The Impact of Malengo

How will the Malengo program affect the lives of the scholars? Will it also affect their families and home communities? Because Malengo is new, there is no rigorous direct evidence on its impact. However, we can make projections.

Impact on recipients

The cost of the program per student net of administration costs is EUR 11,971, which includes living expense stipends, semester fees at the host universities, application process expenses (visas, tests), and travel. Administration and staff costs are not included in this figure to provide a scale-independent cost estimate.

We expect that 80% of students graduate, and that the average time to completion is just under 4 years for a BA and around 6 years for a Master’s. Of those who graduate, we expect 80% to remain in Germany. Of these in turn, we assume that 33% continue to complete a Master’s degree. Of the other 67%, we expect 80% to take a full-time job, 15% to take a part-time job, and 5% to have no job. We assume that it takes students 8 months on average to find work, and that starting salaries are EUR 41,300 for those with a BA, and EUR 46,709 for those with a Master’s. Of those who do not graduate, we assume that 90% to return to Uganda and earn a yearly income of EUR 1,598, and 10% to remain in Germany performing a minimum wage job. We assume that 78% of students have 1.6 children on average in the first 10 years after program entry, and that students leave the labor force entirely for 3 months for each child. We anticipate yearly income growth of 4% in both Germany and Uganda.

Using these assumptions, the average yearly income across the entire pool of students in the first year after graduation is EUR 24,980. Compared to the outside option of taking a job in Uganda with the average yearly income of those with a high school diploma or some college, EUR 1,598, this corresponds to a treatment effect on income of 1,563%.

This compares favorably to the treatment effects of unconditional cash transfers on income, which are typically between 10 and 30%. It is important to note, however, that cash transfers are usually much smaller in magnitude. To provide a better comparison, we can calculate the internal rate of return of the initial program investment of EUR 11,971. During the first 10 years after program entry, the assumptions listed above imply a cumulative return of 1,717%, which corresponds to a yearly internal rate of return of 20.87%. In comparison, unconditional cash transfers generate an internal rate of return around 1.68% over this period.

Note that these calculations do not include spillover effects to family members in Uganda. We do not have good estimates of how much students will send back in remittances, but we expect these amounts to be substantial.

Impact in utility terms

Malengo makes very large transfers to a small number of students, instead of making smaller transfers to a larger number of students. Is this an effective use of money?

One approach to answering this question is to consider the impact of the program in terms of utility. A concave utility function, such as logarithmic utility, reflects the idea that any additional income increase is less impactful in terms of welfare for a person who already has a high income, compared to someone with a lower income.

We can perform a rough calculation to get a sense of the treatment effect of Malengo in utility terms. We assume that during their studies, Malengo students consume EUR 28 per day (this corresponds to their living stipend). After they finish studying, they consume EUR 68 per day, reflecting their average starting salary described above. Without the program, students would consume EUR 1598/365 = EUR 4.38 per day. Malengo therefore has a treatment effect in log terms of log(28) – log(4.38) = 1.87 while studying, and log(68) – log(4.38) = 2.75 afterwards. Assuming a 2% yearly discount rate, and 4% yearly income growth, this corresponds to a treatment effect of 33.24 log points for the Malengo program over the first 12 years after program entry.

Cash transfers are again a useful benchmark. Egger et al. (2021) find a 13% increase in consumption after USD 1000 (EUR 888) cash transfers to low-income families in Kenya. We know less about how this treatment effect evolves over time, but existing studies suggest that it decays relatively quickly. We therefore assume that it remains unchanged for 5 years after the transfer, then goes to 50% magnitude for 2 years, and then to zero. Using the same 2% yearly discount rate, and extrapolating the effect of the USD 1000 cash transfer linearly to a transfer of EUR 11,971, the total treatment effect over the first 12 years after program entry is 9.83 log points for unconditional cash transfers of the same magnitude as the cost of Malengo.

Note that this calculation is somewhat generous to cash transfers: these don’t have to be repaid, while Malengo students are asked to make contributions to future generations of students via an income share agreement. This will decrease their own welfare gains from the program, but increase those of others (who have lower income and therefore higher marginal utility).

