An Update on No Lean Season’s First Cohort: Migration has Begun

The first cohort of “No Lean Season” participants in northern Bangladesh are on their way. More than 4,500 rural workers are travelling temporarily to nearby areas with higher wages and greater labor opportunities to support their family during the lean season. A few thousand more are expected to join the ranks before the end of the lean season around mid-December in Rangpur, the target area for the initial roll-out of the program.

An Innovative Solution to Seasonal Income Insecurity

“No Lean Season” offers travel subsidies of about $15 to poor rural families so that they can send a family member to take advantage of employment opportunities in other areas of the country, including nearby cities.

In the periods between planting and harvesting — when jobs are scarce and wages low in rural areas — rural dwellers experience income insecurity and even hunger in many parts of the world. It is estimated that more than 300 million people globally suffer from this ‘lean season’. Even though temporary migration may be a coping strategy for this period of vulnerability, many poor households cannot take advantage of it even if they stand to gain from migration. However unlikely, the consequences of spending money to migrate but returning empty-handed can be devastating for a household living already close to subsistence.

A travel subsidy changes that by lowering the cost of migration: the household receives the subsidy up front, to be used for any migration trip, and, in the case of a loan, only pays it back if the migrant manages to find a job at destination.

This simple intervention was first proposed and tested with a rigorous randomized control trial (RCT) in Northern Bangladesh. Results showed that migration subsidies successfully help migrants move temporarily to locations where they can earn wages as much as 89% higher than in their village, which in turn increases household incomes by an average 19%. Caloric intake also jumped by 750 kcal per person per day — the equivalent of one extra meal on the table for every member of the household during the three to four months when meals are otherwise frequently skipped by poor households.

From Evidence to Scale

Encouraged by the RCT results, Evidence Action’s Beta team is now exploring whether this solution can be brought to scale. In May 2016, Evidence Action announced a new partnership with RDRS Bangladesh, a leading NGO in the country, to expand “No Lean Season” over the next four years.

At the moment, our goal is to determine exactly what it would take to scale up the program and eventually reach as many as 310,000 recipient families in the next four years. We want to understand whether the intervention would work and continue to show positive impact in different locations in northern Bangladesh, and with a greater number of targeted households — issues that fall into what economists call ‘external validity.’

One of the mandates of our Beta team is to ‘pressure test’ promising development interventions. We deliberately apply a series of tests to see if the impact of a program holds up in a different location or environment; if it creates unintended consequences (good or bad); and if it is politically and operationally feasible. We think this is essential to determine whether a useful proof-of-concept study can be turned into a scaled-up program with the ability to effectively improve the lives of millions.

We are doing exactly this with “No Lean Season”. When choosing geographic areas to offer the subsidy this year, for example, we deliberately picked locations where we felt the program might be less likely to work because there was not an established practice of migration there. We also designed the subsidy as a an interest-free loan (as opposed to a grant) to see if the positive effect on migration would hold in a more sustainable and cost-effective design.

We are also trying to optimize the way we deliver the program to improve cost-effectiveness for when it will be rolled out at a greater scale. For example, we purposefully removed eligibility criteria in some areas to see whether the way the subsidy is offered — as a small loan, with requirements for complying with migration — leads those who would have been eligible under stricter criteria to take out the subsidy, while those who would generally not have been eligible do not take it up. If this ‘self-selection’ does indeed occur, we may be able to forgo an extensive — and expensive — targeting exercise, freeing up more of the budget for actual subsidies. Lastly, we also varied the number of subsidies to be disbursed per village to better understand the capacity of the field staff to handle various work loads.

All along, we are focusing on the design and codification of a clear set of program tools, protocols, checklists, and other supporting documents. Taken together, these will form the basis on which future iterations of the intervention are implemented and will enable a higher degree of fidelity to design — and impact — at scale.

What Are We Learning?

While the lean season is not over yet and more data is yet to come, a few lessons are already emerging:

• We are seeing substantial variations in take-up rates across branch offices (the unit of implementation at RDRS). It will be interesting to investigate whether this is indeed due to the design and delivery tweaks we deliberately introduced or to other factors such as differences in demographic characteristics.
• There appear to be variations in when exactly the lean season occurs, even in communities that are close to each other. In areas that are prone to flooding, for example, the water forces farmers to postpone planting, thereby slightly delaying the onset of the lean season. This information may help RDRS better plan the timing by which its field officers visit each village to offer the travel subsidies in future years.
• People still take the subsidy as an interest-free loan. This is consistent with the original RCT results, and we heard many stories of would-be migrants who had to resort to informal moneylenders charging high interest rates in the absence of the program, indicating demand for our product. Offering the subsidy as a loan instead of a grant helps reduce the cost per recipient, since the money recovered can be reinjected into the program to benefit new people.

We recently also started working with a local ethnographer to help us better understand the forces behind and barriers to seasonal migration, and expect to dig into the results later this year. Administrative data from RDRS has been streaming in as well, and we will be analyzing that to be able to provide much more specific information to migrants next year in terms of the type of jobs, wages, and living conditions they can expect at destination.

What’s Next?

This year represents the first major step in our plan to disburse travel funds to 310,000 households in Bangladesh by 2019, benefiting 1.4 million people.

We are currently planning a larger RCT of the program in Rangpur next year to confirm its impact and cost-effectiveness when it is delivered at scale. Working together with Innovations for Poverty Action, this will be the first time “No Lean Season” is evaluated when run completely independently by an NGO, outside of a tightly controlled research environment. Results of the evaluation, scheduled for early 2018, will allow the group of partners working together on this initiative to make practical decisions about proceeding towards scale in Bangladesh.

We are also working on a replication in Indonesia to explore the applicability of the program in a totally new context, together with the researchers of the original RCT and the support of the Abdul Latif Jameel Poverty Action Lab (J-PAL).

Meanwhile, we are thrilled to see that even as we expand our program and reach other areas of Northern Bangladesh, our small migration incentives continue to improve the opportunities of the rural poor to choose freely what is best for them and their families.