Not only judicial tests and the essence/application dichotomy but also the allocation of the burden of proof affect the strictness of AB&P's review vis-à-vis ETMs. In the following, we analyze whether the AB&P's findings regarding the allocation of the burden of proof are appropriate from an economic viewpoint, i.e. the minimization of the costs involved in the WTO adjudication.
According to Hay, there are two kinds of costs involved in the allocation of the burden of proof: "process costs" and "error costs".215 Process costs are those which are spent by each party in order to obtain favorable results for themselves in the case. If the case is resolved through litigation, the costs for presenting evidence are regarded as process costs. If the case is settled through negotiation, the costs for bargaining constitute process costs. "Error costs" are those which are derived from erroneous results by courts, for example, underdeterrence of harmful behavior for the society.
If we assume that parties have perfect information about whether the plaintiff's claim is meritorious and there is no possibility of settlement, the burden of proof should be assigned to the defendant since the plaintiff has no incentives to bring a meritless claim and, if the plaintiff bears the burden of proof, he has to suffer the costs for presenting evidence.216 However, since parties are not always sure about the results of the cases and there are possibilities of out-of-court settlement, we have to reconsider which party should bear the burden of proof from the viewpoint of the minimizing the total of process and error costs.217 According to Hay, the following five factors affecting a defendant and a plaintiff should be compared in order to optimally allocate the burden of proof: 1) the party's costs of presenting evidence to support its position, 2) the probability that the party's position is correct, 3) the party's estimate of its chance of success, 4) the amount at stake for the party, and 5) the social costs of an erroneous decision against the party.218 As a general proposition, the initial burden of proof should be allocated to plaintiffs because, in normal cases, the probability that plaintiffs make meritorious claims is relatively minor (assuming that the threat of liability normally induces defendants to comply with laws), while both plaintiffs and defendants are usually in a similar position with respect to the four other factors (for example, the amounts at stake for defendants and plaintiffs are usually the same).219 If we allocate the burden of proof to defendants in the cases where the probability that plaintiffs' claims are meritorious is low, two things would happen: the plaintiffs would tend to sue even if the legal grounds of their claims are so weak that their chances of success would be very small; the plaintiffs would tend to sue even if they know that their claims are meritless, hoping that they could extract compensation from the defendants in their out-of-court settlements. (If the amount of compensation is less than the defendants' costs for presenting the evidence which shows that the claims are meritless, the defendants have incentives to agree with an out-of-court settlement.) As a result, the total of process and error costs is likely to be greater when the defendants bear the burden of proof than when the plaintiffs bear it.220
Hay maintains that it might be optimal to allocate the initial burden of proof to defendants in the following four cases: 1) when, based on the information available to the court, it is likely that plaintiffs have meritorious claims, 2) when the plaintiffs' costs of presenting evidence are higher, 3) when the plaintiffs have more at stake, and 4) when errors in the defendants' favor are more socially costly than errors in the plaintiff's favor because such errors dilute the deterrence of harmful behaviors by the defendants.221
From this perspective, it would make sense to allocate the initial burden for establishing a prima facie case to a defending party in the context of GATT Article XX for two reasons: 1) the probability that the defending party has a meritorious claim is relatively low, 2) the defending party's costs of presenting evidence to support its position is normally lower than those of the plaintiffs. The defending parties have motivations to invoke Article XX even if the chance that they can prevail is low because the invocation of Article XX is almost their only way to avoid losing the case under the situation where the complaining parties successfully prove the GATT violations by the defending parties, assuming that additional costs for presenting arguments regarding GATT Article XX justification are low. Therefore, based on the information the AB&P normally have, the probability that defending parties have meritorious cases would be low. Furthermore, in the process of an Article XX analysis, the parties have to adduce evidence regarding the details of the measures in dispute including the information about how the measures have been enforced by the authorities of the defending parties. It would be difficult for complaining parties to present evidence regarding the details of the measures contested.222 As a result, it would be better to allocate the initial burden of establishing a prima facie case to defending parties in the context of Article XX of the GATT.
