4.143 Japan also submitted to the Panel charts which identified the "roughly equal" tax/price ratios, namely a chart showing the Average Retail Prices and Taxes of Liquor and another chart on the Percentage of Taxes in Retail Prices (see Annex VI). According to Japan, the Liquor Tax Law's tax rates are set corresponding to the average net-of-tax price of each category. While net-of-tax prices vary from one category to another, the ratio of the tax over the retail price stays roughly constant between categories. Japan submitted that the ratio of the tax burden over the retail price is roughly the same; as measured in December 1995 vis-à-vis average suggested retail prices, at around 20 per cent. A consumer of any category paid, on average, roughly 20 per cent of the price for the tax, in other words. One category which shows a substantially lower ratio is imported authentic liqueur. Japan argued that this results from the 1989 amendment's integration of authentic liqueur and other liqueur and the application of the lower tax rate. If the pre-1989 rate were applied, the ratio would be roughly in line with the other categories. The second figure uses 20 individual prices which should differ from the weighted average price. Nevertheless, the tax/price ratio of each brand is distributed within a similar range for all categories. If these categories were subject to the same tax rate per alcoholic content as currently applied to whisky, any shochu would bear a heavier tax burden than any imported whisky. Japan referred to the example of the retail prices and taxes of liquor in the UK, as an indication of the result of the abolition of tax categories. In support of its allegation, Japan submitted two charts showing the "Retail Prices and Taxes of Liquors in the UK" and another chart showing the "Percentage of Taxes in Retail Prices in the UK" (see Annex VII). Japan argued that, according to the example, the price of a bottle of whisky which originally cost 20 pounds would increase only 50 per cent to 30 pounds, while that of a bottle of gin would increase three times, from 4 pounds to 12 pounds. For Japan, this tax distorts consumers' choice in favour of whisky, to the detriment of gin. It also runs counter to the notion of equity, because gin consumers must pay two thirds of the consumption expense to the government while whisky consumers are taxed only on third of the expense. For Japan, these are the problems which Japan's liquor tax categories and different tax rates are designed to avoid. To the Community's allegation that the examples are not representative, Japan responded that, even in the promotional leaflet of a supermarket in Brussels which the Community itself submitted to the Panel, any bottle of vodka and any bottle of gin is burdened with a higher tax/price ratio than any bottle of brandy in the shop. Japan also suggested that with the recent expansion of discount operations, actual market prices are, in some cases, substantially different from the suggested prices. However, in the discount outlets, all categories, including shochu, are subject to discount, and the relationship between tax/price ratios of different categories tends to remain the same. For example, for the items listed on the promotional leaflet of a discount outlet sake Ichitba Yamada, which the Community submitted to the Panel, the ratio is 21 per cent for whisky/brandy, and 26 per cent for shochu on the basis of suggested retail prices. At discount prices, it is 35 per cent and 40 per cent, respectively, for whisky/brandy and shochu, resulting in much the same picture. Japan submitted that as an indicator of the overall market situation in Japan, the suggested retail price seems to work better. This is because, while discounts are frequent in the Tokyo metropolitan area, the actual prices in other parts of Japan tend to be close to the suggested prices. In Japan's view, the JETRO survey cited by the Canadian submission confirms this: Old Parr (12 years; suggested retail price of ¥8,500), for example, is selling at ¥4,920 on the average in three outlets in Tokyo; its price in Nagoya and Fukuoka (the fourth and eighth largest cities in Japan) is ¥8,000. In support of its arguments, Japan also submitted its own survey of actual on-shelf prices of all of the relevant items (in aggregate, 5,130 items) displayed at 36 outlets in six cities in Japan.

4.144 Japan also submitted that in the complainants' assessment of the tax burden under the Liquor Tax Law, their criteria of selecting products were not objective. For instance, Japan continued, it is not meaningful to compare a brand of whisky whose tax/price ratio is the highest, and a brand of shochu whose tax/price ratio is the lowest. The complainants' analysis did not adequately capture discounts applied to shochu.

4.145 In view of the clearly protectionist effects attached to the application of lower tax rates to shochu, the Community argued that the aim pursued by Japan is irrelevant for this Panel to rule on the consistency of the measures concerned with Article III:2, first and second sentences. As argued, for the Community, the examination of the Liquor Tax Law's purpose would only become relevant if Japan had claimed that the infringement of Article III:2 is justified under Article XX. In any case, for the Community, the existence of a protectionist purpose is nonetheless manifest. The existence of a protectionist intent results, in the first place, from the apparent arbitrariness and lack of rationality of the product categorization, of which the following are but some of the most egregious examples:

