Thank you, Joel, for letting me join the discussion of Mexico – Soft Drinks. It seems to me that if one simply picks up XX(d) and reads it as a piece of natural language, taking it all together and attending to the illustrative examples, the picture one gets is of a relatively narrow provision (as compared to what many people are suggesting). It seems that the illustrative examples, while they are not exhaustive, are meant to illustrate fairly comprehensively a type of “laws and regulations” that XX(d) is about, a type that does not include such laws as the Alien Tort Claims Act, or Rob’s Mexico 3001 Act, and so on. So I want to offer a text-based argument for such a narrow approach (and incidentally to define the type of laws that XX(d) is about). Many people seem to want to use XX(d) to justify trade sanctions in defense of human rights or labor rights; or alternatively they want to use XX(d) to guarantee that there is room within the WTO system for, say, NAFTA-authorized retaliation for NAFTA violations, or MEA-authorized trade sanctions (against signatories) for MEA violations. These are laudable interpretive goals, which I think should be achieved by other interpretive means than what seems to me an overbroad reading of XX(d). For what it is worth, I think my approach to XX(d) reflects the Appellate Body’s general thinking, although not their reasoning.
XX(d) allows a defense when the measure is “necessary to secure compliance with laws or regulations which are not inconsistent with the provisions of this Agreement [followed by examples].” I want to focus on the phrase “which are not inconsistent with the provisions of this Agreement,” a phrase that everyone else has been ignoring (including the AB, I forget about the Panel). It is easy to ignore this phrase. It is obvious that we cannot allow a defense for measures designed to secure compliance with GATT-inconsistent measures; that would undermine the entire agreement. So it can seem that this phrase merely records that obvious idea, and no more. But I think it does more.
The basic move in my argument is this: First, I suggest that there are some types of laws [I shall hereafter use both “laws” and “regulations” indifferently to mean “laws or regulations”] that can be inconsistent with the GATT, and there are other types of laws that cannot be inconsistent, because they are simply not the sort of thing the GATT addresses at all. (Examples of these types will be given presently.) Then I suggest that the phrase “which are not inconsistent with the provisions of this Agreement” refers to laws that (i) are of such a type that they could be inconsistent with the GATT, but that (ii) are not in fact inconsistent. The crucial move here is (i), which restricts the “target regulations” under XX(d) (that is, the laws compliance with which we are trying to secure) to laws of those types that could be GATT-inconsistent. It might be objected that this move is illogical. After all, if a law is of a type that cannot be GATT-inconsistent, then necessarily it is not itself GATT-inconsistent; and so it may seem that “laws which are not inconsistent with the provisions of this Agreement” must include all laws which are of such a type that they cannot be GATT-inconsistent, and in addition all laws which are of such a type that they could be GATT-inconsistent, but which are not. That is indeed where we are led by standard formal logic that includes material implication. But all speakers of natural languages (that is, all of us) know at least implicitly that natural language often does not live up to the rules of material implication, that it often embodies some version of “relevance logic” (which also has its formalizations). In this case it is a “relevance” idea that suggests that the phrase “which are not inconsistent with the provisions of this Agreement” naturally refers, in this context, only to laws that could be GATT-inconsistent but are not. So worries about formal logic need not deter us. The question is what is the most plausible reading of the text.
That last paragraph was extremely abstract. (Trying to start more concretely created its own problems.) What does the argumentative move I am suggesting – limiting the target regulations to laws that are of types that could be GATT-inconsistent – actually get for us? Well, for a start, it generates quite straightforwardly the conclusion, which we all seem to find intuitive, that the target regulations must be domestic law (including domestic law implementing international law, or even treaties with direct effect in their “domestic law” guise). Here’s the argument: (1) The only types of law that can be inconsistent with the GATT are measures by Members; that is simply all that any of the prohibitory provisions addresses. (2) But no bit of international law is a measure by a Member; every bit of international law results from the joint law-making activity (explicit or implicit) of at least two nations. Of course, each contributing nation engages in legally significant action when it makes its contribution to this joint law-making activity; but still, the international law as such is not a measure of any Member. (3) International law is therefore not the type of law that could be inconsistent with the Agreement. (4) International law is therefore not a possible “target regulation” for XX(d). (5) So, the only possible target regulations are domestic law. QED. (If a treaty has direct effect, then the domestic-effect aspect is the result of unilateral lawmaking, perhaps implicit; it therefore is, for all we have yet said, the sort of thing that could be inconsistent with the Agreement; and thus it might be, for all we have yet said, a target regulation.)
We now have a new argument for the proposition that only domestic regulation can be a target regulation under XX(d). Of course, most of us seemed to accept that already, so have we made any progress? I think so. First, we have seen how the argument form works; we will use it in a moment to deliver a much more interesting conclusion. Second, the new argument for the already-accepted conclusion seems to me stronger than the other arguments that have been suggested. Finally, the fact that the argument form neatly delivers this already-accepted conclusion may provide some validation of the argument form itself, for those who are suspicious of it.
So, what else might we learn from the insight that target regulations must be of such a type that they could be GATT-inconsistent? We have already seen that this limits the field of target regulations to domestic law. But in fact it limits the field still further. Again, if we look at the basic prohibitory provisions of the GATT, we see that all the prohibited measures are of two types: border measures (Art. I in part, Art. II, Art. XI) or internal regulation (in which I include including taxation) (Art. I in part, Art. III). [The proper classification of Art. V, Art. XVI re export subsidies, and Art. XVII is an issue I ignore; none of these threatens my general thesis.] In other words, only border measures and internal regulations can be GATT-inconsistent. So even domestic laws that are neither border measures nor internal regulations cannot be GATT-inconsistent. Hence, domestic laws that are neither border measures nor internal regulations cannot be target regulations under XX(d). Or conversely, target regulations under XX(d) must be domestic-law border measures or internal regulation. (This is confirmed, so far as examples can confirm such a proposition, by the illustrative examples to XX(d).) But the Alien Torts Claims Act (at least in its most characteristic applications – I confess I do not know its precise scope), and the Torture Victims Prevention Act (ditto), and Rob’s Mexico 3001 Act (the primary provisions that condemn foreign behavior), and even a directly-effective NAFTA (insofar as it creates “domestic” obligations for foreign governments not to erect illegal trade barriers, for example) are neither border measures nor internal regulations. One might have been forgiven for thinking in 1947, if one asked oneself the question, that virtually all domestic laws were either border measures or internal regulation. That is less true today. But my fundamental argument is, in effect, that the expansion in what governments feel free to legislate about “domestically” does not ipso facto broaden the scope of XX(d).
I may be accused of “teleological interpretation”. My response is, “Of course, but no more than the inevitable minimum.” I freely concede – I would insist – that I did not reach this interpretation in the first instance by an attempt to construe each word or phrase of the text all by itself and then build up the meaning of the whole from those separate bits. That’s no way to read. But I absolutely started from the text. I started from what the whole thing seemed to mean, and I then looked to see if the most plausible meaning for the whole could be constructed out of plausible readings of the bits. If it could not, I would have had to start over. (This is the search for “reflective equilibrium” in text-construction.) But I think it can.
Don Regan
