Sunday, December 13, 2009

An upgrade to growth diagnostics

One component of the growth diagnostics framework of Haussman, Rodrik, and Velasco involves examining enterprise surveys as part of the search for what is constraining entrepreneurship in an economy. However, as Ricardo Haussman emphasizes in his PED-309 lectures, these surveys will at best give you a clue as to factors constraining enterprises that currently exist, as opposed to factors that prevent enterprises that do not exist from existing in the first place (does anyone have suggestions on how to phrase this better?)

So, for example, surveying construction firms, textile manufacturers, and paint producers will not give you much insight as to why there is no domestic software industry. This is because the needs of the software industry are, for the most part, different from those of the construction, textile and paint industries. So then how do you assess constraints to industries that do not exist?

Traditionally, you look at what you think the requirements of these industries are and look at their shadow values or what else is going on with them. Since software is intensive in human capital, you would run a mincer regression on household survey data to look at returns to education. If they are high, you'd conclude that human capital is relatively scarce and so human capital might be a constraint. There is much more you would do, which may include looking at: the other requirements of a software industry; looking specifically at returns to education in high-tech fields, outward migration rates by specialization, etc. This is all very data intensive, rigorous and requires resourcefulness and creativity. Yet, in the end, how persuasive can your conclusions be? You're trying to find out, using indirect ways, why no software or other higher-productivity industry exists.

Why not try a more direct route, like surveying firms that do not exist?

Here's my contribution: why not survey decision makers in firms that have expanded operations outside of their borders and ask them the following:

1. Besides the country in which you chose to expand your operations, which other countries did you consider, and why did you decide against them?
2. Why did you not consider countries X, Y, Z, etc.?
3. How did you go about getting information that you needed to make your decision?

I think the answers to question #1 above will be the most helpful, even if they are at odds with reality. For example, suppose that Mr. Decider at Octopus Corporation says: we considered expanding to Jamalastan, but we didn't think we could find enough computer scientists. If there was indeed a scarcity of computer scientists, then its likely that you've found a constraint to at least one industry. If it turns out that there a many computer scientists driving taxis, then you've found a different constraint: misconceptions about the country.

For question #2, it is unlikely that you'd get answers that are the result of thorough investigations. But they still might be useful. Question #3 could yield interesting information in that it could tell you what firms consider reliable sources of information (e.g. Heritage Index, Corruption Perceptions, Doing Business Indicators, use of personal networks, conversations with local entrepreneurs and bankers). This could help local governments focus on a set of performance indicators.

I would like to see a survey of hundreds or perhaps thousands of firms every few years asking the above questions. The survey could include firms that have expanded domestically or made acquisitions domestically also.

What other questions would you add to such a survey?