4.147 The Community submitted that it is a commonly held view in Japan that the concessionary tax rates applied to shochu are specifically aimed at protecting the small producers of shochu, even though over 50 per cent of the shochu sold in Japan is now produced by six major companies. Thus, the Japanese media often reflect the view that the lower taxes applied to shochu respond to the pressure exerted by the small manufacturers of shochu, which tend to be concentrated in a few prefectures and enjoy for this reason a disproportionate political influence. A recent confirmation of this was provided by the vigorous opposition of some Japanese legislators to the very modest increase in the tax on shochu proposed by the Government as part of the 1994 tax reform bill. According to a member of the Diet, that increase would have been "a blow to the more than 30% of shochu makers who show a deficit". In order to appease this criticism, the Japanese Government was forced to admit that the tax increases had been kept "to the minimum level possible". Contrary to the Japanese Government's claim that"the only public policy objective pursued by applying lower tax rates to shochu is to maximize the tax revenue while ensuring that the tax burden is distributed among the different categories of consumers in accordance with their respective tax bearing ability, the amount of the tax is unrelated to the actual price of the liquors: a liquor falling within a certain category and having a certain alcoholic content is always taxed at the same rate, regardless of its sales price. Instead, the Japanese authorities make the a priori assumption that certain categories of liquors are more expensive than others and are consumed by more affluent consumers. As a result, it is the amount of the tax that the Japanese authorities arbitrarily decide to impose on each category of liquors which determines its sales price and ultimately its consumption pattern, and not the opposite.

4.148 Against the Community's allegation of political influences based on media articles, Japan referred the panel to a newspaper article conveying a contrary view and submitted that speculative inside stories were not appropriate as the basis of Panel findings. For Japan, the record of Diet deliberation of the 1994 amendment cannot be evidence of a protective intent, as the amendment raised the tax rate on shochu A by 30 per cent and that for shochu B by 44 per cent, while raising the tax on "spirits" by mere 11 per cent and maintaining the tax on whisky/brandy at the same level. Japan also argued that the complaining parties seem to confuse the present Japanese policy with that of 1987. Japan stressed that its current distilled liquor taxation prioritizes neutrality and horizontal equity over vertical equity, and that Japan is not saying that a pursuit of vertical equity can justify a non-trade neutral tax. For Japan, what it is saying is that in so far as a tax system satisfies the trade-neutral requirement, the less regressive it is, the better. In explaining the notion of vertical equity, Japan argued that tax inequity among consumers of different liquors can have a distributional impact when different income groups prefer different categories of liquor. Japan submitted evidence that compares the income, the consumption tax payments, the liquor tax burden in respect of distilled liquor under the current Liquor Tax Law and the hypothetical liquor tax burden under a flat specific rate, of an average household in income quintiles. Under the current Liquor Tax Law, the consumption tax burden or the present liquor tax burden track roughly the income level and are fairly proportional. However, the tax becomes highly regressive under the flat-rate regime.

4.149 For Japan, one of the possible factors contributing to the introduction of the flat-rate tax in Europe and North America seems to be the importance in the tax policy attached to the prevention of alcoholic dependence. Community documents prepared in the course of drafting the Community Council Directives of 1992 indicate that the impact on health has been an important factor. Similarly, the local liquor tax laws of the United States were introduced in the wake of the Prohibition, and some of them earmark the tax revenue specifically for fighting alcoholism. Another possible factor is the high ad valorem consumption tax. The proportionality of the ad valorem consumption tax (e.g., VAT, the state sales tax) could mitigate the overall regressiveness of taxes on liquor. For example in the Community member States’ VAT rates range from 15 per cent to 25 per cent. The Canadian federal government imposes a VAT of 7 per cent, and Ontario levies a 12 per cent sales tax. The rate of State and local sales tax in New York is 8.25 per cent. These taxes serve to ensure equity, enabling the liquor tax to focus on alcohol. These factors are not present in Japan. At the rate of 3 per cent, the ad valorem consumption tax alone cannot ensure the equitable distribution of tax burden among liquor consumers under a flat rate liquor tax system. Japan submitted that the National Diet demands annually an estimate of the ratio of the liquor tax burden relative to the income level for each income group. Japan argued that it was against this background of acute concern over the issue of distributional equity that the Liquor Tax Law incorporated an element of distributional equity in its structure. Different specific rates apply to a variety of products, depending on the degree of value-added. However, Japan argued, the present policy pursues the goal of distributional equity only to the extent compatible with neutrality. For Japan, while the pre-1989 policy assumed that a consumer of imported whisky had a greater tax-bearing ability, and therefore should bear a greater tax burden, Japan's taxation of distilled liquors made a clear departure from that policy with the 1989 amendment following the 1987 Panel Report. Contrary to the claims by other parties to the dispute, the present policy requires an equal burden to be shared among consumers of imported whisky and shochu. For Japan, distortional effects of a tax on consumer choice are most pronounced in the case of flat-rate specific taxes, as currently applied in some countries of Europe and North America. For example, different categories of liquors are produced by different manufacturing processes; some are stored in wooden casks for many years, and others are consumed immediately after distillation. Their value or net-of-tax price differs accordingly. However, the flat-rate tax imposes the same amount of tax across various categories of distilled liquors if the amount of alcohol contained is the same, regardless of product differences. Under the flat-rate taxation, where the amount of tax is the same for all categories of distilled spirits, a bottle of gin the value of which is 4 pounds sterling will cost 12 pounds to purchase after tax. In contrast, a bottle of whisky the value of which is 20 pounds sterling will be priced at 30 pounds. In other words, the tax makes the price of gin three times higher than its value, while making the whisky price only 50 per cent higher. This kind of taxation prejudices consumers' choices against gin in favour of whisky because consumers choose products by comparing their price and their overall value.

