4.55 The Community responded that as far as shochu and "spirits" are concerned, Japan had been able to identify only two main differences in physical characteristics: the alcohol content and the packaging. According to the Community, the differences in alcohol content between shochu and "spirits" are not reflected in their respective legal definitions and, therefore, cannot provide a valid justification for applying different tax rates. There is nothing in the Liquor Tax Law preventing the manufacture of vodka of 25 per cent. In practice, some brands of vodka do have an alcohol strength of 25 per cent as illustrated by the case of Juhyo. On the other hand, shochu B may have an alcoholic strength of up to 45 per cent, whilst the maximum alcohol content of shochu A is set at 36 per cent, i.e., only four degrees below the average strength for spirits. High alcohol shochu is by no means a rarity. In 1994, the sales volume of shochu of 35 per cent was larger than the total sales volume of all types of "spirits". The alleged differences in packaging are irrelevant for a like product determination. The physical properties of shochu remain the same irrespective of the size and the material of the packages in which it is sold.

4.56 Japan submitted that such commonality of sales outlets and advertising styles between "spirits" and shochu as pointed out by the Community are also observed among all alcoholic and non-alcoholic beverages and thus fails to demonstrate that products are "like". Though the Community points out similarity in the shochu-based pre-mixes and the pre-mixes made from other liquors, for Japan, such would not be evidence of "likeness" of shochu and "spirits", just as the similarity among tequila-based, wine-based and beer-based "margaritas" in the United States would not render tequila, wine and beer "like". For Japan, "Juhyo Vodka" and "Juhyo Shochu" are two distinct products with different raw materials and different production methods sold under the same established brand name, and are not "like products". Japan also noted that the 1987 Panel Report failed to deliver a clear-cut conclusion on the issue of "likeness" between shochu A and vodka. Although the panel noted that these "could be considered as like products", these products do not appear on the list of pairs of like products in the Report.

4.57 The Community replied that the survey on drinking styles which it had submitted to the Panel confirmed that the end-uses of shochu and "spirits" were almost the same. Shochu was found in all the end-use categories, with only the exception of tonic water. The examples mentioned by Japan merely pointed to the fact that a limited number of specific end-uses are more typical of certain categories of distilled spirits (e.g., shochu and warm water, rum and cola or gin and tonic). Nevertheless, these end-uses are not exclusive of the liquors in question. Vodka, gin and rum are also consumed with warm water, even if in smaller percentages. On the other hand, there are other end-uses which are as common for shochu as for "spirits". For example, while the percentage of drinkers of shochu and soda is 32 per cent, the percentage of consumers of vodka, rum and gin with soda is 32 per cent, 36 per cent and 21 per cent, respectively. Furthermore, there is a substantial coincidence among the most frequent end-uses of shochu and of each of the main categories of "spirits". The Community also argued that the study submitted by Japan on "Alcohol consumption and meals" was flawed because consumers were asked about their preferences and not about how they had actually drunk each category, which led them to favour in their responses the most traditional consumption patterns. Moreover, the evidence was misleading because, according to the explanations provided orally by Japan, the percentages were not percentages of consumers but percentages of replies. The Community noted that, despite these shortcomings, the survey confirmed that both shochu and spirits were consumed before, during and after meals in substantial quantities.

4.58 The United States stated that under the analysis of origin-neutral taxation in the 1992 Malt Beverages and US Auto Taxes panel reports, Japan�s taxes on distilled spirits are inconsistent with Article III:2, because the products are otherwise similar in terms of their physical properties, production method and end-use by consumers, and the distinction made by the Liquor Tax Law between shochu and other distilled spirits has the aim and effect of affording protection to domestic production of shochu. Since the Community and Canada had presented detailed expositions of the evidence of similarity of physical properties, production method and end-use by consumers, the United States would concentrate on the other elements. The United States noted, however, that shochu A is a colourless, odourless product, typically of 25 degrees alcohol content, resembling vodka. According to US exporters, 100 per cent of shochu A is manufactured by adding water to imported ethyl alcohol and redistilling it in Japan. The United States noted that Japan had confirmed that the ratio of imported material in shochu A has been increasing since the 1960s and is over 90 per cent. Thus, shochu A could be characterized as a product composed of ethyl alcohol imported under a duty-free tariff quota for manufacturers, plus water, made primarily by six large firms, protected by a 17.9 per cent applied tariff and benefiting from excise tax discrimination. The ensemble of government measures ensured that the effective protection for production would be maximized. Shochu B is distilled from fermented rice, sweet potatoes, barley or potatoes. ⁵³ The shochu B industry is composed of a large number of small to medium-sized firms and a few of the larger beverage firms. Geographically concentrated in Kyushu, they have mobilized politically to obtain protection in the form of subsidies and tax discrimination. The United States also contested Japan's allegation that "the crucial difference between whisky and shochu in terms of raw materials is the use of malts", since Bourbon whisky and Tennessee whisky are made from corn and contain no malt. As to Japan�s claim that whisky and shochu "differ in the production process of aging in wooden casks", the United States noted that a number of premium brands of shochu B are aged in casks.

