(e) Prices

60. The European Communities submitted that the consumer survey in the 1998 SM Survey ³⁸⁴ shows that factors that have a direct impact on the prices of other spirits but not on the price of pisco itself, affect the demand for pisco, which shows a directly competitive or substitutable relationship between pisco and those other spirits. According to the European Communities, the survey measured respondent's reaction to changes in the relative price of pisco and the other distilled spirits, and their response indicates a significant degree of cross-price elasticity between pisco and other spirits. In such a scenario, the share of respondents choosing whisky and other spirits instead of pisco would increase from 17.7% to 30.5%. ³⁸⁵
61. The European Communities further argues that the survey showed that pisco consumers, by an overwhelming majority (70%), would opt for other spirits if pisco was unavailable, and only 17% would opt for wine or beer. ³⁸⁶
62. The European Communities asserts that the 1998 SM Survey shows the immediate reaction of consumers to price changes. Bearing in mind that the consumption of distilled spirits is based on habit, which changes very gradually, short-term elasticities are bound to be much lower than long-term elasticities. What this means, according to the European Communities, is that over a period of time, the price changes resulting from the elimination of tax differentials are likely to lead to a shift in consumption from pisco to other spirits by a larger degree.
63. To back up their claims on cross-price elasticity, the European Communities submitted a study conducted on the Chilean distilled spirits market in 1995 by the Gemines consulting group. ³⁸⁷ Chile also submitted a separate study with respect to the cross-price elasticity of pisco and imported distilled spirits. ³⁸⁸ We note at the outset, that both this study and the Chilean data lack a model incorporating the supply side, and as such all the variation in the data is interpreted as movements along, or shifts of, the demand curve.
64. We also note that the studies relied on by both parties rely on small samples. In the case of the Chilean data, the number of observations is particularly low. In the case of the 1995 Gemines study, it lacks data for the key independent variable, the price of whisky, for almost half the sample period. For these reasons we treat these studies with caution.
65. The European Communities argues that the 1995 Gemines study, which was actually done for the Chilean pisco industry, provides further evidence of significant cross-price elasticity between pisco and whisky. The European Communities notes that the Gemines study, which estimated the cross-price elasticity between pisco and whisky on the basis of historical sales and price data in the period 1985-92, found a cross-elasticity rate between pisco and whisky of 0.26%, indicating that if the price of whisky went up by 10%, the sales volume of pisco would increase by 2.6%. The European Communities point out that the Gemines study found that on this basis, pisco and whisky are "substitutes albeit to a limited extent".³⁸⁹
66. The European Communities states that another Gemines study done in 1996, whose findings were widely published, concluded that a reduction in the tax on whisky by 50% would lead to a 47% drop in the price of whisky, which would in turn lead to a 17% drop in the demand for pisco. ³⁹⁰
67. Chile points out that the EC's main evidence, 1998 SM survey, shows a remarkable lack of reliability. For instance, in the case of whisky, Chile argues that a more detailed analysis shows that whisky has a price elasticity of negative 5, meaning that a 10% reduction in price will lead to a 50% increase in volume, a result which is, according to Chile, completely out of range. ³⁹¹ Chile further argues that the elasticities on the other spirits do not deserve any comment because of their unrealistic results. We note, however, that the surveyors compared the envisaged reaction of the survey respondents to preferences expressed after the prices of pisco, whisky, and "other spirits", have changed in response to a hypothetical situation of a tax of 27% ad valorem being applied uniformly to all distilled spirits. Elasticity of substitution is a "partial concept" in that it is a measure of the relationship between one quantity and one price and assumes that all other factors are held constant. For example, a cross-price elasticity measurement could hold the price of pisco constant, decrease the price of whisky and determine the changes in quantity of pisco. In this study, because prices change simultaneously, the outcome of changes in the quantity demanded do not allow the accurate computation of either own-price or cross-price elasticities since the other explanatory variables did not remain unchanged (i.e., the ceteris paribus requirement in this type of analysis has not been provided for, and partial derivatives have not been determined).
68. The question arises, as to how a panel should deal with concepts such as cross-price elasticity in determining whether two classes of products are directly competitive or substitutable. In Japan � Taxes on Alcoholic Beverages II, the Appellate Body affirmed the decision of the panel to look at the economic concept of "substitution" as one means of examining relevant markets. However, the Appellate Body emphasised that this should be considered together with all other legitimate considerations, in the aggregate, in determining direct competitiveness and substitutability, i.e., the use of cross-price elasticity of demand is not the decisive criterion, but merely one among other criteria, such as physical characteristics, common end-uses, etc. ³⁹²
69. The Panel wishes to emphasise that the concept of substitution in markets should not be confused or equated with a numerical measurement of the extent of substitution as found in the co-efficient of cross-elasticity. The existence of substitution between pisco and imported distilled beverages lies at the heart of this dispute and is the fundamental issue in light of both the treaty text and in terms of analytical approach. However, the econometric measurement of the degree of substitution may not, partly for the reasons discussed below, always adequately reflect the extent of substitution.
70. A high coefficient of cross-price elasticity would of course lend more credence to a claim of direct competitiveness or substitutability, although a low coefficient of cross-price elasticity is not necessarily fatal to a claim of direct competitiveness or substitutability. Indeed, a low coefficient of cross-price elasticity may be due to the very measures in question in the dispute. In this regard, the Appellate Body stated that:

