In a misguided attempt to help U.S. memory chip manufacturers, who were competing against low-cost imports from Japan in the depressed chip markets of 1985 and 1986, U.S. trade policy-makers pressured the Japanese into a semiconductor trade agreement in the fall of 1986. A major objective of the agreement was to force memory chip prices up in world markets.
This has left the U.S. computer industry in a steadily worsening fix. As distributors scramble to nail down fast-evaporating memory chip supplies, dynamic random access memories, which early in 1986 retailed for less than $2.50, are currently soaring past $5 -- even $5.50 -- with no end in sight. The increase in the price of DRAMs adds up to almost $100 for a personal computer selling for $600 or $700, vastly more on the souped-up machines wielded by the growing armies of "power users" across America.
This crisis could not come at a worse moment. After a couple of years of sluggish growth, the personal computer market had finally turned around. Some U.S. manufacturers, besieged by low-cost Asian "clones" of the aging IBM PC, had struck back by introducing a new generation of computers using a new operating system, a software product known as OS/2. But along with the increased power promised by OS/2 come vastly increased memory requirements. Just running the basic OS/2 software requires triple the memory found in older generation PCs.
Large increases in memory chip prices will be a significant obstacle to the U.S. computer industry's game plan: to maintain a competitive edge over Asian clones by aggressively introducing new technology. A doubling or tripling of memory cost will greatly slow the acceptance of the next generation of memory-intensive PCs and work stations: a 10 percent increase in computer cost leads to roughly a 15 percent decline in demand. In fact, sales of new machines have been growing more slowly than expected. To many users, the increased cost seems to outweigh the potentially greater power.
For the pact to be effective, chip prices within Japan had to be racheted up (otherwise "gray market" chips could be smuggled out of Japan and offered for sale in export markets). Since its producers accounted for about 90 percent of world sales of these chips, Japan's Ministry of International Trade and Industry was handed the perfect excuse for creating a cartel to fix world memory chip prices, courtesy of our unguided missile of a trade policy. MITI could point out that the only assured method of raising prices in the highly competitive Japanese market was to impose production quotas on Japanese producers. Japanese chip makers cried all the way to the bank as reduced competition and higher prices restored to immense profitability operations that had been awash in red ink.
Had it been in the national interest to force an increase in memory chip prices, it would not necessarily have had to benefit the Japanese chip makers. Simply adding a tariff or countervailing duty to the cost of Japanese chips, whether imported as components or incorporated into assembled equipment, would have diverted the profits from price increases into the U.S. Treasury, not onto Japanese balance sheets.
Increased market prices didn't much inconvenience the Japanese electronic equipment giants, who largely build their own DRAMs. But price hikes spelled disaster for most of their American competitors, who are highly dependent on open-market memory chip purchases. Many U.S. chip users were not made aware of the implications of the pact until after it had been signed, by which time their anguished wails were too late. Even then, the full consequences were not immediately clear, for it took almost 18 months for prices to soar to truly stratospheric heights.
Worse yet, most of the U.S. semiconductor industry did not derive significant benefit from the increased prices. By the time the agreement was signed, only two American chip manufacturers (Texas Instruments and Micron Technology) sold DRAMs, and they accounted for a small share of the world market. An equally small fraction of the increased profits wrung out of the hides of U.S. chip consumers ended up in American pockets. For most U.S. chip makers, the main impact of the price hikes was vastly greater profits strengthening their Japanese competitors.
Some argued that, over the long run, higher prices would encourage U.S. chip producers to reenter the market. This may even prove true for future generations of chips (especially if our cooperative research effort in semiconductor manufacturing technology is successful in lowering U.S. production costs). But for current generations of chips, it is far too late in the product cycle for massive new capital investments to make any economic sense. Texas Instruments has turned to Hyundai, a Korean firm, for additional supplies. Intel now markets Korean producer Samsung's chips. And Motorola packages and sells DRAMs made by Toshiba. True, Motorola has announced its intention to reenter the business and manufacture DRAMs -- but using Japanese designs and technologies, and in exchange for the highly prized technology needed to manufacture its very popular microprocessor designs.
Have there been any other clear winners besides a very small slice of the U.S. semiconductor industry and the Japanese? The answer is yes: the Koreans. Prior to the pact, the fledgling Korean chip industry was facing huge losses. Unable to cover even variable costs, struggling Korean producers had actually shut down production lines for some of their older products. After prices began to climb, these mothballed chip fabrication lines were restarted. There were no restrictions on Korean imports, and thanks to the chip pact, Samsung memory chips are firmly established in the U.S. market today. Hyundai is arriving on the scene, and the Korean chip industry grows stronger by the day. Best of all, the Koreans didn't even pay for the high-priced Washington lawyers who made it all happen: our Semiconductor Industry Association footed the bill.
The writer is an economist at the Brookings Institution.