4.59 Concerning the second step of the test it suggested for the application of the first sentence of Article III:2, the assessment of discriminatory taxation, the Community submitted evidence according to which the tax rate per litre of shochu B is always lower than the rate on the category of "spirits". The tax rate per litre of shochu A is also lower than the rate per litre of "spirits" for beverages below 36 per cent - 37 per cent. Above that strength, the rate on shochu A is higher. Nevertheless, Article 3.5 of the Liquor Tax Law excludes from the definition of shochu A beverages with an alcohol content of more than 36 per cent. Thus, in practice, the rate on shochu A is always lower than the rate on the category of "spirits". More specifically, the Community argued that the tax discrimination index between shochu B of the most common strength (25 per cent) and "spirits" of the most common strength (40 per cent) is 389 per cent. If the tax rates per litre of pure alcohol, instead of the rates per litre of each beverage, are compared, the tax rate applied to the category of "spirits" is still much higher and the tax discrimination index reached 243 per cent. The Community therefore concluded that the liquors falling within the category of "spirits" and the two sub-categories of shochu being "like products", the Liquor Tax Law violates Article III:2, first sentence, by applying a tax rate to the category of "spirits" which is in excess of the tax rates applied to each of the two sub-categories of shochu.

4.60 The United States argued that since Japan�s tax system applicable to distilled spirits has been devised so as to protect domestic production of shochu (in application of the aim-and-effect test detailed in paragraphs 4.24 to 4.32 above and further supported by an additional factual discussion detailed in Section F below) and because white and brown spirits have similar physical characteristics and end-uses, white and brown spirits are "like products" in the sense of the first sentence of Article III:2, and therefore the difference in tax treatment between shochu and vodka, rum, gin, other white spirits, whisky, brandy and other brown spirits is inconsistent with Article III:2, first sentence.

4.61 Japan called the legal test suggested by the Community a "two-step approach", and disagreed with it. It further argued that even if the "two-step approach" should be adopted, the examination of the second step (discriminatory or not) should be made by the comparison of tax/price ratio between imported "spirits" and domestic shochu. For Japan, the tax/price ratio is the superior yardstick for an examination of the tax burden since it indicates better the impact on consumer choice (and therefore discrimination) than the ratio of tax over volume product or alcohol content. A consumer usually does not buy a product exclusively on the basis of the size of the bottle or on the basis of the alcoholic strength. Consumers choose products by comparing the price and the overall value of a product, which rests upon the taste, flavour and other features and is not confined to the volume and strength. This is why, Japan argued, the tax/price ratio is a better criterion to evaluate the effects of taxes on competitive conditions; and neutrality is achieved when the tax/price ratio is equalized, as is the case with the Japanese tax. Japan submitted that the weighted average of liquor-tax/price ratios for the 20 best-selling brands of domestic shochu A, shochu B, imported vodka, imported rum, and imported gin are 22 per cent, 13 per cent, 18 per cent, 12 per cent, and 18 per cent, respectively. Japan concluded that, even under the "two-step approach", taxes on "spirits" would not be found discriminatory against shochu when an appropriate yardstick is applied.

4.62 The Community argued that the 1987 Panel Report, Article III:2, second sentence, the first paragraph of Article III to which reference is made in the second sentence of Article III:2; and the Note ad Article III:2, all establish a two-step test for the examination of the conformity of a system of internal taxation with Article III:2, second sentence. First, it must be determined whether the taxed imported and domestic products are "directly competitive or substitutable"; and secondly, it must be established whether the taxation is "protective". For the Community, all liquors falling within the categories of "whisky/brandy" and of "liqueurs" and the two sub-categories of "shochu" are "directly competitive or substitutable" among each other. Should any of the liquors falling within the category of "spirits" be found by the Panel not to be a "like product" to shochu, the Community submitted that shochu and "spirits" are, at the very least, "directly competitive or substitutable" products. By applying higher tax rates to the categories of "whisky/brandy", "liqueurs" and "spirits" than to each of the two shochu sub-categories, the Liquor Tax Law affords protection to the domestic production of shochu, thereby violating Article III:2, second sentence.

4.63 For the Community, the concept of "directly competitive or substitutable product" - the first step of the test - is wider than the concept of "like product" and may include products with different physical characteristics but substitutable in terms of uses such as, for instance, skimmed milk powder and vegetable proteins; ⁵⁴ apples and oranges; ⁵⁵ butter and oleomargarine; ⁵⁶ tung oil and linseed oil; ⁵⁷ or natural rubber and synthetic rubber. ⁵⁸ In order to determine whether two products are directly competitive or substitutable, the following criteria may be relevant: the aptitude of the two products to serve the same uses (it is not necessary, however, that the two products are substitutable in respect of all their potential uses); the extent and the form in which the two products are available to the public; the respective prices of the products and the responsiveness of the demand for one of the products to the changes in the price of the other.

4.64 As for the second step of the test it suggested for the second sentence of Article III:2, the Community argued that the following criteria may be relevant in order to determine whether a difference in taxation is "protective" of domestic production. (1) The level of the tax differential, but contrary to the first sentence of Article III:2, a tax difference does not lead automatically to a violation of the second sentence of Article III:2. On the other hand, even small tax differences may be protective. Nonetheless a de minimis differential may, in certain cases, be found not to afford protection. (2) The degree of substitutability and competition between the two products. Logically, the protective effect of a system of taxation increases with the degree of substitutability and competition. (3) Whether the less taxed product is produced in other countries. A system of taxation is protective of the domestic production if the less taxed category is almost exclusively produced in the country imposing the taxes. In contrast, the fact that the more taxed product is produced also in the country applying the internal taxes is irrelevant. The Community also submitted that since Article III:2 protects trade expectations on the competitive relationship between imported and domestic products rather than expectations on trade volumes, it is not necessary, in order to establish a violation of Article III:2, second sentence, to show that the difference in taxation has had an actual effect on the volume of trade.

