"At War with Japan"

On the morning of July 2, 1987, Japanese viewers watching the breakfast news saw something that did not look like American politics. On a patch of grass on the west side of the Capitol, six members of Congress stood in shirtsleeves around a small Toshiba radio set on a wooden block. They took turns swinging a sledgehammer. Wood splintered, plastic flew, the radio collapsed into a cluster of green circuit boards and broken speaker cones. Don Ritter, a Pennsylvania Republican, swung first. Helen Bentley of Maryland, a former maritime journalist with no patience for diplomatic euphemism, told the cameras that “treachery by any other name is still treachery,” and added that the other name, today, was Toshiba. Elton Gallegly took a swing. Duncan Hunter, a hawkish Californian who would spend a decade trying to revoke the company’s right to sell anything in the United States, lined up behind him. Staffers and reporters pressed closer for the photo op. Somewhere a Japanese network correspondent was patching it through to Tokyo.

The radio that died on the Capitol lawn cost less than a tank of gas. The grievance behind the sledgehammers was not really about radios. It was about submarines, propeller blades, computer-controlled milling machines, and a larger anxiety the radio happened to embody: that an American ally, sheltering inside the American security umbrella, had spent two decades extracting the United States’ most strategic industry from under it, and that the political class was finally going to do something about it. The radio was a prop. The story was the policy.

That story had begun two years earlier, on June 14, 1985, when the Semiconductor Industry Association filed a petition with the Office of the United States Trade Representative under Section 301 of the 1974 Trade Act. Section 301 was an obscure provision designed to give the executive branch a hammer against foreign trade practices it deemed “unreasonable” or “discriminatory.” It had almost never been used against a major ally. The SIA petition argued that Japan’s semiconductor market was structurally closed to foreign producers, that the keiretsu trading networks systematically locked Americans out, and that the result of three decades of MITI’s industrial policy was a market in which foreign chips held about ten percent share while Japanese chips held nearly a quarter of the American market. The petition asked the Reagan administration to use Section 301 to pry the Japanese market open.

Behind the filing was a small group of CEOs who had spent the previous two years quietly turning themselves into a Washington lobby. Robert Galvin, the second-generation chairman of Motorola, had begun warning anyone who would listen that Japanese pricing was not a market accident; he told one Washington audience that anyone in the electronics business who did not think the bell was tolling for them was deaf. Sporck had thrown himself into the lobbying with such intensity that his board worried he was neglecting his own company. Jerry Junkins, who had taken over as president of Texas Instruments in 1985, brought to the cause the only one of the big American firms still making commodity DRAM in volume. Grove of Intel, who had decided privately that Intel was getting out of memory entirely, nevertheless backed the petition: a closed Japanese market hurt every American producer, including the ones moving up the food chain into microprocessors.

The SIA framed the petition carefully. It did not ask for tariffs. It did not ask for protection. It asked for reciprocity: open access to a market that, the petition argued, the Japanese government had spent twenty years closing. This was a politically shrewd choice. Reagan had spent his first term as a vocal opponent of protectionism. His trade representative, Clayton Yeutter, was a free trader by instinct who had been initially skeptical of the SIA’s case. But by 1985 the trade deficit with Japan had blown past $40 billion, the manufacturing belt was bleeding jobs, and Congress was drafting tariff bills that the administration regarded as worse than any negotiated solution. A targeted Section 301 action, focused on a single strategically resonant industry, gave Yeutter a way to take the political pressure off the broader trade relationship by giving Congress a win.

Ten days after the SIA filing, Micron Technology, a small Idaho memory-chip maker nobody outside Boise had previously paid much attention to, filed an antidumping petition of its own, accusing six Japanese firms (Fujitsu, Hitachi, NEC, Oki, Toshiba, and Mitsubishi) of selling 64-kilobit DRAMs in the United States below cost. In September, Intel, Advanced Micro Devices, and National Semiconductor filed a parallel petition on EPROMs. In December, the Commerce Department took the unusual step of self-initiating an antidumping investigation against Japanese 256K DRAMs. The cases stacked on top of one another. They were technically distinct, Section 301 being about market access while antidumping was about predatory pricing, but in Tokyo they read as a single coordinated assault.

Inside the Reagan administration the negotiations fell to Clyde Prestowitz, a counselor to Commerce Secretary Malcolm Baldrige whose Trading Places would, the following year, become one of the canonical accounts of the era. Prestowitz had spent his early career in Japan as a businessman, spoke fluent Japanese, and arrived at Commerce convinced that the Japanese trade bureaucracy operated on rules American negotiators had never bothered to learn. He found a counterpart in MITI in Makoto Kuroda, a tough, voluble vice minister who openly mocked American negotiators and was happy to let conversations bog down for months. Trading Places reads as a diplomatic procedural in which the American side gradually realized that no concession by Tokyo was self-executing, that “we’ll try our best” meant nothing, and that the Japanese delegation was unwilling to put market-share commitments in any document anyone would ever read.

