Semiconductor Industrial Policy: Why India Should Follow The Eastern Route

Part-1 briefed the “60s European semiconductor landscape and associated policies at the inception of their sovereign electronics” industries. In this part, the gaze turns East. A contextual prelude is in order and it is worth mentioning that primarily there are two generic types of industrial policies to drive technological change and economic growth.

The first seeks to do it directly by encouraging scientific and/or technological advance. The second operates indirectly by transforming the facilitating infrastructure such as building roads or contributing to the health and education of the labour force.

First order of business is to dispel the myth that industries develop due to philanthropic outpouring of developed nations for global good, (Lipsey, 2021) summarises it succinctly:

Until recently, it was nearly impossible for an industry to begin producing an established product in a currently non-industrialised country because of many disadvantages namely: small initial scale, the absence of both sufficient human capital in labour and management and the research and development (R&D) capacity needed to compete in the fierce global competition where ability to produce a string of new products is vital.

Fortunately, this has changed with disintegration of production to international supply chains due to revolutions in communication and transportation of late twentieth century. These revolutions allow a non-industrialised country to produce a manufactured product that is one small part of a world-wide supply chain with minimum ingredients unlike the past barrier that compelled the set-up of a full local industry. Few notable Asian countries stand out in capitalising on this opportunity to graduate to robust semiconductor manufacturing economies. Let’s begin the journey through the east starting from Japan.

Stop-1: Japan Junction: A Follower To The Leader

Japanese catchup policies encouraged licensing agreements for established technologies developed in other countries. Japan chose established technologies and diminished uncertainty, while identified niches that could be filled. Incremental development of an industry to grow organically using these niches (ex, low cost germanium radios), by expediting facilitating structure can often be superior compared to attempts at leaping quickly to a full grown, multifaceted industry holding its own against established international competitors.

Japan started work on semiconductor devices under American licences in early 1950s and produced the first transistor radio in 1955 termed — ‘Sony TR 55’ using germanium transistors at Tokyo Tsushin Kogyu Ltd — a Western Electric licensee. By 1956, there were five firms licensing American patents namely Hitachi, Mitsubshi Electric, Tokyo Shibauro Electric, Kote Kogyo and Tokyo Tsushin Kogyu Ltd producing products like table radio receivers, battery operated tape recorders, pocket receivers and hearing aids. By 1959, Japanese semiconductor industry had captured 50 per cent of the American market for portable radios. In 1970, Japan was estimated to be paying 10 per cent of their sales as royalties to American companies.

In 1971, MITI devised a programme with emphasis on VLSI (very large-scale integration) circuit development and fabrication with investments of $65 million per annum. The result of these heavily financed programmes was twofold. First, it gave the Japanese significant lead over US in large-scale memories and second this lead in turn propelled Japanese world-domination in VLSI. Walllish (P, 1990) summarises the scenario in 1983 writing for IEEE Spectrum as follows:

There were tax incentives and subsidies for successful firms with some protection of the home market from foreign competition. Pre-commercial R&D was coordinated centrally. The Japanese policy was focused on facilitating co-operation framework for the transfer of knowledge between firms, it encouraged co-operation that accelerated research while preventing wasteful duplication. The subsequent growth of the Japanese industry and its movement to higher value products illustrated the success of this early infant industry programme.

Although, the Japanese project to develop breakthroughs in semiconductors began in 1976 by targeting pre-commercial research. The project failed to create genuine breakthroughs but it did produce a considerable number of technological spinoffs that helped Japanese firms in their competition with American firms (unlike the European counterparts as discussed here). A few key differentiators compared to US and EU were:

a) Technology Import: Japanese government agencies demanded tough terms right from the start including technology transfers from foreign companies like IBM and Texas Instruments that wanted access to growing Japanese market. Japanese companies donned a proactive role in modulating the pattern of technology importation instead of passive receptors. They followed a prudent strategy to be a leader from a follower that can be summarised briefly in three phases: 1) Take-off by securing internal demand. 2) Sequential competency-building by mastering process technologies and 3) Leapfrogging for technological leadership with R&D emphasis and initiating industry standards.