4.65 Canada's claim was limited to the second sentence of Article III:2. Canada submitted that taxes imposed on Canadian whisky as compared to those imposed on domestically-produced shochu are inconsistent with the provisions of Article III:2, second sentence. For Canada, the Liquor Tax Law that in 1987 was determined to be inconsistent with Article III:2, second sentence, continues, even as amended, to be inconsistent with Article III:2, second sentence.

4.66 For Canada, Article 3.2 of the DSU makes clear that it is the express wording of the WTO Agreement that ultimately defines the rights and obligations of Members and thus, whether the Liquor Tax Law is inconsistent with Article III:2. To the same effect, Canada cited Professor Lauterpacht regarding the role of treaties in defining the rights and obligations of States:

4.67 Canada referred the Panel to the exact wording of Article III:2, second sentence, which provides that "No Member shall otherwise apply internal taxes or other internal charges to imported or domestic products in a manner contrary to the principles set forth in paragraph III:1". The "principles set forth in Article III:1" provide that "Members recognize that internal taxes and other internal charges ... should not be applied to imported or domestic products so as to afford protection to domestic production". Canada recalled that the Note ad Article III, Paragraph 2 to Article III:2, second sentence, was added pursuant to the Havana Conference Report of Sub-Committee A of the Third Committee on Tariff Negotiations, Internal Taxation and Regulation "so that it would be easier for Members to ascertain the precise scope of their obligations under this Article". ⁶⁰ The report stated that "A tax conforming to the requirements of the first sentence of paragraph 2 would be considered to be inconsistent with the provisions of the second sentence only in cases where competition was involved between, on the one hand, the taxed product and, on the other hand, a directly competitive or substitutable product which was not "similarly taxed". Canada argued, therefore, that a clear reading of these provisions establishes that under Article III:2 second sentence, four criteria must be satisfied for Japan's Liquor Tax Law to be found to be inconsistent with this provision:

4.68 The United States submitted that since Japan�s tax system applicable to distilled spirits has been devised so as to protect domestic production of shochu and because all distilled spirits have similar physical characteristics and end-uses, they are "directly competitive and substitutable" in terms of Article III:2, second sentence. Therefore, the United States considered that the difference in taxation between distilled spirits exceeds any de minimis level because that difference materially alters the conditions of competition between domestic and imported products. In the present case, the United States submitted that the change in conditions of competition is illustrated by factors such as the demonstrated effect on consumption choices and the cross-price elasticity of demand discussed further below.

4.69 The United States reiterated that the plain language of Article III:2, first sentence, condemns measures that explicitly target foreign products and accord less favourable treatment. This, in the US view, makes sense because the discriminatory aim of such measures is apparent. However, when a measure is origin-neutral and therefore such an aim cannot be presumed, it does not make sense to say that the purpose of the measure becomes irrelevant. The United States noted that Article III is designed to protect against discrimination, not to create a per se rule of absolute liability for any greater burden or restriction on international trade. All direct and indirect regulation of goods has domestic and international trade-restricting effects, because by its nature regulation imposes burdens. For the United States, the rule proposed by the Community would mean that a government could not adopt any measure, irrespective of its purpose, if the measure had the effect at some point of burdening foreign more than domestic products. Such a "pure effects" test would give no guidance or certainty to legislators or to their legal advisers, because in any situation its application could change from day to day based on international and domestic factors that could not be anticipated at the time a measure is adopted.

4.70 In the view of the United States, the Community had recognized that its "effects" rule would position perfectly desirable, non-discriminatory governmental regulation - including measures maintained by the Community and its member States - under a legal guillotine. To address the obvious overbreadth of its own legal theory, the Community had invented two arbitrary "flexibilities". The United States asked the following questions: What if two automobiles with different engine displacement have identical fuel economy and emissions - why should not these automobiles be "like"? In the US view, this demonstrates that the Community, while claiming to disregard the issue of legitimacy of policy purpose, cannot really do it. The Community simply wants panels to make an ad hoc judgment that the engine displacement criterion used in EC auto tax schemes would remove these taxes from the "guillotine rule" of the first sentence of Article III:2. How many more flexibilities would future panels need to invent to deal with the fact that the EC rule simply overreaches, asked the United States. As for the Community�s analysis of the second sentence, the United States argued that the second sentence does not involve a "straight effects test". For the United States the effect of a legislation may be determined based on various elements, as further detailed in paragraphs 4.24 to 4.32 above and in Section F below.

4.71 Japan suggested, as similarly argued with regard to the first sentence of Article III:2 in paragraphs 4.33 and following, that the consistency of a different treatment of products with Article III:2, second sentence, should be judged in light of paragraph 1 of the Article, in particular the language "not be applied ... so as to afford protection to domestic production", and that whether or not the tax at issue is designed "so as to afford protection to domestic production" should then be judged by the "aim" and "effect" of affording protection. Japan further argued that all the parties to the dispute agree that the second sentence of Article III:2 should be interpreted in the light of whether a measure at issue is applied "so as to afford protection to domestic production", and that the crucial difference between its approach, or the aim-and-effect test approach, and the approach taken by the Community and Canada, or the "two-step" approach with regard to the second sentence of Article III:2, exists in relation to the interpretation of the language "so as to afford protection to domestic production" of paragraph 1 of Article III. The latter approach's criterion is the protective effect alone, while Japan's approach considers not only the effect but also the aim of the measure in question. Japan added that even the Community had come to agree that the analysis, according to the text of the second sentence of Article III:2, should also proceed on the basis of protective purpose and effect. For Japan, the aim of the tax distinction made among categories of distilled liquors by the Liquor Tax Law is neutrality attained through a constant tax/price ratio amongst tax categories. Japan reiterated that the Liquor Tax Law does not have the effect of protecting domestic production of shochu since it does not distort the competitive relationships between imported and domestic products based on the following three cumulative criteria: (1) the neutrality of the tax burden among categories of the legislation under examination, (2) the production of the allegedly protected products outside the imported country and of the allegedly "imported" products in the country, and (3) the absence of directly competitive and substitutable relationship (cross-price elasticity) between the imported and domestically produced products. Japan argued that if there is no difference in tax burden, the system does not distort trade; if there does not exist directly competitive and substitutable relationships, differences in tax burden do not matter; if the products at issue are produced in and out of the country, the tax differentiation should not be construed to afford protection to domestic production; and thus protective distortion can be shown only when all of the three requirements are met. Japan further argued that as one examines the relative tax burden between products in question, the tax burden should be measured by the tax/price ratio, a yardstick which best captures the impact on consumers' behaviour, and in examining whether or not the category in question is almost exclusively domestic, what needs to be examined is not import ratios, but whether or not an allegedly "domestic" category is produced in other countries and whether or not "imported" products in question are also domestically produced.

