So far, the asymmetry requirement has been described in rather simple terms. When applied, it raises many problematic details, some of which will be discussed in this section.
In Article III, the asymmetry test asks whether the impact of the national measure falls to a larger extent on imports than on domestic goods. If the burden arising from the measure is not qualitatively heavier for imports for some structural reason (asymmetric intensity 195), asymmetry can result from a horizontal imbalance in the distribution of goods between imports and the domestic group of all like products. This horizontal asymmetry is the antithesis to the diagonal approach and will be the focus of the following assessment. While this discussion will use national treatment as an example, the same considerations apply to most-favored-nation treatment. The only conceptual modification is that domestic goods in Article III become imports from one (any) country and imported products become imports from other WTO Members.
A factual asymmetry will exist if imports are underrepresented in the favored sub-category or if domestic goods are underrepresented in the disfavored sub-category. Underrepresented means that the quantitative ratio between favored and disfavored goods is lower than within the group of like products of the other origin. It is therefore insufficient to observe that the beneficiaries of the scheme are "almost exclusively domestic producers", 196 because the ratio between favored and disfavored products could still be equivalent for imports and domestic goods. For instance, the disadvantaged goods could also be almost exclusively of domestic origin and the few imports of like products could reflect the same ratio between favored and disfavored with much lower absolute numbers. Conversely, the fact alone that some domestic products fall within the less favorable category, as defendants regularly claim, is insufficient to disprove asymmetry. In contrast, it has been correctly stated that there is asymmetry where the lower tax bracket covers predominantly domestic goods and the higher tax bracket predominantly imported products. 197 However, unlike the distribution test between the categories of goods of either origin suggested above, this cumulative test fails to cover all asymmetric situations. For instance, an even distribution of imports and domestic goods in one of the regulatory sub-categories does not preclude a finding of asymmetry. Even a predominance of goods from the same origin in both sub-categories does not exclude that most domestic goods benefit and most of the few imports suffer from the distinction. The most correct definition of the test, therefore, does not relate to the composition of the regulatory sub-categories, but to the different regulatory destiny of imports as opposed to domestic goods. In other words, it consists of a comparison between products of different origin in the ratio resulting from their distribution across the regulatory sub-categories. 198 Symmetry requires that this ratio be roughly equivalent:
An alternative means of detecting an asymmetric impact of taxes and tariffs is to calculate the weighted average of the tax imposed on imports and like domestic products respectively. This means simultaneously identifies the precise extent of the differential treatment, but it cannot as easily be applied to regulation as to taxes, since non-tariff internal laws are more difficult to quantify. As far as the examination of asymmetry is concerned, this test is equivalent to the simple formula above which is applicable to taxes, tariffs and regulation. 199 In more complicated cases, however, for instance where there are three or more tax brackets, the weighted average formula is superior since it takes account of both the degree of the asymmetric distribution and the extent of the differentiation, i.e. the amount of the difference between the various tax rates. The combination of these two factors expresses whether or not and to what extent partial asymmetries that point in opposite directions offset each other, in other words, whether or not there is an overall asymmetry.
An appreciable imbalance in the distribution of goods will qualify as asymmetry. It is clear that a ratio between disfavored and favored products of 90/10 among imports versus 30/70 among domestic goods counts as asymmetry, whereas 59/41 versus 55/45 does not. There is obviously no clear line between those two scenarios. 200 Thus, the non-discrimination rules do not have the clarity they would enjoy under a diagonal test in relation to "less favorable treatment" in GATT Article III and the corresponding requirements in the other provisions. 201 One must, however, consider the current level of legal certainty for WTO Members, where it is unclear whether there is an asymmetry requirement. Knowing that roughly symmetric situations are safe may, therefore, improve the predictability of outcomes.
In borderline cases of moderate factual asymmetry, it could make sense to consider also the level of the differential treatment. 202 One may express the degree of competitive disadvantage for the whole group of like imports as a function of both the comparative share of imports affected and the degree of the disadvantage. This is admittedly a delicate proposal as far as Article III:2, first sentence, is concerned, since it is established jurisprudence that "taxation in excess" contains no de minimis exception. 203 In the area of regulation, one might believe such a rationale to be unsuitable, since many differentiations (e.g. safety or environmental standards) take the form of prohibitions. The sale of goods with certain characteristics is either allowed or prohibited. Nevertheless, that differentiation can be of different intensity, because the economic burden associated with compliance may vary significantly. For example, a requirement to sterilize milk could impose a lesser burden than a maximum amount of dioxins in milk, over which the producer has little control. One might require a stronger asymmetry where compliance with the regulation is simple and cheap, than where it is very cost intensive or hardly possible for the affected producer.