The result, signed on September 2, 1986, was the United States–Japan Semiconductor Trade Agreement. The public text contained two large concessions. First, Japanese firms would no longer sell DRAMs and EPROMs in the United States below “fair market values” calculated by the Commerce Department, effectively imposing floor prices on Japanese memory chips in the U.S. market. Second, Japan would monitor its producers’ export prices to third countries, including Europe, Korea, and Hong Kong, to prevent them from dumping there what they could no longer dump in the United States.

The third concession was nowhere in the public text. In a side letter Tokyo would for years refuse to acknowledge, the Japanese government recognized the U.S. industry’s “expectation” that foreign-affiliated companies’ share of the Japanese semiconductor market should grow to “slightly above 20 percent” within five years. The number was famously hedged. It was an expectation, not a guarantee. Tokyo was recognizing the American expectation, not committing to deliver it. The Japanese negotiators considered the language ambiguous enough to be deniable. The American negotiators considered it concrete enough to be enforceable. Both sides walked out believing they had outwitted the other. Prestowitz, who watched the language being drafted, later wrote that the side letter was the moment the United States crossed into managed trade, the open admission that markets, left alone, were not going to deliver the outcomes Washington required.

Within weeks of signing, the agreement began to fail. Japanese producers, sitting on enormous DRAM inventories and facing a depressed global memory market, kept shipping chips to Singapore and Hong Kong at prices well below the new American floor. Brokers bought them there and routed them into the United States as gray-market imports. Floor prices in the U.S. ratcheted American computer-makers’ costs up while Japanese gray-market chips priced at a fraction of those floors leaked in through the back door, often through subsidiaries of the very firms that had signed the deal. Meanwhile, foreign-share data inside Japan refused to budge. Through 1986 and into early 1987, the share of foreign semiconductors in the Japanese market hovered around 8.5 percent, well below the level it had been when negotiations began.

By the spring of 1987 the political calculation in Washington had shifted. Reagan had spent capital on the agreement; Congress was now demanding to see results. On March 27, in a memo to his cabinet, Reagan announced that the United States would impose 100 percent ad valorem tariffs on $300 million worth of Japanese electronics, including laptop computers, color televisions, and certain power tools, in retaliation for what the administration described as Japan’s failure to enforce both halves of the agreement. The tariffs took effect on April 17, 1987. They were the first significant tariffs the Reagan administration had imposed on any major trading partner. They were calibrated to roughly match the export revenue Japanese firms had gained by violating the deal, a sum the USTR had calculated by extrapolating from third-country dumping data, and they applied to consumer electronics rather than to chips themselves, on the theory that a semiconductor tariff would just punish American computer-makers.

Reagan’s statement was characteristically unembarrassed. The administration, he said, remained committed to free trade. The tariffs were not a repudiation of free trade but its defense, taken with great reluctance after Japan had failed to meet commitments it had freely entered. The line was rehearsed. The reality was that the most ideologically free-market president of the postwar era was, for the first time in his presidency, slapping a hundred percent tariff on a treaty ally over a dispute about computer memory.

Then came the Toshiba scandal, which detonated underneath the negotiations like an unrelated mine.

In late 1986 a former employee of a Japanese trading company called Wako Koeki walked into the U.S. embassy in Tokyo and explained why American sonar operators had been mystified for the past two years by the sudden quietness of the new Soviet submarine fleet. Toshiba Machine, a subsidiary of Toshiba, had between 1982 and 1984 sold the Soviet Union eight large nine-axis computer-controlled milling machines, including its MBP-110 model, paired with digital controllers from the Norwegian state-owned firm Kongsberg Vaapenfabrikk. The combination was sophisticated enough to mill the curved, multi-axis propeller blades that determined a submarine’s acoustic signature. With the new machines in service, NATO hydrophone arrays began picking up Soviet boats far closer to coastlines than they had ever been able to detect them. American defense officials would later estimate that detection range against the affected Soviet classes had been cut roughly in half. The sale was a flagrant violation of the Coordinating Committee for Multilateral Export Controls, the Cold War regime through which the Western alliance restricted dual-use exports to the Soviet bloc, and Toshiba and Kongsberg had falsified the export paperwork to conceal the equipment’s capabilities.

