2.1 The Panel's decision: general exposition

The focus of the dispute, which was triggered by complaints from Argentina, Canada, and the US, was the legality of various measures, introduced by the EC as a whole and independently by several Member States, with respect to the import and placing on the market of certain bio-products.Footnote ⁶ In particular, the complainants challenged three categories of EC measures:Footnote ⁷ a general EC moratorium on approvals of biotech products, various product-specific EC measures affecting the approval of specific biotech products, and various EC member State safeguard measures prohibiting the import and/or marketing of specific biotech products. I will focus in this article on the third category, which is more interesting because it brings to fore the problematic of the precautionary principle within the context of the SPS Agreement.Footnote ⁸ The safeguard measures were adopted by EC member States on the basis of Article 16 of Directive 90/220 (later replaced by Article 23 of Directive 2001/18) and Article 12 of Regulation 258/97.Footnote ⁹ The measures targeted several types of genetically modified (‘GM’) crops, such as T25 Maize, Bt0176 Maize, and Topas oilseed rape, which provided the crops with novel qualities, in particular, an increased tolerance to certain herbicides and a natural insecticide property.Footnote ¹⁰ These provisions authorize Member States to provisionally adopt safeguard measures that prohibit trade in biotech products, despite the existence of Community-wide marketing approval. The safeguard measure must be supported by a new risk assessment demonstrating that the bio-product constitutes a risk to human health or the environment.Footnote ¹¹ For each of the disputed safeguard measures, the relevant EC scientific committee reaffirmed the earlier assessment that the relevant products did not present any risk.Footnote ¹²

2.2 The Panel's view of the precautionary principle

The Panel referred to the precautionary principle (‘PP’) in two contexts: first, in an inquiry into the status of the PP as a general principle of international law; second, in its analysis of article 5.7 of the SPS Agreement – the SPS ‘version’ of the PP. Consider, first, the Panel's general remarks about the PP.Footnote ¹³ Following the Appellate Body's ruling in EC–Hormones,Footnote ¹⁴ the Panel noted the continuing disagreement with respect to the meaning of the PP among legal scholars and the lack of authoritative decision by an international tribunal recognizing the PP as a principle of general or customary international law.Footnote ¹⁵ Given that background, the Panel decided that it does not need to take a position on the status of the precautionary principle for the purposes of disposing the current dispute.Footnote ¹⁶ I agree with the Panel that the current formulas of the PP are too vague to establish a general principle of international law.Footnote ¹⁷ The numerous references to the PP in various environmental treaties and declarationsFootnote ¹⁸ have made little to remove this vagueness. However, the indisputable vagueness and inconsistency of the PP does not absolve the WTO tribunals, I will argue, from the duty to develop an intelligible interpretation of this principle within the WTO realm. As we shall see, the Panel has used a highly literal interpretative technique – trailing closely the text of Article 5.7 and the other relevant provisions of the SPS Agreement (2.2 and 5.1).

Let me start by briefly describing the Panel's analysis of Article 5.7, moving then to my critique.Footnote ¹⁹ Article 5.7 of the SPS Agreement reads as follows:

Following the Appellate Body jurisprudenceFootnote ²⁰ the Panel noted that that there are four requirements a Member must meet in order to adopt and maintain a provisional SPS measure as contemplated in Article 5.7:Footnote ²¹

1. (a) the measure is imposed in respect of a situation where ‘relevant scientific evidence is insufficient’;

2. (b) the measure is adopted ‘on the basis of available pertinent information’;

3. (c) the Member seeks ‘to obtain the additional information necessary for a more objective assessment of risk’; and

4. (d) the Member reviews the measure ‘within a reasonable period of time’.

The Appellate Body emphasized that these four requirements are cumulative in nature. Thus, ‘Whenever one of these four requirements is not met, the measure at issue is inconsistent with Article 5.7.’Footnote ²² In Japan–Apples the Appellate Body distinguished between the first two requirements and the next two. While the Appellate Body does not say so explicitly, the separation between these two sets of requirements seems to be based on a distinction between substantial and procedural understandings of the precautionary principle. Article 5.7 first sentence deals with the substantive element; it defines a set of circumstances in which it may be justified to use precautionary measures. Article 5.7 second sentence deals with the procedural element. It prescribes several actions, which should be taken once it is concluded that the use of precautionary measures is justified.