Furthermore, the necessity to produce numbers on production, (potential) import or market share ratios may in itself seem cumbersome and detrimental. As already explained, this is not to be confused with a "trade effects" test. From a conceptual viewpoint, it is unspectacular that facts such as the composition of the market matter in the realm of de facto discrimination. 204
Past disputes have illustrated, however, that the numbers adduced by the parties can differ or be contentious. What period of time would the data have to refer to? It is open to question whether panels and the Appellate Body are sufficiently well equipped to collect or assess such data. Another concern is that the practical burden of WTO litigation and the inequality between trading nations would increase given the different levels of legal infrastructure and other resources both at the governmental level and among the interested commercial operators.
Related to this is the concern that the asymmetry requirement could bring about an inappropriate shift of the burden of proof to the detriment of the complainant. The proposed solution to make lacking asymmetry a defense, 205 would however only impose the same burden on the defending party. Given that past practice has applied elements of an asymmetry requirement, this solution may also result in the more significant shift. Compared to a hypothetical diagonal approach, as it does not truly exist in practice, the obstacle for the complainant naturally increases. Taken to its logical conclusion, however, the diagonal approach does not really require the complainant to prove anything other than likeness. A solution could consist in a balanced allocation of the burden of proof. This could serve to protect panels and the Appellate Body from having to gather data themselves. The party alleging a violation needs to adduce conclusive data and the defendant needs to rebut this prima facie case.
On the other hand, there is the belief that it is very easy to use data in order to show a disadvantage for imports. 206 The past jurisprudence at least illustrates that proving asymmetry is not an insurmountable obstacle for the complainant. Panels and the Appellate Body seem to become quite strict once there is a suspicion of protective effect. All the cases reviewed in the section on the jurisprudence involved quantitative notions of the composition of the industries and imports at issue, even though the reports did not stipulate an explicit asymmetry requirement. Hence, the significance of a detrimental effect on predictability and costs of litigation should chiefly depend on whether WTO litigation would continue to operate with rough, but practicable estimates under an asymmetry requirement or deteriorate into data wars.
The existence of asymmetric effects depends on production, trade and consumption patterns, as they (would) result from undistorted market forces and decisions of producers, marketers and consumers. It therefore seems exaggerated to dismiss asymmetry claims as the panel in U.S. - Taxes on Automobiles did,207 just because foreign manufacturers have the capability to produce the more favorably treated type and domestic manufacturers are able to produce the disfavored type. 208 This would ultimately limit asymmetry to situations beyond the will of market participants. 209 An example would be a regulatory distinction referring to a raw material or technology not commonly available. This approach also disregards the economic costs related to shifting from one production to another because of the laws of another country.210 Furthermore, it undermines the economic premise of the world trading system - the mutual benefits of comparative advantage, since the foreign manufacturers' comparative advantage might precisely be in producing the disfavored type of goods. 211
In contrast, the effects of the very tax or regulation on the distribution of goods across the differently treated categories must be considered separately and cannot simply be relied on in support of an asymmetry claim.212 If a tax intends to affect behavior and is effective, domestic producers are likely to modify their production, although other countries do not impose it. The home market may be of relatively higher importance to them than that single market for a producer abroad. Depending on the level of the tax differentiation there may eventually be hardly any domestic production left in the disfavored category, while such imports continue to exist (in a larger proportion). Regulation very often prohibits the sale of goods not meeting the legal requirements. Only to the extent that export markets are sufficiently important, will that kind of production remain. France prohibited importation, production and sale of asbestos products, so the ban obviously prevented exclusively foreign actual suppliers from selling asbestos in France because domestic production ceased to exist.
