I have watched the development of the `SemIndia' project with some curiosity. It looks like a situation where there is an `industrial policy' push in favour of one project (SemIndia) in one industry (semiconductor manufacturing). There is a raft of laudatory stories in the press, and various arms of government seem to be keen to dole out equity capital, `viability gap funding', land, etc. [Story 1] [Story 2] [Story 3]
I wrote a column for Business Standard arguing that there is no role for the State in this business. There is no public good being produced by State involvement. I argue that if the State does want to foster top end research and knowledge, then the best way to do it is through a system of defence contracts akin to that which was operated by DARPA in the US.
Getting a fab into India will be an occasion to celebrate, but it should be based on the true merits of India as an economy. Getting a "fake" fab in based on government subsidies is as silly as trying to get "fake" exports growth based on either export subsidies, or on a government-distorted exchange rate. We mix up means and ends when we start going after the goal of a fab, rather than seeing a fab as a litmus test of how well India is faring on the core public goods.This debate is about `industrial policy', where the State picks winners. Sometimes, there is a view that India should attempt industrial policy, as has been done in South Korea [recent ambivalent NYT article on South Korean industrial policy]. I am willing to believe that there are countries where a big State-driven push into `good' industries and `good' companies can work. But in India today, I deeply believe that it can't work. The Indian governance puzzle is about keeping the State focused on public goods, when the political system would all the time love to goof off playing the patronage game with either industrial policy or social programs.
One facet of Semindia that is particularly disturbing is a political allocation of land - 1200 acres of it. I argue that this is way beyond the ordinary land requirements for a fab. This is part of the deeper problems of the land market in India. Of the four factors of production - land, labour, capital and enterprise - the State is still a very big player in the land market.
My views on these issues were greatly shaped by the late 1980s evolution of the global semiconductor business, and particularly the evolution of the last piece of industrial policy in the US, Sematech. Back in 1995, there was an excellent article by T. J. Rodgers titled Technology traps: Defunding the megaprograms on Sematech. Rodgers was CEO of Cypress, and stood to gain from public money being poured in his industry. That makes his logic all the more credible. You will also find this interview with him to be fascinating.
On the `mercatus' blog, I found an interesting pointer to a PDF file describing site selection for a fab at Intel. That links up to this discussion in the sense of answering `what would it take for Intel to voluntarily choose to build a fab in India?'.