During the COVID-19 pandemic many countries experienced difficulty obtaining the semiconductors that are vital for smartphones, computers, cars, artificial intelligence, cybersecurity, and many other applications. This column looks at how Asia gained comparative advantage in this sector and identifies lessons for countries seeking to promote domestic semiconductor manufacturing.
Semiconductors are vital for smartphones, computers, cars, artificial intelligence, quantum computing, cybersecurity, and other applications. An economy without access to semiconductors would seize up and a military without access would be vulnerable. Semiconductor manufacturing is concentrated in East Asia. During the COVID-19 pandemic many countries experienced difficulty obtaining semiconductors and they now desire to relocate production closer to home. How did Asia gain comparative advantage in this sector? Are there lessons for countries seeking to promote domestic semiconductor manufacturing?
Protectionism and the Japanese and Korean semiconductor industries
John Bardeen and Walter Brattain invented the transistor in New Jersey in 1947. Tadashi Sasaki was in New Jersey at the time and had worked on precursors to the transistor with Stanford Professor Karl Spangenberg (Aspray 1994). Sasaki envisioned adding transistors to an integrated circuit to produce pocket calculators for his company Sharp. His engineers studied calculator technology at Osaka University. Sasaki sought complementary metal oxide semiconductors (CMOS) chips to save power. Japanese companies declined to supply these, so Sasaki convinced the American company Autonetics to produce them. Autonetics had lucrative military contracts, and manufacturing chips for calculators generated low margins. Sasaki told Autonetics they would learn by doing and earn increasing profits (Johnstone 1999). In 1969 Sharp introduced a calculator with four integrated circuits produced by Autonetics.
Sharp and other Japanese firms produced millions of calculators with American microchips. Japanese semiconductor companies that had previously refused to supply Sharp now complained and the Japanese government prohibited Japanese companies from purchasing US chips (Johnstone 1999). Autonetics, after being promised to benefit from the learning curve, was no longer able to produce for Sharp or other Japanese firms.
Japanese companies then mastered the CMOS technology that was developed in the US. Many US firms missed the advantages of CMOS for consumer goods and manufactured PMOS (positive-MOS) and NMOS (negative-MOS) chips for military applications. When the potential of CMOS to save power and facilitate miniaturization became evident, Japanese firms were in the pole position. Three of the four leading semiconductor firms in 1980 were American while three of the top four in 1990 were Japanese (Bown 2020). Irwin (1996) noted that Japanese firms increased their share in the market for Dynamic Random Access memory (DRAM) from less than 30% in 1978 to over 75% in 1986.
US firms responded to the Japanese juggernaut by demanding protectionism (Bown 2020). They complained that the Japanese government had excluded US firms such as Autonetics from the Japanese market. US plaintiffs filed antidumping and Section 301 cases against Japan. The US and Japan reached an agreement in 1986 implying that 20% of the Japanese semiconductor market would go to US companies and that Japanese firms would raise prices and limit exports. Determining market shares by government fiat represented a sea change for US trade policy (Irwin 1996).
The US-Japan Semiconductor Agreement focused on DRAMs and provided an opportunity for South Korea. South Korea faced the danger of invasion from the north and prioritized economic development to face this threat. The government allocated bank loans to firms in order to export, and only continued extending loans to successful exporters. Pecht at al. (1997) observed that Korean workers, who recognised the development imperative, were hard-working and patriotic.
In the 1980s Samsung chairman Byung-Chull Lee identified DRAM chips as a promising export sector. Samsung could meet the heavy investment requirements for semiconductor manufacturing because of the government loan guarantees. In 1983 Samsung obtained DRAM technology from the US and Samsung's engineers studied the technology day and night. As Samsung's prowess grew, Japanese firms were constrained to sell at higher prices and in limited quantities. Samsung could sell without restrictions at these higher prices. It channelled these revenues into research and development and capital formation. By the early 1990s Samsung became the leading DRAM producer and remains so in 2021.
The Taiwanese and Malaysian semiconductor industries
While Korea dominates in manufacturing memory chips such as DRAM, Taiwan leads in producing logic chips such as central processing units and graphical processing units. As South Korea faces a menace from the north, Taiwan faces the threat of war with China. Inflation in Taiwan also rose to 22.9% in 1973 and 40.6% in 1974 and Taiwan lost access to Japanese capital goods in 1974. Facing multiple crises, Taiwan prioritised economic development in order to survive. Many overseas Chinese offered to help Taiwan free of charge. For instance, Wen-yuan Pan, head of the renowned David Sarnoff Laboratories, chaired a Technical Advisory Committee (TAC) of leading overseas Chinese researchers.