Impact on non-recipients

This calculation also neglects another important feature of both cash transfers and the Malengo program: both are likely to have spillover effects to non-recipients. In the case of cash transfers, we have evidence on the magnitude of these effects: Egger et al. (2021) calculate a transfer multiplier of about 2.4 after $1000 unconditional cash transfers in Kenya. This implies that we should scale up the treatment effect for cash transfers by that factor. (This is slightly generous; Egger et al. find this multiplier over the first 18 months or so after transfers, and it is likely smaller later.) On this calculation, the total treatment effect of unconditional cash transfers including spillovers is 9.83 x 2.4 = 23.58 log points. Note that this is still substantially smaller than the treatment effect of Malengo without spillovers (33.24 log points).

The spillover effects of Malengo are difficult to predict. However, several studies suggest strong human capital responses to migrant exposure. Bedasso et al. (2020) find an effect of migrant exposure on completing secondary education of 14–17%, and on own future migration of 22%. Abarcar and Theoharides (2021) show that for each nurse who migrates from the Philippines to the USA, nine additional nurses are licensed in the Philippines. Dinkelman and Mariotti (2016) show that migration opportunities to South African mines increased human capital by 5–7% in Malawi. Batista et al. (2011) estimate an elasticity of secondary school completion likelihood with respect to migration likelihood of 0.4 in Cape Verde. Thus, Malengo students are likely to have significant positive human capital externalities on their families and home communities.

Likely remittance flows

In addition, Malengo students are likely to send back remittances, further increasing welfare in their families and home communities. Note that such remittances will accrue to lower-income individuals (e.g., family members staying behind) and thus have a larger treatment effect in log utility terms.

To estimate the magnitude of such remittances, a useful statistic is that the average Ugandan migrant in the UK sends USD 4,000 per year back to Uganda (Cooper et al., “Remittances in Uganda”, Centre for Financial Regulation & Inclusion, 2018). While we don’t have data for the remittances sent by the 2,600 Ugandans living in Germany, it is likely that the magnitude is similar due to the relatively similar household incomes in Germany compared to the UK. (The per capita GDP of Germany is 14% higher than that of the UK.) Note that this amount is four times as large as a typical unconditional cash transfer; and that these remittances flow every year, rather than being one-off payments (as is often true of cash transfers). Thus, it is likely that Malengo students who remain in Germany contribute significantly to the economic well-being of their families and home communities in Uganda.

Impact evaluation

To evaluate the effects of the program both on students themselves as well as on their home communities, we will conduct a randomized controlled trial which takes advantage of the lottery element of the program: after students are screened for eligibility based on their scholastic achievements and their family wealth, a lottery determines who amongst those who are “above bar” is offered participation in the program. This lottery creates randomly chosen treatment and control groups. We will follow both groups over time using surveys and interviews, and comparison of the two groups can then be used to determine program impact. Importantly, we will survey not only the students themselves, but also their families and home communities, in order to assess spillover and general equilibrium effects of the program. In addition, a qualitative research element will provide a richer picture of program impacts. A power calculation shows that a treatment group of 450 program recipients and 1600 non-recipients (controls) allows us to detect treatment effects of 0.15 standard deviations (SD) with 80% probability. 

References

Abarcar, Paolo, and Caroline Theoharides. 2021. “Medical Worker Migration and Origin-Country Human Capital: Evidence from U.S. Visa Policy.” The Review of Economics and Statistics, October, 1–46. https://doi.org/10.1162/rest_a_01131.

Batista, Catia, Aitor Lacuesta, and Pedro C. Vicente. 2012. “Testing the ‘Brain Gain’ Hypothesis: Micro Evidence from Cape Verde.” Journal of Development Economics 97 (1): 32–45. https://doi.org/10.1016/j.jdeveco.2011.01.005.

Bedasso, Biniam, Ermias Gebru Weldesenbet, and Nonso Obikili. 2020. “Emigration and Education: The Schooling of the Left behind in Nigeria.” Migration and Development: 1–17. https://doi.org/10.1080/21632324.2020.1806605.

Cooper, Barry, Antonia Esser, Rose Tuyeni Peter, and Shazeaa Lal Mohamod. 2018. Remittances in Uganda. Centre for Financial Regulation & Inclusion.

Dinkelman, Taryn, and Martine Mariotti. 2016. “The Long-Run Effects of Labor Migration on Human Capital Formation in Communities of Origin.” American Economic Journal: Applied Economics 8 (4): 1–35. https://doi.org/10.1257/app.20150405.

Egger, Dennis, Johannes Haushofer, Edward Miguel, Paul Niehaus, and Michael Walker (in press). General Equilibrium Effects of Unconditional Cash Transfers: Experimental Evidence from Kenya. Econometrica.

Independent Evaluation Team