Published on Development Impact

Are uncertain urban labor markets deterring internal migration? Evidence from a field experiment in Kenya: Guest post by Gwyneth Miner

This page in:
Are uncertain urban labor markets deterring internal migration? Evidence from a field experiment in Kenya: Guest post by Gwyneth Miner

This is the fifth in this year’s series of posts by PhD students on the job market

Migration to cities is often a gamble for the rural poor: a move has the upsides of improved standard of living and higher income potential but downsides of missed rural earnings, relocation costs, and separation from home communities. These downside risks may explain why we don’t see greater migration to cities in African countries, motivated by better job opportunities, greater income growth, and higher possible consumption.

The difficulties of finding urban employment are exacerbated when rural migrants have little to no savings to smooth wage loss while job searching upon arrival in cities. Individuals living in their home villages are deterred from moving by the uncertainty of employment, especially when they lack an urban job contract waiting for their arrival. However, it’s unlikely that a person will develop a strong urban network, capable of providing a viable long-term job, if they remain in their village. Thus a dilemma arises where migrants do not want to move without urban job security, but job security is only possible by moving and strengthening urban job skills and networks.

Conditional unemployment benefits: a safety net for migrants

In my job market paper, I investigate whether a migration conditional unemployment benefits program can encourage rural-urban migration and support migrants as they transition to urban employment. The goal of this program was to help smooth income during periods of unemployment for migrants in Nairobi to prolong their move and result in more permanent migration.

I evaluate the impacts of this unemployment benefit program using a randomized controlled trial. Specifically, I conduct a large-scale experiment with 1,300 young adult male workers in 111 villages in rural Western Kenya and track their migration, employment and income over the course of a year. Participants were invited to the study under the qualification that they were male, between the ages of 18-35, owned a mobile phone, were not currently enrolled in school, and had some interest in migrating. On average, the villages were about 9 hours away by bus from Nairobi.

In the villages selected for the conditional unemployment benefits program, participants were informed that they could collect 500 KSH (about 4 USD, the daily urban wage for this sample) on the weekdays they were unemployed at the research team’s office in Nairobi. Participants could collect at most 15 times and typically spent about 100 KSH to travel to and from the collection point within Nairobi.

In addition to pure control, I include a study arm where rural workers are given an unconditional cash transfer, as another benchmark in this study. Individuals in this group were given a lump sum of 6,500 KSH (50 USD) with strong encouragement to use the funds for transport and initial relocation costs of moving. The size of the transfer was equalized to the maximum net gain an individual in the unemployment benefits group could expect from the study.

All study participants were surveyed via phone every four to six weeks for eight months. The unemployment program was available for the seven months following the baseline survey and the cash transfer was dispersed at the beginning of that period. A follow-up survey was conducted one month after the close of the unemployment program, and another survey round is currently underway, four months after the end of the program. Tracking rates for these surveys are high: 99% of the sample was surveyed at least once since baseline. Moving to a city has no effect on the likelihood of attrition.

Participants motivated by guarantee of support

The migration rate to Nairobi tripled over the course of the study in the unemployment benefit group, compared to control. The unemployment benefit treatment increases the likelihood of ever moving to Nairobi during the study period by 10.9 percentage points, over a mean of 4.7% in the control group (a 200% increase in the probability of moving to Nairobi). One month following the conclusion of the unemployment benefits program, the treatment effect of moving to Nairobi persists: 70% more migrants in the benefits group remained in Nairobi compared to the control group. The unconditional cash transfer also increases the probability of ever moving to Nairobi during the study, but to a lesser extent: a treatment effect of 4.5 percentage points (almost a 100% increase, over a control mean of 4.7%).

Figure 1 shows the proportion of each study arm living in Nairobi from the start of the study through the latest round of data collection. Members of the unemployment benefit group began moving immediately (although the unemployment program did not begin until January 2024), and at a constant rate for four months. Participants in the unconditional cash transfer group also quickly move to Nairobi, but at a lower rate. Migration to Nairobi among the control group is much more gradual. Ten months from the start of the study, migration in the control group begins to catch up to the level in the cash transfer group. This shrinking gap is an indication that the cash transfer may have moved forward the timing of migration for migrants in this group, whereas the unemployment benefits induced some individuals who would not have otherwise moved.

Figure 1: Proportion of each study arm in Nairobi over time

Image

 

Migrants using the unemployment benefit program as designed

Surprisingly only 59% of Nairobi migrants in the unemployment benefit group came to collect at least once. The partial take-up is not from lack of information about the program, instead those who moved to Nairobi but never collected reported that they did not have a need for the money. For those that did collect, the most common pattern was to collect once or twice a week and pause collections after securing a longer-term job contract. There were no instances of an individual collecting 15 times consecutively and immediately returning to their home village. Respondents report using their benefits for transportation costs to look for jobs, training fees, and living expenses (food, rent). These take-up patterns reflect that respondents were using the program as intended: smoothing their Nairobi income during times without work, but only when they needed it.

The tale of migration and opportunity costs

Migrants in all groups were on average younger, less likely to be married, and more likely to have finished secondary school than their rural counterparts. Figure 2 displays the propensity to migrate within income brackets, separately for each treatment group. To accommodate for outliers, the last income bracket is for those that make 16,000 KSH or higher, about 10% of the sample. The Nairobi unemployment benefits induced the most migration, across all income groups. In the control group, Nairobi migration was observed at the lowest and highest incomes. The unconditional cash transfer induced migration somewhere between the levels of control and unemployment benefits, at the lower end of the income distribution.

Figure 2: Propensity to Migrate to Nairobi by Income and Treatment Group

Image

The change in propensity to migrate by income group in the unemployment benefits arm demonstrates that the program brought down the opportunity cost of moving. If a worker makes a high (relative) income at home in their village, but moves to the city without a job, they stand to lose more income (a higher opportunity cost) than a person who was unemployed in the village but considering an urban move. The program was designed to reduce potential loss by mitigating dips in earnings with guaranteed support. The observed migration response is evidence that the unemployment benefits reduced the opportunity cost of moving to Nairobi, by guaranteeing income in an uncertain labor market.

Effects of migration

One concern about rural-urban migration is that a migrant’s quality of life may suffer due to worse living conditions or strains on physical and mental health. I find a positive treatment effect on housing quality in the unemployment benefit group; urban migrant housing is more likely to have finished walls, finished floors, and piped water connection. However, I also observe a 10% increase in reported stress, affirming that moves to cities can be mentally taxing.

Data collected in November 2024 will shed further light on the persistence of migration effects. At the time of writing, the data shows Nairobi migrants make more than their rural baseline income after several months in the city, not instantaneously after moving. This underscores a motivation for the conditional unemployment benefit program: migrants need time to become successful in a new city, but often lack the runway to do so.

Gwyneth Miner is a PhD candidate at UC Berkeley. 


Join the Conversation

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