[A] tax system that discriminates against imports has the consequence of creating and even freezing preferences for domestic goods. In the Panel's view, this meant that the consumer surveys in a country with such a tax system would likely understate the degree of potential competitiveness between substitutable products. ³⁹³

71. It is our view that in general economic terms, a high estimated coefficient of elasticity would be important evidence to demonstrate that products are directly competitive or substitutable provided that the quality of the statistical analysis is high. In this case, in the 1995 Gemines study, for example, a cross-price elasticity of demand for pisco with respect to whisky of 0.26 was obtained, which is low (a 10% rise in the price of whisky will lead to an increase of only 2.6% in the demand for pisco). Nonetheless, we accept this as evidence of substitutability, recognising that some of the other factors discussed below are also indirectly influenced by the price of whisky and may have an impact on the market. We continue to recognise the above-mentioned need for caution in light of concerns about the quality of a particular statistical study.
72. Customs, traditions and consumer preferences embodied in brand loyalties could render demand less sensitive in the short run to changes in price and income than would otherwise be expected. It has been noted that alcoholic beverages are "experience goods" for which demand changes only slowly as consumers become gradually more familiar with new products. ³⁹⁴ We also think that the estimated elasticity would have been higher, had whisky been less expensive and we note that the whisky imported into Chile tends to be at the higher end of the price range for distilled spirits generally. Whisky is taxed at a much higher rate than pisco and we also note that tariffs on imported spirits were high for a long period of time. These factors, which include the very measures at issue here, can have a noticeable impact on the price of the imports, both directly and indirectly. For instance, the tax and tariff structure can change product offerings towards more expensive items and partly as a result, increase other costs such as distribution. If the retail price of whisky were lower than what it is now, or other normally lower priced products were readily available in a neutral tax setting, it is not unreasonable to expect based on the information before the panel that new customer groups could be attracted and the actual cross-price elasticity would be higher than the current estimates. ³⁹⁵
73. All these factors point to the logic in the Appellate Body's reasoning in the Japan � Taxes on Alcoholic Beverages II case, that undue weight should not be placed on estimated price-elasticity, demand-elasticity or cross-price elasticity. Studies that attempt to measure the relationship between dependent and independent variables and are only part of the totality of factors a panel should take into account in determining the question of direct competitiveness or substitutability. ³⁹⁶
74. As we have already noted, Chileans consider pisco as a traditional or national drink. We further noted that this does not mean that pisco cannot be substituted for at any time and at any occasion. In this regard, we note that Chile submitted that pisco, in any of its categories, has a pre-tax price that is substantially lower than whisky, and that consequently taxation has no real effect on retail prices. ³⁹⁷ We also note the EC response that Chile is comparing the price of a relatively expensive brand of whisky such as Johnnie Walker, with an inexpensive brand of pisco. The European Communities also points out that the differences in prices between different varieties of pisco can be as large as differences between prices of a particular variety of pisco and say, whisky.
75. We agree with the European Communities on the question of price differences between pisco and imported distilled spirits. From the evidence before us, it is clear that one cannot speak of a price difference between pisco and all imported distilled spirits. The price differences are only relative, depending on which variety of pisco is being compared with which imported distilled spirit. To assert that all pisco is within a certain price range and all imported spirits in another price range is to go against the weight of evidence. ³⁹⁸ What is important is the effect of relative price movements, since it is relative prices and their changes that influence consumer behaviour in the dynamic situation of changes in the demand for directly competitive or substitutable products. ³⁹⁹
76. It is, therefore, arguable that the retail price levels of imported distilled spirits, such as whisky, could have been influenced by the taxes at issue in this dispute. Despite this, as we have noted, there is some degree of cross-price elasticity between pisco and the imported distilled spirits including whisky (which as we have earlier noted, Chile finds most difficult to envisage being substitutable with pisco).
77. As we noted, earlier cases have stated that the concept of cross-price elasticity is a strong indicator of substitutability. As we also observed above, a low estimated coefficient, as determined in the study submitted by the European Communities and the data from Chile, is not in itself conclusive that substitutability does not exist. We further note that studies that measure elasticity applied to historical data may reveal little about the potential for substitution and competition.
78. In the case before us, we find it significant that the studies indicate some degree of cross-price elasticity, indicating a potential for substitutability. The level of the elasticity may be a function of actual retail price levels, which could be influenced by taxation and other factors such as past measures, including tariffs, as well as higher distribution costs and other factors resulting from lower volumes. ⁴⁰⁰ In our view, therefore, the evidence presented by the data in the various studies and surveys is supportive of the other factors in arriving at a determination that pisco and the imported distilled spirits are directly competitive or substitutable. In this regard, we agree with the panel in Korea � Taxes on Alcoholic Beverages when it stated that:

