In many parts of the rural developing world, a lack of jobs for agricultural workers in the months between planting and harvest creates a lean season at a predictable time each year. The nonprofit Evidence Action estimates that each year, some 600 million people experience this seasonal hunger.
Government and charitable programs can temper the impact, but they are costly and aren’t a solution to the core issue—a lack of work opportunities. Mushfiq Mobarak, a development economist at Yale SOM who often uses field experiments as a way to develop and test welfare-improving interventions, designed a program that offered round-trip bus tickets to agricultural workers, so they could migrate temporarily to cities where there was work. In Bangladesh, the $11 investment provided remarkable returns for the workers, their families, and their communities.
Yale Insights talked with Mobarak about the experiment, its implications, and plans to expand this intervention and adapt it to address to seasonal poverty in other countries.
Q: Would you describe briefly the problem of seasonal hunger?
In South Asia and many developing countries, most people who engage in agriculture are actually landless. They supply their labor on other people’s farms to earn an income. This can lead to a seasonal poverty problem. As an example, in Bangladesh, farmers plant rice in August and harvest in January. When there is planting activity, there’s a lot of demand for their work, wages are high, and there are job opportunities. During harvest, again, there is a lot of work and wages are high. But the period in between—September to December—is a lean season when labor demand is low, and therefore wages fall. It unfortunately happens to be the time when the price of rice also spikes.
This coincidence of high prices and low wages leads to much lower caloric intake. It ends up being a hungry season when people have to go with less food. This can have longer-term adverse consequences: if children don’t get adequate nutrition for three months of the year, then that might lead to stunting which is associated with lower productivity later in life.
Q: Would you describe your idea to smooth out the shock of seasonal poverty?
The lean period is such an expected trial that it has a name in Bangladesh: monga, which roughly translates to a period of high prices. The government and many NGOs that operate in rural Bangladesh know about the problem. They have tried to address it with free food distribution, food-for-work programs, and targeted microcredit. But these programs may not be very cost-effective, because they are not working with the underlying economic structure: there are no jobs in rural areas during these periods. So rather than forcing job creation in an economy that doesn’t support it, perhaps it would be cheaper and more effective to encourage people to move to where the jobs are.
We went about testing the idea using a randomized control trial. That’s the gold standard for rigorously evaluating whether a program works. We provided grants of about $11 to pay for round-trip bus fare plus a few days of food to landless households in rural areas suffering lean season deprivation.
The grants were conditional on one member of the household migrating for some period during that lean season to a city, in search of employment. We provided some general information on where people could go and the wages they might expect. We didn’t put any restrictions on which household member should travel, how long should they go for, or where they should go.
We ran the grant program in one set of randomly selected villages. In a second set of randomly selected villages, we ran the same program, but rather than a grant we offered a zero-interest loan that they were expected to pay back after the harvest. In a third set of villages, we just provided the information without offering the loan or the grant. A final set of villages served as a control group, a comparison point.
We found the migration subsidy, either in the form of a grant or a loan, induced many people to go. In the control group, about 36% of households traveled. In the information-only group, 36% of households traveled, which means that information was not the constraint that they were facing. However, if you offered the loan or the grant, the migration rate jumps to 56% and 59%.
Q: What did the experiment tell you?
The first result is, many people can be induced to go and try this out. About a quarter of households try migration on the basis of this transfer. The second result is that if one person, typically the adult male, migrates, that entire family consumes 600 to 700 calories more per person per day. Because the rest of the family stayed in the rural area, we could measure caloric intake very precisely. The third result is that three years later—we continued tracking what happened in these villages, without repeating the program—many people choose to go back to the city during lean seasons on their own, and they pay for those subsequent trips themselves.
Q: What are the implications of these results?
The program is quite successful helping people cope with the lean period in a cost-effective way. On the basis of a one-time transfer of $11, about 80% of people are successful at finding work. For the average migrant this produced about $110 to $120 in income in the city. That pays for a lot more calories at home.
Q: Could you explain how risk plays into the decision about whether to migrate for work?
Risk is a big deterrent. Imagine that I offer you a lottery ticket that costs $11, and I tell you that 80% of the time you’ll get a $60 return. But 20% of the time, you’re going to be unsuccessful and you lose your money. It sounds like a fantastic deal.
However, the $11 transfer is equivalent to 10 to 12 days of lean-period wages. Paying the $11 out of pocket and coming back home empty-handed if you’re unable to find a job in the city could be devastating. When people are under the threat of famine, being $11 poorer might drop the family below subsistence. In those circumstances, it becomes very difficult to go after that upside opportunity, even with the 80% success rate, because they cannot manage that 20% downside risk.