Hence, even if the measure has no asymmetric effects at the outset, it is likely to produce them over time. 213 Relying on 214 or accepting 215 such effects in support of a discrimination claim disregards the burden the measure places on domestic manufacturers who would otherwise produce the disfavored type (in larger proportions). 216 Competitive opportunities rather than actual trade flows matter, and no one would argue similarly that a measure like the French asbestos ban could not be discriminatory simply because there are in fact no (more) asbestos imports. Potential production and trade flows count equally, 217 so should potential domestic production be irrelevant only because these producers ceased to exist unlike their foreign counterparts? Arguing protectionism on grounds of the effects of the measure itself would also penalize successful policies and favor unsuccessful attempts to affect consumer or producer behavior.218
One way of eliminating these effects of the measure on the distribution of goods across regulatory categories, and of taking account of potential imports and production, is to return in time to the moment the measure was introduced. 219 This would increase the predictability and stability of GATT constraints on the legislator and also better reflect the original purpose of the measure. 220 The historic test offers the advantage that the measure at issue has not yet distorted the distribution of the domestic production. This is also true for imports, should the economy be large enough to have an effect on production patterns abroad.
However, the prohibition of discrimination is a continuous obligation protecting effectively equal competitive opportunities at any, not only one, point in time. Article III is expressly concerned with the treatment actually accorded and the taxes applied, which relate to the present. National measures must, therefore, also remain non-discriminatory and comply with the substantive requirements of legality at all times. The conformity of a measure with substantive legal requirements does not depend on whether it has been in place for a long time or whether it has just been enacted, 221 lest Members should be in a different legal situation when they repeal and reenact a law than when they maintain the law. A panel typically has the function of assessing a measure's conformity with a covered agreement at the moment of the dispute. The historic examination can only answer the question about discriminatory effects at the moment of the measure's introduction. Circumstances may change due to modified production or consumption patterns throughout the world or the development of new products. In the context of de facto discrimination, it is intuitive that the evolution of the facts matters. The effects of a rule may, therefore, subsequently become or cease to be discriminatory due to changes in the factual circumstances other than those produced by the measure itself.
It may, therefore, be necessary to assess the composition of actual and potential imports at the time which is relevant in the dispute. The markets of other countries not applying a similar measure could provide useful insights about potential competition from imports and possibly about the hypothetical shape of domestic production in the absence of the measure. 222 But it remains difficult for judges to guess the marketplace. Even if econometrics can provide for additional data, the reliability of the evidence and the predictability of the findings made on the basis of such evidence would inevitably suffer. 223 Panels and the Appellate Body would ultimately have to make use of their normative judgment. For practical purposes, however, many cases should be clear-cut and unproblematic as the relevant GATT and WTO disputes of the past show.
The discussion so far has mostly referred to "like imports" in general, without specifying whether this stands for the group of like imports from all sources, taken together, or only those originating in an individual WTO Member, for instance the complainant in a given case. It appears that the dispute settlement practice is not entirely consistent in this regard, when it comes to the comparison between imported and like domestic products as to whether imports receive "less favorable treatment". Panels sometimes draw this comparison between the group of like domestic products and the group of all like imported products from any foreign origin,224 but sometimes take into consideration the group of like imports only from the complaining WTO Member. 225 In most cases, panels do not seem to take a stand at all.
Under the asymmetry requirement, this question can be of decisive importance, because the relative distribution of like imports across the regulatory sub-categories may be different for imports from one Member, compared to the distribution of imports from all foreign sources, taken together. What is more, if imports from individual WTO Members were to be considered individually, it would be quite difficult for the importing Member to satisfy the requirement of symmetric impact vis-à-vis each of the other Members, whose production patterns are likely to differ. 226 Under the diagonal approach, this whole question naturally does not arise, since, in the absence of an asymmetry requirement, the configuration of imports is irrelevant, as long as some (potential) imports fall within the disadvantaged sub-category.
The opening language of GATT Articles III:2 and III:4 ("The products of ... any contracting party imported ...") suggests that the asymmetry test has to occur between the group of like domestic products and the group of like imports only from the single WTO Member whose national treatment right it at issue. In contrast, the second sentence of Article III:2 and Articles III:1 both refer to imports in general. 227 The relevant part of Article I:1 ("... any advantage ... granted ... to any product originating in ... any other country shall be accorded ... to the like product originating in ... all other contracting parties.") could be read as providing for a comparison between the imports from one country and the like imports from all other Members, taken together. The most-favored-nation provision in GATS Article II, however, corresponds to GATT Article III, and not to Article I:1 in this respect.
