When the US-Japan Semiconductor Agreement was signed in July 1986, there were analysts who said that it was too late. In 1985-86, US memory producers lost $2,000m while their Japanese competitors reportedly lost $4,000m. Intel, Mostek and Motorola left the dynamic memory chip market and the US commitment to future generation products was seriously jeopardised. The semiconductor agreement was unprecedented in both the way it was negotiated and the approach that was taken. The agreement was negotiated with the active participation of both the users and the producers. The agreement uses the exist-ing trade policy tools to respond to to Japanese targeting practices that threatened the health of both the semiconductor industry and the entire electronics industry food chain. Since the agreement was signed, it has become popular to attribute everything that happens to the semiconductor market to the agreement or its effects. Critics of the agreement range from those who argue that dumping is something good that should be encouraged – not discouraged – to those who say that the agreement has caused an increase in prices and a shortage of product. The reality is that the agreement is neither a panacea for all the ills of the US semiconductor industry nor the cause of all phenomena, natural and unnatural, that affect semiconductor users and producers. Before addressing the critics of the agreement, it is useful to recall why we want a healthy merchant semiconductor industry. What do we ask the semiconductor industry to do? Invest over 12% of sales revenues in research and development and bring new technologies to market as rapidly as they are available. Build fabrication lines that cost up to $200m apiece in order to produce a new generation of product with a development time of four to five years and a product life as short as three years. Bring down prices by 25% per year, year in and year out. Respond to demand fluctuations that can be as great as 25% up or down in one year. Live up to user expectations that every new generation of memory will decline by over 1,500% to cost $2 or less by the end of a product cycle. Bear all of the losses during times of excess capacity and maintain price reductions during periods of short supply. There is nothing sacred about the agreement. We have challenged every economist and every critic to come up with a better idea and no-one has succeeded. What do we ask the semiconductor industry to do? Do nothing? This was simply not an option in light of US law and international agreements that mandate action against dumping. More importantly, this would have caused an irreparable loss to the US economy, the loss of our semiconductor industry to unfair competition. We have done little to disguise our dislike of the US-Japanese Semicond uctor Agreement: it int erferes with free market forces, favours chipmak ers to the detriment of a much larger population of users, and is bilateral rather than multilateral. It is therefore only fair that the other side of the argument should be put, and we present here in full the Semiconductor Industry Association’s arguments, with a comment appended at the end. The most obvious alternative was to slap anti-dumping duties of up to 188% the highest margin of dumping found by the Department of Commerce. Now, with dynamic RAM imports running at over $1,000m, that would be a useful solution to the budget deficit, but it would not have helped semiconductor consumers. It would have made the United States an island of high prices and hastened the offshore movement of computer manufacturing. The agreement takes a different approach, designed to accommodate the interests of both users and producers. As long as imports are priced above cost, the Agreement has no impact on prices. In the current environment, it’s generally recognised that prices are will above fair market values. It is fair to ask what can be done about the current shortages of dynamics. If we abandoned the agreement today, would it increase the supply of RAMs by one unit? The answer is no because the shortage
of dynamics – in the first half of 1988, in common with the shortages that occurred in 1984, 1980 and 1975 – are attributable to the terribly complicated mechanism by which supply and demand of semi.pl 71 conductors are matched. On the supply side, the fab lines that exist today to make 1 megabit dynamics were built in 1985-86 when memory chip makers were being seriously hurt by overcapacity and dumping. While demand appears robust today, producers are justifiably leery of the possibility that some of it is phantom demand that arises when users place two or three orders for the same shipment in times of perceived shortage. Whatever the causes of the current problems, the outlook for memort chip availability is positive based on the projected ramp up of 1-Megabit parts, and the broad base of Japanese, US, Korean and European manufacturers that are either in production or coming on stream. In short, the semiconductor agreement appears to have encouraged, not discouraged, new suppliers to enter the memory chip market by restoring normal, competitive market forces to dynamics RAMs. Japanese semiconductor industries is that while the US industry consists almost entirely of dedicated chip manufacturers, the Japanese industry is composed entirely of major chip users who choose to sell their surplus on the world market. That being the case, the Japanese firms can afford to bear losses in times of oversupply where US manufacturers cannot. As a result, it is well-nigh impossible for the two sides to find a common interest. The Agreement clearly pilloried the Japanese as the Bad Boys, and it is difficult to believe that they are not enjoying hugely the pain that the memory chip shortage is causing US companies now, and would be the whitest of saints were they not to be taking the opportunity here and there to exacerbate the problem. The most vocal of the users who are said to support the Agreement is the US military, which fears that it may be forced to rely on imports of key chips. The Semiconductor Industry Association asks for an alternative to the agreement: the most effective alternative would be for the US chip industry to become more like the Japanese. If it is right for the US government to interfere in international trade by signing such bilateral agreements, would it be any worse for the US to order the biggest US chip users – led by IBM, General Electric and AT&T – to expand output of commodity chips to the point where they matched the Japanese in the merchant market?