The fear of that big loss keeps people from pursuing opportunities that can take them out of poverty. In that sense, this operates like a poverty trap, where people are so poor that they cannot go after good economic opportunities.
Q: Are there ways people find to mitigate the risk?
Yes, in further research we find that when people migrate, it’s because they are already operating in informal risk-sharing networks with other members of the village. When some network members have an income source that’s less correlated with the income sources within in the village, then they can insure each other much better. Not only does it benefit the people who migrate, it indirectly benefits others who live in the same village and happen to be sharing risk with the migrants.
In another round of experiments, we also find that providing a larger number of subsidies to many people in the village simultaneously makes the program work better. People travel together to help mitigate the risk of migration.
Q: What are the next steps for this research?
We are pursuing two streams of research. One is whether this program ought to be scaled up in Bangladesh. We’re working with two NGOs, Evidence Action and RDRS, to scale this eventually to 310,000 households. As we go, we will be testing: does the program work with 200,000 people in the same way that it did at the pilot scale of about 2,000 people? Will there still be an 80% success rate, or would something fundamentally change? For example, if the most productive people migrate, could those left in the village end up much worse off? We’d like to be careful about unintended consequences of migration. Similarly, when you move to a scale of hundreds of thousands of households, perhaps the city will run out of its absorptive capacity for taking migrants. It might lower wages or cause other problems in slums, and that’s another thing that we’d like to track.
Separate from all the economic factors, we’d also like to make sure that there aren’t unintended consequences of migration in other spheres of life. For example, maybe it changes the relationship between the men who migrate and their wives. Maybe it leads to more divorces, or there might be more disease transmission back and forth between the village and the city. Or their political beliefs might change.
We kept track of all of this as we scaled up the program to about 36,000 individuals. Thus far, we find an additional bonus for the village economy, because encouraging people to migrate during the lean season creates some slack in the village labor market, and wages get bid up because employers have to try harder to find workers. Otherwise, the migration doesn’t lead to fundamental changes in terms of how the household operates, long-term migration patterns, or disease prevalence, at least in Bangladesh.
As we think about moving the program to other countries, we might have to take more care to look at those factors. For example, sexually transmitted disease rates are much higher in sub-Saharan Africa than in Bangladesh. If this program were to port to Africa, we would want to track that more closely.
Q: You mentioned a second stream of additional research?
In another track of our research, we’re now considering whether this program would replicate in other contexts. The experimentation in Bangladesh, as well as the theory-building that we’ve done, helps us identify the precise conditions under which a program like this is most likely to be helpful and successful. For example, we’d want to operate in a country where there is a predictable lean period and cities with vibrant labor markets that can absorb an influx of temporary migrants into unskilled jobs that even rural farmers can do. We would want to focus on areas from which it’s very risky for people to travel. For a rural area within an hour of the city, people would have experimented with migration themselves; there’s no reason for us to subsidize that travel. Whereas, a rural areas five hours away from a city, where people are extremely poor and face the same type of risk that we observed in Bangladesh, a migration subsidy program could help people learn whether the city is for them.
We looked for those criteria in Malawi, but there isn’t a city with a labor market that can absorb lots of migrants, so we decided not to proceed. In the eastern islands of Indonesia, there is a lean season pattern that’s similar to Bangladesh, and there are cities that offer job opportunities in industries like construction that can take in lots of rural workers during the lean period. Our next step is to move to a full-scale experiment phase in Indonesia.
We are also experimenting with this idea in Nepal, where the border with India is open, so rural Nepali workers can move to Indian cities that offer much larger wage gains than you see within the same country.
Q: How does this work fit with larger patterns of urbanization and rural development?
The world is urbanizing, with or without this program. Historically, in the United States and Canada, and more recently in Brazil and China, cultivators start mechanizing when they saw labor shortages as people moved to cities. They invest in labor-saving technologies where you are substituting capital for labor.
The process of structural transformation from a very rural, agricultural-focused economy to a very urban, high-growth economy took about 150 years in the U.S. In China that same amount of rural-urban movement happened in about 30 years. We need to be prepared for these types of large-scale movement in other developing countries, and design policies that can facilitate and enable high-growth outcomes.
The next phase of experimentation that I plan to do in Nepal will be to combine these types of migration subsidy programs directed to landless laborers with access to labor-saving technologies directed to the cultivators who employ them.
Additionally, while we figured out that migration may not be the right solution in Malawi, seasonal poverty is still omnipresent. We need other innovative strategies. Seasonal hunger doesn’t have a one-size-fits-all solution. We’d like to experiment with a suite of context-relevant policies to address these problems.
Evidence Action's No Lean Season program was recently named to the list of Top Charities recommended by GiveWell, a nonprofit that evaluates the impact of philanthropic organizations. Learn more and donate on the Evidence Action website.