It is open to question whether the wording truly intends to prejudge this question. Similarly to the question of asymmetry versus diagonal test, it might be wrong to give dispositive weight to the treaty text and to read into those words an answer that those words may not contemplate. 228 Where the GATT refers to "any contracting party", "any product" or "any advantage", it is clear that, in the first place, it intends to emphasize that, without exception, all WTO Members enjoy those rights with regard to all products.
In the realm of origin-neutral rules, one may argue further that every WTO Member is entitled to have its products protected against de facto less favorable treatment than that accorded to the group of like domestic products.229 One could also argue, however, that the purpose of the national treatment obligation is to prevent Members from granting preferences to their national products to the detriment of imports as a whole. As a consequence, instances of fortuitous de facto discrimination against one national group of like imports are harmless.
In order to precisely assess the balance between imported and domestic goods, it is also necessary to ensure that the comparison includes the totality of like (substitutable) products on both sides. It seems that, in practice, panels and the Appellate Body mostly decide only on the likeness or substitutability of the differently treated products the parties refer to. This could be insufficient because the respective groups of like products might contain additional goods and, therefore, result in a different balance than with just the goods raised in the dispute. Not considering these other goods for reasons of practicality would seem little less selective and manipulative than the diagonal approach. A comprehensive market analysis ex officio, however, would certainly be difficult and cumbersome. It would not only require the common case by case likeness test of the goods at issue but, in addition, a determination of the limits of likeness with a view to other products. Already today, litigators and dispute settlement reports must pay significant attention to the likeness or substitutability test. A requirement of delimiting these concepts, or the possibility of changing the picture by introducing additional like products, would make that test even more complicated. 230
A possible solution would be for a panel to assess the likeness only of those goods which are mentioned in the complaint and to include in this assessment the products which the defendant relies on in an attempt to demonstrate an overall balance. Goods referred to by neither party would not enter the asymmetry test. Both in the past GATT/WTO jurisprudence and within the European Communities, where a more or less strict asymmetry requirement applied, this issue does not seem to have posed problems.
195 See supra section III.A.2, note 10.
196 Panel Report, Korea - Taxes on Alcoholic Beverages, supra note 67, para. 10.102.
197 William J. Davey & Joost Pauwelyn, supra note 8, at 39. Appellate Body Report, Korea - Taxes on Alcoholic Beverages, supra note 68, para. 150.
198 Correct therefore: Appellate Body Report, Chile - Alcoholic Beverages, supra note 72, paras. 64, 67.
199 The weighted average tax imposed on imports is: (favored imports/total like imports) x lower tax rate + (disfavored imports/total like imports) x higher tax rate. An equation composed of the two weighted averages can mathematically be transformed into the above-mentioned equation. Both equations therefore produce the same results as regards the main question of whether the distribution of products among the regulatory sub-categories is symmetric or asymmetric.
200 William J. Davey & Joost Pauwelyn, supra note 8, at 50, note 117; Kazumochi Kometani, supra note 185, at 463.
201 See also Ole K. Fauchald, supra note 9, at 225.
202 Panel Report, U.S. - Taxes on Automobiles, supra note 81, para. 5.14: "no sudden transition to a higher tax at the threshold". See also Case 112/84, Humblot v. Directeur des services fiscaux,  ECR 1367, paras. 14-16 and Massimiliano Danusso & Ross Denton, supra note 123 at 81.
203 Appellate Body Report, Japan - Alcoholic Beverages, supra note 63, at 115 (section H.1.(b)), Panel Report, 1987 Japan - Alcoholic Beverages, supra note 49, para. 5.8, 5.11, Panel Report, U.S. - Malt Beverages, supra note 6, para. 5.26.
204 See supra section IX.C, text accompanying notes 186-187.
205 William J. Davey & Joost Pauwelyn, supra note 8, at 41.
206 Massimiliano Danusso & Ross Denton, supra note 123, at 80: "easy to manipulate the data".
207 See Panel Report, U.S. - Taxes on Automobiles, supra note 81, para. 5.14, as well as the summary of the panel's reasoning supra in section IV.C.