i) Physical characteristics, end-uses, tariff line and availability to the public

4.72 For the Community, the two categories of shochu and the liquors falling within the categories of "spirits", "whisky/brandy" and "liqueurs" are directly competitive and substitutable since they share the same essential physical characteristics, have similar end-uses, are similarly available to the public and are marketed in a similar way. Furthermore the prices of shochu and of the other distilled spirits and liqueurs are within a close range, once the liquor taxes are deducted. Moreover, there is evidence that, despite the distorting effects on competition of the Liquor Tax Law, the demand for shochu is largely influenced by the fluctuations in the prices of other types of distilled spirits and liqueurs.

4.73 Canada and the Community referred the Panel to the 1987 Panel Report where it was determined that shochu Groups A and B were directly competitive or substitutable with all grades of whisky, including high-quality, high-priced whisky products, for the following reasons:

4.75 The Community argued that the fact that all distilled spirits and liqueurs have the same basic properties and are objectively apt to serve the same end-uses is confirmed by the consumption patterns observed in the Japanese market. In support of its allegation, the Community submitted the results of a research conducted by the Japanese whisky industry on the presence of shochu and whisky in "snack bars". All of them are drunk "straight", "on the rocks" or, more frequently, diluted with water or other non-alcoholic beverages. The drinking styles of the various distilled spirits (including shochu) and liqueurs are virtually the same. Furthermore, the advertising of the different types of spirits and liquors tends to promote the same drinking styles. Both shochu and the other types of distilled spirits and liqueurs are widely drunk across all categories of consumers, regardless of their age, sex and occupation. A very high proportion of shochu consumers are also regular consumers of whisky and other spirits and liqueurs and this proportion is higher than among consumers of alcoholic beverages in general. The same pattern has been observed with respect to the consumption of premium brands of western-style spirits and liqueurs sold at the highest prices. In the last few years, a new market has emerged, especially among young consumers, for bottled or canned pre-mixed drinks combining spirits and soft drinks. As shown by the advertising materials submitted by the Community, the executional style, target market and drinking style of the shochu based pre-mixes and the pre-mixes made from other liquors are identical.

4.76 The Community further argued that shochu and the other types of spirits and liqueurs are directly competitive since they are available in the same trade channels and are promoted and advertised in a similar way. All of them are sold at the same outlets, both for on-premise consumption and for home consumption. Although in the past, there may have been a certain specialization among on-premise outlets, in recent years this specialisation has disappeared. The Community submitted that the results of a recent survey shows that in the Tokyo area 71 per cent of the "snack bars" (a category of outlet where western-style liquors were traditionally predominant and which represents approximately 40 per cent of the total on-premise market) now serve both whisky and shochu. Similarly, an increasing number of "izakayas" (a once "traditional" shochu/sake/beer style of outlet which accounts for approximately 20 per cent of the on-premise market) are now serving whisky and other western-style spirits and liqueurs. Both at on-premise outlets and at outlets selling for home consumption, shochu is positioned and promoted side-by-side with the other types of spirits and liqueurs, thus evidencing that both the retailers and the public regard them as substitutable and competitive products. The level of advertising spending on shochu brands is comparable to the advertising spending on the brands of other spirits and liqueurs. The advertising of shochu and of other spirits and liqueurs is very similar in executional style, is targeted towards the same categories of customers (young consumers and "salarymen") and aims at projecting similar images, regardless of whether the products are of "traditional Japanese origin", like shochu, or western-style. The distribution of advertising spending among the different media is similar for shochu and other spirits and liqueurs, a sign that similar markets are being targeted.

4.77 For Canada, the evidence makes abundantly clear that in Japan, shochu and whisky continue to be directly competitive or substitutable products in that shochu and whisky have many common characteristics and are commonly consumed at similar diluted alcoholic strengths and manufacturers of shochu in Japan capitalize on these common characteristics by marketing some shochu products on the strength of their similarity to whisky and on the basis that they can be consumed in the same manner as whisky. Canada suggested that the processes and raw materials used in the production of shochu and Canadian whisky are very similar. Canada submitted that both Canadian whisky and shochu are produced from a variety of grain sources such as wheat, barley, rye and corn, although a slightly broader range of agricultural raw materials, such as rice, can be used to manufacture shochu. Many varieties of shochu use exactly the same raw grains that are used in whisky production. The enzymes and yeasts used in the production of whisky and shochu are also similar. With respect to production methods, the processes of milling, cooking and conversion are common to both whisky and shochu. The same equipment is used and the fermentation processes are similar. Whisky and shochu are both produced using continuous and pot distillation methods, or through a blend of the two methods. Up to the end of the distillation process, Canadian whisky and shochu can be identical. The main distinction between Canadian whisky and shochu is that Canadian whisky must be aged. This is not so for shochu, although some shochu is currently being aged. ⁶⁶