The Panel then made several observations about the interpretation of Article 5.7 and its relation to Articles 5.1 and 2.2 of the SPS Agreement,Footnote ²³ starting with the relationship between Article 5.7 and Article 2.2. Article 2.2 provides:

The Panel noted that Article 5.7 is not an exception in the nature of an affirmative defence but a qualified right.Footnote ²⁴ Characterizing Article 5.7 as a qualified right rather than an exception means that if a challenged SPS measure was adopted and is maintained consistently with the requirements of Article 5.7, the obligations in Article 2.2 would not be applicable.Footnote ²⁵ This characterization, the Panel noted, also has implications for the allocation of the burden of proof. Where a complaining party alleges that an SPS measure is inconsistent with Article 2.2's requirement not to maintain SPS measures without sufficient scientific evidence, it is incumbent on the complaining party to establish a prima facie case of incompliance with Article 5.7. Article 2.2 will apply only if such inconsistency is demonstratedFootnote ²⁶ (this burden arises only when the responding party invokes Article 5.7).Footnote ²⁷

The Panel then considered the relationship between Article 5.1 and Article 5.7. Article 5.1, which should be read together with the Annex A(4) definition of risk assessment, provides:

The panel noted, based on the second sentence of Article 5.7, that SPS measures which have been provisionally adopted pursuant to Article 5.7, are not subject to the risk assessment requirement of Article 5.1 (as defined in Annex A(4)).Footnote ²⁸ The Panel noted however, that Article 5.7 second sentence, which refers to ‘a more objective assessment of risk’, should be understood as requiring provisional measures adopted under Article 5.7 to be based on a risk assessment, which is ‘different in nature from the kind of risk assessment envisaged in Annex A(4)’, namely, ‘a risk assessment which takes into account available pertinent information’.Footnote ²⁹ With respect to the question of burden of proof the Panel placed again the burden of demonstrating that the challenged measure is inconsistent with Article 5.7 on the complaining party.Footnote ³⁰ The burden arises, however, only when the responding party invokes Article 5.7 explicitly in response to the claim of violation under Article 5.1.Footnote ³¹

The Panel does not provide us, however, with a general framework for understanding Article 5.7, clinging to its casuistic interpretative technique. One can draw, though, two general ideas from the Panel's decision. First, the Panel is of the view, so it seems, that Article 5.7 requires Members seeking to invoke it to perform a unique type of risk assessment, which is ‘different in nature from the kind of risk assessment envisaged in Annex A(4)’, namely, ‘a risk assessment which takes into account available pertinent information’.Footnote ³² Second, the Panel also explained what is meant by the term ‘insufficient scientific evidence’, which is referred to in article 5.7 first sentence. It might be worthwhile to quote the Panel's explanation, since it plays a crucial role in its subsequent analysis:Footnote ³³

After explicating the relationship between article 5.7 and articles 2.2 and 5.1 the Panel moved to apply this framework to the case at hand. The Panel began its analysis by examining whether the relevant safeguard measures meet the requirements of Article 5.1, notably the requirement to base SPS measures on a risk assessment.Footnote ³⁵ If the relevant safeguard measure meets the requirements set out in Article 5.1, the Panel notes, there would be no need to examine the Complaining Parties' claims under Article 5.7. If it is found, however, that the safeguard measures do not meet the requirements set out in Article 5.1, the Panel would need to proceed to examine whether the measures are consistent with Article 5.7. If the safeguard measures were consistent with Article 5.7, Article 5.1 would not be applicable, leading to the conclusion that the EC has not acted inconsistently with Article 5.1. Conversely, if the safeguard measures were inconsistent with Article 5.7, Article 5.1 would be applicable and, in view of the assumed fact that the measures do not meet the requirements of Article 5.1, the Panel would need to conclude that the EC has acted inconsistently with its obligations under Article 5.1.Footnote ³⁶

The Panel began its analysis, then, by considering the consistency of the different safeguard measures taken by several EC Member States with Article 5.1 and the definition of ‘risk assessment’ in Annex A(4). The Panel divided its analysis into two parts. First, it examined whether a risk assessment within the meaning of Annex A(4) and Article 5.1 has been conducted, with respect of all the products subject to the safeguard measures.Footnote ³⁷ Second, to the extent that a risk assessment was found to exist, the Panel went on to examine whether each measure was ‘based on’ the relevant risk assessment. This requirement, the Panel noted, has been interpreted as meaning that ‘there must be a rational relationship between a risk assessment and the SPS measure taken, or in other words, that the results of the risk assessment must “sufficiently warrant” or “reasonably support” the SPS measure at issue’.Footnote ³⁸

It is beyond the scope of this article to elaborate on the Panel's detailed assessments of each of the safeguard measures. Let me just summarize its general conclusions. The Panel has found that at least some of the documents which were relied upon by the Member States were risks assessments in the meaning of Annex A(4) and Article 5.1.Footnote ³⁹ However, it found that none of the disputed safeguard measures, which imposed a complete prohibition on the introduction of the relevant bio-products to the territories of the Member States in question, satisfied the rational relationship requirement of Article 5.1.Footnote ⁴⁰