[T]he question is not of the degree of competitive overlap, but its nature. Is there a competitive relationship and is it direct? It is for this reason, among others, that quantitative studies of cross-price elasticity are relevant, but not exclusive or even decisive in nature. ⁴⁰¹

79. We find, therefore, that the existence of cross-price elasticity between pisco and the imported distilled spirits, although at a low level in the studies, is further indication of a directly competitive and substitutable relationship between the two.

4. Conclusions with respect to "directly competitive or substitutable"

80. Substitutability and competitiveness refer to the ability of products that may be dissimilar in some respects to satisfy a particular consumer want. This definition would suggest that "end-use" is a very important indicator of substitution.
81. In our view, studies or surveys that reveal the following all serve as evidence of substitutability in end-uses:

1. a tendency among consumers to regard products as substitutes in satisfying a particular need;
2. that the nature and content of marketing strategies of producers indicate that they are competing for the expenditure of potential consumers in a particular market segment; and
3. that distribution channels are shared with other goods;

82. In evaluating substitutability in end-use, it may be useful in this regard, to refer to an approach in consumer theory which has been gaining ground. ⁴⁰² According to the theory, goods are, in the eyes of consumers, never really perceived as commodities that are in themselves direct objects of utility; rather, it is the properties or characteristics of the goods from which utility is derived that are the relevant considerations. It is these characteristics or attributes that yield satisfaction and not the goods as such. Goods may share a common characteristic but may have other characteristics that are qualitatively different, or they may have the same characteristics but in quantitatively different combinations. Substitution possibilities arise because of these shared characteristics. The oft-cited hypothetical textbook example of butter and margarine may be instructive. Butter and milk are both dairy products that share important characteristics that margarine does not have. However, butter and margarine each have combinations of characteristics that make them good substitutes as compliments for bread, which is not the case with milk. The characteristics of butter and margarine can be expressed as physical properties such as spreadability, taste, colour and consistency. These physical characteristics combine to render both products good substitutes as bread complements. The latter represents the end-use of the commodities as determined by their combination of characteristics derived from certain physical characteristics.
83. In our view, the same type of reasoning can be applied to the substitutability of pisco and other spirits such as whisky, brandy, cognac, etc. With respect to whisky and pisco, although the two spirits are distilled from different substances, namely barley and grapes respectively, they share the characteristics of being potable liquids with high alcohol content, which is the product of distillation, as well as being receptive to mixing with non-alcoholic beverages. In any event, even the differences in ingredients between whisky and pisco is not sufficient to render these two distilled alcoholic spirits, both of which have a high alcohol content and more or less satisfy a similar need, incapable of being substituted with each other. As for brandy, cognac and some other spirits, we have already noted that the differences in physical characteristics are only post-distillation differences such as colour and smell which are not sufficiently significant to change the basic character of spirits essentially made from grapes or other fruits.
