We had so much fun talking about discriminatory purpose last week, I thought I'd bring up the subject again. This time, I want to focus on exactly how one might evaluate a measure's legitimacy, which you may or may not think involves a determination as to whether a measure has a protectionist intent. Let's look at two examples from Appellate Body reasoning.
First up is Chile - Alcohol, with reasoning under the "so as to afford protection" element of GATT Article III:2, second sentence:
63. We turn, therefore, to the design, the architecture and the structure of the New Chilean System itself. That system taxes all alcoholic beverages with an alcohol content of 35° or below on a linear basis, at a fixed rate of 27 per cent ad valorem. Thereafter, the rate of taxation increases steeply, by 4 percentage points for every additional degree of alcohol content, until a maximum rate of 47 per cent ad valorem is reached. This fixed tax rate of 47 per cent applies, once more on a linear basis, to all beverages with an alcohol content in excess of 39°, irrespective of how much in excess of 39° the alcohol content of the beverage is.
64. We note, furthermore, that, according to the Panel, approximately 75 per cent of all domestic production has an alcohol content of 35° or less and is, therefore, taxed at the lowest rate of 27 per cent ad valorem. Moreover, according to figures supplied to the Panel by Chile, approximately half of all domestic production has an alcohol content of 35° and is, therefore, located on the line of the progression of the tax at the point immediately before the steep increase in tax rates from 27 per cent ad valorem. The start of the highest tax bracket, with a rate of 47 per cent ad valorem, coincides with the point at which most imported beverages are found. Indeed, according to the Panel, that tax bracket contains approximately 95 per cent of all directly competitive or substitutable imports.
65. Although the tax rates increase steeply for beverages with an alcohol content of more than 35° and up to 39°, there are, in fact, very few beverages on the Chilean market, either domestic or imported, with an alcohol content of between 35° and 39°. The graduation of the rates for beverages with an alcohol content of between 35° and 39° does not, therefore, serve to tax distilled alcoholic beverages on a progressive basis. Indeed, the steeply graduated progression of the tax rates between 35° and 39° alcohol content seems anomalous and at odds with the otherwise linear nature of the tax system. With the exception of the progression of rates between 35° and 39° alcohol content, this system simply applies one of two fixed rates of taxation, either 27 per cent ad valorem or 47 per cent ad valorem, each of which applies to distilled alcoholic beverages with a broad range of alcohol content, that is, 27 per cent for beverages with an alcoholic content of up to 35° and 47 per cent for beverages with an alcohol content of more than 39°.
66. In practice, therefore, the New Chilean System will operate largely as if there were only two tax brackets: the first applying a rate of 27 per cent ad valorem which ends at the point at which most domestic beverages, by volume, are found, and the second applying a rate of 47 per cent ad valorem which begins at the point at which most imports, by volume, are found. The magnitude of the difference between these two rates is also considerable. The absolute difference of 20 percentage points between the two rates represents a 74 per cent increase in the lowest rate of 27 per cent ad valorem. Accordingly, examination of the design, architecture and structure of the New Chilean System tends to reveal that the application of dissimilar taxation of directly competitive or substitutable products will "afford protection to domestic production."
What I see here is a thorough consideration of how the measure is designed and operates with respect to its treatment of imported and domestic products. The Appellate Body looks at the particular tax categories that were established, and says, in effect, that setting up those categories in such a way that imported products mostly get the high tax rate, and domestic products mostly get the low one, is good evidence of protection of domestic products.
The second case is U.S. - COOL, with reasoning under TBT Article 2.1
343. As designed and applied, however, the COOL measure does not impose labelling requirements for meat that provide consumers with origin information commensurate with the type of origin information that upstream livestock producers and processors are required to maintain and transmit. Rather, the origin information that must be conveyed to consumers is less detailed, and will often be less accurate. This is because the COOL measure requires the labels to list the country or countries of origin, but does not require the labels to mention production steps at all. If, for example, the relevant production steps took place in more than one country, the relevant label (B or C) will identify more than one country, but will not identify which production step took place in which of those countries. Under the labelling rules, labels for Category B meat may also list countries of origin in any order, such that the order of countries listed on the labels cannot be relied upon to indicate where certain production steps took place. Furthermore, due to the additional labelling flexibilities allowed for commingled meat, a retail label may indicate that meat is of mixed origin when in fact it is of exclusively US origin, or that it has three countries of origin when in fact it has only one or two. For Category D meat, the COOL measure requires only that the customs designation of origin be indicated. Given that the United States does not use the same definition of "origin" for customs purposes as it does for the COOL measure, a D Label will not convey information on the countries of birth or raising of the livestock from which the imported meat was derived. Even Label A, indicating "Product of the USA", which the Panel found to be the only label that provides "meaningful information for consumers", is not required to refer explicitly to the productions steps of birth, raising, and slaughter.
