Memory Loss: A U.S.-Japanese DRAMa

What lessons from the 1980s computer chip rivalry between the United States and Japan can be applied to today?

May 28, 2002

What lessons from the 1980s computer chip rivalry between the United States and Japan can be applied to today?

And the news is not limited to Japan. South Korea — Asia's other leading producer of DRAMs — is also feeling the heat. The world's number three DRAM producers, Hynix, is going broke — and Samsung, the world's DRAM champion, has no intention of buying it. When Micron, the U.S. market leader and global number two, walked away from negotiations, the Korean government told Hynix, as they say in Silicon Valley, to “get real.”

These developments are all the more remarkable since as recently as a decade ago the DRAM market was supposed to be dominated by Asia. This dramatic turnaround should remind us of some valuable lessons about how economies really work.

The inroads made by Japanese DRAM producers in the 1980s were the cause celebre of global trade policy in that decade. A variety of large U.S. industries had complaints about Japan in the 1980’s. But chip producers used their technological allure — and moved to the front of the line. Japanese memory chip producers — companies such as Toshiba, Fujitsu, NEC and the other big_name keiretsu conglomerates — were taking market share rapidly from Intel, Motorola and other U.S. companies.

Armed with cheap access to capital funds, a relatively closed economy, non-existent shareholder rights and complete vertical integration, the keiretsu could churn out chips and stick them in their end-use products until the cows came home.

The U.S. semiconductor industry yelped — and their complaint resonated with a vengeance within the political circles of Washington, D.C. Thanks primarily to large federal deficits, the U.S. trade balance had worsened significantly.

Many U.S. industries, including auto and steel, suffered. But the chip producers used their technological allure to move to the front of the line.

In fact, they became a favorite example for a variety of new theories about how economies worked. Three leading theories emerged.

One was that some industries, like memory chips, were “strategic”: DRAMs would be a platform from which the Japanese would capture the market for microprocessors — and then the production of computers themselves.

A second, and more general, theory cited the DRAM “crisis” as key evidence that “manufacturing really mattered.” In other words, America needed to have a manufacturing base (including DRAMs) — if it wanted to be a key player in the new service economy.

You couldn't design cars you didn't build — and you couldn't write software for devices you didn't manufacture, the logic went.

A third theory was that the Japanese had found a kind of magic wand — a new system of public-private partnership that was irreconcilable with American capitalism.

As a matter of sheer survival, the proponents of this view argued that Japan Inc.'s machinations had to be sealed off at the border.

Armed with all these powerful rationales, the U.S. semiconductor industry got the Reagan Administration to pressure the Japanese into a “voluntary” restraint agreement on their memory product shipments to the United States. Now, 15 years later, there isn't a new Japanese memory chip left on planet Earth. What happened?

The first faulty assumption was that DRAM chips were a Trojan Horse that would ultimately lead to the demise of the U.S. computer industry.

Still, I vividly remember being told in the late 1980s, when I was chief economist of Unisys Corporation, the high-end computer manufacturer, the following saga by the representative of a large, U.S. chip maker:

They — meaning the Japanese — had their sights initially set on taking over computer memory. Then they would swallow up the microprocessor business (his company's forte), then the low-end computers (meaning desktops) — and, finally, the high-end computers — meaning my company at the time and other U.S. mainframe makers.

Of course, the proponent of this industrial domino theory was perfectly, if sublimely, wrong. Producing a rapidly evolving product such as memory chips turned out to be like riding a tiger.

Not only was the product a commodity, but every two years or so, a new generation of more powerful chips required the Japanese to spend billions on new factories to make them.

These big overheads forced the producers to sell as many memory chips as possible for whatever price they could get — which flooded the market and kept profits low.

When the Japanese bubble burst a decade ago, and companies were suddenly pressed for cash and its close cousin, profitability, they quit the business.

Today, as the business has matured, two of the three largest DRAM producers are Korean — the aforementioned Samsung and Hynix. One is American — Micron. The fourth, Infineon, is German. DRAMs, like automobile components and other examples of vertical integration, proved to be not an advantage, but an albatross for Japanese companies.

The second breakdown of the new thinking was that, despite all the doomsday warnings, globally interconnected markets provided all the strategic linkage the U.S. economy needed after all.

Cheap Japanese computer memory allowed the U.S. information technology industry — not its Japanese counterpart — to rule the world. Unencumbered by the large investments tied up in chip production, U.S. computer companies invested in using Japanese chips to make computers cheap and ubiquitous.

Cheap Japanese computer memory also allowed U.S. software companies to write more powerful, productive and entertaining software. And it helped establish the U.S. monopoly in operating systems.

These developments illustrated a new principle — and, at the same time, the restatement of an older one — of an interconnected, information-intense world economy.

In a world in which you can have access to any resource, product, service, good, input, component or factor, the key to corporate competitiveness is to determine how best to add value to the assortment. It matters far less to control all the steps along the value chain — as long as there is easy access to the various components.

And a corollary is that, contrary to the hype of the 1980s, manufacturing doesn't really matter that much. That's not to denigrate manufacturing. But we have now seen that a country does not actually need to make memory chips in order to write and sell software — or, for that matter, produce things in order to finance or design them.

The Internet's ability to provide costless information anywhere allows manufacturing to go wherever it wants. Sure, labor costs still count.

But increasingly, manufacturing goes where there's an infrastructure of services to support it — meaning telecommunications, transportation, finance, information processing and the like.

In the information age, services matter at least as much as manufacturing does. That, coincidentally, is a pressing reason why forward-looking developing countries need to liberalize service trade and lower the cost of doing business in their locales in their very own interest.

And then, as the big finale, came the demise of Japan as Number One. A variety of forces have led to its decade-long torpor. But the most important was the very lack of openness that Japan's economic disciples initially found so enticing.

While Japan's incestuous conglomerates were attempting to dominate the entire computer “stack” through over-investment and production-at-a-loss, U.S. specialists such as Microsoft, Oracle, Cisco and Dell were efficiently attacking individual slices of that stack and dominating them.

In essence, they took the vertically-integrated “stack” — and turned it 90 degrees, by specializing in horizontal segments of the stack instead of competing to provide the entire vertical affair. The more open U.S. approach — to take what the market offers and find a way to add value to it — has run roughshod over the Japanese public-private partner and plan approach.

The failure of the economic domino theory, which led to semiconductor protection in the United States in the 1980s, reminds us that protecting domestic industries rarely does anything worthwhile — other than save the few jobs that remain. In short, one man's meat is another man's poison.

Cheap foreign steel leads to low-cost domestic cars, construction equipment and appliances. Cheap foreign textiles lead to low-cost U.S. fashion products. Cheap computer memory leads to a plethora of new electronic devices and their applications.

An open, competitive economy takes what the world offers — and turns it into something more valuable. Like the Japanese chips that no longer exist, cheap imports are not something smart economies need protection from.

— Everett Ehrlich is the Senior Vice President and Director of Research for the Committee for Economic Development.