Migration of Doctors and Nurses

This post/review of evidence was requested by Charlie Goldsmith. I take requests for post topics, especially if there is a policy-related reason you would like a post to exist; just email me at lagilbert@gmail.com.

Skilled professionals - especially those with globally valuable skills - often seek to leave low income countries. Generally, the more educated you are, the more able you are to leave. Health workers are the classic example of this. There is often much public concern about doctors leaving the country; in many low-income countries, the majority of medical personnel want to leave the country.¹

In 2020, 3.9% of physicians trained in a LMIC practiced abroad, though this is highly heterogenous by region. Brain drain is a problem in Africa, where 11.3% of physicians trained there practice abroad; it is not in the western Pacific, where only 0.9% of physicians practice abroad. Most that go abroad go to a handful of rich countries - some 25% of US physicians were trained outside the US, and 17% of NHS staff (all cadres) report a non-UK nationality.

For the individual, migration makes a lot of sense; it is not uncommon to 10x one’s salary by moving to the US as a nurse. For the countries they leave behind, it’s less clear. There is much public policy concern that low income countries invest large amounts into training health personnel, but do not receive the benefits. It is not difficult to find papers expressing concern about the extent of international brain drain - here are just a few.

This literature tends to focus on the most highly skilled health professionals; doctors, primarily, and to a lesser extent nurses. Public policy, too, focuses on doctors - for instance, Nigeria considered forbidding doctors from practicing outside the country for five years after completing their training.

It is, however, worth noting that physicians are a relatively small minority of health personnel in most developing countries. Less than 10% of the health workforce are MDs. There are about 2 physicians per 10,000 people in sub-Saharan Africa; there are about 12 nurses or midwives per 10,000 people. There are also a relatively large number of health workers with less formal training, including community health workers - perhaps three times as there are doctors.²

Much of the below discussion will still focus on physicians because that’s where academic interest has tended to focus. There are a handful of papers on nurse migration - some of which I will discuss below - but I am not aware of any papers on international migration of other health workers. If you are, please bring them to my attention! This is a living lit review, and I will edit.

The Classic View of Brain Drain

Fiscal Cost

In most cases, educating doctors costs the state money. While the level (and existence) state subsidy for medical education varies substantially across countries, I’ll come up with a reasonable guesstimate based on a country frequently concerned about brain drain.

In Nigeria, private medical school in Nigeria can cost up to $10,000 a year.³ Let us say that is the true cost of educating a doctor (though obviously this does include some profit margin). Since state medical schools are <$500 a year, the state is therefore subsidizing public medical schools by nearly $10,000 per student per year. The length of medical training also varies across countries but it seems reasonable to suppose that a low-income country government might spend perhaps $30,000 training a nurse or doctor.

There are currently 5,204 physicians trained in low income countries working abroad, so this equates to a fiscal cost of $150M for physicians alone. It seems likely that nurses practicing abroad result in a larger fiscal cost, since nursing training is not that much shorter than physician training and there are significantly more nurses in low-income countries than there are doctors.

Still, this doesn’t seem huge; low-income countries currently have a GDP of $660B. The fiscal loss from doctors migrating abroad is perhaps 0.02% of GDP; this seems small enough it shouldn’t be a major public policy issue.

Not Enough Doctors = People Die

Of course, the fiscal cost is not generally the main reason for concern about migration of health personnel. There are also more difficulties delivering care if there are fewer nurses or doctors available.

The basic case for brain drain being bad is fairly obvious: if you hold the number of medical personnel constant and their skills constant, it is better to have more people in a country than fewer people. In the short run, it would be more advantageous to countries if all graduates stayed in country.

Indeed, given the dearth of medical personnel in many LICs, people will die due to lack of doctors. One paper (Saluja et al 2020) estimates⁴ that the average physician in a low-income country saves 4.6 lives per year. As there were (at time of paper publication) 5,204 physicians trained in LICs abroad, this paper estimates that there are 23,938 avertable deaths in LICs per year due to physician brain drain.

This is a relatively small percentage of total deaths. LICs see about 5.3 million deaths per year, so this is 0.45% percent of deaths per year. That being said, many deaths are not avertable; 20,000 deaths per year is roughly the same number of deaths per year one would expect to avert with 10 million doses of R21, a malaria vaccine, so it’s not a trivial number either.

Why low-income countries instead of low and middle income countries?

You may have noticed I specified only low-income countries above.

The numbers look much larger if you consider middle-income countries. There are only 26 low-income countries; there are 105 middle-income countries. Similarly, there are not 5,000 physicians that are abroad; there are 254,250 from middle income countries that practice abroad.

