ml per month of soju compared with 11,702 ml of whisky. That is, more soju was sold than whisky in this allegedly western-style beverage channel of distribution. This seems to detract from the Korean claim that this type of on-premise channel of distribution overwhelmingly favoured whisky.
86. Korea asserted that western-style beverages are limited for on-premise consumption to "classy" establishments such as "high-class" bars, karaoke bars and expensive restaurants. ⁴⁰¹ In response to these arguments, the United States sent its embassy personnel in Seoul in search of large, traditional Korean-style restaurants to test the hypothesis. They claim to have found nine such establishments in the vicinity of the US Embassy that sold both whisky and soju. This prompted a discussion among the parties as to whether the identified restaurants would be typical or more expensive than normal. The resolution of the question of whether these restaurants were representative or were too expensive to qualify as "traditional Korean-style" is less important than the nature of the discussion itself. We do not think that a product distinction in a dispute under Article III:2, second sentence, can turn on such a thin and changeable distinction as Korea has attempted to make based on whether a restaurant is "high-class" or "expensive" or not. The only meaningful distinction in channels of distribution and points of sale that came to light in this case was the distinction between on-premise and off-premise distribution, but that distinction does not appear to distinguish between the imported and domestic products at issue. We find that, overall, there is considerable evidence of overlap in channels of distribution and points of sale of these products and such evidence is supportive of a finding that the identified imported and domestic products are directly competitive or substitutable.

