La Porta and Shleifer released a working paper recently on the informal economy (which I believe is a draft for a future issue of the Journal of Economic Perspectives, but I could be wrong). They give an overview of what we currently know about the size and characteristics of informal firms.
The thing that stuck out most after reading this was the strong evidence that big, formal firms do not grow out of small, informal ones. In other words, small informal firms tend to stay small and informal. Big formal firms are created as formal firms, and while they may start relatively small, they generally start out bigger than most informal firms will ever be. While institutions/regulations may make incent some people to run informal firms, these regulations are not preventing informal firms from becoming formal. There is essentially no transition in any country of informal firms into formal ones.
This is important because big formal firms are much, much more productive (per worker, or in terms of total factor productivity) than small informal ones. So big formal firms are the source of nearly all the significant gains in aggregate productivity within countries. You don't see any highly developed nations dominated by small, informal firms. And fostering the growth of big formal firms is different (according the La Porta and Shleifer) from fostering the growth of small informal ones.
A similar sentiment can be found in a recent column by Daniel Altman, titled "Please Don't Teach this Woman to Fish". As the tag line to the article says: poor countries have too many entrepreneurs and too few factory workers. Promoting small (almost universally informal) firms can improve living standards slightly, but does not lead to the massive productivity gains that generate big gains in GDP per capita.
So what does it take to promote big formal firm growth? La Porta and Shleifer suggest that a big constraint is highly trained managers and/or entrepreneurs that can handle running a large firm. Improving the average level of education is less important, in this case, than extending the tail of the education distribution. Nearly all big formal firms are run by college-educated managers, so developing countries need to generate more of those kind of people. Getting everyone to go from 6 to 7 years of education won't do it - it would be better to leave nearly everyone at 6 years, but add a few extra people with 16 years or 18 years of education.
Yes, you also need an institutional/regulatory structure that makes it low-cost for those college-educated managers to open and operate firms, obviously. But apparently having a good regulatory structure won't buy you anything without the stable of potential managers.
So here's a question(s) related to education policy in developing countries. Would they be better off spending their budget providing scholarships for students to got to college (perhaps abroad) and/or paying for high-achieving students to intern or work abroad at large firms for a while. If you could get GE to hire 100 students into their managerial program, would that ultimately be better for development than achieving universal secondary schooling? Is it worth it to the whole country to create an elite cadre of managers who own/run large formal firms?