It is sometimes pointed out that allocating the initial burden of Art. XX justification to ETM-invoking countries means that the WTO puts higher value on free trade than on environmental protection. However, as explained above, the low likelihood of meritorious Article XX claims and the informational asymmetry between complaining and defending parties, not the preference for trade values, may be the reasons why the defending parties should bear the initial burden.
With regard to Article 5 of the SPS Agreement, the initial burden of establishing a prima facie case of the violation of the SPS Agreement is imposed on the complaining country. At first glance, allocating the burden of proof to complaining parties looks odd because both the SPS Agreement and Article XX of the GATT deal with the same kind of ETMs (at the very least, a measure which falls under the scope of GATT might be also subject to the SPS Agreement). However, if we take into account the economic considerations discussed above, it makes sense that a complainant should bear the initial burden, in the context of Article 5 of the SPS Agreement. In this context, the probability that a complaining party makes a meritorious claim is not high because to some extent there exists uncertainty regarding the result of the case. Imposing the initial burden of establishing a prima facie case of consistency with the SPS Agreement to a defending party would provoke an increase in meritless complaints of violations of the SPS Agreement since, if a defending party bears the initial burden and fails to prove the compatibility of its ETM with the SPS Agreement, the AB&P decide in favor of the complaining party. That is, exporting countries would be tempted to bring complaints to the DSB, hoping that the defending countries fail to satisfy the threshold for establishing the prima facie case of non-violation or that the defending countries agree on a settlement prior to the establishment of the Panel in order to avoid losing in the DSB, even if the grounds of the complaining parties' argument are not strong.223 Therefore, it makes sense that, in Hormones, the AB found that a complaining party bears the initial burden of establishing a prima facie case of inconsistency with the SPS Agreement.
However, it should be noted that a defending party has better access to evidence on the details of the SPS measures in dispute. It would be difficult for a complaining party to adduce sufficient evidence to support their argument if the threshold for shifting the burden of proof is too high. Because of this asymmetrical access to the information regarding the SPS measure concerned, the threshold for establishing a prima facie case of inconsistency with the SPS Agreement should be relatively low. In this regard, it makes sense that, in Apple, the AB found that the initial burden of a complaining party can be satisfied merely by showing the non-existence of relevant studies or researches in the contexts of Article 2.2 and Article 5.1 of the SPS Agreement. Article 5.8 obliges SPS-imposing countries to offer information regarding the scientific justification of their SPS measures when exporting countries request it to do so. The AB attempted to lower the level of the threshold for Article 2.2 to the level which can be accomplished by the evidence obtained through an Article 5.8 request. It could be said that the likelihood that a contested measure is based on sufficient scientific evidence is low, provided that the measure-imposing country fails to offer information about the scientific justification of the measure. Therefore, the burden of proof should shift to the measure-imposing country.
Furthermore, this low threshold could be applied for Article 5.6, the LTRA standard. If we assign complaining parties the initial burden of establishing a prima facie case of defending parties' failure to satisfy the LTRA test, the burden would be very difficult to be satisfied since the complaining parties have to show all of the following three things: 1) there is at least one alternative which is reasonably available, taking into account technical and economic feasibility; 2) the alternatives can achieve the Member's appropriate level of protection; and 3) the alternatives are significantly less restrictive to trade than the SPS measure in dispute.224 As a matter of fact, there have been no cases where the complaining parties were able to show all of these factors. It could be said that this very high threshold for showing the violation of Article 5.6 almost emasculates the LTRA requirement under the SPS Agreement. Therefore, the burden of showing consistency with the SPS Agreement should shift to defending parities if complaining parties are able to show the non-existence of relevant studies or researches. By so doing, the AB&P can reduce error costs. That is, it is unlikely that the defending parties have deliberated on whether the measures concerned are the least trade restrictive among available ones in the situation where the measure-imposing countries fail to show relevant studies and researches to countries which make Article 5.8 requests. The defending parties should bear the burden of proof because they have better access to relevant information, and because the possibility that their measures are the least trade restrictive among reasonably available ones is low.