(iv) Prices

87. Complainants have submitted a study of Korean consumer behaviour (the Dodwell study) related to relative price movements of soju and various western-style beverages, including premium Scotch whisky, standard whisky, cognac, vodka, gin, rum, tequila and liqueurs. The Dodwell study purports to show what happens when either the price of soju increases or the price of western-style beverages decreases, both done in specific increments. The survey also attempted to determine whether there was any evidence of cross-price elasticity. In response to Korea's challenges to the data and methodology, the complainants responded that the study was not attempting to show actual calculations of cross-price elasticity ratios because of difficulties inherent in the situation. Complainants said that the imported products had not been available in sufficient quantities to provide adequate consumer familiarity with the products and, furthermore, the Korean tax measures at issue also skewed the pricing and product availability structure such that it would be difficult to calculate actual cross-price elasticity ratios. As noted above, the complainants argued that alcoholic beverages were experience goods. People tend to consume what they are familiar with. Brand and product loyalty are strong and consumers will change their patterns only slowly over a long period of time following significant marketing activity and dependent upon plentiful product availability. Complainants emphasized the statement in the study that it was intended to "determine whether any evidence exists of cross-price elasticity between different spirits categories" rather than actually calculating such elasticities. Complainants referred to this as a more modest goal that was achievable and all that was possible in the circumstances.
88. Complainants stated that the evidence of substitutability was quite strong when the two separate trends of lowering import prices and raising soju prices occurred. The United States summarized this in charts which showed the Dodwell results concluding that the respondents would chose imported brown spirits rather than soju 15.22% of the time under current price conditions, but 28.4% of the time when the price of diluted soju was raised 20% and the price of brown spirits was lowered to its lowest point in the survey. Similarly, with respect to white spirits the choice went from 13.8% for imports at current levels to 23.8% when soju was again increased 20% in price and imports were at their lowest survey levels. ⁴⁰²
89. Korea provided considerable criticism of the Dodwell study. Citing a EC Commission notice on submissions related to EC competition law, Korea argued that any market study done for the purposes of influencing decision makers must be suspect. Korea also noted that surveys based on asking consumers hypothetical questions about opinions rather than direct factual questions are inherently untrustworthy because, among other things, there may be ambiguity in the questions and there was a need to infer factual results from opinions. Korea also noted that the firm retained to do such a survey has an incentive to try to provide answers consistent with the clients desires so that it might be retained again in the future. Specifically, Korea criticized the complexity of the questions and the unrepresentative nature of the respondents. Korea pointed out a number of anomalies in the results such as increases in soju consumption when the price increased from 1,100 won to 1,200 won. Korea also complained that premium diluted soju was included in the alternative samples along with imported beverages rather than included along with standard diluted soju as the base for comparison. According to Korea, this skewed the results. Finally, Korea was critical of the formulation of the questions, which Korea argued could be taken by the survey respondents to mean that they were being asked if they would try a bottle of imported beverages as a one-time purchase if offered a special cut rate price.
90. The complainants repeated that the Dodwell study had much more modest purposes than calculating cross-price elasticities for alcoholic beverages in the Korean market. Complainants noted that the Korean market had only been even partially deregulated for a few years and cited the findings of both panels examining Japanese alcoholic beverage taxes to the effect that government regulations and taxes often can freeze consumer preferences. In light of this, according to complainants, it stands to reason that the Dodwell study must be based on a selection of persons who have tried western-style beverages in order that they might have a frame of reference. Because of the recent arrival of western-style beverages in the market, they had to be asked a series of hypothetical questions rather than asking merely for factual information about current behaviour. Also, given the nature of alcoholic beverage purchases as on-going decisions on relatively low cost consumables, it was correct to ask whether the respondents to the survey would be willing to purchase some western-style beverages if prices changed rather than asking them if they would change their fundamental drinking habits. Complainants noted that there always will be statistical anomalies in any survey, but that in the case of the Dodwell study the overall trends were clear even if there was an occasional negative correlation in the data. Finally, complainants have noted that the Dodwell study used the same research methods as the ASI study cited by the panel in Japan � Taxes on Alcoholic Beverages II. ⁴⁰³
91. Korea correctly identifies some of the weaknesses and anomalies in the Dodwell study. The responses move in unexpected directions in some instances. However, on balance, we consider that the Dodwell study provided useful information regarding at least the potential competitiveness of the imported and domestic products. We also do not agree that some of the issues highlighted by Korea are detrimental to the results. We do not find it a flaw that the chosen respondents were not an accurate cross-section of all of Korean society. The surveyors selected 500 men between the ages of 20 and 49 from 3 Korean cities who had purchased soju in the past month and whisky in the past 3 months. ⁴⁰⁴ The age, gender and geographic profiles make sense. It is illogical to ask someone if they would shift to another consumer product -- particularly a food or drink item -- following a price change if such a person had never sampled such a product, or a similar one, before and the group chosen seems to be the most likely to have done so. We also agree with complainants about the prospective nature of the questions. If one is asking about the response to potential price changes, it is difficult to understand how a question about current behaviour will elicit a useful response. Also, when dealing with a consumable product which has a low price relative to income, it is not necessary that a respondent will permanently change drink preferences. The willingness to occasionally substitute one product for another when there is a relatively high frequency of purchase should be sufficient.
92. We must also note a general concern with some of the Korean criticisms of the Dodwell study. Article III serves to protect the expectations of competitive opportunities. Requiring a survey based on current, actual behaviour would prevent a potential market entrant from ever challenging government restrictions. ⁴⁰⁵ Indeed, it must be recalled that the Appellate Body confirmed that such surveys are not the decisive factor in decision making under Article III:2, second sentence. We do not find the Dodwell study decisive, but it is consistent with other information and is therefore helpful evidence. When dealing with an inquiry in the nature of the competitive relationship between products, quantitative analysis is helpful, but not necessary.
93. There was also considerable disagreement between the parties on the level of price differences between soju and imported western-style beverages. Korea used weighted averages and claimed that whisky was nearly 11 times more expensive than soju, making the effect of the taxes negligible. Complainants responded that the price of standard Scotch whisky was only about three times more expensive than premium diluted soju. On the other hand there were even greater variations with categories such as whisky, for instance between bulk blended-in Korea brands and fine malt Scotches, but Scotch whisky nevertheless was generally considered a single category of beverages. The complainants also argued that because of the high taxes and duties, the imports had tended to be of the higher priced brands thus skewing the numbers used by Korea. Korea further argued that the high priced brand argument was illogical because, unlike the Japanese system of specific duties, the Korean tax system was ad valorem. Complainants said that in such a restrictive market, it was not unusual for firms to lead entry into the market with higher priced niche brands to build awareness and sell with an exclusive cachet in segments where premiums could be charged and the consumers had higher incomes and therefore would be relatively less affected by tax levels. Complainants offered some evidence to support their claim by showing consumption patterns of various brands in other selected markets where there was a heavier relative weighting of sales towards lower priced brands than in Korea. ⁴⁰⁶
94. In examining the evidence before us, we found that, while there currently are significant price differences between the imported and domestic products, overall the differences were not decisive. Korea presented prices as weighted averages which obscured the higher prices for premium diluted soju, which was the small but fast growing category created specifically by Korean manufacturers to be most competitive with imports. The price of premium diluted soju appears to be approximately two times the price of standard diluted soju, while vodka was four times the price and standard whisky four and a half times the price of premium diluted soju. ⁴⁰⁷ Distilled soju was twice the price of standard whisky. ⁴⁰⁸ There are greater price differences within some categories, e.g., whisky, ⁴⁰⁹ which none of the parties argued rendered such subcategories of products not directly competitive or substitutable. Furthermore, we agree with the complainants that absolute price differences are less important than behavioural changes that occur due to relative price movements. ⁴¹⁰ When examined as a whole, the price differences are not so large as to refute the other evidence of potential competitiveness and substitutability, and there was evidence that relative price movements are likely to result in changes in consumption patterns. Overall, we found that the data on prices and the potential for changes in consumer behavior based on relative price changes, to be supportive of a finding that the identified imported and domestic products are directly competitive or substitutable.