The Panel then went on to examine whether the disputed measures satisfy the requirements of Article 5.7, and reached the conclusion that none does. The Panel's reasoning was quite simple, although, as I will demonstrate in the next section, it leads to an incoherent conception of Article 5.7 and the Precautionary Principle. The Panel based its conclusion on Article 5.7 first prong, which requires that the measure in question will be imposed in respect of a situation where ‘relevant scientific information is insufficient’. The Panel recalled its earlier observation that ‘relevant scientific evidence’ will be ‘insufficient’ ‘if the body of available scientific evidence does not allow, in quantitative or qualitative terms, the performance of an adequate assessment of risks as required under Article 5.1 and as defined in Annex A to the SPS Agreement’.Footnote ⁴¹ The Panel noted that since in each of the cases it has found that there was at least one adequate risk assessment, ipso facto, one cannot argue that the information – with respect to each of the disputed GM products – was insufficient. Since, as the Appellate Body ruled, whenever one of Article 5.7's four requirements is not met, the measure at issue should be seen as inconsistent with Article 5.7,Footnote ⁴² the measures were found to be inconsistent with Article 5.7.Footnote ⁴³

2.3 A critique of the Panel's decision

The Panel's protracted and highly literal reasoning makes it somewhat difficult to decode and criticize its decision. Reading it one has the feeling of entering a literary labyrinth. Indeed, the key problem of the Panel's decision, which ultimately makes it both internally incoherent and unhelpful as a guideline for regulators, is that it is not driven by a comprehensive conception of the precautionary principle. There is a compelling force in the Panel's initial comments about the vagueness and inconsistent usage of the PP.Footnote ⁴⁴ The argument that Article 5.7 should be interpreted as a stand-alone – WTO unique – realization of the PP is also persuasive. However, these statements do not release the Panel, as I argued above, from the onus of developing an independent conception of the PP within the WTO universe. The Panel's failure to develop such independent vision, and its casuistic interpretative approach, have generated an internally incoherent portrait of the PP.

The Panel's view of the possible applications of the PP within the SPS Agreement – whether in the context of Article 5.1 or Article 5.7 – does not provide a basis for invoking the PP. Let me explain. The Panel's ultimate rejection of the EC Article 5.7 argument was based, as noted above, on the first prong of Article 5.7, the requirement of ‘insufficient scientific information’. The Panel's interpretation has emptied Article 5.7 of any practical meaning. On the one hand, the Panel noted that Article 5.7 does not license arbitrary or capricious actions, since it still requires governments to perform some kind of risk assessment, although of a different nature.Footnote ⁴⁵ Presumably, such special-type risk assessment should be based on some information – otherwise it will constitute an act of pure (unsubstantiated) regulatory force. However, the Panel does not provide an explanation of what such special type risk assessment may look like, or what kind of information is required, other than saying that it is not the kind of information required to perform ‘normal’ risks assessment.Footnote ⁴⁶ Lacking a clear understanding of the meaning of these special categories of information and risk assessments, one can reasonably assume that almost any risk assessment performed by regulators would qualify as a risk assessment under article 5.1, causing the state invoking Article 5.7 to lose, instantly, its Article 5.7 rights. Indeed, that is exactly what happened on this case. This conclusion is strengthened when one reads the Panel's comments in Annex K, where it reminds us that Article 5.1 has been broadly interpreted by the Appellate Body allowing both quantitative and qualitative analyses to qualify as ‘proper’ risks assessments, and emphasizing that a proper risk assessment must also be sensitive to the fact that the risk to be evaluated is ‘not only risk ascertainable in a science laboratory operating under strictly controlled conditions, but also risk in human societies as they actually exist … in the real world where people live and work and die’.Footnote ⁴⁷ Applying this definition to the case before it the Panel notes that it was ultimately not persuaded that the scientific evidence available at the relevant time did not allow (was insufficient) for the performance of risk assessment – so defined.Footnote ⁴⁸

What about using the precautionary principle within the context of Article 5.1, without relying on Article 5.7? The Panel seemed open to that idea. It notes, in response to the EC assertion that each of the safeguard measures at issue were based on the precautionary principle, that the fact that a Member has decided to follow a precautionary approach could have a bearing on the assessment of whether an SPS measure is ‘based on’ a risk assessment as required by Article 5.1.Footnote ⁴⁹ It then goes on to explain what this ‘bearing’ might mean:Footnote ⁵⁰