The story broke in the American press in the spring of 1987, almost simultaneously with the imposition of the tariffs. To members of Congress who had spent two years watching Japanese chip-makers dump memory at American firms, the Toshiba revelation was confirmation of a worldview. The same Japan that was strangling Silicon Valley was also arming the Soviet submarine fleet. Lobbyists for the American semiconductor industry did not need to draw the connection out loud; the press did it for them. Don Ritter, the Pennsylvania congressman who would lead the sledgehammer ceremony, told reporters that Toshiba had “hurt every American sailor and submariner.”

On June 30, 1987, the Senate passed a measure banning all Toshiba imports into the United States for two to five years. The vote was 92 to 5. The House was expected to follow. The next morning, Ritter, Bentley, Gallegly, Hunter, and a handful of colleagues walked out onto the Capitol’s west lawn with their sledgehammers. The footage played in Tokyo for months.

Toshiba responded with one of the largest corporate lobbying campaigns Washington had ever seen. The company’s chairman and president both resigned, an act of public penance that startled American observers unfamiliar with Japanese corporate ritual. Toshiba paid for full-page apology ads in fifty American newspapers, reading “Toshiba Corporation extends its deepest regrets to the American people.” It hired a battalion of former senators and congressmen as lobbyists. It mobilized its American customers, including hospitals using Toshiba CT scanners and major buyers like Bell Labs, Caterpillar, and Honeywell, to flood Capitol Hill with letters arguing that punishing the parent company would devastate American businesses. By the time the trade bill that contained the Toshiba sanctions reached Reagan’s desk in 1988, the punitive language had been narrowed dramatically: the import ban applied only to Toshiba Machine, the subsidiary directly involved, not to Toshiba Corporation as a whole. The Washington Post called it the most successful foreign-corporate lobbying campaign in American history.

The sledgehammers had done their work. They had not killed Toshiba. They had done something more lasting. They had fused, in the American political imagination, two stories that had been operating on separate tracks. Story one: Japanese industrial policy was destroying an American strategic industry. Story two: Japan, for all its claims of pacifism and alliance loyalty, would sell anything to anyone if the price was right. After the summer of 1987 those stories were the same story. The semiconductor agreement, which had begun as a narrow trade-policy exercise about pricing and market access, was now sitting inside a larger frame in which chips, submarines, and national security were inseparable.

By November 1987, the Reagan administration could announce, with relief, that Japanese third-country dumping had largely ceased and that an additional $84 million of the original tariffs would be suspended. A first tranche of $51 million had been lifted in June. Roughly half the original $300 million in tariffs remained in place, and the foreign-share clause of the side letter would not actually be met until a single quarter at the end of 1992. The immediate crisis had passed. The agreement had stabilized at the price of admitting publicly that the United States now treated semiconductors as a strategic industry whose market outcomes the government would intervene to manage.

That admission was the deeper consequence of the period from 1985 to 1987. The economist Kenneth Flamm, who would write the most rigorous reconstruction of the dispute in his book Mismanaged Trade?, argued that the agreement and its enforcement were as much symbolic as economic. The deal did not noticeably restore American DRAM market share. It raised costs for American computer-makers. It probably accelerated the rise of Korean memory firms who happily took the demand the Japanese were now constrained from filling. The symbolic act mattered. The Reagan administration, the institutional citadel of free-market ideology, had stood up and declared that semiconductors were different. They were what the Pentagon was building its post-Vietnam offset strategy on. They were what kept American allies’ weapons more accurate than Soviet ones. A country that could not make them on its own was a country that could not, in the long run, defend itself.

Once that frame was set in place, a series of subsequent moves became thinkable that would not have been thinkable a few years earlier. A government-industry consortium to subsidize American chip manufacturing went from a fringe idea to a Pentagon line item in less than two years. DARPA began funding semiconductor manufacturing equipment at a level it had never previously contemplated. Congress began holding hearings on the health of specific American suppliers in the chip equipment industry, an exercise that would, within months, fix Washington’s gaze on a small Massachusetts company called GCA, the last American maker of advanced lithography steppers, whose collapse would soon teach an unforgettable lesson about how a single chokepoint could be lost and never recovered.

The radio shards that the Capitol Hill custodial staff swept up on the afternoon of July 2 went into a trash bag and from there into history. The chip industry the radio belonged to had crossed a line. From now on, when an American executive walked into a meeting with the Department of Defense, or a Japanese trade negotiator sat down across from the USTR, or a startup in a Massachusetts mill town sized up its competitors in Tokyo, all of them would be operating inside a different kind of marketplace. It was a marketplace in which a memory chip and a submarine propeller could be argued, in front of a congressional committee, to be the same kind of thing.