(v) Conclusions with respect to "directly competitive or substitutable"

95. We are of the view that the weight of the evidence overall supports a finding that the imported and domestic products at issue are directly competitive or substitutable. Complainants have sustained their burden of proof in this case by showing that there is some degree of current competition as well as trends towards relative shifts in consumption from soju to the identified imported distilled beverages. The production and marketing decisions of the Korean beverage companies reflect a realization of this in a very concrete manner by the development and rapid success of premium diluted soju. There clearly is an attempt to develop an image of certain types of soju that shows a direct competitive relationship with imported alcoholic beverages. It is probable that there are different marketing focuses (e.g., whether identifying accompaniment with food as a favored mode of consumption) by the importers compared to standard diluted soju; however, marketing strategies alone should not be the basis for finding products not potentially competitive. Marketing strategies can be changed quickly and if there is substantial other evidence that products are potentially directly competitive, it would be incorrect to find them otherwise based on transitory factors such as marketing strategies especially when such strategies can be shaped by the very government policies in question. On the other hand, when two products which have some present market differentiation begin to be marketed in similar fashions, as is happening in the case of the Korean soju makers, it is strong evidence of potential competition. Again, the purpose of Article III is to protect competitive opportunities, not protect actual market shares. Competitive opportunities should encompass the ability to change marketing strategies without the need for beginning a new dispute settlement case. A mere change in marketing strategy cannot be all that distinguishes success from failure of a complaint pursuant to Article III:2. That clearly would be an overly narrow interpretation of the term directly competitive or substitutable.
96. The levels of overlapping end-uses are currently relatively low if end-uses are defined as narrowly as suggested by Korea. However, even within such overly-narrow end-use categories, the evidence must be viewed in light of the relatively recent introduction of the western-style beverages to the market. Furthermore, we do not agree with Korea's argument about the distinctness of current market differentiation. We think Korea has drawn too fine a distinction between products for purposes of Article III:2, second sentence. Again we recall the examples of substitutable products offered by the drafters, which included apples and oranges. This also can be seen in the significantly overlapping channels of distribution both for off-premise sales and on-premise sales. In our view, the only meaningful distinction with respect to channels of distribution in this case is the distinction between on-premise and off-premise consumption and both imports and soju are distributed through both channels.
97. There is evidence both of some level of current actual competition and significant potential competition. However, complainants do not have to prove that there is a complete overlap in their analysis of substitutability. Moreover, we take guidance from the earlier panel findings that the current market conditions may be skewed by government tax and regulatory policies which tend to freeze consumer preferences in favour of the domestic products. The current price levels are probably the most telling evidence contrary to complainants assertions. In our view, the Dodwell study is a useful piece of evidence showing the potential competitive relationship between the domestic and imported products under various pricing scenarios. It is not perfect evidence, but we do not find the Korean critique conclusive in rebutting its basic premises. Indeed, we find confirmation of some of the basic points about potential competitiveness in both the Korean end-use survey conducted by the AC Nielsen study and the Trendscope survey. Furthermore, we do not accept that the price differences in the present case establish that the products in question are not even potential competitors. Prices are subject to change by extraneous factors such as exchange rates.
98. We are of the view that there is sufficient unrebutted evidence in this case to show present direct competition between the products. Furthermore, we are of the view that the complainants also have shown a strong potentially direct competitive relationship Thus, on balance, we find that the evidence concerning physical characteristics, end-uses, channels of distribution and pricing, leads us to conclude that the imported and domestic products are directly competitive or substitutable.

