Addressing poverty traps through migration, in the small and in the large

by Ajay Shah.

People who were pessimistic about the future of BIMARU states and other poverty traps have long felt that migration out of these bad places was a good way out. In recent months, an empirical literature in the US has suggested that there are large positive effects upon poor people when taken out of poverty traps. Let's think about policies that assist or retard migration, and about downstream implications. Some things that make a difference (positive or negative) to migration are well known:

• `Workfare' programs like NREGA may hamper migration.
• A malfunctioning land market makes it harder for people who own land to leave. Laws that make it difficult for SCs or STs to sell land are particularly harmful in locking people to the land.
• Improvements in urban governance will help increase migration flows to cities.
• A better criminal justice system will increase the safety of migrants and thus assist migration flows.
• Divergence of per capita income across states will spur migration flows.
• More use of English in education and in government creates a nationwide labour market where people from all over the country are able to move, get jobs, access government services, file a complaint in a police station, etc. We should aim to become more like the US (a nationwide labour market with English) and less like Europe (where barriers of language reduce migration flows).

While leaving is good for the household or the individual who leaves, it's interesting to think about the consequences of departure. In the US, the standard intuition is expressed in the Tiebout model, where labour mobility creates local neighbourhoods for people with homogeneous preferences. I worry that we have a different problem in India. We should think more carefully about the consequences for a place like Uttar Pradesh when many people choose to leave.

We may conjecture that the more capable people leave. I'm reminded of the HMY model: there's a fixed cost of migration, and paying this activation energy is justified when the NPV of future cashflows is large enough. In the HMY model, only more productive firms export or do outbound FDI. In similar fashion here, more capable people are likely to make the jump.

In the small, this is not a problem. But what if a sufficiently large phenomenon of migration-by-the-best takes place. Could this rob a poverty trap of the people who can exercise leadership in starting a business, teach, do research, criticise wrong-doing, or engage in politics? For a sufficiently strong selectivity process, and for a sufficiently large exit-by-the-elite, we could get an explosive process where the departure of the elite leads to a reduction in the quality of government and the local economy, which encourages more elite flight, and so on. Badlands with warlords might become worse if there is a systematic bias where people who could fix things tend to leave.

For this unpleasant dynamics, migration has to take place on a large enough scale to adversely affect the talent pool at the source location. As an example, in the 1960s, Jagdish Bhagwati worried about the `brain drain' out of India. With the benefit of hindsight, we know this was not a big issue. There has been no real problem in the talent pool within India for the purpose of teaching, doing research, criticising wrong-doing, leading firms, participating in politics, etc. The scale of out-migration from India was not large enough to damage the upper tail in India.

For example, from my batch at IIT Bombay, which graduated in 1987, roughly half are in India today. The selection process for departure has not strongly encouraged departure of the upper tail. I have often felt that leaving India is efficient for people below $2\sigma$ (by ability) but by the time you get to $4\sigma$ (by ability), life in India is more interesting.

Could conditions in Bihar or Uttar Pradesh be different? Could the selection process and the magnitude of out-migration create an explosive process which knocks out the upper tail? If such a vicious cycle is present, it changs the way we think about federalism. In a happy Tiebout world, politicians compete to offer local public goods that are attractive to the median voter, and the tails of the distribution of preferences leave. At the same time, there would be in-migration of people who are attracted to the package of local public goods that the politicians are offering. In this benign world, more federalism is undoubtedly a good thing. If, on the other hand, we get feedback loops through which poverty traps get trapped with warlords, this raises concerns about the outcomes associated with the democratic process in poverty traps.

We should be careful to distinguish between two distinct issues. The subsidiarity principle suggests that the best accountability for local public goods is obtained by placing the votes and the voice in the hands of people who are at the lowest possible level of government. This favours decentralisation. However, if there are conditions where an explosive process can remove the upper tail of the talent pool in a poverty trap, this can adversely affect the quality of local politics, and the ability to create productive firms. Under these conditions, we should worry about decentralisation.