208 Highly critical of the panel's findings ("results-oriented, opportunistic manipulation of facts" and "indirect discrimination by origin") is also Richard A. Westin, supra note 77, at 128, 130. See also Aaditya Mattoo & Arvind Subramanian, supra note 146, at 310-312; and the emphasis on "how the United States defined the vehicles which are subject to the tax" by Charles T. Haag, Legitimizing "Environmental" Legislation Under the GATT in Light of the CAFE Panel Report: More Fuel for Protectionists?, 57 U. Pitt. L. Rev. 79, 101 (1995).
209 In support however David M. Driesen, supra note 16, at 349-350, 351, 356, 357.
210 James H. Snelson, Can GATT Article III Recover from Its Head-On Collision With United States - Taxes on Automobiles, 5 Minn. J. Global Trade 467, 492 (1996).
211 Eric Phillips, World Trade and the Environment: The CAFE Case, 17 Mich. J. Int'l L. 827, 846 (1996); Aaditya Mattoo & Arvind Subramanian, supra note 146, at 312.
212 See also Edmond McGovern, supra note 43, p. 8.22-8.
213 See also Ole K. Fauchald, supra note 9, at 217, using the term "self-reinforcing discriminatory effects".
214 The EEC relied on such effects in the disputes 1987 Japan - Alcoholic Beverages, see Panel Report, supra note 49, para. 3.2 (e); U.S. - Taxes on Automobiles, see Panel Report, supra note 81, para. 3.111.
215 In principle Ole K. Fauchald, supra note 9, at 217-218.
216 The European Court of Justice recognized this effect of the measure at issue and relied on it in order to consider the existing asymmetry as irrelevant in Case 140/79, Chemial Farmaceutici,  ECR 1, paras. 16 and 18, and in Case 46/80, Vinal v. Orbat,  ECR 77, paras. 15 and 18.
217 Panel Report, U.S. - Section 337, supra note 10, paras. 5.13, 5.23; Appellate Body Report, Canada - Periodicals, supra note 71, at 464, 466 (sections IV and V.A). William J. Davey & Joost Pauwelyn, supra note 8, at 40, 50 note 117; Ole K. Fauchald, supra note 9, at 218, 219.
218 Massimiliano Danusso & Ross Denton, supra note 123, at 79.
219 Simon Simon Lester & Kara Leitner, supra note 74, at 13.
220 Ole K. Fauchald, supra note 9, at 232-233; Panel Report, U.S. - Taxes on Automobiles, supra note 81, para. 5.24.
221 This legal principle is familiar from national constitutional regimes: a limitation of fundamental right must also remain consistent with the constitution. See the recent example of a limitation of the professional freedom that is no longer necessary, and hence disproportionate, in view of changed circumstances, Judgment of the Bundesverfassungsgericht, 1 BvR 335/97, 13 December 2000, 54 Neue Juristische Wochenschrift 353, para. 41 (2001).
222 In the assessment of likeness and substitutability the Appellate Body includes latent demand and evidence from other markets. See Appellate Body Report, EC - Asbestos, supra note 5, para. 123; Appellate Body Report, Korea - Taxes on Alcoholic Beverages, supra note 68, paras. 115, 120, 137.
223 See also the concern of William J. Davey & Joost Pauwelyn, supra note 8, at 41.
224 Taking account of "imports from other countries" than the complainant e.g.: Panel Report, U.S. - Taxes on Automobiles, supra note 81, para. 5.12.
225 Panel Report, 1987 Japan - Alcoholic Beverages, supra note 49, para. 5.9 a); Panel Report, EC - Bananas III, supra note 46, para. 7.334 in the context of services in GATS Article XVII.
226 See supra note 151 and accompanying text.
227 See, with such opposite conclusions for the two sentences of Article III:2: Kazumochi Kometani, supra note 185, at 465, note 57.
228 See supra section IX.A. and text accompanying notes 170-173.
229 See also James H. Snelson, supra note 210, at 493.
230 As antitrust law suggests, it would, however, not be impossible to delineate the market on which goods compete. In favor of applying this market definition under Article III: Damien J. Neven, How Should "Protection" Be Evaluated in Article III GATT Disputes?, 17 European Journal of Political Economy 421, 423, 440 (2001).