In conclusion, in the contexts of GATT Article XX and the SPS Agreement, the allocation of the burden of proof by the AB&P has been generally appropriate. However, the threshold for shifting the burden of proof to defending countries in the context of Article 5.6 of the SPS Agreement is too high. The threshold should be lowered by shifting the burden of proof to defendants when complaining parties are able to show that there are no relevant studies or researches.
The burden-of-pleading rule defines "the set of issues that each party must assert in order to put those issues into play."225 That is, unless the party who bears the burden of pleading for an issue asserts the availability of the issue in the case, the court does not have to examine the issue. In most cases, the party who bears the burden of proof about an issue has to bear the burden of pleading about the issue.226 However, that is not always the case. For example, under U.S. tort law, although defendants must plead contributory negligence, some states impose the burden of proof on plaintiffs.227 According to Lee, the "divergence" of allocation between the pleading of proof and the burden of proof can be economically justified. Lee maintains that the burden-of-pleading rule functions as a device for reducing process costs in two ways. First, process costs could be saved by placing the burden of pleading to defendants about those issues which are unlikely to arise. Otherwise, process costs increase since plaintiffs have to raise every possible issue including the ones which arise infrequently. Second, through imposing the burden of pleading on the party to whose "version of the issue in question is more discrete", the process costs could be saved.228 By so doing, the other party does not have to anticipate and produce evidence contravening the indefinite number of possible claims and arguments.229 For example, in the context of contributory negligence in the U.S. tort law, the allegations of defendants would be more "focused" on certain elements of the case than the allegations of plaintiffs. In other words, if plaintiffs have to bear the burden of pleading, they have to assert that they were not negligent in all the aspects of their behavior in the case. Therefore, the burden of pleading is imposed on defendants in the context of the U.S. tort law. However, because plaintiffs have better access to the information with regard to the level of care of themselves at the time that the case (for example, a traffic accident) occurred, some courts impose the burden of proof on plaintiffs.230 In sum, since the burden-of-proof and burden-of-pleading rules reduce process and error costs in different ways, allocating burdens to different parties under burden-of-proof and burden-of-pleading rules could be justified.
In the context of the WTO, the AB&P have seemed to be aware of the issue of the divergence between burden-of-proof and burden-of-pleading rules. For example, in Asbestos, in applying the LTRA test and the chapeau test, the Panel seemed to allocate the burden of pleading to Canada, the complaining country, even though the Panel clearly pronounced that the EC, the respondent, bore the burden of proof. That is, in applying the LTRA test in its GATT Article XX (b) analysis, the Panel only examined the alternative courses of action which were asserted by Canada in spite of the possible availability of other alternatives. 231 Furthermore, in its chapeau analysis, the Panel found that the application of the French measure did not result in "unjustifiable discrimination", "arbitrary discrimination", or "disguised restrictions on trade" since Canada did not argue that the application of the French measure amounted to according less favorable treatment to foreign products.232 These allocations of the burden of pleading under the LTRA test and the chapeau test make sense economically. In applying these tests, it is the complaining countries that can "narrow" the scope of their claims. If the AB&P impose on defending countries the burden of pleading, the defending countries would have to anticipate all possible claims and produce evidence refuting all of them. The AB&P would also have to examine all the points. On the other hand, if the burden of pleading is imposed on the complaining countries, the defending countries merely have to refute the arguments made by the complainants; in addition, the issues which the AB&P have to examine are also limited to these points. In sum, process costs can be saved by imposing the burden of pleading onto the plaintiffs. On the other hand, as we discussed above, the burden of proof should be assigned to defending countries in the context of GATT Article XX analysis since it is unlikely that the arguments of the defending countries are meritorious and the costs of obtaining evidence regarding the issue involving GATT Article XX by defending countries would be less than those of the complaining countries. In this way, allocating the burden-of-proof and the burden-of-pleading on different parties could be justified in the context of GATT Article XX analysis.