84. We should stress that the complainants did not have to prove that there is a complete overlap in their analysis of substitutability. We take note in this regard, as discussed above, that the negotiators of what became Article III:2, second sentence, considered, inter alia, that apples and oranges were directly competitive or substitutable products. Moreover, we take guidance from the earlier panel findings which found that current market conditions may be distorted by government tax and regulatory policies which tend to freeze consumer preferences in favour of the domestic products. We do not agree that there are pre-tax price differences in the present case of such a level as to establish that the products in question are not even potentially competitive. In addition, after-tax price differences are partly a function of the tax system both in direct impact and indirectly through other factors such as product availability and distribution costs.
85. The complainants have established that, as between pisco and the imported distilled spirits, there is certainly a degree of current competition and the likelihood of greater competition in future. We also note that the production and marketing decisions of the pisco producers also reflect such a realisation. The evidence submitted ⁴⁰³ clearly shows their desire to convey an image of pisco as a drink that competes with the best imported distilled spirits. We believe that when a product is being marketed in ways that suggest that it is in competition with the most up-market imported distilled spirits, this is evidence of at least potential competition with those imports. In addition, evidence of actual or potential competition must be viewed in the context of the fact that consumer habits on consumption of distilled spirits only change gradually over time. ⁴⁰⁴ We are of the view that the totality of the evidence presented supports a finding that the imported distilled spirits, and pisco are directly competitive or substitutable.
86. We also note that our findings are consistent with a recent opinion of the Chilean Central Preventive Commission ⁴⁰⁵ ("the Commission") which, in deciding on a merger between two major pisco producers, stated that:

[T]here are two conditions that must be met for a monopoly to exist:

1. Non-availability of substitute products for pisco; and
2. Existence of barriers to entry preventing new pisco growers from entering the market.

With respect to the first condition, the Commission was satisfied that this was not the position in the Chilean market because:

The broad range of substitute products for pisco available in the liquor segment of the alcoholic beverages market should allow a full substitution in view of the "premium" that consumers are willing to pay for the specific features offered by pisco. ⁴⁰⁶

The Commission also observed that:

[T]he market for alcoholic beverages is the pertinent market for pisco. This market includes beer, wine and liquor. Specifically, pisco participates in the segment formed by liquor, brandies and distilled liquor, in a particular niche covered by appellation regulations. ⁴⁰⁷

Ultimately, the Commission concluded that:

[P]isco faces major competition from other alcoholic beverages, such as wine, beer and whisky, due to the usual practice of drinking pisco diluted with non-alcoholic beverages. Therefore, in the market for alcoholic beverages,��.there are alternative products which consumers of alcoholic beverages can choose to drink (emphasis supplied). ⁴⁰⁸

87. We note that the Commission was dealing with the question of competition from an anti-trust perspective, which generally utilises narrower market definitions than used when analyzing markets pursuant to Article III:2, second sentence. ⁴⁰⁹ Consequently, it seems logical that competitive conditions sufficient for defining an appropriate market with respect to anti-trust analysis would a fortiori suffice for an Article III analysis. We would, therefore, regard the findings of the Commission as tending to confirm the finding that in the Chilean market, pisco and imported distilled spirits are directly competitive or substitutable products.
88. In summary, we are of the view that there is sufficient unrebutted evidence in this case to show both present and potential direct competition between the two categories of products. Accordingly, we find that the evidence concerning physical characteristics, end-uses, channels of distribution and pricing (including cross-price elasticity), ⁴¹⁰ leads us to conclude that the imported and domestic products at issue in this case are directly competitive or substitutable.