Despite its seeming openness to precautionary action, this paragraph stops short of explicating the conditions under which the PP could be invoked within the context of Article 5.1. Thus, for example, what are the factors which may affect scientists' level of confidence? In a footnoteFootnote ⁵¹ the Panel mentions as a possible factor ‘a limited body of relevant scientific evidence’ – but is that not that the kind of circumstances falling under Article 5.7? Further, the Panel notes that even ‘if a Member follows a precautionary approach, its SPS measures need to be “based on” (i.e., ‘sufficiently warranted or “reasonably supported” by) a risk assessment’.Footnote ⁵² However, in its reflections on Article 5.7 the Panel observed that this article requires unique ‘risk assessment’ and reasoning structure – it seems reasonable that these unique constructs should also apply in the context of a precautionary action within the domain of Article 5.1.Footnote ⁵³ But if so – how exactly are we to understand these concepts? The Panel's further comments in Annex K do not resolve this puzzle.Footnote ⁵⁴

The upshot of all this is that we are left in the dark with respect to the conditions under which governments might successfully defend precautionary measures within the SPS Agreement, whether within the context of Article 5.7 or Article 5.1.

2.4 An alternative vision of the precautionary principle

Before exploring an alternative understanding of the PP, I would like to briefly consider an additional interpretive path, which could also resolve the apparent inconsistency in the Panel's ruling. It is possible to view the Panel's ruling as a subtle argument, demonstrating the inherent vacuity of the precautionary principle. According to this view, the PP adds nothing to existing health and environmental regulations. Current regulatory law asks regulators to follow a simple imperative: do what is best. And in doing so they have to take into account the distinct contexts in which regulatory decisions take place. Since, as the Panel emphasizes, ‘normal’ risk-assessment process can rely on qualitative data with varying degrees of confidence (not just on quantitative measurements),Footnote ⁵⁵ and can also legitimately lead to different conclusions based on the same data,Footnote ⁵⁶ it should also be able to cope with the conditions supposedly falling under the wings of the PP.Footnote ⁵⁷

This interpretation seems at odds, however, with the Panel's insistent attempts to decode the meaning of the PP within the context of the SPS Agreement. In the following I will try to offer an alternative framework for understanding the PP. The starting point for interpreting Article 5.7 is to abandon the distinction between sufficient and insufficient information as developed by the panel. The Panel's view is inconsistent with the contemporary vision of the scientific endeavor in philosophy. To a large extent the current view of science is overshadowed by Hume's deep challenge to the method of induction. Hume argued that any attempt to justify the extension of our conclusions from past observations to the future must ultimately fail. Such extension cannot be justified through demonstrative reasoning since we can always conceive of a change in the course of nature (which cannot be ruled out a-priori). Neither can we rely on probable reasoning, since any inference from experience is based on the supposition that nature is uniform – that the future will be like the past. This supposition cannot be proved by appeal to experience, because such appeal will be question begging or circular.Footnote ⁵⁸ Hume's critique, which despite being more than 200 years old has not been refuted, means that the scientific endeavor is by its very nature incomplete (Binmore, Reference Binmore, Binmore, Kirman and Tani1993; Rescher, Reference Rescher2000: 31–34). It implies, as Nicholas Rescher notes, that the content of our objective factual claims about such things as GM crops, neutrino, and Sequoia trees, will always outrun the evidence for making them. The endorsement of any such claim always involves ‘some element of evidence-transcending conjecture’ (Rescher, Reference Rescher2000: 33). The cognitive incompleteness of the scientific project means that our theories of the world are inherently fallible – there is always the risk that the discovery of new data or some conceptual innovation will bring us to revise our view of the world.Footnote ⁵⁹ The scientific propositions that are invoked in the context of the SPS Agreement are thus – by the very nature of the scientific endeavour – ‘insufficient’.Footnote ⁶⁰ Thus, the problem facing the law is not to distinguish between conditions of ‘full knowledge’ and ‘insufficient knowledge’, but between different levels of insufficiency.

