Antidumping in Supercomputers or Supercomputing in Antidumping? The Cray-Nec Case

The U.S. antidumping case in vector supercomputers - the Cray-NEC case - resulted in the imposition of the highest antidumping duties in the entire U.S. antidumping history (454% for NEC, the main defendant). This, along with the considerable attention attracted by the dispute, would be sufficient ground for a closer examination of the case. Three additional aspects of the Cray-NEC case, however, render it particularly relevant to the economics and law of antidumping: (1) the Cray-NEC case offers an opportunity to look at cases of dumping in the context of a product where antidumping could be a likely policy option - though such a policy is probably a second best option. There are, indeed, two situations where dumping could be considered as a serious source of concern and cause of policy intervention: predatory and strategic dumping. Predatory dumping is more likely to occur where only a few firms are involved, such as in the supercomputer industry. Strategic dumping requires static or dynamic scale economies which seem also likely in this case; (2) the Cray-NEC case offers an opportunity to look at antidumping policy as a component of industrial policy. The supercomputer industry has so far been shaped by two other industrial policy tools: active subsidy through R&D funding and massive public procurement policies. The antidumping action could have been lodged as a substitute to these two instruments; (3) the supercomputer case raises key issues about antidumping procedures per se, from the point of view of both determination and enforcement. Supercomputers as a product do not seem really subject to markets. The number of transactions, realized through auctioning, is very small, and the products sold are highly diversified. In these circumstances, is it meaningful to talk about a «market» for supercomputers - hence to apply GATT antidumping rules which presuppose that there is a market? Turning to enforcement, the Cray-NEC case offers an interesting development: if Japanese supercomputers cannot be sold in the U.S. market, they can be located in Japan, and their services sold to U.S. potential buyers from there (through appropriate telecommunications). In other words, the Cray-NEC case may offer the first clear example of trade in services as a substitute to trade in goods - raising a host of issues regarding antidumping procedures