The early development followed the familiar trajectory of transfer and imitate foreign technologies, adopt, adapt and indigenise foreign technology, advance from pure imitation to innovative imitation and finally to creative innovation.

b) The financing method: A large portion of the capital for firms was through banks belonging to common business group (keiretsu) with bare minimum requirement to earn enough to cover interest on their debt and no need to finance their growth out of retained earnings.

c) Diverse product portfolio: The Japanese companies (Hitachi, Fujitsu) which entered semiconductor manufacturing also produced a wide variety of other goods unlike their US or EU counterparts, who exclusively manufactured semiconductors providing them the much-needed latitude.

d) Strategy of automation: Unlike US and EU companies setting up fabrication facilities in Asia, Japanese semiconductor manufacturers concentrated on automating their production facilities, which resulted in substantially higher quality standards compared to American semiconductor companies.

e) Skilling workforce: Unlike US, where personal advancement was through changing companies or education outside the firm, Japanese firms offered stability and career-long training in-house and additionally through technical societies like Japan Union of Scientists and Engineers and the Japanese Institute of Standards. For ex. Toshiba allowed graduates to study an allied discipline one day a week. A mechanical engineer could study electronics a day per week and vice-versa. The highly educated workforce has been a significant factor of Japan’s industrial success. For example, in 1977 (on per capita), Japan had three times more electrical and electronics engineers than US, more than four times in Britain, six times more than France and 70 per cent more than West Germany (G, 1985)

The 1990s saw the decline of Japanese market share due to management problems termed as — “strong factory-weak headquarters” syndrome(Fujimoto, 2004) and lack of focus, revitalisation of the US semiconductor industry to pull back the lost ground in semiconductors (termed as “leveling the playfield” in the US playbooks) and emergence of Korean companies. Today, Japan still holds 50 per cent market share in silicon wafer supply, is a significant player in supplying critical chemicals and gases used in Fabs (ex, Shin-Etsu, Sumitomo, Mitsui) and plays an important role as Fab equipment vendor (ex, Tokyo Electron).

Stop 2: Korean Kiosk: The Laggard To Trailblazer

The Korean electronics industry is an interesting case for several reasons because both India and South Korea secured self-administration at about the same time with more or less equal levels of development in 1950. Korea has had extraordinary progress in comparison to India. Three primary reasons have been attributed to this performance (Datta-Chaudhui, 1990):

1) India inherited an efficient bureaucracy from the British which was not wired toward commerce instead perfected for licence and rent-seeking, while Korea developed its industrial policy bureaucracy from scratch and staffed it mainly with members of the business elite.

2) Indian industrial policy had a large number of often opposing objectives, which required heavy control of the activities of the individual firms. In contrast, Korean policies were narrowly defined on export promotion and gave more autonomy to individual firms.

3) India’s industrial policy allowed losses to be absorbed by the public sector and had no strong mechanisms for rewarding success contrary to Korea’s strong export promotion policy.

Even though Korean government policy for electronics industry has origins in 1969, it remained a minor player mostly as an offshore assembly area for semiconductors until the early 1980s, analogous to India’s standing in 2021. Korea relied heavily on the foreign direct investment (Fairchild & Motorola), which capitalised on low-wage labour force. Even though the 1970s saw the entry of the Japanese firms (Sanyo & Toshiba), they too opted for the same pattern of exploiting labour-intensive assembly. These foreign subsidiaries remained as exporting enclaves importing all necessary materials and production equipment without any linkages with the local Korean economy and restricting Korean participation to providing export zones in Masan and Kumi with cheap labour force.

Korea’s first serious attempt at upgrading semiconductor industry began with the launching of Heavy and Chemical Industry Programme in 1975. The long-awaited breakthrough (from sweat-shop economy relying on foreign direct investment) came as the result of a major institutional change in the form of ease of control by state over the banking sector enabling a politics of reciprocity between Korean state and the Chaebol (industrial conglomerates). The Chaebol took advantage of the state subsidised loans and tax breaks to grow very quickly creating vertical monopolies (ie, the company and its subsidiaries control most steps in production, from acquisition of raw materials to fabrication). Four most prominent chaebols were Samsung, LG, Hyundai and Daewoo.

In spite of the establishment of Korean Institute of Science and Technology (KIST) and Ministry of Science and Technology (MOST) in 1966 and 1967 respectively, it was only after 1980 that government intensified R&D efforts focusing on generic technologies and bridging university research labs and businesses. Korea’s centralised innovation system offered a consistent long-term (over two decades) strategy for development and commercialisation of specific technologies and products. The government systematically evolved the human capital essential for technological upgrading by dispatching scientists and engineers to US, Japan and Europe for further study or work experience. The government also created a network to connect with Korean scientists and engineers working abroad, who became a human capital brain bank to support technology upgrading. The government’s highly advanced national (HAN) programme spending $4.7 billion on R&D activities focused on strategic technologies and linking many disciplines together.

