Many people from poor rural areas migrate to urban areas for work to cope with seasonal deprivation. In Bangladesh, however, researchers observed that many vulnerable households, who could potentially reap large benefits from temporary migration, didn’t send anyone away to work, thereby risking hunger. Why weren’t more people migrating? Would these households improve their food security if they were to send a migrant to these areas during the lean season?

SuMMary Findings

A research team from Yale University, the London School of Economics, the University of Sydney, and Innovations for Poverty Action tested between 2008-2011 whether providing information or small financial incentives, worth about the cost of a bus ticket, increased migration and in turn, improved household welfare.

They found that households offered either a grant or loan to migrate were substantially more likely to send someone to work outside the village during the lean season, and those families increased caloric intake relative to those not offered the incentives. Many of those households chose to re-migrate on their own a year later.

A replication and expansion of the study in 2014-2016 not only confirmed these findings, it also showed that larger scale emigration increases wages and work hours in the village of origin, indirectly benefiting other residents who stay back.

the research in more detail 

In 2008-2011, the research team measured the impact of information, small loans, and small cash grants on migration, food security and income. Researchers randomly assigned 100 villages (1900 households) to one of four groups:

  • Information (16 villages): Potential migrants received information about the types of jobs available in cities, the likelihood of getting each job, and approximate wages.
  • Grant (37 villages): In addition to the same job information, households in this group were offered a grant of 800 Bangladeshi taka (US$11.50) conditional on one member migrating. Six-hundred taka (US $8.50, which covers the round-trip travel cost) was provided in advance with a promise of 200 taka more given once the migrant checked in at the destination.
  • Loan (31 villages): Same as the grant treatment, except that the 800 taka was offered as a zero interest loan with implicit limited liability, conditional on migrating.
  • Comparison (16 villages): Households in this group did not receive any information or incentives.

Effects of Program on Migration Rates

The research team collected data on the migration patterns of household members during the 2008 lean season in response to the information or incentives, and also any re-migration (absent any further incentive) during the 2009 lean season and a milder lean season in 2011. 36 percent of households in the comparison group reported that at least one person migrated in 2008, the same level as the information group. In contrast, receiving the subsidy in the form of either a conditional grant or a loan had a substantial effect on the propensity to migrate: 59 percent of households offered the cash and 57 percent of households offered the loan sent a migrant in 2008.

In subsequent years, researchers found a persistently higher re-migration rate among those offered the grant or loan incentive, even absent any further subsidies.

This higher migration rate among incentivized households reveals a barrier preventing poor rural households from taking advantage of seasonal migration– along with a tool for overcoming it. Notably, there is no effect in the information group, which indicates that the reluctance to migrate does not rest on the poor being misinformed about the average profitability of migrating.

Impacts on Household Welfare

The seasonal migration induced by this program was highly profitable on average, producing increases of 30-35 percent in food and non-food expenditures, and 550-700 more calories consumed per person per day, relative to the comparison group (2008, 2009 data).

This is equivalent to an extra meal per person during a period when meals are regularly skipped among the poor.


In 2014, the research team conducted a second randomized evaluation in partnership with Evidence Action. With an interest in delivery at scale, researchers designed an evaluation that not only re-examined the direct benefits of the approach, using new methods and outcome measures, but also explored the indirect spillover effects accruing to non-beneficiaries.

During the 2014 lean season, researchers randomly assigned 133 villages to either a comparison group (38 villages), a “low-intensity” treatment in which 10 percent of the landless population were offered a 1000 taka loan to migrate (48 villages), or a “high-intensity” treatment (47 villages) where 50 percent of the landless population was offered the same loan.

Offering the loans again led to increased seasonal migration. Each household receiving an incentive in the “high intensity” villages was 40 percentage points more likely to send a migrant than those in comparison villages. A household receiving that exact same incentive in “low-intensity” villages was 25 percentage points more likely to send a migrant. The higher take-up in the high-intensity villages is indicative of some benefits of coordinated travel, when many neighbors simultaneously receive offers. This benefit even reaches other residents of the high intensity villages not receiving the subsidies themselves: They become 10 percentage points more likely to send a migrant (without any incentive).

Household income increased by an average of 19 percent during the lean season for households offered a loan in high-intensity villages, and poor households in the same village not offered loans indirectly benefited, experiencing a 10 percent increase in income.

Households not offered the incentives benefited in two ways. Not only were they more likely to migrate as their neighbors migrated, but those who chose not to migrate also gained from the intervention as they faced fewer competitors for the scarce jobs available locally. Through the incentives, researchers induced an additional 30 percent of households to send a migrant from high-intensity villages. This led to an increase in available work hours for non-migrants resident in those villages, along with an 8-9% increase in the agricultural wage rate at home.

These direct and indirect spillover benefits persisted after a year, during the 2015 lean season. Households that had been offered migration subsidies a year before in the high-intensity villages were 29 percentage points more likely to re-migrate, and even non-offered households in those same villages were 12 percentage points more likely to re-migrate. This led to substantial increases in income earned in cities by those households. A working paper on this research is forthcoming. 

What's next? 

We are now preparing for a related evaluation in Indonesia. We are also testing the Bangladesh program in an operational RCT to assess a number of operational dimensions, and to assure that the impact holds as No Lean Season grows.

In addition, we are examining the potential for any negative effects of seasonal migration that may emerge in the future, such as a strain on urban infrastructure or a saturation of the labor market, or other unintended social consequences which could dampen the program's impacts. We expect that the effect of seasonal migration on the employment prospects of the urban poor is minimal, but are exploring that rigorously in 2017.