The TAC advised Taiwan to develop a semiconductor industry. Taiwan established the Industrial Technology Research Institute (ITRI) to oversee its development and purchased semiconductor technology from RCA. The ITRI recruited 40 engineers, some with US PhDs, who diligently studied the technology. The ITRI spun off United Microelectronics Corporation in 1979 and Taiwan Semiconductor Manufacturing Company (TSMC) in 1987. TSMC does not design integrated circuits but manufactures these based on customers' specifications. It is now the world's third largest semiconductor firm (Bown 2020). Many related firms emerged in Taiwan.
The Asian economies had another advantage. The semiconductor industry requires massive and continuous investment in research and development and capital formation (Baldwin 2018). East Asian economies had high private saving rates and disciplined fiscal policy (Yoshitomi 2003). High national saving facilitated investment, enabling chipmakers to remain competitive.
Impressed by Korea and Taiwan's success, Malaysia also employed industrial policy to pursue a cutting-edge semiconductor industry. Unlike Korea and Taiwan, however, Malaysia faced no existential threats. To parallel the ITRI, Malaysia established the Malaysian Institute of Microelectronics Systems (MIMOS) in 1985. MIMOS spun off the semiconductor firm Silterra in 2000. However, as Rasiah (2017) reported, Malaysia did not choose the most qualified candidates to lead Silterra and other institutions. The government's affirmative action initiatives favour indigenous citizens (Bumiputera) over Malaysians who are ethnically Indian or Chinese. Rasiah noted that Malaysia did not choose Loh Kin Wah, the managing director of the German company Qimonda, to head Silterra. The government also withheld grants from dynamic electronic firms if they were not headed by Bumiputera and continued supporting indigenous firms even when their performance was poor. Malaysia's emphasis on redistribution generated rent seeking losses. Its semiconductor industry never advanced to higher value added activities such as design, R&D, and manufacturing.
Several policy lessons flow from the East Asian experience for countries seeking to nurture domestic semiconductor manufacturing.
- First, government largesse matters less than facing competition. While US researchers invented transistors, CMOS chips, LCD displays, and other breakthrough technologies, Asian firms often profited from these. US electronics firms were coddled by defence contracts and lacked incentives to convert new technologies into marketable products. Asian firms, competing in demanding consumer markets, had to choose technologies carefully and employ them to produce desirable products.
- Second, entrepreneurs are essential. Sasaki's vision to use CMOS integrated circuits to miniaturize calculators led to millions and millions of calculators being sold. Byung-Chull Lee identified DRAM chips as promising. Samsung's subsequent success in memory chips contributed to its market capitalisation of over $10 trillion in 2021.
- Third, industrial policy functions better when the polity unites to confront a threat. Willett (1997) noted that the unitary actor model, where agents work together for the common good, can explain outcomes when a nation's survival is at risk. On the other hand, the interest group model, where pressure groups pursue rents at the expense of others, better explains democratic outcomes in peacetime. Government officials, entrepreneurs, workers, and others in Korea and Taiwan viewed economic development as vital for survival and united to achieve it. Malaysia, however, lacked existential threats and focused on redistribution. In Malaysia industrial policy led to rent-seeking waste and did not increase efficiency.
- Fourth, education and technology transfer are vital. Asia invested in education (Yoshitomi 2003). Sasaki was well trained and quickly sensed the technologies that his firm should invest in. Korean and Taiwanese engineers were capable and assimilated technology from American firms. A well educated workforce is better able to absorb knowhow from abroad (Urata 2007).
- Fifth, protectionism can be counter-productive. Japan's refusal to let firms like Autonetics continue selling semiconductors to Japan and America's actions against chip firms in Japan ultimately weakened the semiconductor industries in both countries.
- Sixth, high national saving rates facilitate the massive investment required to remain at the cutting edge. Countries running large budget deficits risk crowding out the requisite investment.
- Seventh, harnessing incentives is important. Hausmann and Rodrik (2003) noted that Korea stopped extending loans to firms that failed at exporting. On the other hand, Malaysia kept supporting Bumiputera firms whose performance was poor.
As Eric Schmidt observed, countries seeking to strengthen their technology sectors are unlikely to succeed by throwing money at the problem (Tanaka 2021). Rather, they should align incentives, encourage entrepreneurs, educate their workforce, and employ industrial policy judiciously.
This article first appeared on www.VoxEU.org on October 15, 2021. Reproduced with permission.