This is a writeup of a shallow investigation, a brief look at an area that we use to decide how to prioritize further research.
In a nutshell
- What is the problem? Domestic seasonal migration may carry large benefits for migrants and their families, but some people who might benefit from migration may not be taking part.
- What are possible interventions? A funder might support or expand the micro-credit or conditional cash transfer migration support programs that have been studied in Bangladesh, or fund more studies of the impact of such programs in different settings to help address the question of generalizability.
- Who else is working on it? The authors of the one randomized study to date are planning replications in other countries. It is unclear whether the partner organization from the study will be expanding the program studied to other parts of Bangladesh.
1. What is the problem?
Seasonal migration from rural to urban areas in developing countries may carry large benefits for the migrants and their families under certain circumstances.1 However, some people who might be able to benefit from domestic migration may not be taking part.
We know of one randomized controlled trial on this topic, which has driven our interest in the area. It was conducted in Rangpur, a region of rural Bangladesh that persistently suffers from pre-harvest famines.
1.1 A randomized study of incentives to migrate
Bryan, Chowdhury, and Mobarak ran a multi-arm cluster-randomized trial with 100 villages:2
- 16 villages received information about jobs in potential destination areas, but no subsidies of any kind.3
- 37 villages were offered cash worth $8.50 (a little more than the cost of round-trip transportation to the destinations) at the origin if they decided to migrate, with another $3 paid if the migrant checked in with the researchers at the destination, and were given the same information as the group above.4
- 31 villages were offered 0% interest limited-liability loans in the same amounts as the cash transfers, along with the same information.5
- 16 villages served as the control group.6
Both the 0% interest loans and the conditional cash transfers led to large increases in seasonal migration, while “information only” treatment had no apparent effect,7 leading the authors to combine the cash and credit groups and information only and control groups into “incentivized” and “non-incentivized” groups, respectively, for the remainder of their analysis.8
In the year the study was conducted, incentives led to a 22 percentage point increase in migration on a baseline level of 36% in the non-incentivized group, a 60% increase.9 In the following year, when incentives were no longer provided, migration was still 10 percentage points higher in the villages that had previously received cash or credit incentives. Three years later, an 8 percentage point increase persisted in these villages.10 This increase in seasonal migration led to large and well-identified gains for the families of migrants at the origin, who benefit from remittances:11
- 37% increase in expenditures on food12
- 37% increase in total expenditures (including food expenditures)13
- 38% increase in calories consumed (from a control mean of 2,061/day)14
All of these effects are statistically significant at the 5% level, and persist (statistically significantly, though with attenuation) in the following year even after incentives are removed.15
The authors are unable to estimate a direct financial return on investment from the incentives because individuals who did not migrate typically do not work for wages (instead working in self-employment, agriculture or entrepreneurship, which makes estimating control group income difficult).16 They nonetheless find migrants induced by the experiment earn an average of $105 at the destination, most of which is saved or remitted, potentially suggesting high rates of return to the original $8.50+$3 incentive to migrate.17 Another approach to estimating the financial returns to the migration incentive is to impute such returns from consumption gains: the authors report an increase in consumption at the source worth about $20/family/month,18 though it’s unclear whether these results were measured during the famine season and how much they might generalize beyond the month of the survey.19 Finally, non-experimentally comparing earnings between incentivized migrants and non-migrants who earned wages or profits, the authors observe a gap of approximately $36 in income.20 Summarizing across these disparate estimates, the $8.5+$3 investment appears to carry a ~2-6x financial return, not counting any benefits from migration during later seasons.21
The authors argue that it was risk aversion on the part of potential migrants that prevented them from realizing these benefits prior to the intervention.22
Although the returns to induced migration appear to have been large in this case, we are not sure how generalizable the results are, as the study took place in an area that may have been unusually appropriate for the intervention:
- Despite the frequency of seasonal famines, agricultural laborers from the region of the study may seasonally migrate at lower rates than those from other regions of Bangladesh (though this appears to be contested).23
- The variation in income and poverty between Rangpur and other areas of Bangladesh appears to be much higher than the variation across seasons within Rangpur, which means that migration may be a more appropriate approach than services like microfinance that facilitate consumption smoothing.24
- Existing government and NGO efforts in the region focused on direct subsidy programs rather than migration support.25
2. What are possible interventions?
We see several avenues for a potential funder interested in this area to get involved:
- Funding further experimental research on the returns to migration or mechanisms for encouraging migration in other settings. We haven’t investigated this issue, but have the impression that such a study might cost on the order of $500,000-$1,000,000.
- Supporting or scaling the program in Bangladesh that was studied (including conditional cash transfers or micro-credit loans).
- Advocating to countries, funders, or NGOs to scale up the program to other settings.
We don’t have a sense of how these opportunities might stack up in cost or likely returns, though our intuition is that supporting further research is likely the appropriate first step.
3. Who else is working on this?
We spoke with Mushfiq Mobarak, one of the authors of the paper describing the randomized controlled trial in Bangladesh, about what other activities are being undertaken. He reported:26
Replication of Professor Mobarak’s study on seasonal migration in Bangladesh We are pursuing opportunities to replicate our studies in other areas. AusAID has expressed an interest in doing a similar study in Indonesia, and we’ve had early conversations with a foundation about funding a replication in a couple countries in southern Africa, which also have hunger seasons, though those conversations have been more preliminary…
Scaling within Bangladesh
In Bangladesh, the program from the original study was implemented in partnership with the Palli Karma Shohayok Foundation – the umbrella organization for micro-credit NGOs in Bangladesh. PKSF has a very positive approach to all of this because it has their stamp on it. The microfinance groups are generally busy with their main business of doing microcredit with frequent repayment, which forces borrowers to stay at the origin rather than migrate. The kinds of loan programs we’re talking about, with less regular repayment periods, are outside of their regular way of doing business, so I think it will require a little more pushing.
On the government side, I haven’t seen as much interest.
Professor Mobarak directed us to some other scholarship on the returns to seasonal migration but was not aware of any other experiments or NGO efforts aimed at supporting domestic seasonal migration in low-income countries.27
4. Questions for further investigation
Our research in this area has been relatively limited, and many important questions remain unanswered by our investigation.
Amongst other topics, our further research on this cause might address:
- non-experimental estimates of the returns to seasonal migration from other settings (to see whether they comport with the findings from Bangladesh).
- financial costs and potential humanitarian benefits of replicating or scaling the research conducted by Bryan, Chowdhury, and Mobarak.
- the existing level of funding that is available for research in this area.
- Bryan, Gharad, Shyamal Chowdhury, and A. Mushfiq Mobarak. 2011. Seasonal Migration and Risk Aversion (PDF). Working Paper.
- GiveWell. Notes from Phone Conversation with Mushfiq Mobarak (11/08/12) (PDF).
- Khandker, Shahidur R. and Wahiduddin Mahmud. 2012. Seasonal Hunger and Public Policies: Evidence from Northwest Bangladesh (PDF). Washington, D.C.: The World Bank.