let my people go...but busfare certainly helps: seasonal migration in Bangladesh


This page in:

This week I want to talk about some interesting work that Gharad Bryan, Shyamal Chowdhury and Mushfiq Mobarak are doing in Bangladesh (policy note and presentation are online, paper coming shortly).

So the story goes like this.   There is a region in north-western Bangladesh which is called Rangpur. In this poor region, things get worse after planting and before harvest (Sept-Nov) – rice prices go up, income falls (wages go down and there aren’t a lot of jobs) and expenditure falls. This causes a seasonal famine. 

Now, there is money to be made elsewhere in Bangladesh during this time of the year, but only about 1/3 of the people are going elsewhere. So what Bryan and colleagues do is incentivize folks to migrate. They have three treatment arms: one group got information only on the likelihood of getting a job and the wages at different destinations, a second group got the information plus cash roughly equivalent to bus fare to the nearby towns (about $8 plus $3 if they reported at the destination) and the last group got information plus a zero interest loan for the same amount. 

So what happened? The cash and credit had the same effect – six months after the incentives, these folks had migrated at the rate of 58%, compared to 36% for those without an incentive. The information? No effect.  

And migration pays off. Although about 20% of migrants don’t find a job, the average migrant earns about $110 during the lean season and sends more than half of that back home. This translates into a boost for consumption of around 30-35%.   One interesting wrinkle: the incentivized migrants were making (and remitting) less than the non-incentivized migrants.   Despite this the folks getting the incentive were more likely to migrate the next year, even without the incentive – their migration rate was 47% against 37% for the non-incentivized

So what’s the story? This could be consistent with liquidity constraints, but that’s a bit hard to believe – the transfer is very low relative to the returns and, moreover, there is enough variation in the expenditure and consumption of these households that it would be hard to believe they couldn’t come up with the cash. Bryan, et. al’s preferred story: this is a migration-poverty trap where people don’t migrate because they are afraid of the consequences if they go. Indeed, the folks who are “induced” to migrate by this transfer are those closer to subsistence, those with lower weaker social networks in the destination community, and those with fewer job leads. 

So this result shows us that a small financial push can have big dividends where information alone won’t do the job.   It also broadens the scope for thinking about microfinance or small scale insurance in terms of labor mobility rather than just for starting businesses.   Of course, if we think about taking it to scale we run into another problem – what would happen to labor markets if this were really widespread? 



Markus Goldstein

Lead Economist, Africa Gender Innovation Lab and Chief Economists Office

May 24, 2011

Interesting article about an important topic.

Nevertheless, (in what I found) the authors only look at average treatment effect, not any of the distribution. Migration is a particularly risky decision opening the migrant to dangers of being robbed or taken advantage of in an environment where they do not know anyone and may not know the social customs (e.g. What street not to walk down). Therefore individuals may indeed have been making rational decisions in choosing not to migrate because of the risk of bad outcomes. Given that the information treatment did not work and the incentive treatment did work, I think the burden is on the authors to show that the initial choice not to migrate was an irrational one and that the migration was welfare-improving.

In short, the incentive may have induced people to make the WRONG choice.

The fact that the treatment induced 10 percentage points higher migration in the following year is good, but does not rule out the fact that some people may have had very bad outcomes from migrating. The ones that do choose to migrate in the second year are clearly selected--those that assimilated well into their new home and had good outcomes in the first year.

Finally, it'd be interesting to know how the authors dealt with attrition in their survey sample.

Asif Dowla
May 24, 2011

There is a simple explanation: the migrants do not want to leave their family behind during the silent hunger season (Monga). It is an interesting research. I am not sure about the policy implication: the government should do conditional cash transfer?. Wouldn't it be better if the government create jobs in the area such as food for work scheme. This way the men do not have to migrate. Grameen Bank had a special project for the area where they trained the young girls and men to produce products and do small scale assembly. The products were sold in the stores of BRAC in Dhaka. I am not sure about the status of the project. Now with Yunus being pushed out and the government taking over the bank, this types of social safety net project by the bank will be dead.

May 24, 2011

I agree with Michael. It would be interesting to look at the distribution of the treatment effects.

Besides, I would be interested in more subjective assessments of the program. Did happiness or subjective well-being also improve for the beneficiaries? Migration might have positive effects on income, but there are certainly costs involved as well.