The first challenge in contemplating a coherent version of the PP is to set forth a hierarchy of insufficient knowledge. Such hierarchy was proposed, for example, by Charles Weiss (Reference Weiss2003). Weiss distinguishes between 11 categories of (subjective) levels of scientific uncertainty: fundamental, rigorously proven, very probable, probable, more likely than not, attractive but unproven, plausible, possible, unlikely, and impossible (Weiss, Reference Weiss2003: 36–38). Underlying this hierarchy is an anthropological vision of scientific knowledge. Arguably this scale reflects the discourse scientists are using when ‘gauging the likelihood that a given scientific proposition will “turn out to be true”’ (Weiss, 2003: 26).Footnote ⁶¹ The precautionary principle should be invoked at some point on this schedule where scientific confidence is low. At which point exactly? This is something that science does not have an answer to. It is a decision that calls for political debate and collective decision making. The PP does not seem to apply, however, in situations of complete ignorance, i.e. where nothing can be said about the possible likelihood of an adverse event (not even with the lowest possible confidence). A further point, which should be emphasized at this junction, concerns the role of other disciplines in making precautionary judgments. If, as is argued by some philosophers,Footnote ⁶² the natural sciences do not have a privileged role in generating factual descriptions, the views of other disciplines (e.g., economics, sociology, anthropology, esthetics, etc.) should be taken into account in the regulatory process of GMOs. And this requires further reflection on these fields' unique hierarchies of insufficient knowledge.

This discussion facilitates the development of a prescriptive interpretation of the PP. The PP prescribes, under this view, the conditions under which it may be legitimate to base regulatory measures stalling the introduction of new technology on more ‘risky’ inferences, i.e. inferences that are more likely to prove wrong in the future. There are two ways in which this general formulation can be interpreted. First the precautionary principle can be understood as a principle that assigns high negative weights to (sufficiently) undesirable outcomes in situations in which the scientific understanding of the causal paths associated with a new technology is (sufficiently) low (Peterson, Reference Peterson2006: 600). This view means, however, that if the potential benefits of a proposed activity/product are high enough, the PP will allow it to proceed. A second interpretation of the PP conceives it as setting certain thresholds of scientific uncertainty and hazardous potential, which, once met, would allow regulators to take appropriate regulatory actions (e.g., impose a ban on the introduction of a new technology), irrespective of the costs – measured in terms of benefits foregone (Ervin and Welsh, forthcoming: 9; Kriebel et al., Reference Kriebel, Tickner, Epstein, Lemons, Levins, Loechler, Quinn, Rudel, Schettler and Stoto2001: 875). The second interpretation of the PP is more radical, then, in terms of the sacrifices it is willing to make for safety (measured by the willingness to forego potential benefits). Science does not have an answer to the questions of ‘what weights’ or ‘what thresholds’. These questions belong to the political domain. In taking any of these approaches, the possibility that GMOs may be found not to be risky should be explicitly acknowledge. And this means that by adopting precautionary measures society may miss significant benefits. There is a price for being careful; it is in this sense that the precautionary approach is risky.Footnote ⁶³

One of the difficult questions raised by the PP's second interpretation concerns the determination of the ‘hazard’ threshold. There is some discussion of this question in the Appellate Body's EC–Hormones ruling. The Appellate Body noted that a panel charged with determining whether ‘sufficient scientific evidence’ exists to warrant a particular SPS measure ‘may … and should, bear in mind that responsible, representative governments commonly act from perspectives of prudence and precaution where risks of irreversible, e.g. life-terminating, damage to human health are concerned’.Footnote ⁶⁴ The Equator Principles (2006) provide another formulation, in a different context, by imposing stricter EIA requirements on projects ‘with potential significant adverse social or environmental impacts that are diverse, irreversible or unprecedented’.Footnote ⁶⁵

Article 5.7 does not determine which of the foregoing interpretations of the PP is the ‘correct’ one or how they should be implemented in practice. However, its language is broad enough to accommodate both. Importantly, both interpretations emphasize the role of political deliberation in the decision-making process regarding GMOs. To clarify this argument it may be useful to apply the foregoing interpretations to our case. Dr Snow, one of the experts that gave testimony before the Panel, highlighted several risks that may be associated with the GM crops discussed in this case. First, they may exacerbate weed problems. Second, they may cause direct or indirect harm to non-target species. Finally, they may lead to the use of more herbicides or insecticides, including more toxic ones, potentially harming flora, fauna, and human health.Footnote ⁶⁶

These risks were recognized as scientifically based risks, rather than mere scaremongering.Footnote ⁶⁷ Given a certain allocation of negative weights (PP's first interpretation), or certain thresholds of scientific uncertainty and hazardous potential (PP's second interpretation), this risk-portfolio could have provided a ‘sufficient’ or ‘reasonable’ support to the regulatory actions of the EU Member States (the introduction of the safeguard measures). Note that under this interpretation the Member States' decision could not be portrayed as irrational, because there would be no flaw in its deductive structure or internal consistency. To be legitimate such a decision would have to satisfy a different kind of requirement: political openness. This requires the regulatory decision to build on some political deliberation. It is true that such decision may not be optimal and in that sense could be inconsistent with the precepts of cost–benefit analysis – but this is not the test imposed by Articles 2.2 and 5.1