`Industrial policy' in action

Modern thinking on economic policy emphasises the dangers of `industrial policy', where a government picks winners, trying to identify industries or firms which will do well. Will the government be able to correctly identify winners? Will the political economy go all wrong, and the government ends up fostering the wrong firms / industries? In recent decades, a pragmatic sense, that has become the consensus view, is that there is no role for `industrial policy'; that a government must just stick to doing public goods.

A while ago, I had written about these issues in the context of a semiconductor fab, where I had criticised many features of the proposed government involvement for SemIndia. Yesterday Business Standard had an editorial that points out that even if you wanted to do industrial policy, the right place to target is IC design and not the fab:

India may be getting excited about domestic semiconductor manufacturing, but there are two analyst reports—from J P Morgan and Gartner—in the last one week which state that the country should not go ahead with chip manufacturing since it’s not economically feasible without a large subsidy from the government. The subsidy that is actually on offer (expected to be 22-25 per cent of project cost) falls short of industry expectations. China, for instance, gave newly-built semiconductor players a 100 per cent tax break for the first five years, and then a 50 per cent discount for the next five years. To get into the business of making chips, India needs to compete with China, Ireland, Israel and Malaysia in this regard.

Why, then, is SemIndia gung-ho about the $3 billion (Rs 13,200 crore) Fab City project? The answer, as has long been argued, is that setting up a semiconductor plant will create a semiconductor “eco-system”, which in turn will help Indian companies move up the global value chain. India’s consumption of electronic equipment is expected to touch $363 billion by 2015, up from $28.2 billion in 2005, at a compound annual growth rate (CAGR) of 30 per cent. Of this, the market for the semiconductor industry is expected to be around $36.3 billion.

But why should any company come to India for manufacturing chips when Taiwan can execute the same work at a fraction of the price? Besides, foundry revenue has slowed the world over due to price competition from new entrants like China. The growth of the “fab-less” industry (comprising companies that do not manufacture silicon wafers, and concentrate instead on the design and development of semiconductor chips) too has slowed. Taiwan captures 65 per cent of worldwide foundry revenue. One daunting point is that the gestation period for a chip-manufacturing plant is around one and a half years, by which time the technology involved could change, something that could leave a new entrant stranded.

Rather, India’s current play with integrated chip (IC) assembly (labelled AMTP to denote assembly, mark, test and package) makes more sense. This requires much lower investment—in the range of $250-400 million. India can also become a powerhouse in chip design. It already has a large and growing pool of experienced IC design engineers, and hundreds of expat engineers are returning to India every year (as happened with Taiwan in the 1980s). India also scores over China in this regard with more engineers, a greater English-speaking workforce and better protection laws for intellectual property. It already has around 125 companies doing design. In 2005, J P Morgan reveals, multinationals like Texas Instrument, Intel, Cypress, Infineon and STMicroelectronics comprised around 70 per cent of the semiconductor design industry in India. The industry’s turnover was $3.2 billion in 2005, with an engineering workforce of around 75,000. Those numbers are expected to reach $43 billion and 780,000 engineers by 2015. This would therefore seem to be the bigger apple to shoot for.