I focus on low-income countries for the following reasons:

1. An additional physician makes much more of a difference in a low-income country than a middle-income country. The above paper estimates that the average physician in a low-income country saves 4.5 lives a year; the average physician in a middle-income country saves 0.168 lives per year. This implies that brain drain costs 25,000 lives per year in low-income countries (total population 736M) but only 40,000 lives per year in middle-income countries (total population 5.9 billion).

That result is also consistent with other evidence - providing more physicians in Nigeria (not quite low income but close) improved infant mortality outcomes⁵ substantially, while the effects of providing more physicians in Brazil (edging towards high-income) was quite modest. This seems pretty intuitive; the first doctor in a town probably does make a huge difference to survival rates; the fifth… eh, not so much.⁶ Medical personnel, like many other things, appear to have diminishing marginal returns.

2. More doctors (in percentage terms) leave from low-income countries than from middle-income ones. Saluja et al 2020 estimates 11.3% of physicians trained in low-income countries practice abroad; 3.8% of those trained in middle-income countries do.⁷

Given these two facts, brain drain really is a low income country problem. Given the low number of avertable DALYs and lower rate of migration, brain drain is close to a rounding error in middle income countries.

There is also a policy reason to focus on low-income countries. The list of low-income countries is an almost perfect subset of the WHO red list - that is, the list of countries where the WHO is so concerned about health brain drain that international health recruitment is banned there.⁸ International medical recruiting is allowed in 75% of middle-income countries; it is banned in nearly every low-income one.⁹

For most middle-income countries, brain drain might be a topic of concern, but it’s probably not existential. If there are countries should worry about brain drain, it should be low-income countries.¹⁰

However, there remains the question: does brain drain actually cause 25,000 deaths a year in low-income countries?

Correlations

For a start, countries that send more physicians abroad do not have worse public health outcomes. If anything, the opposite is true - countries that send many doctors abroad have better public health outcomes (see figure 5 here).¹¹

(Figure 5 here)

This doesn’t tell us all that much, of course; perhaps countries that produce more physicians are better at healthcare in general. That could lead to both better domestic outcomes and more migration. But it is at least suggestive that health migration isn’t catastrophic for the sending country, or we would be able to see a correlation.

Remittances

Why not? Well, perhaps the most obvious reason might be that migrants do contribute to their country even while abroad. Because health professionals are relatively high income, they can send back larger amounts in remittances than in lower-paid professions. For instance, nurses’ remittances make up 25% of the remittances sent back to the Philippines, despite making up only ~15% of overseas Filipino workers.¹²

In 2011, the average African physician practicing abroad sent back $5,400. Since (horrifyingly) that was 14 years ago, we can inflation adjust that to estimate that the average African physician practicing abroad probably sends back around $7,200 a year today.¹³

At this rate, it would take about 4.16 years for a physician to remit the entire cost of their education to their home country. While I have been unable to find a good current estimate of the career life expectancy of an immigrant physician practicing in a rich country, it is likely to be >4 years.¹⁴ If an immigrant physician practices abroad for 20 years, they will remit >$100,000 more than they cost their country to train.

When one considers nurse remittances as well, it seems likely that doctor and nurse migration is a considerable source of income in low-income countries. Remittances also improve health in those countries; families of those remitting money use that money to purchase more medical services.

Are the positive health impacts of remittances to enough to make up for the negative impacts from losing doctors and nurses? That is difficult to tell without better numbers on 1) the amount of remittances, 2) the elasticity between DALYs and remittances, and 3) estimates of lives saved per health worker. But it is sufficient to say that simply considering the negative impacts of health worker migration - and not the benefits - will significantly overstate the costs of brain drain.

We cannot say - as one well-cited paper does - that brain drain costs a total of $15B a year if that number is not subtracting off the gains from remittances.¹⁵

Further Complicating The Brain Drain Narrative

Furthermore, in general equilibrium, the number of health workers is not fixed. Indeed, one might imagine that more people become doctors or nurses because it is an internationally valuable skill, always intending to go abroad.

The classic case of health workers training to migrate is nurses from the Philippines. There is an extremely long history of nurses from the Philippines migrating to the US, and the Filipino government encourages workers to go abroad. About 40% of foreign-born nurses in the US are from the Philippines.

It is true that the Philippines is an unusual case here. The government has actively encouraged migration, rather than trying to discourage it. Still, it provides a case study of “made for export” nursing.

This is the scenario examined in Abarcar and Theoharides 2024. This paper uses a natural experiment - in 2000, the US drastically expanded access to visa for foreign nurses; in 2007, they contracted it back down to pre-2000s levels.

Since different provinces produced different levels of nursing graduates, these changes had differential effects across the Philippines. They exploit this variation to see how people responded to the changes in policy and likelihood of migration and find that:

• Nursing school enrollment tripled when migration was easier.

• Many more nursing schools opened (about a 10% increase year-on-year).

• Considering only the population of “migration-opportunity-created nurses”, only about 1 in 10 ended up going abroad.