344. In comparing the origin information requirements imposed on upstream producers with the origin information conveyed to consumers, we also consider relevant the fact that the COOL measure exempts from its labelling requirements muscle cuts of beef and pork that are "ingredient[s] in a processed food item", or are sold in a "food service establishment" or in an establishment that is not a "retailer". As noted above, upstream producers do not, and likely could not, distinguish between livestock that will be used to produce a product exempt from the labelling requirements, and livestock that will be used to produce covered commodities that must be labelled when sold at retail. This is because "the ultimate disposition of a meat product is often not known at any particular stage of the production chain". This means that, generally speaking, information regarding the origin of all livestock will have to be identified, tracked, and transmitted through the chain of production by upstream producers in accordance with the recordkeeping and verification requirements of the COOL measure, even though "a considerable proportion" of the beef and pork derived from that livestock will ultimately be exempt from the COOL requirements and therefore carry no COOL label at all.
...
346. Taking account of the overall architecture of the COOL measure and the way in which it operates and is applied, we consider the detail and accuracy of the origin information that upstream producers are required to track and transmit to be significantly greater than the origin information that retailers of muscle cuts of beef and pork are required to convey to their customers. That is, the labels prescribed by the COOL measure reflect origin information in significantly less detail than the information regarding the countries in which the livestock were born, raised, and slaughtered, which upstream producers and processors are required to be able to identify in their records and transmit to their customers. Furthermore, upstream producers will be subject to the COOL measure's recordkeeping and verification requirements even when the meat derived from their animals is ultimately exempt from the labelling requirements of the COOL measure, for example, due to the type of establishment in which the meat is sold. Lastly, a processor's decision to use livestock of different origins rather than exclusively US origin livestock will not only be more costly, it will also lead to confusing information being conveyed to consumers.
347. For all of these reasons, the informational requirements imposed on upstream producers under the COOL measure are disproportionate as compared to the level of information communicated to consumers through the mandatory retail labels. That is, a large amount of information is tracked and transmitted by upstream producers for purposes of providing consumers with information on origin, but only a small amount of this information is actually communicated to consumers in an understandable manner, if it is communicated at all. Yet, nothing in the Panel's findings or on the Panel record explains or supplies a rational basis for this disconnect. Therefore, we consider the manner in which the COOL measure seeks to provide information to consumers on origin, through the regulatory distinctions described above, to be arbitrary, and the disproportionate burden imposed on upstream producers and processors to be unjustifiable.
Here, it seems to me, the Appellate Body is doing something quite different. Instead of thinking about the design and operation of the measure as it relates to imported and domestic products, it looks at the reasonableness of the measure itself, focusing on the proportionality of the information collected in relation to the information passed on to consumers.
As noted, the Chile - Alcohol analysis was done under GATT Article III:2, second sentence; the COOL analysis was under TBT Article 2.1. But let's put aside where you might do this kind of analysis (this provision or that provision; obligation vs. exception), and just think about what the analysis should look like.
To me, the Chile - Alcohol approach seems closely tied to the idea of nationality-based discrimination, as it looks deep into the measure and tries to determine, objectively of course, its intent. In that case, it seems pretty clear that the intent was protectionism: The higher tax rate appears to target imports.
By contrast, in the excerpted passage from COOL, the nationality issues are not there, at least not as clearly. It seems more like an identification of arbitrariness in the measure: Not all the information that is collected is made use of in the labelling. This may be true, but it's not clear how that ties into the broader issue of discrimination.
Now, there is more to the reasoning in both cases, but I'm just trying to highlight certain parts, to illustrate what I see as particularly helpful analysis. Broadly speaking with regard to discrimination, as long as the right kind of analysis will be done somewhere, I'm not too concerned about the scope of these rules being overbroad. But I worry that this will not always happen when applying the exceptions, and that we will get distracted with arbitrary aspects of a measure, rather than trying to look at whether the measure's intent was protectionist or not.