5. Not similarly taxed

99. The Appellate Body in Japan � Taxes on Alcoholic Beverages II summed up its findings with respect to this element of the decision as follows:

Thus, to be "not similarly taxed", the tax burden on imported products must be heavier than on "directly competitive or substitutable" domestic products, and that burden must be more than de minimis in any given case. ⁴¹¹

100. In the present case, the Liquor Taxes on diluted soju are 35 percent and 50 percent on distilled soju. The Education Tax is surtax of 10 percent levied on soju. With respect to imported alcoholic beverages, the Liquor Tax ranges from 50 percent for liqueurs to 100 percent for whisky and brandy. The Education Tax is 30 percent for all imported alcoholic beverages except liqueurs which have a 10 percent rate. Thus the total tax on diluted soju is 38.5 percent; on distilled soju and liqueurs it is 55 percent; on vodka, gin, rum, tequila and ad-mixtures it is 104 percent; on whisky, brandy and cognac it is 130 percent. Thus the tax rate on imported whisky, for example, is more than three times the ad valorem rate on diluted soju. These differentials are clearly in excess of de minimis levels. ⁴¹²

6. So as to afford protection

101. The Appellate Body in the Japan Alcoholic Beverages case stated that the focus of this portion of the inquiry should be on the objective factors underlying the tax measure in question including its design, architecture and the revealing structure. ⁴¹³ In that case, the Panel and the Appellate Body found that the very magnitude of the dissimilar taxation supported a finding that it was applied so as to afford protection. In the present case, the Korean tax law also has very large differences in levels of taxation, large enough, in our view, also to support such a finding.
102. In addition to the very large levels of tax differentials, we also note that the structures of the Liquor Tax Law and the Education Tax Law are consistent with this finding. The structure of the Liquor Tax Law itself is discriminatory. It is based on a very broad generic definition which is defined as soju and then there are specific exceptions corresponding very closely to one or more characteristics of imported beverages that are used to identify products which receive higher tax rates. There is virtually no imported soju so the beneficiaries of this structure are almost exclusively domestic producers. ⁴¹⁴ Thus, in our view, the design, architecture and structure of the Korean alcoholic beverages tax laws (including the Education Tax as it is applied in a differential manner to imported and domestic products) afford protection to domestic production. We therefore conclude that there is nullification or impairment of the benefits accruing to the complainants under GATT 1994 within the meaning of Article 3.8 of the DSU.

7. Like Product

103. The complainants in this case argued that vodka is like soju. ⁴¹⁵ Korea disagreed. ⁴¹⁶ We note that there are many similarities between vodka and soju and that these are sufficient to establish that the products are directly competitive or substitutable. However, as the Appellate Body found in Japan � Taxes on Alcoholic Beverages II, the concept of "likeness" in Article III:2, first sentence, is to be narrowly construed. ⁴¹⁷ The question is whether the products are sufficiently close in nature that they fit within this narrow category.
104. We find that there is insufficient evidence in this case to make a determination that vodka and soju are like products. We do not find that they are "unlike". Rather we find that there is insufficient evidence in the record of this case to establish that they are like. In making this finding, we recall that the Appellate Body also noted that a determination of whether vodka was like shochu or was instead only directly competitive or substitutable did "not materially affect the outcome of [the] case." ⁴¹⁸ We find this conclusion equally valid in the facts of the case at hand. Thus, while we have found that vodka and the other identified imported distilled alcoholic beverages and the domestic products are directly competitive or substitutable, we are unable to conclude that the imported products, or any subcategory of them, are like the domestic products.