Stop 3: Taiwan Terminus: The Pure Play Pioneer

In the 1960s, semiconductor industry development in both Korea and Taiwan followed the same trajectory of import-substituting industrialisation, inward foreign direct investment and technology transfer. Both governments in the 1970s shifted the policy to export-led industrialisation with strong emphasis on R&D setting-up institutes like Korean Advanced Institute of Science and Technology (KAIST) and Industrial Technology Research Institute of Taiwan (ITRI). The divergence began in the early 1980s with Korea opting for private Chaebol driven promotion of technological development, while Taiwan choosing public sector-driven path.

The creation of new industry was through government agencies, instead of government support of private firms. Taiwanese government established Hsinchu Science based Industrial Park (HSIP) in 1980. Government organisation Electronic Research and Service Organization (ERSO) licensed foreign technologies (RCA)and then sub-licensed them to local industry (UMC). The industry was fostered by public assistance until mature enough to be commercially viable. Much of it was then transferred to the private sector. In 1987, the government facilitated the joint venture of Phillips with several small Taiwanese manufacturing firms forming Taiwan Semi-Conductor Manufacturing Corporation (TSMC) under Morris Chang as detailed in this article. In 2010, HSIP had provided an assistance of close to $90 million through subsidies to conduct IC R&D, additionally $84 million and $120 million was allocated for building human capital through semiconductor design courses and semiconductor manufacturing process courses respectively.

A few Taiwanese ingenuities are worth highlighting in comparison to Japanese and Korean approaches:

• Vertical disintegration: Moving away from the traditional Integrated Device Manufacturing (IDM) model and splitting supply chain into three distinct areas of design, manufacturing, packaging and testing, which allowed each area to develop independently. UMC & TSMC pioneered the pure-play foundry services. Today, TSMC is the largest foundry in the world serving more than 500 customers and manufacturing more than 10,000 products. It has nine factories in Taiwan alone, with expansion plans in Japan, US and (potentially) Europe in the near future.

• Joint-venture and external strategic linkages: In spite of mushrooming out of Taiwan’s public sector all major companies were set up as joint ventures and during their early years were encouraged to compete globally and establish their own external linkages for partnership and development. UMC acquired equity in startup firms in the US Silicon Valley. TSMC expanded into microprocessor production in agreement with Advanced Micro Devices (AMD) and Winbond reached out to Hewlett-Packard and Toshiba.

a. Construct a network of foreign advisers to monitor bureaucrats — Taiwan managed to find foreign advisers with technological expertise and moral integrity with enthusiasm about Taiwan’s semiconductor industry. The presence of these advisers as monitors meant that bureaucrats had less incentive to mislead or deviate from achieving government’s objectives.

b. Use private entities in the joint-ventures to serve as ‘fire alarm’: The state-owned firms face the problem of incentive misalignment, where state appointed representatives do not have strong incentive to make the joint venture a success. The presence of private parties served as a positive influence to compel state-designated managers to perform and deliver efficiently as against a wholly government-owned enterprise with no countervailing force.

c. Encourage bureaucrats to start their own business: A unique policy employed by Taiwanese leaders was to encourage technocrats in public institutions to set up their own semiconductor-related business. The ITRI/ERSO even provided technical and personnel support. By encouraging startups and spin-offs, the government aimed for a faster diffusion of design and technology knowledge among Taiwan’s local IC businesses.

After pioneering pure play foundry space, few key spin-offs from UMC in IC Design include Mediatek, Novatek and Faraday Technology basking the PC boom of 2000s and a decade later they have shifted focus to mobile device platforms. In 2019, Taiwan had 238 IC design companies with a workforce of over 40,000 and a revenue totaling $22 billion. Taiwan also dominates in the backend space of packaging and testing (OSAT) with ASE claiming over 50 per cent of this market share.

Industrial policy has no simple rules. Taking on the established giant as US would have been a route to failure for all these Asian dilettantes. However, several policies worked towards success against heavy odds for failure. Emphasis was on success in the international market and support was withdrawn if that was not secured.

Once the firms were transferred to the private sector, they had to compete with each other along with succeeding internationally. The decisive influence of specific government organisations of science, technology and innovation is apparent be it in establishing science and technology parks or research universities focused on semiconductors or developing human capital.

As a new entrant, it is an easy step for India to target semiconductor production through inviting foreign companies or support the launching of national firms.

The difficult steps are ahead which include the imperative to create conditions for persistent technology upgrading to support the transformation from low to high value-added activities, attracting and developing R&D human capital from both abroad and within the country, as well as developing research universities that bolster the industry with knowledge-intensive activities.

It is a hope that these Asian hikes to peak of semiconductor industry are useful markers in the Indian pursuits. Godspeed!