To continue with "not similarly taxed"

³⁸⁴ EC Exhibit 22.

³⁸⁵ See EC First Submission, Table 19.

³⁸⁶ Ibid., Table 17.

³⁸⁷ The Possible Effects on the Pisco Industry of a Reduction in the Tax on Whisky, Gemines, August 1995 (the 1995 "Gemines Study"). This study apparently was commissioned by the Chilean pisco industry.

³⁸⁸ See Annex II, Chile First Submission.

³⁸⁹ 1995 Gemines study, p.61, EC Exhibit 20.

³⁹⁰ 1996 Gemines study, as reported in El Diario, 2 July 1996, EC Exhibit 30. This evidence submitted by the European Communities indicating a significant cross-price elasticity is all the information we have about this report. We note that Chile was requested to submit a copy of this report. Chile was unable to do so. According to Chile, the study is the property of the pisco industry which refused to make it available due to alleged flaws in the results as well as confidential business information contained therein. Chile argues that neither it nor its industry is compelled to submit such information. We find the decision of Chile and its industry regrettable. Confidential business information can be protected. If there are alleged flaws in an otherwise relevant study, the party can submit comments in that regard. Chile has done so with respect to the 1995 Gemines study which was also a study commissioned by the Chilean industry. It is true that there is no compulsory discovery process in WTO dispute settlement proceedings. However, the overall dispute settlement process cannot work fairly and efficiently either at the consultation or panel stage if relevant evidence is withheld. In this case, the Chilean pisco industry decided to withhold this evidence. While it is the Chilean government which is party to this case, it would be unrealistic and artificial to argue that the panel should not address the issue based on this distinction given the direct underlying economic interest of the Chilean industry. Thus, Chile did not avail itself of the opportunity to rebut the evidence presented by the European Community. (See also the Panel Report on Indonesia � Certain Measures Affecting the Automobile Industry, adopted on 23 July 1998, WT/DS54, WT/DS55, WT/DS59, WT/DS64, paras. 14.230-14.235. We note that this case involved the failure of a complainant's industry to submit evidence supporting complainant's case, but we agree generally with the point that parties and their industries should not be able to withhold relevant evidence and expect panels to view it favourably.).

³⁹¹ See Chilean Statement at Second Meeting at pp. 12-14.

³⁹² Appellate Body Report on Japan � Taxes on Alcoholic Beverages II, supra., p.25.

³⁹³ Panel Report on Japan � Taxes on Alcoholic Beverages II, supra, para. 6.28, citing Panel Report on Japan � Taxes on Alcoholic Beverages I, supra., para. 5.9.

³⁹⁴ Appellate Body Report on Korea � Taxes on Alcoholic Beverages, supra., para. 123 citing the Panel Report on Korea on Alcoholic Beverages, paras. 10.44, 10.50 and 10.73.

³⁹⁵ The Panel in Japan � Taxes on Alcoholic Beverages II specifically noted that calculated cross-price elasticities in cases such as these will generally underestimate the actual degree of substitutability. Panel Report on Japan � Taxes on Alcoholic Beverages II, paras. 6.28 and 6.31. This was expressly approved by the Appellate Body in that case and reiterated in the Appellate Body decision in Korea � Taxes on Alcoholic Beverages. Appellate Body Report on Korea � Taxes on Alcoholic Beverages, supra., para. 120.

³⁹⁶ See Appellate Body Report in Japan � Taxes on Alcoholic Beverages II, supra., p. 25 and Appellate Body Report on Korea � Taxes on Alcoholic Beverages, supra., paras. 120-123 and 134.