4.150 Japan further argued that in order to ensure neutrality to consumers' choice, a liquor tax should not be based exclusively on one element of the overall value: alcoholic strength. The Liquor Tax Law, on the other hand, captures other elements of value by classifying liquors on the basis of the difference in value added by post-distillation processing. Indeed, this results in a broadly constant value/price ratio across categories (80 per cent, as measured in December 1995 vis-à-vis average suggested retail prices) and a tax/price ratio (20 per cent as measured in the same manner). For example, prices of a bottle of whisky (¥3,000) and of a bottle of shochu (¥1,000) increase by a similar percentage after tax. Categorization of distilled liquors under the Liquor Tax Law thus serves the goal of neutrality, a legitimate tax policy objective. In fact, this role of varied specific rates is not peculiar to Japan. Eight countries out of 15 Member States of the Community apply a tax rate to sparkling wine several times higher than the one levied on still wine. Japan also submitted that the US tax on sparkling wine per litre of alcohol is 3.2 times higher than that on still wine, compared with the 3.9-to-1 tax differential per litre of alcohol between whisky and shochu A in Japan. Also, the import ratio of still wine in the United States is lower than that of sparkling wine, as is the import ratio of shochu, which is lower than that for whisky. Moreover, in Japan, wine is wine and the same tax rate is applied to both sparkling and still wines, whereas in the United States, alcohol is alcohol and the same tax rate is applied to all distilled liquors.

4.151 The Community, Canada and the United States reiterated that the taxation systems of the countries other than Japan were outside the terms of reference of the present Panel.

4.152 The Community stated that although Japan argued that "equity" requires a "fair distribution of the tax burden among consumers", it did not provide any clear explanation of what is considered as a "fair distribution" of the tax burden. In the Community's view, Japan argues that "equity" means that tax rates on distilled spirits should be progressive or at least proportional to the consumers' income level: "Tax inequity among consumers of different categories can have a distributional impact when different income groups prefer different categories of liquors". The Community submitted that different income groups do not have fixed and inherently different tastes. Consumers with lower incomes drink less than consumers with higher incomes simply because whisky is more expensive. In turn, it is more expensive because it bears higher taxes. Thus, through the application of higher tax rates on the categories of liquors deemed a priori to be preferred by the rich, a groundless assumption is turned into a self-fulfilling prophecy. The assumption that certain income groups prefer certain liquors was also at the heart of the "tax bearing ability" principle invoked by Japan as a justification before 1987. This justification was rejected by the 1987 Panel Report. "Equity" is but another tag name for "tax bearing ability" and should be condemned on identical grounds. The Community concluded by stating that the above considerations led the 1987 Panel Report to dismiss in categorical terms a similar justification advanced by the Japanese Government:

4.153 Japan reiterated that the "effect" of "so as to afford protection" must be judged by whether the tax distorts the competitive relationship between imported and domestic products. For Japan, the Liquor Tax Law does not distort the competitive relation between imported and domestic products for the following reasons. 1) The tax/price ratios of all tax categories are roughly the same. In terms of the examination of the tax burden, the tax/price ratio is the superior yardstick and better indicates the impact on consumers' choice than the ratio of tax over product volume or alcohol content and it is common practice to employ a tax/price ratio in comparing the burden of an excise tax. 2) Shochu is produced outside Japan; indeed in examining whether or not the category in question is almost exclusively domestic, what needs to be examined is not import ratios but rather whether the "domestic" product is produced in other countries, and whether the "imported product" is also domestically produced. 3) There is no directly competitive or substitutable relationship between domestic products and imports, precluding therefore any possibility of protective effects. Since in Japan's view, protective distortion exists only when the three above-mentioned cumulative requirements are met, it is clear that the Liquor Tax Law does not distort any competitive relationship and is therefore consistent with Article III:2.