4.78 Canada also argued that in Japan, both whisky and shochu can be consumed in common styles, i.e., "straight", with water or "on ice", and submitted evidence of current liquor advertising in Japan. As distilled liquors, whisky and shochu are both sold to the public at alcohol strengths significantly above the strengths common for other liquors such as beer and wine. Canadian whisky is sold at the retail level in Japan at an alcohol strength of at least 40 per cent. Shochu can be sold at the retail level up to an alcohol strength of 36 per cent for shochu Group A and 45 per cent for shochu Group B, (although the most common retail strength is 25 per cent for both shochu groups). Distilled liquors such as whisky and shochu are commonly consumed in a diluted style, with the resulting alcohol content being similar for both products. Pre-mixed, i.e., diluted, products sold in Japan contain whisky at an alcoholic strength of between five per cent and eight per cent and contain shochu at an alcoholic strength of between four per cent and six per cent. Advertisements for the sale of shochu refer variously to the similarity in the raw materials used to produce whisky and the advertised shochu, the similarity in the production processes that produce whisky and the advertised shochu product and the similarity in the physical appearance (e.g., coloration) of whisky and the advertised shochu product. In addition, Canada argued that recently, recognizing consumer perceptions that whisky and shochu are alternative choices, the Japan Spirits and Liquor Maker's Association stated that "the drinking patterns of whisky/brandy and shochu [are] becoming alike and [that] they are competing with each other [in the] market".

4.79 Japan argued that "spirits" and "shochu" differ in physical characteristics, end-use, and in tariff lines as is described in paragraph 4.54 above. Japan also argued that whisky/brandy and shochu differ in materials (with malts versus without malts; Bourbon, Tennessee, and Canadian whiskies without malts are classified as "spirits" under the Liquor Tax Law), in the post-distillation processing (aged in wooden casks versus over 99 per cent not aged in wooden casks), in alcoholic strength (around 40 per cent versus 20 to 25 per cent), in colour (0.2 to 0.8 of optical density versus 0.08 of optical density) and in containers (0.7 litre glass bottles versus bulky plastic, glass and paper bottles over 1.8 litres). For Japan, they also differ in end-uses: according to a study in Japan, 60 per cent of shochu consumers drink shochu during meals, while 72 per cent of whisky consumers drink whisky after meals; and according to a study submitted by the Community, only eight per cent of consumers of shochu drink the beverage "on the rocks" while 68 per cent of bourbon whisky consumers do. None of bourbon whisky consumers mix such whisky with hot water or juice, while 42 per cent and 37 per cent of shochu consumers do respectively. They also differ in tariff lines: whisky is classified as "2208.30 whisky" while shochu is classified as "2208.90 Other". Japan also argued that the commonality in availability to the public mentioned by the Community exists only to the extent applicable to all alcoholic and non-alcoholic beverages: the menus and promotion leaflets submitted by the Community list not only whisky(ies) and shochu but also sake, wine, beer, juice, coffee and tea side by side. As to the Canada's comment that pre-mixed shochu and pre-mixed whisky are taxed at the same rate in Japan, Japan argued that the fact that pre-mixed wine and pre-mixed spirits are taxed at the same rate in Canada does not imply a directly competitive or substitutable relationship between wine and spirits, the two products taxed at completely different rates in Canada.

4.80 Japan also noted that the aptitude of the two products to serve the same uses raises the issue of the extent of the sameness. Since the use for quenching the thirst, for example, is common to all beverages, and since the use for enjoying alcohol is common to all alcoholic beverages, the concept of "sameness" should be understood in a narrower sense. According to Japan, the Community argues that sameness in drinking habits between shochu and other distilled liquors is sufficient to meet the criteria. However, Japan�s evidence shows a good degree of divergence in drinking habits not only between shochu and spirits but between shochu and Bourbon whisky as well. The aptitude to serve the same uses does not seem to exist beyond what would apply to all alcoholic beverages.

4.81 The Community argued that Japan's criticisms were unfounded. The study on drinking styles submitted to the panel showed that the end-uses of shochu and bourbon whisky were the same except that shochu is not drunk with tonic water and bourbon is not mixed with warm water or juices. Moreover, the study showed that three out of the five most frequent end-uses of shochu were also found among the five most frequent end-uses of bourbon.

