Questionable economic modelling in Byrd amendment arbitration
Geneva, 1 Sep (Chakravarthi Raghavan) - The World Trade Organization authorized eight members (Brazil, Canada, Chile, the EC, India, Japan, Korea and Mexico) to impose trade sanctions - additional import duties above the bound customs duties on products from the US, as well as suspension of application of obligations under the Anti-Dumping Agreement and the Subsidies and Countervailing Measures Agreement.
The sanctions were authorized over the US failure to implement the WTO ruling calling for change in the US law on anti-dumping to remove the Byrd amendment, which asks the customs authorities to distribute the proceeds of anti-dumping and anti-subsidy duties to those parties in the US who had initiated the complaints.
(The arbitrator’s award was briefly reported in SUNS #5637).
The original panel, meeting as arbitrators, used a complicated formula using an econometric model, to set the level of retaliation by each country as compensation for nullification and impairment, at a figure based on the anti-dumping or countervailing duties collected on the imports from the country in the previous year, multiplied by a coefficient of 0.72 or 72 percent of the amounts collected and disbursed.
The requesting parties produced a model, at the request of the arbitrators, showing a coefficient of 1.54, and the US started with a zero, and conceded up to 0.25.
According to figures provided by the US, a total of a little over $190 million was collected in the US fiscal year 2003, and of this the imports from the eight probably accounted for some $140 million. The largest share of this, about $80.5 million was on imports from Japan; the duties on imports from the EU was about $13.5 million.
According to some projections, the total levy and collections in US fiscal 2004 (ending 30 September), which will be distributed in October will probably run to a figure higher than the $190 million.
Of the original complainants and parties to the dispute, three did not seek authorization - Australia, Indonesia and Thailand.
Excepting in some rare cases, where the panellists have acted on their own (even asking the secretariat to withdraw when they met to consider their conclusions), in most disputes the secretariat, both the legal division and some of the substantive divisions, provide guidance to panels, but behind the back of the disputants.
None of the panellists are known for their particular expertise in econometric modelling, and presumably the arbitrators have relied on the WTO secretariat’s inputs and advice on this. It is not clear from the arbitration award whether this was done in the presence of the parties, thus giving them an opportunity to react and or rebut, or behind their backs after the parties had presented their case and arguments and withdrawn.
The econometric model and formula used (to calculate how much of the compensation received by a complainant would be passed through to inhibit further imports) is complicated enough. In most of the modelling exercises the socalled GTAP model, or variations of it, are used and this is a defective model for calculating such matters as the ‘compensation’. Econometric models and modelling at best may be of some use in understanding broad trends. But they cannot be mistaken for actual facts, and are unreliable for estimations, more so of counter-factual estimations of what would have happened without the restrictions.
Unlike most economists who are cautious in using modelling for trends, the WTO used the GTAP model recently in a discussion paper for a study of the effects of the end to the Agreement on Textiles and Clothing and the discriminatory quota regime that had been authorized under it. As a result, the modelling came under scrutiny.
After studying the model and the WTO’s use of it, in the textiles and clothing study, the Executive Director of the International Textiles and Clothing Bureau, Mr. Munir Ahmad, told the SUNS that the model is “incorrect and misleading at its very core”. Other models based on GTAP, or variations are equally misleading in terms of their disaggregation and classification of products and sub-products.
The models used for the arbitration are of questionable utility - in terms of the disaggregation of products on which the US customs duties have been collected. Equally questionable is the view of the arbitrators (not subject to appeal) that only a part of the funds distributed to the complainants would have been ‘passed through’ by those who received compensation to further inhibit imports and thus only this portion should be sanctioned.
That an enterprise or firm initiating a subsidy or dumping complaint, and being compensated, may use the money received to initiate other complaints and thus engage in trade harassment - a well-known practice, particularly used by the US parties to inhibit competing imports - does not seem to have figured.
The requesting parties in the arbitration seeking authorization, sought compensation for ‘nullification and impairment’ of their WTO rights.
The parties who had raised the original dispute collectively and won the case, argued that (a) the Byrd amendment, or Continued Dumping and Subsidies Offset Act of 2000 (CDSOA) as the US law is known by its title on the statute books, is a violation of the US WTO obligations, and the rights of the other members, and thus ‘a form of nullification and impairment’; ( b) that the notion of ‘benefit’ under Art. XXIII of GATT 1994 and the DSU encompassed their rights under the WTO; and ( c) they had a right to expect that the CDSOA should not exist and that each of them, together with other requesting parties, had a right to suspend concessions or other obligations up to the full amount of disbursements under the CDSOA.
The panel made a distinction between ‘violation’ of a right, and the nullification and impairment of benefits. While a violation might affect all WTO members, it does not ipso facto result in nullification or impairment of a given member’s benefits up to the ‘value’ of the violation, it said.
Outlining the two stages of a dispute settlement - establishment of nullification or impairment - and a subsequent or separate process for determining the level of benefit nullified or impaired, the arbitrators said there could be no assimilation of the violation of a right and nullification or impairment of benefit.
The arbitrators disagreed with the US that the effects of the CDSOA application in particular instances could not be taken into account in assessing the trade effects.
Citing earlier arbitration awards to take account of ‘trade effects’, while noting that previous arbitrator’s decisions based on direct trade impact are not binding precedents, they add: “However ... the ‘trade effect’ approach has been regularly applied in other Article 22.6 arbitrations and seem to be generally accepted as a correct application of Article 22 of the DSU.”
The awards of arbitrators are not appealable, and are not adopted by the DSB.
Thus, it is not clear how the three panellists, now acting as arbitrators, have come to the view that earlier awards of arbitrators are “generally accepted as a correct application” of Art.22 of the DSU.