XI. Conclusions

1. In light of the findings above, we reached the conclusion that soju (diluted and distilled), whiskies, brandies, cognac, rum, gin, vodka, tequila, liqueurs and ad-mixtures are directly competitive or substitutable products. Korea has taxed the imported products in a dissimilar manner and the tax differential is more than de minimis. Finally, the dissimilar taxation is applied in a manner so as to afford protection to domestic production.
2. We recommend that the Dispute Settlement Body request Korea to bring the Liquor Tax Law and the Education Tax Law into conformity with its obligations under the General Agreement on Tariffs and Trade 1994.

⁴⁰⁰ US Exhibit G; EC First Submission, Annex 11. See also paras. 6.93-6.94 and 6.188-6.189. We note that this evidence also shows that imports are sold in a similar manner to each other and supports our conclusion in section 3, above.

⁴⁰¹ Statement of Korea at First Meeting of the Panel at p. 8; Statement of Korea at Second Meeting of the Panel at p .20.

⁴⁰² First Submission of the United States at paras. 78-85.

⁴⁰³ See Report of the Panel on Japan �Taxes on Alcoholic Beverages II, supra., at para. 6.32.

⁴⁰⁴ EC Annex 13 at p. 3.

⁴⁰⁵ As noted above in our discussion of potential competition, requiring surveys exclusively on current actual behavior would make it even more difficult for less wealthy complainants to establish sufficient actual market presence to establish a prima facie case of nullification or impairment.

⁴⁰⁶ See Annex 1 to EC Answers to Questions from the Second meeting of the Panel.

⁴⁰⁷ Thus, vodka was approximately eight times the price of standard diluted soju and standard whisky was approximately nine times the price of standard diluted soju.

⁴⁰⁸ EC Second Submission to the Panel at Annex 7. Also, from the Dodwell study and other evidence it appears that the following are the relationships between the prices of the other products: Gin is approximately 3.25 times the price of premium diluted soju and 6.5 times the price of standard diluted. Tequila is approximately 5.5 times the price of premium diluted soju and 11 times the price of standard diluted. Liqueur is approximately 5 times the price of premium diluted and 10 times the price of standard diluted. Cognac is approximately 12 times the price of premium diluted soju and 24 times the price of standard diluted soju. We note that Korea offered some lower price comparisons in its Interim Review comments to the effect that vodka was 5.7 times more expensive than diluted soju, while gin was 5 times more expensive and rum 6.2 times more expensive. Korea also stated that cognac/brandy was 19.2 times the price of diluted soju. Korea did not indicate if weighted averaging caused these differences from the figures above.

We note that in the decision in Japan � Taxes on Alcoholic Beverages II, adjustments were made for alcohol content. Panel Report in Japan -Taxes on Alcoholic Beverages II, supra., Annex VI, Figure 10. When such adjustments are removed, it seems that the absolute price ratio differentials in the present case are more similar to those in Japan than otherwise appears.

⁴⁰⁹ See para.6.105; Dodwell Study.

⁴¹⁰ Indeed, we must reiterate that caution must be exercised in relying on absolute price ratio differences in making product distinctions in a market such as this. Prices can respond to extraneous factors such as exchange rates or can be affected through product mix or high overhead or distribution costs possibly caused in part by the government policies at issue.

⁴¹¹ Appellate Body Report on Japan -- Taxes on Alcoholic Beverages, supra., at p. 27. It should be noted that in the case of Japan, the duties were specific and there was an ultimately unsuccessful argument by Japan that the tax/price ratios were not dissimilar. Because Korea's taxes are strictly ad valorem, the rates are more easily comparable and there is no such issue.

⁴¹² The fact that distilled soju and liqueurs are taxed at the same rate does not detract from this finding with respect to the other products and with respect to liqueurs compared to diluted soju.

⁴¹³ Appellate Body Report, on Japan -- Taxes on Alcoholic Beverages II, supra., at p. 29. See also Canada - Certain Measures Concerning Periodicals, supra., at pp. 30-32.

⁴¹⁴ The only domestic product which falls into a higher category that corresponds to one type of imported beverage is distilled soju which represents less than one percent of Korean production.

⁴¹⁵ See para. 5.100et. seq.

⁴¹⁶ See para. 5.264et. seq. and para. 5.296 et. seq.

⁴¹⁷ Appellate Body Report on Japan � Taxes on Alcoholic Beverages II, supra., at p. 21.

⁴¹⁸ Ibid.