ii) Cross-price elasticity

4.82 Continuing on the issue as to whether shochu and other imported liquors are directly competitive and substitutable, the Community argued that the retail prices of shochu and of the other distilled spirits and liqueurs are within a relatively short range once the liquor taxes and the ad valorem consumption taxes are deducted. This, in the Community's view, confirms that all of them are, at least potentially, competitive in terms of price. The retail prices net of taxes per litre of pure alcohol of most western-style liquors are much lower than the corresponding prices for shochu but both shochu and western-style liquors are frequently diluted with non-alcoholic beverages and drunk at roughly the same strength. Therefore, it may be concluded that, but for the discriminatory taxes imposed pursuant to the Liquor Tax Law, many western-style liquors would be less expensive than shochu in real terms. Price competition between shochu and the other spirits and liqueurs is therefore distorted by the lower taxes applied to shochu. Despite these distortions, there are clear indications that the demand for shochu is largely influenced by the fluctuations in the prices of the other distilled spirits and liqueurs. The Community argued that this is demonstrated, in particular, by the evolution of the sales of shochu and of the other categories of liquors following the drastic changes in prices provoked by the 1989 tax reform: The tax reform of 1989 abolished the sub-categorization of whisky into "premium whisky", "first grade whisky" and "second grade whisky" and introduced a single tax rate for all types of whisky. As a result, the tax rate applicable to those brands which until then had been classified as second grade increased more than threefold and their retail prices almost doubled. No separate market share data are available for second grade whisky. Nevertheless, the impact of the 1989 reform is reflected in the market share of domestically-produced whisky, which includes virtually all brands previously classified as second grade whisky. The market share of domestically-produced whisky fell from 26.7 per cent in 1988 to 19.6 per cent in 1990, i.e., by more than seven percentage points in only two years. This downward trend has persisted after the 1989 tax reform. In 1994 the market share of domestically produced whisky dropped to only 13.2 per cent. In contrast, the 1994 tax reform provided for only a very modest increase in the taxes on shochu. This allowed the manufacturers of shochu to take advantage of the dramatic increase in the price of whisky brands formerly classified as second grade. As a result, sales of shochu, which had been declining during the preceding years, started to grow again in 1990. Since then, sales of shochu have continued to increase in both absolute and relative terms. Thus, in 1994 the market share of shochu reached 74.2 per cent compared to 61.6 per cent in 1989. The 1989 tax reform lowered the taxes applied to "whisky/brandy" (other than second grade), "spirits" and "authentic liqueurs", thus making possible a substantial reduction of the retail prices of these categories. Thereafter, the sales of these categories enjoyed an instant but short-ived increase. In 1992 the Japanese economy entered into a deep recession that made consumers much more price sensitive and provoked a shift of demand towards the less expensive categories of liquors. This development has hit hardest the sales of "whisky/brandy" (of all grades), "spirits" and "authentic liqueurs" which continue to be much more heavily taxed and, for this reason, still retail at higher prices despite substantial further price reductions. As a result, since 1992 sales of these categories have fallen in both absolute and relative terms. In contrast, sales of the considerably less taxed shochu have continued to grow during the same period at the expense of the sales of the more heavily taxed categories.

4.83 In response to the Community's allegation of cross-price elasticity, supported by Canada and the United States' claims, Japan submitted a rebuttal to the Community's arguments on the changes in consumption of whisky and shochu since 1989, the response of consumers to questions asked by Shakai-Chosa Kenkyujo (Institute for Social Studies), and the result of the econometric analysis of national household survey statistics. Japan raised the following points in its rebuttal to the Community's arguments on the changes in the consumption of whisky and shochu since 1989. First, Japan responded to the Community's argument in stating that the decline of whisky consumption and the rising shochu price, a phenomenon since 1992 (Japan referred to the Community's own submission), indicates the lack of responsiveness of demand for one to the price of the other. Second, Japan submitted that when the complainants discuss market shares of product categories in the sales volume of all distilled liquors, they refer to distilled liquor sales as the denominator, which, according to Japan, rests on an a priori assumption of a competitive or substitutable relationship between these distilled categories. ⁶⁷ Third, Japan also argued that some of the evidence submitted by the Community does not demonstrate the cross-price elasticity of demand between shochu and whisky (i.e., the responsiveness of the demand for one product to the price of another), but merely points to the responsiveness of the demand for imported whisky in Japan to changes in its own price. The fact that the demand curve for imported whisky shows a downward slope does not prove it is competing with shochu. Fourth, non-price factors such as the recent consumer tendency toward drinks of less alcoholic strength ought to be considered as well. Fifth, for Japan, the recent decline in whisky consumption seems to result, at least partly, from a pricing policy of the Scotch whisky industry which expanded its profit margin following the liquor tax reduction. While the import of whisky from the United States increased from 7,000 kl to 15,000 kl, and the import from Canada from 1,000 kl to 2,000 kl during 1987 to 1994, the figure for the United Kingdom stayed roughly the same at 23,000 kl to 24,000 kl, after a temporary surge in 1989. The temporary surge in 1989 and the ensuing decline is apparently a result of the liquor tax reduction in 1989 and the increase in the Cost-Insurance-Freight (CIF) prices in 1990. As the export prices increased, the manufacturers� profit margin doubled from 10 pounds sterling per case in 1989 to 20.

4.84 The Community responded that this assertion was unwarranted. The Community argued that it had shown that the fall in sales of former second grade whisky in the wake of the 1989 tax reform was accompanied by a simultaneous increase in the market share of shochu. Likewise, the decrease in sales of whisky (other than former second grade), brandy, spirits and authentic liqueurs after 1992 took place in parallel with an increase in sales of shochu. Moreover, sales of shochu did in fact decrease in absolute terms during 1989 and 1990, in the context of an overall decline in demand for distilled spirits. In relative terms, the market share of shochu increased from 61.2 per cent in 1989 to 63.1 per cent in 1991. During the same period, the share of domestic whisky (which includes virtually all former second grade whisky) fell from 23.4 per cent to 19.6 per cent. In the Community's view, the combinations described by Japan are the result of the deep recession which has affected the Japanese economy since 1992. The recession has made consumers more price sensitive and provoked a shift of demand towards less expensive liquors. Despite substantial reductions in the prices of "whisky/brandy", "spirits" and "authentic liqueur", shochu still retails at lower prices due to the differences in taxation. This has enabled shochu manufacturers to take advantage of the shift in demand notwithstanding a modest price increase.