4.154 For Japan the lack of protective effect of the Law is demonstrated by the neutrality of the tax. The application of varying rates dependent on average net-of-tax prices of distilled liquor categories results in a fairly stable tax/price ratio. Thus the liquor tax is far less distortional than taxes of other countries, and far less likely to alter consumer choice.

4.155 The Community responded more specifically that the Liquor Tax Law is not, de facto, "neutral" even according to the standard of "neutrality" defined by Japan. The tax/price ratios, as calculated by Japan, are far from being "roughly constant". Moreover, the calculation method followed by Japan leads to the systematic overestimation of the ratios for shochu and the underestimation of the ratios for the other categories. Even if the tax/retail prices were exactly the same (quod non), the Japanese tax system would still not be truly neutral. The specific tax rates on a litre of whisky/brandy, spirits and liqueurs are from three to nine times higher than the specific tax rates on a litre of shochu. For example, the specific tax rate on a 0.75 litre bottle of whisky of 40 per cent is ¥737, i.e., more than the weighted average retail price net of taxes for an equivalent bottle of shochu A which, according to Japan's own calculations, is ¥631. As a result, whisky is bound to be much more expensive than shochu, regardless of whether its pre-tax price is lower or higher. The considerably much higher taxes imposed on "whisky/brandy", "spirits" and "liqueurs" have excluded the less expensive brands of these categories from the lower segments of the Japanese market for distilled spirits where price is the main competition factor. On the other hand, these brands cannot compete in the upper segments where quality and brand image are as important as price. Confirmation of this is provided by the spectacular decline in the sales of former second grade whisky in the wake of the 1989 tax reform. The distorting effects of the Liquor Tax Law are also evident in the composition of Community exports. Premium brands of scotch such as Chivas or Johnnie Walker Black Label represent a disproportionate share of imports. Likewise, imports of Community brandy consist almost exclusively of very expensive brands of X.O. and V.S.O.P cognac. For the Community, the exclusion of the less expensive brands of "whisky/brandy", "spirits" and "liqueurs" from the Japanese market has the consequence that the overall tax/price ratio for these categories is much lower than it would be if the possibility to sell these brands was not precluded through dissuasive taxation. For this reason, even if the tax/price ratios for shochu and the other categories of distilled spirits were "roughly constant", this would not be proof that the Japanese tax system is "neutral". Rather, it would be the consequence of its lack of neutrality.

4.156 In response to the Community's claim that differences in net-of-tax prices among categories are the results of tax differences, Japan raised the following points. First, since 1987, the tax rate on whisky has been halved, but the share of premium whisky in the imported Scotch whisky market expanded from 33 per cent in 1987 to 51 per cent in 1994. Japan argued that, contrary to the allegation of the Community, the decrease of tax promoted premium whisky sales. Also, during the same period, in which the shochu tax rate was increased twofold, containers of shochu A shifted from a medium size to a large size. Again, contrary to the Community's allegation, the higher tax has led to an increase in the share of low-cost brands. The Japanese market configuration shifted in a direction opposite to the Community's hypothesis. Second, according to Japan, differentials in net-of-tax prices between categories similar to those in the Japanese market do exist in countries which apply a flat-rate tax on distilled liquors as well. For example, calculation of the average net-of-tax prices of all items appearing in the leaflet of a supermarket in Brussels, which the Community submitted to the Panel, results in significant price differentials among the categories of distilled liquors. Also, according to a survey done by Business International, Inc., an affiliate of The Economist magazine, brandy is more expensive than whisky and whisky is more expensive than gin in 32 cities of the world where a flat tax rate is applied. For Japan, net of tax prices of distilled liquor differ from one category to another, reflecting differences in such factors as materials, production methods and alcoholic strength, as agreed by a delegate from the Community during the Panel deliberations. Japan argued that the price differences exist under a flat rate excise tax as well as under multiple rates, and that the tax cannot be responsible for the universal phenomenon.

