good question — these are coefficients on the treatment dummy, so they represent the *difference* between the treatment and control groups. So the average caregiver salaries in the treatment group are given by the sum of the control group mean and the coefficients on treatment.
Thanks so much Nick, this is really helpful and interesting. We do base eligibility on objective criteria, such as asset ownership, housing situation, consumption, etc. We hesitate to exclude folks based on very broad categories such as location (although we do restrict eligibility by district for research and logistics reasons). So far this has been working reasonably well, as you can see from the data above on income amongst the selected students. Very interesting point about the government schools, we'll discuss that.
On the fraud, yes I'd be grateful for any advice you have!
Great job bot! Just one small correction — we actually haven't had any visa rejections so far! Once someone has a university spot and our funding commitment, it's pretty automatic.
Hello dear coauthor! Thank you so much for the thoughtful questions. Some responses below, marked with [JH]:
In what countries do educational migrants end up? How many stay in Germany after their degree versus returning home or elsewhere? This article says about half of the international students stay in Germany, but I can't find the study it links to. Also, what's the dropout rate of Malengo students?
[JH] We haven't been able to find good data on this so far. Our internal assumption (which is as much an estimate as an aspiration) is that 80% of students graduate, and 80% of those who graduate remain in Germany. But this isn't based on data. Anecdotally all students in the pilot cohort want to stay.
How much displacement of other students can we expect? I assume the cost-effectiveness will look quite different if it's closer to zero versus 100%, even if it mainly displaces middle-income country students.
[JH] In our view this is not a big worry, for three reasons that we described in the article above. Pasting again here:
First, many university programs in high-income countries are “open admission”, i.e. they are not selective, and anyone who fulfills the admission criteria is admitted. In fact, universities often have incentives to admit as many students as they can (while maintaining quality), as public funding is linked to the number of students.
Second, most current international students in high-income countries are from middle-income rather than low-income countries (DAAD, Statistisches Bundesamt 2019). Thus, any displacement would likely affect higher-income students.
Perhaps most importantly, it is likely that the number of available slots will grow substantially over time, both in response to demand from international students, and in response to demand for highly skilled labor in high-income countries. In this context, we note that an estimated 26.1m students start a Bachelor's program in high- and upper-middle income countries each year. Recall from above that 22.2m secondary school graduates from low-income countries may want to study abroad each year. Thus, existing capacity in high- and upper-middle income countries would have to increase by a relatively modest 85% to accommodate all potentially interested students from low-income countries. (About 9% of students in high-income countries are from abroad.)
How positively viewed is having to graduate international students stay in Germany? How much does this vary by country of origin? I'm concerned about potential political backlash, and I think the cost-effectiveness is probably quite a bit lower if most students return to their home country.
[JH] Like I mentioned above, we do expect a large share of students to stay. Potential political backlash is of course a worry. One hope I have for Malengo is that it can provide a forum where a respectful and evidence-based conversation about this can be had, similar to the kind of discourse GiveDirectly catalyzed around cash transfers.
What are the total costs, including staff and overhead costs? What are the scale-dependent costs? I understand the organisations will likely be able to exploit economies of scale if it grows, but it'd be nice to know what are the total costs of Malengo to send a student now and what it would be if you were sending 10,000 and 100,000 students, etc
[JH] We currently assume 36.9% overhead (see malengo.org/scale). This is realistic for the somewhat smaller numbers of students we'll send in the near term. Our current projections use the same rate for the longer term as well, but this is conservative; I hope that the rate can come down much closer to the twenties with scale and technology.
Is your team planning any RCTs or quasi-experimental studies? It seems like you could be placed to do some valuable evidence generation. As you pointed out, the causal evidence of immigration's effects on immigrants is very sparse.
[JH] Yes! There's an RCT built into the program, led by a great team (Matthias Sutter, Merve Demirel, Toman Barsbai, Marcello Perez Alvarez, Emmanuel Rukundo).
Hi Milli,
good question — these are coefficients on the treatment dummy, so they represent the *difference* between the treatment and control groups. So the average caregiver salaries in the treatment group are given by the sum of the control group mean and the coefficients on treatment.
Best
Johannes
Thanks so much Nick, this is really helpful and interesting. We do base eligibility on objective criteria, such as asset ownership, housing situation, consumption, etc. We hesitate to exclude folks based on very broad categories such as location (although we do restrict eligibility by district for research and logistics reasons). So far this has been working reasonably well, as you can see from the data above on income amongst the selected students. Very interesting point about the government schools, we'll discuss that.
On the fraud, yes I'd be grateful for any advice you have!
Johannes
Great job bot! Just one small correction — we actually haven't had any visa rejections so far! Once someone has a university spot and our funding commitment, it's pretty automatic.
Hello dear coauthor! Thank you so much for the thoughtful questions. Some responses below, marked with [JH]:
[JH] We haven't been able to find good data on this so far. Our internal assumption (which is as much an estimate as an aspiration) is that 80% of students graduate, and 80% of those who graduate remain in Germany. But this isn't based on data. Anecdotally all students in the pilot cohort want to stay.
[JH] In our view this is not a big worry, for three reasons that we described in the article above. Pasting again here:
First, many university programs in high-income countries are “open admission”, i.e. they are not selective, and anyone who fulfills the admission criteria is admitted. In fact, universities often have incentives to admit as many students as they can (while maintaining quality), as public funding is linked to the number of students.
Second, most current international students in high-income countries are from middle-income rather than low-income countries (DAAD, Statistisches Bundesamt 2019). Thus, any displacement would likely affect higher-income students.
Perhaps most importantly, it is likely that the number of available slots will grow substantially over time, both in response to demand from international students, and in response to demand for highly skilled labor in high-income countries. In this context, we note that an estimated 26.1m students start a Bachelor's program in high- and upper-middle income countries each year. Recall from above that 22.2m secondary school graduates from low-income countries may want to study abroad each year. Thus, existing capacity in high- and upper-middle income countries would have to increase by a relatively modest 85% to accommodate all potentially interested students from low-income countries. (About 9% of students in high-income countries are from abroad.)
[JH] Like I mentioned above, we do expect a large share of students to stay. Potential political backlash is of course a worry. One hope I have for Malengo is that it can provide a forum where a respectful and evidence-based conversation about this can be had, similar to the kind of discourse GiveDirectly catalyzed around cash transfers.
[JH] We currently assume 36.9% overhead (see malengo.org/scale). This is realistic for the somewhat smaller numbers of students we'll send in the near term. Our current projections use the same rate for the longer term as well, but this is conservative; I hope that the rate can come down much closer to the twenties with scale and technology.
[JH] Yes! There's an RCT built into the program, led by a great team (Matthias Sutter, Merve Demirel, Toman Barsbai, Marcello Perez Alvarez, Emmanuel Rukundo).
Thanks again for the great questions!
Johannes