4.85 Japan also cited the response of the consumers to questions asked by Shakai-Chosa Kenkyujo (Institute for Social Studies). Japan argued that according to the survey, if whisky were not available, 32 per cent of the consumers would choose beer, 32 per cent would choose brandy and only 10 per cent would choose shochu. If shochu were not available, 35 per cent would choose beer, 30 per cent would choose sake and only six per cent and four per cent, respectively, would choose "spirits" (e.g., gin, rum, vodka) and Scotch whisky. Japan also submitted the result of the econometric analysis of national household survey statistics. The study applied the statistical method used in the Bossard study commissioned by the European Commission in 1994, ⁶⁸ to Japanese consumption data for the past 20 years, based on household surveys by the Bureau of Statistics of the Japanese Management and Coordination Agency. Using prices of shochu, whisky, beer, wine and sake, the household consumption expenditures, and the trend factor as seven explanatory variables, 16 equations were developed in order to explain the shochu consumption and the whisky consumption, respectively. The result was striking. Neither the impact of the whisky price on the shochu consumption nor the impact of the shochu price on the whisky consumption was proven significant: the whisky price cannot logically explain the shochu demand; equations containing this variable lead to either the relationship that the higher is the shochu price, the larger is the shochu consumption or that of the higher is the whisky price, the lower is the shochu consumption. In contrast, prices of shochu and beer explain the shochu demand in a highly significant manner. Whisky consumption equations containing the shochu price as an explanatory variable either lead to the non-logical relationship that the higher the shochu price, the lower the whisky consumption, or result in a lower absolute value of t-statistics, or the lack of significance. In other words, shochu and whisky are not competing with each other in the Japanese market. Thus, the distinction between shochu and whisky in Japan should be less distortional to the market than the distinction between beer, wine and distilled liquors in the European markets.

4.86 The United States criticised the Japanese econometric study. The first point related to the conclusions of the Japanese study. Using the volume of shochu consumed as the dependent variable, the Japanese model found that shochu�s own price elasticity is positive and the cross-price elasticity with whisky is negative; it suggested that when the price of shochu goes up the volume consumed proportionately increases by two to three times as much, and when the price of whisky increases the consumption of shochu decreases. The t-statistics corresponding to the model�s estimated coefficients were greater than 2 or less than -2, meaning that the model as specified attaches a high degree of statistical significance to these results. Yet a finding that consumers react to an increase in the price of a product by increasing consumption is contrary to one of the fundamental tenets of microeconomic theory -- a downward sloping demand curve. No credence could be given to a model that states, with a high degree of statistical robustness, that demand for a good increases because its price increases; at a minimum such a counter-intuitive result casts serious doubts on the validity of the regression procedure. Secondly, looking further at the regression results in the Japanese study, the United States noted that although the model generally produced elasticities of the expected sign (i.e., negative own elasticity and positive cross-elasticity), the low t-statistics for the cross-elasticity estimates indicated that the variables, as specified, do not significantly explain movements in the consumption of whisky. The United States concluded that in no way does the model support the conclusions cited by Japan. To the contrary, in the US view, the results of the regression analysis only showed that the underlying model is mis-specified and the methodology is flawed. Thirdly, the United States pointed out the Japanese model�s failure to correct for basic problems in estimation of time-series models, such as serial correlation and auto-correlation. Because serial correlation biases the standard error of the regression, a naive analysis will draw two erroneous conclusions: (1) the conclusion that the parameter estimates are more precise than they actually are -- the t-statistics will be higher than they should be in a correctly specified model, and (2) a high R² statistic that gives an overly optimistic picture of the success of the regression model in estimating relationships between variables. Japan had failed to use well-recognized techniques for correcting these problems such as "first-differencing" (using the change in a variable from the preceding period) or the Cochrane-Orcutt estimation procedure (used in the Bossard study referred to by Japan). In a correctly specified model, a trend variable would not be needed because the effect of increased consumption due to increased income would be picked up by the coefficient of the consumption function variable. An additional benefit of specifying the model in this form is that the coefficients of the price variables would be the actual own- and cross-price elasticities. Finally, the United States raised concerns with the Japanese data itself. The model utilized annual data since 1974. The implicit assumption that no other factors have affected the consumption of shochu or whisky (such as a change in consumer tastes) was unwarranted. For the United States, it would be preferable to use monthly or quarterly data over a shorter time period. Finally, Japan had failed to use real (i.e., inflation-adjusted) prices to avoid an inflationary bias. The United States concluded that the results of the regression study cited by Japan do not support the Japanese Government�s contention that whisky and shochu are not directly competitive or substitutable products. The methodology used in the study was flawed, differing from both the Bossard study and accepted econometric practice. The model should be re-estimated as outlined above, preferably utilizing more detailed consumption and pricing data.