³⁹⁷ Table I, Annex I, Chile First Submission.

³⁹⁸ We also note that there are other distilled spirits imported into Chile in relatively smaller quantities that are often sold at lower prices than many brands of whisky. We also note the EC's argument that small import volumes, due in part to the tax differentials, can lead to a concentration of imports in higher price categories.

³⁹⁹ See Panel Report on Korea � Taxes on Alcoholic Beverages, supra., para. 10.94.

⁴⁰⁰ See Appellate Body Report on Korea � Taxes on Alcoholic Beverages, supra., paras. 122-123.

⁴⁰¹ Panel Report on Korea � Taxes on Alcoholic Beverages, supra., para. 10.44. See also Appellate Body Report on Japan � Taxes on Alcoholic Beverages II, supra., p. 25, and Appellate Body Report on Korea � Taxes on Alcoholic Beverages, supra., paras. 120-123 and 134.

⁴⁰² See Kelvin J. Lancaster, "A New Approach to Consumer Theory", in Journal of Political Economy, Vol. LXXIV (April 1966), pp. 132-157.

⁴⁰³ See EC Exhibits 50-54.

⁴⁰⁴ Some of the imported products have an extensive history in the marketplace. Whisky, for example, has long been present in the Chilean market and subject to high duties and/or taxes. However, we note that the taxes and duties on whisky were increased substantially in the 1980's, which appeared to result in a substantial and sustained loss of market share.

⁴⁰⁵ In its submission dated 23 November 1998, the European Communities provided some commentary on the Commission paper. Subsequently, Chile argued that the EC had not requested nor been granted permission to file such comments. Chile requested permission to file further comments itself or, in the alternative requested the Panel disregard the EC's comments. The Panel decided not to grant Chile's request for further time for commentary, given that the Commission paper was a document of the Chilean government that was issued 11 days prior to the second substantive meeting of the Panel with the parties. Chile, in fact, did make some comments on the Commission paper at that meeting. We agreed to take into further consideration Chile's alternative request and have decided to grant it. The European Communities did not ask for time to make further comments on the Commission paper and chose to submit its comments in a submission that was explicitly limited by the Panel to commentary on other documents. Accordingly, we will disregard the EC's comments on the Commission paper.

⁴⁰⁶ Legal Opinion adopted by the Central Preventive Commission, Merger of Pisco-Manufacturing Cooperative Societies CAPEL and Control, Case Record No. 107-97, pp. 4-5.

⁴⁰⁷ Ibid., p. 2.

⁴⁰⁸ Ibid., p. 8.

⁴⁰⁹ The panel in Korea � Taxes on Alcoholic Beverages was of the view that the definition of a market in anti-trust is not the same as under Article III. The panel felt that because the two concepts (anti-trust and competitive opportunities under Article III) are designed to address different concerns their definitions of the market need not be the same. According to the panel, since Article III is primarily concerned with competitive opportunities it defines markets more broadly than anti-trust which is designed to protect the actual mechanisms of competition. See Panel Report on Korea � Taxes on Alcoholic Beverages, supra., para. 10.81. We agree with that panel's conclusions on the general relationship of the definition of markets in anti-trust cases and under Article III.

⁴¹⁰ The European Communities argued that tariff classification can be important in determining whether products are directly competitive or substitutable. In this case, the European Communities referred to the 4-digit Harmonized System category 2208 as such evidence. Chile responded that 4-digit classifications are too general to be of much analytical use and in some cases contained products which Chile maintained clearly were not substitutable. We note that the Appellate Body and previous panels have referred to the tariff classifications of products in making determinations. However, two issues must be taken into account. First, is the classification sufficiently narrow to be of much probative value? Chile has a valid point in urging caution in this regard. Second, is relying on tariff classifications merely duplicative of examinations made in greater detail for the specific market in question in regard to such factors as physical characteristics and similar end-uses? In our view, with respect to the Chilean market, there is not a great amount of probative value in referring to HS 2208 in light of other evidence available, but we note that the classification is consistent with our conclusions.