4.157 The Community submitted that for the purposes of applying Article III:2, taxes must be compared on the same basis as they are levied. Accordingly, in the present case the only relevant comparison is the comparison of rates per volume of beverage. A comparison of rates per volume of alcohol might also be relevant if the specific taxes, though assessed on the volume of beverage, were aimed at taxing the alcohol content. However, Japan has not claimed that the tax rates are proportional to the alcohol content. According to Japan, alcoholic strength is only one of the criteria for classifying distilled spirits under the Liquor Tax Law. However, alcoholic content is not even mentioned in the legal definitions of categories other than shochu. Moreover, a comparison of tax/price ratios, as suggested by Japan, is irrelevant. An accurate comparison of tax price/ratios is, in practice, unfeasible. Unlike tax/volume ratios or ad valorem tax rates, tax/price ratios are not transparent. They are not known in advance to the producers, the consumers or even the Government. They can only be estimated retrospectively on the basis of necessarily selective price data which can be easily manipulated. Furthermore, tax/price ratios are subject to constant change due to variations in prices. The Community argued that an adjustment a posteriori of the tax rates to take into account the variations in tax/price ratios, even if carried out on a regular and consistent basis, would not render the Liquor Tax Law neutral. If a higher rate is imposed on a certain category, the less expensive products within that category will be excluded from the market. This will drive down the ratios for the entire category, thus creating the false impression that an upward adjustment, instead of a downward adjustment, is warranted. In any event, there is no legal provision requiring the Japanese authorities to readjust the rates to take into account changes in prices. The Japanese Government is not even required to survey periodically the evolution of the ratios. The periodical readjustment of rates has merely been envisaged by two non-binding recommendations of the Tax Commission, an advisory body, the first of which dates from 1993 only. In practice, the readjustment of rates has been carried out sporadically and in response to foreign pressure. Thus, in order to keep pace with price changes and preserve the neutrality of the system, it would be necessary to readjust the tax rates continuously. This would not only deprive international trade of necessary stability but, in addition, would risk turning the application of the Liquor Tax Law into a matter subject for permanent review by WTO panels. The Community submitted further that the argument that the Liquor Tax Law is "neutral" in terms of tax/price ratios is not new. It was put forward by Japan before the 1987 Panel Report in order to justify the differences in taxation between, on the one hand, special grade whisky and, on the other hand, first and second grade whisky. This argument was rightly disregarded by the 1987 Panel Report which only took into account the differences between the rates per volume.

4.158 In response to a question how one can be assured that the tax/price ratios would be roughly constant across the categories in the future as well, Japan argued that one can be assured of it since i) equalization of tax/price ratios across the categories of distilled liquors is firmly rooted in the fundamental principles and philosophy of Japan's tax policy; ii) the Tax Commission in its reports repeatedly requested the Prime Minister to watch tax/price ratios as a continuous exercise, and iii) the division in charge carefully monitors the price developments. Japan argued further that it is a common practice to employ the tax/price ratio in comparing the burden of an excise tax: the British Chancellor of the Exchequer, Mr. Clark, based his tax rate decisions on the "share of the tax in costs" in his budget address; the European Commission's "Excise Duty Rate Tables" - compares the "tax burden in the retail price"; the US Congressional Budget Office report on excise taxes (August 1990) includes a section titled "Federal Tax Rates in Relation to Product Prices"; the Community documents prepared in the course of drafting the liquor tax directive state that a higher tax is necessary for sparkling wine, which is more expensive than still wine, in order not to distort competitive conditions between the two products; and a cartoon indicating the tax ratio in the retail price plays a prominent role in the Scotch Whisky Association's call for a lower tax. Japan noted that the Community initially rejected this approach and embraced both "tax per litre of beverage" and "tax per litre of pure alcohol". For Japan, the Community has thereafter changed its position and argues now that the "tax per litre of beverage" is the only meaningful yardstick because the Japanese tax is levied on the basis of volume of the beverage. The Community's new approach, which allows comparison only on the same basis as taxes are levied, however, would deny any effort to compare those taxes which have different bases. For example, the UK beer tax, which is levied on the basis of the volume of pure alcohol, cannot be compared with the UK wine tax, which is levied on the basis of the volume of beverage. Taxes per tax base cannot be the only yardstick to measure the relative tax burden. Japan reiterated that the purpose of comparing the tax burden here is to judge whether or not the tax has a distortive effect on competitive conditions. The terms of comparison ought to be, therefore, such that would correctly capture the tax's impact on consumer behaviour. Japan submitted the example of a bottle of imported whisky and a bottle of domestic shochu both priced at ¥2,000 before tax. The whisky is contained in a 0.7 litre bottle at 40 per cent alcohol, while the shochu is marketed in a four litre plastic bottle at a 25 per cent strength. If the amount of tax per volume of beverage is equalized, the tax levied on the bottle of shochu must be 5.7 times the tax on the bottle of whisky. Thus, if the whisky is to be taxed ¥700, the shochu must be taxed ¥4,000. Alternatively, if the tax per a quantity of alcohol contained is equalized, the tax on the shochu will be ¥2,500, or 3.6 times the ¥700 tax. In either case, a potential whisky customer is likely to remain undeterred. However, a shochu customer would be forced to alter its choice toward other beverages such as beer or sake. For Japan, this tax would distort consumers' behaviour; in fact the least distortive method is to levy the identical amount if products' before-tax prices are equal; ¥700 tax both on ¥2,000 shochu and ¥2,000 whisky.