The panel has not accepted the contention of the complainants that the purpose of the authority for suspension would be to induce compliance with the ruling, and has argued that no part of the DSU or the WTO supports an approach where the purpose of suspension is to induce compliance.
The parties requesting authorization, the arbitrators said, have used the total disbursements under the CDSOA as a proxy for an economic analysis of the CDSOA disbursements on their exports, and more generally on the competitive situation of the concerned businesses. The arbitrators have not accepted the argument that total disbursements should be used as a benchmark, rather than a ‘classical’ benchmark based on lost trade.
While recognising the use of an ‘economic model’ as “breaking new ground”, the arbitrators have claimed that based on the data and relevant economic literature of the two sides, an economic model to find the trade effects was feasible.
The US suggested a model that the arbitrators said was straightforward, but based on some assumptions that affect the input of the model, the values of elasticities and treatment of unavailable data. The US argued for a zero pass-through effect, but conceded a 25% effect as reasonable.
The requesting parties produced a model resulting in a coefficient of 1.54.
The US model was rejected because of the various assumptions involved and the lack of disaggregated data for the 7 products subject to the countervailing duties from the requesting parties in the years 2001, 2002 and 2003 - alloy magnesium, ball bearings, pasta, preserved mushrooms, silicon metal, stainless steel butt-welded fittings, and top of stove stainless steel cookware.
While the comments of each side on the other’s model was sought, the report of the arbitrators does not give any indication that its own model was posed to the parties and their views obtained.
However, the DSU and its panel system, and even more the arbitrators, function like the English medieval Star Chamber - and legal or judicial norms don’t apply!
The arbitrators have cited, for comparisons, the effects under the GTAP model, and the US applied GTAP model (neither of which they have used). The GTAP model has 44 sub-headings or sub-sectors, the North American Industry Classification (NAIC) has eleven.
No concordance can be established between the two models - unless one can get at the data at 8-digit level. The WTO uses the 6-digit Harmonised System. Only the US could have provided at a disaggregated level, in terms of the sub-products on which dumping duties were levied, data that would have been of use.
But the arbitrators chose not to demand this, or draw an adverse inference if the US did not produce it.
Instead, using the NAIC at 3-digit level, they have identified (from data provided by the US?) 17 sub-categories where the CDSOA has been redistributed: miscellaneous manufactures, n.e.s (not elsewhere specified), primary metal, food, fabricated metal, chemicals, non-metallic mineral, transport equipment, plastics and rubber, paper, fish products, non-electrical machinery, agriculture, textiles mill products, textiles and fabrics, wood production and furniture and fixtures.
They then applied their own modelling, to estimate the trade damage.
It has used the concept of ‘pass-through’, the extent to which a beneficiary will use the payments to reduce its own price on the market, and thus affect competitivity, and the compensation is set on this basis.
Twelve values for elasticities and pass throughs for three years have been used to get an averaging and the arbitrators say, “based on our analysis, we have estimated the trade coefficient of the disbursements to be 0.72.”
One charitable way perhaps of looking at the arbitrator’s award is that the requesting parties suggested a model and an exercise that would produce a 1.54 coefficient. The US started with zero and then conceded a 0.25. The arbitrators struck a mean of sorts and gave a 0.72 coefficient - cloaking it in some mumbo-jumbo of econometric modelling and calculations that most trade ambassadors and negotiators won’t even understand!
In some concluding remarks, the arbitrators say:
· the DSU does not expressly explain the purpose behind the authorization of the suspension of concessions or other obligations. The general obligation to comply with DSB recommendations and rulings seems to imply that the suspension of concessions or other obligations is intended to induce compliance. However, what exactly may induce compliance is likely to vary in each case in the light of a number of factors including, but not limited to, the level of suspension of obligations authorized.
· but the requirement that the level of such suspensions remain equivalent to the level of nullification or impairment suffered by the complaining party seems to imply that the suspension of concessions or other obligations is only a means of obtaining some form of temporary compensation, even when the negotiation of compensation has failed.
Thus, it is not completely clear what role is to be played by the suspension of obligations in the DSU and a large part of the conceptual debate in these proceedings could have been avoided if a clear ‘object and purpose’ were identified.
· The WTO dispute settlement system authorizes Members to challenge a law as such, irrespective of whether it has been applied or not. The ‘classical’ approach based on an assessment of trade effects of a measure may not always contribute to the identification of the actual level of nullification or impairment, in particular if no instances of application had arisen at the time. This may be because the trade impact of a measure may be difficult to assess due to the lack of verifiable figures. While parties share a duty to cooperate with the Arbitrator in the establishment of facts, there is no reason a priori to sanction the requesting party or the respondent if supporting figures are difficult or impossible to find.
“We believe this is a situation that has to be addressed in order to reach a decision on what may be achievable through recourse to suspension of obligations in such cases.”
*In this arbitration, we have interpreted the concept of nullification or impairment, inter alia, from the terms of Art.XXIII of GATT 1994 and Art.3.8 of the DSU.
We believe however, from the extensive discussion of this concept by the parties, that the actual meaning of this provision is disputed and needs to be addressed in the appropriate forum.”
The arbitrators add that “as a consequence of following the approach that each party would be granted the right to suspend obligations exclusively in relation to its own exports is that there will remain disbursements under the CDSOA in respect of goods from other Members and non-WTO Members for which no suspension of concessions or other obligations have been authorized.”
Perhaps this observation may become an inducement for the US Congress not to act to comply: there is a residual benefit to the US! – SUNS5638a
[c] 2004, SUNS - All rights reserved. May not be reproduced, reprinted or posted to any system or service without specific permission from SUNS. This limitation includes incorporation into a database, distribution via Usenet News, bulletin board systems, mailing lists, print media or broadcast. For information about reproduction or multi-user subscriptions please contact: email@example.com