Published on Development Impact

A successful intervention to boost international migration, and what it did for attitudes towards redistribution

This page in:

In “if it needs a power calculation, does it matter for poverty reduction?”, I conclude by saying “The distinction above between large transformative policies that dramatically reduce poverty and incrementalistic policies that raise the incomes of poor people a modest amount raises the question of why we don’t just put all our policy and research efforts on the former. One reason is that even if we know what we want to do (e.g. move people to richer places, generate vibrant and growing firms who hire lots of workers), we often have much less idea on how to go about doing this.”

One of the areas where this is clearest is with international migration. Thanks to a series of studies using oversubscribed lotteries of government migration programs, we know that moving to a richer country leads to massive increases in income that dwarf the impacts of pretty much any other development intervention we can think of. However, in the absence of new government migration agreements, efforts to simply facilitate more international migration within the existing system have proven much more difficult. For example, Emily Beam, Dean Yang and I tried a whole host of interventions in the Philippines including information, an online recruitment website, and assistance with documentation, and while this resulted in individuals taking more steps towards migrating, it had a precise zero effect on actual international migration.

A new experiment in Mizoram, India

Given this background, I was impressed to see the results in a new working paper by three political scientists, Nikhar Gaikwad, Kolby Hanson, and Aliz Tóth. They conduct an experiment that connected members of Scheduled Tribe (ST) groups in the Northeast Indian state of Mizoram with employment opportunities in the Persian Gulf region’s hospitality sector. Their main goal of inducing new migration is to use this to then measure the impacts of both the option and experience of migrating on political preferences for taxation and redistribution. But to do this requires doing better than most previous efforts in getting people to migrate.

Targeting those interested in migrating but who can’t do it on their own

International migration is a rare event. One of the challenges facing previous work has been identifying a set of individuals who really are interested in going abroad, but who also do not already have the network and contacts to allow them to do this. The authors overcome these issues by geographic targeting, self-selection, and screening. They chose Mizoram, a geographically isolated border area of India, with few existing connections to foreign employers. Their target population is English-speaking and relatively well educated, but face discrimination as minorities in India, making them particularly receptive to going abroad (the authors document that they think employers in the Gulf would value their skills more, treat them better, and discriminate against them less than do Indian employers). They then work with a state government office and local NGO to advertise the program, and have those interested in working abroad self-select into applying. Further screening is done on English-skills and education. This gives a population of young (average age 23), well-educated (>70% have passed grade 12), mostly unemployed youth who are interested in working abroad.


The sample size of 392 is then randomized into treatment and control groups, each of 196 individuals. The treatment consists of two components:

·       Basic training and interview preparation: a free five-week training program designed to prepare individuals for hospitality jobs in the Gulf. This started with classroom training, with modules on restaurant food service (e.g. food preparation, safety and hygiene standards), beverage and counter service (e.g. customer interaction, order-taking, billing), and housekeeping (e.g. managing household equipment, making beds). It then included on-the-job training in hotels, restaurants, and fast food chains. They also got help preparing resumes and practicing their interview skills. Employers give more intensive training at destination, so the goal was to equip them with the basic knowledge to pass at interview and to understand what the jobs abroad would be like. 58% of those assigned to treatment attended training.

·       Connecting to job opportunities and helping with paperwork: they worked with a list of vetted potential employers (e.g. international chains like Pizza Hut, Costa Coffee and Mandarin Oriental). Everyone assigned to treatment was invited to interviews, typically multiple times. Those with job offers then got logistical assistance in getting visas and medical certificates, and checking paperwork.

The authors estimate that this assistance cost approximately US$200 per person.


The authors measure results with two surveys. First, a midline survey was done after training but before individuals received job offers or moved abroad. This was completed by 290 of the 392 (74%). An endline survey was then done in January-March 2021, about two and a half years after the beginning of the study. This was answered by 248 of the 392 (63%). In both cases attrition rates were about 5% higher for control than treatment, but uncorrelated with baseline characteristics. Both surveys were done by phone – I imagine the pandemic prevented in-person surveying at endline.

This assistance actually did increase international migration, crowding out domestic migration

According to the endline survey, 23 percent of the treatment group had lived overseas at some point in the past 2 years, compared to only 3 percent of the control group. This 20 percentage point increase in migration is a large impact, and provides a proof-of-concept that this training and placement assistance to a targeted sample can induce more international migration. The majority of those that moved went to Kuwait and the U.A.E., about half on one-year contracts, and the others on multi-year contracts. Note that while these impacts are large in percentage point terms, given the sample size at endline, this equates to only 26 additional migrants (or 39 if extrapolated to the full treatment sample). So we should be cautious in interpreting impacts of migration itself given the numbers who actually move.

This increase in international migration is almost entirely offset by a decline in domestic migration: 32 percent of the control group move domestically, compared to 13 percent of the treatment group. But international migration pays a lot more: an average of $540 per month for those employed abroad, versus $250 for domestic migrants and $170 for those who stay put. The ITT on monthly wages at endline is for a doubling of wages, for an increase of $75 per month. Note that only 43% of treatment and 39% of control are employed, some of those treated have returned already, and not all treated migrate, which is why this ITT impact is less than the simple comparison of incomes at home and abroad.

This migration also has benefits for the family of the migrants, who report higher family income, and more ownership of durables like computers, refrigerators and motorbikes.

Even ignoring the benefit to the family, in cost-benefit terms, the intervention costs $200 per person assigned to training, and increases income by an average of $900 in a year, so pays for itself in less than three months. This raises the question of what the market failure is here – that is, why don’t individuals borrow against the income to be earned abroad (as happens in many settings, with many middlemen offering loans)? The authors suggest that while their skills training program has some close local substitutes, the placement assistance they provide does not-  that is, even with $200, an individual who wanted to migrate is unlikely to get the connections and interviews with vetted employers the program provided. This raises the question of how much extra demand for workers is out there that such a program could be scaled up to serve?

This increase in migrant opportunities is accompanied by less support for taxation and redistribution

The authors find that treated individuals are more likely to oppose high taxes to support social spending, more likely to disagree with the government trying to reduce income inequality, and more likely to agree that the poor can advance through hard work alone. The authors explore mechanisms for this effect, to see whether it is a direct effect of experience abroad (other research has argued that exposure to foreign institutions can change migrants’ views, but this may apply more when moving to e.g. Western countries with strong and efficient social services than to Gulf countries), an effect of the higher income per se, or also an effect of the expectations and exit option that migration provides.

The authors use a couple of nice approaches here. The first is to use the dynamics of their surveys. By documenting changes in these political attitudes in their midline survey, they can demonstrate that this increase in the possibility of migrating (before it has occurred or is certain) itself leads to changes in attitudes. Then, they use machine learning (BART) to see what baseline characteristics predict the likelihood of migrating in the treatment group, and compare impacts for likely migrants and likely non-migrants (another solution to the endogenous stratification approach blogged here). They find that while impacts are bigger for the likely migrants, even those less likely to migrate change their preferences, providing support for the exit option. In contrast, they find little support for the idea that exposure to institutions abroad was what changed preferences.

A final note on ethics

International migration offers the potential to dramatically improve well-being. But there are also risks of employer exploitation, concerns about separating families, and other risks involved. The paper has a very detailed ethics appendix. Appendix A.5 provides a 5 and a half page detailed discussion of ethical issues, going through each of the 12 principles and guidance for human subjects research set out by the American Political Science Association. Here they discuss potential concerns with employment conditions in the Gulf countries, and efforts they took to protect subjects from harm.


David McKenzie

Lead Economist, Development Research Group, World Bank

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000