IF SEMATECH WORKS, it is likely to become an influential precedent for cooperation between the federal government and private companies in research and development. A consortium of semiconductor producers, Sematech is being established to do something about the industry's weak point -- its inadequate manufacturing technology. It is to cost $1.5 billion over the next six years, half of it public money and half from the companies. The appropriations are now making their way through Congress, which asked its Congressional Budget Office to look into the wisdom of this investment.

The American producers still dominate the world market for semiconductors, but their Japanese competitors are gaining rapidly. It's a pattern that runs through many industries. The Americans are unbeatable in engineering design, but the Japanese are far ahead in manufacturing. The Japanese firms are now outspending American companies on semiconductor research, CBO says. Normal market economics doesn't work well in the fragmented American industry, because much of this investment pays no special return to the company that makes it. When a concept is developed, word spreads fast. Sematech is being set up to achieve manufacturing processes that can set the world standard in cost and reliability. CBO's description makes it pretty clear that any risks in this public investment are clearly outweighed by the risks of doing nothing.

The federal money would come from the Defense Department, not an entirely ideal arrangement. The department is already spending several hundred million dollars a year on semiconductor research, but its highly specialized requirements do not reliably contribute to commercial efficiency. It will be important to keep Sematech insulated from the immediate interests of Pentagon procurement officials. The money can be more than justified by Defense's broad interest in a competitive American industry.

The alternative to funding Sematech, as the CBO suggests, is plain old protectionism. The administration's current attempts to protect the semiconductor industry are not working well, and will make more trouble as time passes. Putting money into technology is vastly preferable to imposing import quotas, and in more industries than this one. There should always be three criteria for federal support. The industry has to be a crucial one (sorry, shoemakers). It has to be able to draw up its own agenda for research. And it has to be willing to put up half of the money from its own pockets. Sematech meets all three conditions