215 See Bruce L. Hay, Allocating the Burden of Proof, 72 Ind. L. J. 651, 654 (1997). See also Lee, supra note 91,at 4-5 (1997).
216 See Hay, supra note 215, at 657-8.
217 See id. at 659-60.
218 See id. at 660-73
219 See id. at 676-7. See also Bruce L. Hay & Kathryn E. Spier, Burdens of Proof in Civil Litigation: An Economic Perspective, 26 J. Legal Stud. 413, 425 (1997).
220 See Hay, supra note 215, at 677-8. Lee points out another possible reason why the default rule is to impose the burden of proof on a plaintiff. He points out that a judgment for the plaintiff provokes the increase in process costs which does not occur when the judgment is for the defendant; the court has to calculate the amount of damages and enforce the judgment. Therefore, he argues that the defendant should bear the burden of proof only when the decrease in error costs outweighs the increase in process costs when courts impose the burden on the plaintiff. See Lee, supra note 91, at 12-14. See also William Chris. Sanchirico, The Burden of Proof in Civil Litigation: A Simple Model of Mechanism Design, 17 Int'l Rev. L. & Econ. 431 (1997) (arguing that the burden-of-proof rule functions as a single by which all potential plaintiffs, who have superior information regarding their cases, self-select to bring the only cases worth hearing for the society).
221 See Hay supra note 215, at 677. See also Lee, supra note 91, at 15-27.
222 Lee notes that `there are at least two categories of issues where defendant's access to proof is likely to be superior. First are issues involving conduct of the defendant where the plaintiff may not have been involved. ... The second category of issues for which defendant has greater access to proof involves interactions between the two parties for which the defendant has superior incentives to keep records of the transaction." ETMs normally fall into the first category. See Lee, supra note 91, at 16-17.
223 For example, imagine a case where Country A estimates that it can increase the amount of its exportation to Country B by $10 million if Country A successfully strikes down one of Country B's SPS measures. The probability that Country A prevails before the AB&P is 10 percent and that the amount at stake is $10 million. Country A is willing to complain about Country B's SPS measure to the DSB if the cost for the "litigation" is less than $1 million ($10 million x 10%). Furthermore, if Country B also estimates that the probability that they lose is 10 percent and the costs of Country B for presenting evidence which shows the consistency of its measure with the SPS Agreement is more than $1 million, Country B has an incentive to settle the case prior to the establishment of the Panel. This shows that a Member might have motivation to bring a complaint to the WTO even if its claim is not so meritorious.
224 See Salmon AB Report, supra note 107, at para. 192.
225 See Lee, supra note 91, at 6.
226 See id. at 28.
227 See id. at 28-29. See also Hay and Spier supra note 219, at 428
228 See Lee, supra note 91, at 7.
229 See id. at 7.
230 According to Hay and Spier, the reason why courts are divided on the issue of who should bear the burden of proof in the context of contributory negligence would be the "timing" of scrutinizing plaintiffs' level of care. If the courts examine both the plaintiffs' and defendants' levels of care simultaneously, there is no reason to assume that the plaintiff took due care. However, if the courts should examine the parties' levels of care sequentially (first, the defendants' levels, then the plaintiffs' levels), it could be assumed that the plaintiffs took due care when the incidents happened. For it is unlikely that the plaintiff was not careful, given that the courts find that the defendants were negligent. See Hay and Spier, supra note 219, at 428-9.
231 See Asbestos Panel Report, supra note 28, at para.8.204-8.222.
232 Id. at para.8.228.