4.87 Japan responded that this first criticism is equal to saying that it is improbable that shochu consumption is not explained by prices of shochu and whisky but by prices of shochu and beer. Similarly, in response to the Community's criticism that it was improbable that many regressions did not yield a valid result whilst one variable less or more suddenly produced very strong results in one or two regressions, Japan argued that if the beer price is a real factor and the whisky price is a mere noise, "one variable less or more" should totally alter the result. According to Japan, what the Community and the United States are requesting is the result "any hypothesis can explain shochu consumption", a result which is truly improbable. Japan submitted that it is a basic econometrics principle that the level of significance depends on the combination of explanatory variables. In the present analysis, the level of significance attained by an identical set of explanatory variables did not fluctuate widely depending on methods of conversion or estimation. For Japan, the aim of an econometric analysis is to examine the validity of hypotheses. If a hypothesis led to a finding contrary to accepted economic theory under any of the standard models and methods -- linear, log-inverse, log-log, Cochrane-Orcutt, maximum likelihood -- and another led to a meaningful result, what should be rejected is not the models but the former hypothesis. Japan explained that it rejected the hypothesis that "the consumption of shochu is affected by the shochu price and the whisky price" because it produced a result contrary to accepted economic theory. It chose instead the "shochu price and the beer price" hypothesis, because the result accorded with general economic principles. In response to the US criticism that the Japanese study failed to follow the basic point of econometrics technique by failing to use log-linear transformation, which the Bossard study used, Japan noted that there is no reference in the Bossard study to the use of log-linear models, although its use of the linear models, log-log models and log-inverse models is noted. The Japanese study also used linear models, log-log models and log-inverse models, but did not use log-linear models. This is because log-linear models are regarded as more appropriate at the initial stage of introduction of new products into a given market, and are not effective for the analysis of established products such as liquor. Contrary to the US criticism that the Bossard study used the Cochrane-Orcutt method of estimation to correct for autocorrelation and that the Japanese study did not, Japan responded that the Japanese study used not only the ordinary least squares but the Cochrane-Orcutt method as well. Additionally, the maximum likelihood method, which is believed to be a better corrector for autocorrelation, was used. In response to the US criticism against Japan's use of nominal prices, Japan submitted that the Bossard study used nominal price indices. Moreover, Japan submitted that reliability of conversion of nominal prices into real prices by the use of the Consumer Price index (CPI), for example, hinges on whether or not a change in the CPI affects consumption to the same extent as a change in the nominal price of the product at issue. If consumers tend to be influenced more heavily by a change in the price of the product than a change in the price level in general, nominal price indices are more reliable. Moreover, introduction of the CPI could add noise to the analysis because of the issues of weight given to product categories or the reference year. Since prices were generally stable during the years used for the study, it is more appropriate to explain consumer behaviour on the basis of nominal prices. Japan additionally submitted an analysis using real (deflated) prices, which, according to Japan, led to similar results as the analysis using nominal prices. Japan also noted that according to the analysis submitted by the United States, which utilized real (deflated) prices and the technique of "first-differentiating", "the annual price indices of whisky and shochu and annual household expenditures do not account for movements in the quantity consumed of these products". Japan argued that this result also supports Japan's rejection of the hypothesis that the prices of shochu and whisky affect the shochu consumption. Japan thus concluded that the criticism raised by the United States was off the mark.

TO CONTINUE WITH JAPAN - TAXES ON ALCOHOLIC BEVERAGES

⁵³ The shochu sample submitted by the United States is a barley-based shochu B and the sample from the EC is a sweet-potato-based shochu B.

⁵⁴ Panel report op. cit note 16 (EEC - Measures on Animal Feed Proteins), adopted on 14 March 1978, BISD 25S/49, para. 4.3.

⁵⁵ EPCT/A/PV/9, p.7.

⁵⁶ Reports of Committees and Principal Subcommittees, UN Conference on Trade and Employment, 1948, p. 61.

⁵⁷ E/CONF.2/C.3/SR.11, p. 1 and Corr.2.

⁵⁸ E/CONF.2/C.3/SR.11, p. 3.

⁵⁹ Lauterpacht, International Law: Collected Papers, 86-87 (1970).

⁶⁰ Havana Reports, at p. 61, para. 36.

⁶¹ Ibid, at p. 41, para. 7.

⁶² Panel report adopted on 7 November 1989, BISD 36S/345, para. 5.10.

⁶³ Panel report adopted on 2 December 1970, BISD 18S/97, para. 9.

⁶⁴ 1987 Panel Report, para. 5.7.

⁶⁵ Thus, in May 1988 (i.e., shortly after the adoption of the 1987 Panel Report), the Japanese manufacturer Takara started marketing "Jun Legend", a light amber coloured brand of shochu produced by blending two types of alcohol distilled from barley and corn and maturing them in charred white oak barrels for one to five years. According to Takara, "the most noticeable characteristic of this brand is a flavour and taste similar to whisky". When the new brand was launched, Takara announced its expectations that the new product would appeal to former consumers of second grade whisky which, as a result of the 1987 Panel Report, was expected to become subject to much higher tax rates as from 1989.

⁶⁶ In support of its allegation Canada submitted a letter from Hiram Walker & Sons Ltd which confirmed: "The process of milling, cooking and conversion are common to both spirits. The same equipment, grains, and enzymes are used in the production of both shochu and Canadian whisky. The details of these processes will vary by distiller and formula but the basic process is the same for both types of spirits. The same similarities hold true for the fermentation process".

⁶⁷ Japan submitted an example: For the sake of illustration, let us assume that there is no competitive relationship between products A and B. If you introduce a concept P which is the sum of the sales of the two products and calculate the shares of the products in P, you can readily create an impression that A and B are competing with each other. For example, if the price of A increase and its sales decrease, the market share of B in P will increase even if its sales column is not affected at all. Then you have a result which ostensibly indicates the responsiveness in B�s share to an increase in A�s price. Japan concluded that this kind of argument is, to say the least, inappropriate".

⁶⁸ On the question of the cross-price elasticity of demand between alcoholic beverages, Japan had referred the Panel to a statistical analysis made by Bossard Consultants, Competition Between Difference Categories of Alcoholic Drinks (1994) which was commissioned by the European Commission in 1994. This study by Bossard Consultants resulted in a series of findings for the European markets: 1) When the price of wine rises 1%, the consumption of distilled liquors will increase 1.4%, 0.55-0.9% and 0.4%, respectively, in Spain, the United Kingdom and West Germany. 2) One percentage rise in the beer price will lead to the expansion of the distilled liquors by 1.3%, 1.2% and 0.9% respectively, in West Germany, the Netherlands and Denmark.