4.159 Canada recalled that Japan repeatedly asserts that the tax/price ratio between all alcoholic beverages is "roughly constant" and that, accordingly, the Liquor Tax Law does not distort conditions of competition between whisky and shochu. Thus, it follows, in Canada's view that tax/price ratios that are not "roughly constant" show that the Liquor Tax Law distorts conditions of competition between whisky and shochu in favour of domestic shochu production. On the basis of Japan's own evidence that ostensibly summarizes tax/price ratios using suggested retail prices, the Liquor Tax Law does not yield "roughly" equivalent tax/price ratios between distilled liquors. This is made even more clear using tax/price ratios based on on-shelf retail prices. Put simply, in the consumer marketplace where whisky and shochu compete, the tax/price ratios show that the tax differentials between whisky and shochu are clearly skewed in favour of domestic shochu production. For Canada, in stating that "[a] consumer of any category must pay roughly 20 per cent of the expense for the tax", Japan acknowledges that on-shelf retail prices -- the prices that consumers actually pay -- are the appropriate price variable in assessing tax/price equivalency between categories of distilled liquors. In Canada's view, Japan relies on the prices offered at the discount retail outlet "Sake Ichiba Yamada", to show that "at discount prices [the tax/price ratio] is 35 per cent and 40 per cent, respectively, for whisky/brandy and shochu" and thus that "the relation between tax/price ratio of different categories tends to remain the same". However, Canada argued, the prices submitted by Japan reflect only a single discount retail outlet; significantly, in comparison to the whisky and brandy sold at this store in volumes of either 700 or 750 ml sizes, eight of the nine shochu products listed by Japan are sold in very large sizes ranging from 2.7 litres to 5.0 litres that maximize volume discounts in pricing, thereby "driving down the per unit cost of shochu and consequently "grossing-up" the tax/price ratios of shochu". The one shochu product selected having a volume of 720 ml has a tax/price ratio of 25.4 per cent that compares to a tax/price ratio for whisky of similar volume that reaches a high of 52.7 per cent. Indeed, tax/price ratios based on on-shelf retail prices in Japan show that whisky and shochu are subject to a substantial tax differential in favour of shochu:

4.160 The Community also argued that, according to the data submitted by Japan the percentage of taxes on retail prices may vary from only 5 per cent to as much as 22 per cent . Whether it is shochu or the other categories of distilled spirits that bear a higher tax/price ratio lacks any relevance. The evidence shows that the Japanese tax system is far from being "neutral" even in the terms defined by Japan and on the basis of Japan's own price data. Moreover, the method followed by Japan to calculate the tax/retail price ratios grossly and systematically underestimates the ratios for liquors other than shochu while overestimating the ratio for shochu:

4.162 Japan responded that it is correct for the complainants to argue that there is a difference in tax/price ratios among categories and that they are not fully equalized. A discrepancy, however, does not necessarily impair competitive conditions of imported distilled liquors vis-a-vis domestic shochu. For example, if the ratio is to be strictly equalized at the level of shochu A, the heaviest increase in burden will fall on brandy, and then rum. Although shochu B will have to bear a greater burden, so will whisky, vodka and gin. The burden on shochu A, on the other hand, will remain unchanged. This "full equalization" will not improve competitive conditions of imported distilled liquors vis-a-vis domestic shochu. An inevitable fate for a tax policy maker is to make continuous effort to catch up with the market reality. Though an ad valorem tax is supposed to automatically reflect the market price developments, regrettably, it usually is outwitted by sophisticated tax evaders. Thus, liquor taxes are not able to fully equalize the tax/price ratio. However, the existing lack of full equalization benefits imported distilled liquors, not domestic shochu, thus resulting in no protective effect. Concerning the argument that Japan's data on tax/price ratio is based on manufacturers’ suggested retail prices, which differ from actual on-shelf prices, Japan responded that on-shelf prices in major cities other than Tokyo are fairly close to suggested retail prices. In local cities of a medium to small size, little difference exists between suggested prices and on-shelf prices. After all, the number of "discount" liquor stores in Japan is estimated to be only a few thousand out of a total of 170,000 liquor stores. Moreover, these discount outlets offer both whisky and shochu at discount prices. Japan argued therefore that suggested retail prices are the reliable denominator by which tax/price ratios can be calculated and compared between beverage categories for the Japanese market as a whole. In order to examine the above point, Japan submitted the results of an additional survey which took place on February 14 to 16 of 1996:

4.163 In explaining the relevance of tax/price ratios, Japan referred to the example of a bottle of imported whisky (0.7 litre, 40 per cent alcohol, pre-tax price of ¥2,000) and a bottle of domestic shochu A (4 litre, 25 per cent alcohol, pre-tax price of ¥,2000). The Community criticized this example as a case of selective reference, and referring to the on-shelf price data in an exhibit submitted by Japan, requested Japan to reformulate the example into that of bottles of imported whisky and domestic shochu both containing 0.7 litre and at ¥1,000 pre-tax. Japan explained that the price shown in the exhibit is after-tax price per a quantity containing the same amount of alcohol as a 0.75 litre, 40 per cent bottle. A pre-tax price of ¥1,000 for a bottle of shochu A containing 0.7 litre at 25 per cent corresponds to ¥1,958. Among 475 items surveyed, no item was at or above ¥1,958. Similarly, among the total of 1,578 imported whisky items surveyed, only 128 bottles, or 8.1 per cent of the total, were with prices at or below the level corresponding to a 0.7 litre, 40 per cent bottle of pre-tax price of ¥1000. Japan noted that the example was used just to illustrate that tax/price ratio is a better indicator than tax per litre of beverage or alcohol, but Japan argued that it did not mean that use of one example alone can demonstrate the appropriateness of its tax rates. Japan also argued that the comparison of tax burden should be made, not by an item by item comparison, but by a comparison between the product groups created by the tax distinction, for example, between the group of imported whisky as a whole and the group of domestic shochu A as a whole. For example, under a flat rate specific tax system, a low-priced imported item has a higher tax/price ratio than a high-priced domestic item. This, however, does not mean that the tax is creating a discriminatory difference in tax burden between imported and domestic products. In the case of a flat rate tax, comparison should be made between the product group of imported distilled liquors as a whole and the product group of domestic distilled liquors as a whole. In response to the allegation by the Community and Canada that the comparison of tax/price ratios was inappropriate since it is based on average price data, Japan responded that the distribution for each category overlaps with each other, indicating roughly constant tax/price ratios across categories. Japan also responded to Canada's argument that tax/price ratios are not fully equalized. Since an ad valorem tax cannot be adopted due to the difficulty in securing compliance, the only available choices are 1) a multiple rate specific tax, which Japan currently uses, and 2) a flat rate specific tax, which the federal government of Canada currently uses. Under a multiple rate specific tax, the distribution of tax/price ratios for different categories substantially overlap one another, though perfect equivalence cannot be attained. Under a flat rate specific tax, most domestic shochu would be burdened with higher tax/price ratios than most imported whisky/brandy, creating an extreme discrepancy in burden across categories. According to Japan, all tax authorities can do is to choose from available alternatives the structure which they deem best, and to continue their best efforts to maintain the broad equivalence of tax/price ratios across categories.

4.164 The Community, Canada and the United States responded that they did not have enough time to study this additional survey presented to them on the day of the second meeting of the Panel with the parties. The Panel then allowed the complainants to respond in writing to the conclusions of the additional survey submitted by Japan. Then, the Community criticized the example used by Japan in the analytical part of its survey and requested Japan to reformulate the example into that of bottles of imported whisky and domestic shochu both containing 0.7 litre and at ¥2000 pre-tax. The Community also argued that the group of products selected for the comparison was inappropriate. The Community argued that even if it was correct that on average the tax rate on imported liquor is roughly the same as that for domestic liquor, as argued by Japan, one can show that the figure is irrelevant for purchasing decisions, and the figure is not the result of the wisdom of the Japanese government, but a simple economic process. The expert of the Community argued that the figure submitted by Japan is irrelevant, because it hides discrimination within market segments, and then submitted a figure which according to him demonstrates for instance that a bottle of shochu with a net price of ¥500 has a tax of ¥76, which implies that the tax increases the price by 15 per cent. By the same token whisky with a net price of ¥500 is burdened with an additional tax burden of 158 per cent. According to the expert of the Community, there exist a tax discrimination in every single market segment. This is true in particular for the low and medium quality brands of whisky and brandy, while for the premium brands the effect is less important. According to the expert of the Community, an important possible counter argument against the evidence submitted by the Japanese could be that the differences in taxes are justified by the different alcohol content of shochu (25 per cent), imported brandy (43 per cent) and whisky (40 per cent). This argument may not be unreasonable and could be taken to justify a tax on whisky that is 1.6 times higher than that on shochu to take this correction factor into account. This correction does not diminish the highly discriminatory nature of the tax system. For the complainants, the average tax rate for brandy and whisky is low not because the Japanese government set it low, but because expensive brands are what is left on the shelf, while low and medium quality brandies are driven out of the market through a tax system that very effectively discriminates against this market segment.

4.165 Canada argued that the portions of the survey relating to imported whisky are particularly revealing of the fundamental failures in the survey. According to Canada, the whisky samples represented in the survey cover an enormous price range, running upwards from ¥999 to over ¥15,000. Canada noted that since whisky is commonly sold in containers of similar volumes and within a similar range of alcoholic strengths, the tax applicable to each item will be very similar in most cases. Canada stated that with the applicable tax being roughly constant and the prices ranging over a magnitude of 15, the result is necessarily an equally vast range of tax/price ratios. Canada noted by way of example that a 750 ml bottle of 40 degree whisky priced at ¥1500 will have a tax/price ratio of 49 per cent whereas an equivalent premium whisky priced at ¥15000 will have a tax/price ratio of 4.9 per cent. Canada argued that the survey masks this by the artificial imposition of a single tax/price ratio of 17 per cent that purportedly applies to each and every one of the whisky samples in the survey. According to Canada, when examined on an item-by-item basis, the degree to which tax/price ratios diverge in actual situations from Japan's claimed "rough constant" of 20 per cent becomes obvious. Canada concluded, therefore, that Japan's attempt to use the survey to support its use of suggested retail price falls apart.

4.166 Japan also argued that a flat tax rate runs counter to the notion of tax equity: a consumer of gin must pay most of his expenditure as liquor tax, while a consumer of whisky is taxed only a small portion of his expenditure. Similar inequity would result if Japan were to apply a flat-rate tax. Japan recalled that under the Liquor Tax Law consumers of shochu and imported whisky are both paying about one fifth of their expenditure as liquor tax. If a flat rate tax were to be applied to these products at the present whisky tax level, however, shochu consumers would be forced to pay half their expenditure as the liquor tax, while whisky consumers' burden would remain at one fifth of their expenditure. The Tax Commission's following recommendation to the Prime Minister stresses the importance of the tax/price ratio as a yardstick in achieving an appropriate level of the tax burden: "The liquor tax has specific tax rates. When the prices rise, the level of the burden becomes lower. Therefore, it is necessary to review the level from time to time, and to ensure an appropriate level of the burden".

4.167 Japan submitted the results of the survey mentioned in paragraphs 4.83 to 4.89 above, namely a first survey conducted in March 1995 by "Shakai-Chosa Kenkyujo" (Institute for Social Studies) as well as a Summary of Findings of Statistical Analysis based on data taken from the national household survey (a government survey conducted by the Census Bureau of the General Affairs Agency since 1962 on the Japanese household revenues and expenditures) for the 20-year period of 1975 through 1994, which, in Japan's view, confirmed that there is no cross-price elasticity between shochu and other distilled imported liquors and if shochu has any cross-price elasticity with another liquor, one may argue that there is cross-price elasticity between shochu and beer.

4.168 For the Community, the Japanese statistical analysis suffered from such problems relating to the data underlying it, the common trends to which they are subject and the limited number of data points (20), that statistical correction techniques reduce the degrees of freedom in the data to such an extent that it becomes highly unlikely that one could ever statistically prove that there is cross-price elasticity of demand between whisky and shochu, even if it existed in reality. The Community also argued that in these circumstances Japan should be required to disprove that there is such cross-elasticity of demand. For the Community, the most important error of the study is that it uses nominal prices instead of deflated ones. This means that statistically the analysis does not distinguish between the price of shochu in 1984 and the price in 1993, which are practically identical in the table. In real terms, however, the price in 1993 was actually substantially lower than in 1984. Taking the average Japanese inflation rate in the 1980s (=1.9 per cent p.a.), there was a real decrease in the price of shochu by roughly 20 per cent. Evidently, the failure to discern a real price change from a nominal price change can lead to spurious results. A proper analysis should therefore use real (i.e., deflated) values to correct for this problem. A major problem of time series analysis in general is that time series are dominated by trends. In the Community's view, if one looks at the raw data, one can see for instance that the consumption of sake steadily decreased while that of beer steadily increased. There could be multiple reasons for this, e.g., changes in taste or successful advertising. In the present case, price plays a relatively small role compared to those factors that are not included in the analysis, except for what is labelled under "trend". The Community continued by arguing that most time series have such an underlying trend. If one took a random time series, e.g., of cucumber production in France, the population of India and shochu consumption in Japan, a naive statistical analysis could "prove" that for the last 20 years shochu consumption in Japan was positively related to cucumber production in France. Evidently, there is no causal relationship just because all three variables follow a common trend. This type of logic is, however, to be found in the Japanese study: Because the consumption of shochu rose at the same time as its price rose (due to inflation), a naive regression will tell you that the higher the price of shochu, the higher will be shochu consumption. This is evidently economic nonsense, but is predictably the result of the regression (expressed as positive own price elasticities in the table).

TO CONTINUE WITH JAPAN - TAXES ON ALCOHOLIC BEVERAGES

⁷⁴ 1987 Panel Report, para. 5.13.