• This drastically increased the supply of nurses in the Philippines.

The expansion of nursing was not entirely a free lunch; the quality of nurses (as measured by exam pass rates) did decline. However, the Philippines simply did not allow nurses that failed their exams to practice as nurses, so I’m not overly worried that education for migration meant that the Philippines was overrun with dangerously incompetent nurses. And, honestly, I suspect nine sort of mediocre nurses can provide better care than one really awesome nurse. Quality matters in nursing, but so too does volume.

But to return to our primary question. If we are worried there will be fewer doctors or nurses available at home if migration is allowed, this paper would suggest… not necessarily. If we assume the supply of doctors and nurses is fixed, yes, sure - but in real life, it’s not.

When there is the option to migrate, investing the time and effort into becoming a nurse or doctor seems more attractive. If one can only work in the Philippines, you’re looking at four years of education to make around $60 a week. This is below average wage in the Philippines.

Instead, it appears that people decide to become nurses or doctors in hopes of being able to move to the US. But then life happens, and people get married and have kids and actually end up deciding that moving halfway across the world doesn’t sound so great after all. Some stay in that profession in their home country; some that were never all that suited to it in the first place (and only did it in order to migrate) do something else instead.

The Philippines ultimately did not experience brain drain from sending nurses abroad; it experienced brain gain. It seems plausible other countries can repeat their experience.

There are two important caveats to this relatively positive story about brain gain.

1. The Philippines allowed nursing schools to expand in response to increased demand. Not all countries do this; some countries severely limit training places.

However, this would seem to be a relatively easy policy lever to pull. Almost all countries have at least some private universities; simply allowing more private universities to educate medical personnel would suffice. There are also examples of states requiring state-funded institutions to increase intake; recently, Nigeria has doubled their medical school intake.

There are also ways that sending countries can encourage this. Lee Crawfurd and Helen Dempster highlight bilateral labor agreements where receiving countries make investments in the country of origin. One could imagine the UK funding medical school places in proportion to the number of doctors from that country practicing in the NHS.

2. Nursing is a very useful thing. It appears that wage gains from migration will cause people who otherwise would not choose that field to go into it (see also IT workers in India). For health professions in low-income countries, this seems generally good. One could imagine that for other professions, vastly increasing the number of people doing it might not be helpful.

Let us say, for instance, that there was suddenly huge market demand for philosophy professors. One could imagine that, actually, a country would not be better off with a suddenly increased stock of philosophy professors; probably there wouldn’t be good domestic jobs, it wouldn’t improve health or growth much; the country would not be much better off.¹⁶

When we are talking about health workers, generally we want there to be more domestic supply, so this kind of overproduction relative to the base case is good. But that may not be true of all professions.

Conclusions

For all the public policy concern about brain drain, the quality - and quantity - of evidence here isn’t great.

Here’s what we do and don’t know:

1. We know nothing about brain drain of the most common types of health workers in low-income countries; the existing literature tends to focus on doctors, which are quite rare in many developing countries!

2. If brain drain is a problem, it is a problem in low-income countries. In terms of deaths caused (and physicians migrated), brain drain is pretty close to a rounding error in middle-income countries.

3. All else equal, it’s probably true that more doctors are better than fewer doctors. This is especially true in low-resource contexts; it seems like additional doctors are probably not that useful in higher-resourced countries.

4. That being said, more opportunities to migrate probably means more people decide to become doctors, and that is also probably good.

5. There are probably policy levers to make it more likely that you produce more nurses or doctors both for the international and domestic market, such as expanding medical schools.

6. Remittances from health workers abroad are significant, but there’s little work on how one can trade off that extra income vs. having them work in their home countries.

1. Though this is not entirely a LMIC problem; a third of doctors in the UK also want to leave.

2. Grouping the CHW and other health worker categories together.

3. I use this number because these schools are not subsidized by the state and thus closer to the true cost of education than tuition paid at state-run institutions.

4. I do not like this estimate very much, though. This paper uses a multivariate regression model with no identification strategy to speak of, and is thus incredibly prone to omitted variable bias. I use it here because it gives us some place to start.

5. Eagle-eyed readers may wish to know why I use a “regression soup” paper I don’t particularly like for my estimate of deaths averted instead of this RCT. This is for two reasons: 1) this paper looks at seven-day infant mortality, which is a significant subset of all mortality, 2) honestly, I think the external validity of the Nigerian medical system is pretty limited. I will not elaborate on the particular political dysfunctions of the Nigerian state here, but suffice to say it is badly run even by the standards of the region.

6. Middle-income countries have about 4.5x the doctors per capita as low-income countries.

7. This is a weighted average of the lower middle-income and upper middle-income percentages of table 1.

8. Though this does not prevent individuals from applying to jobs abroad.

9. The only two exceptions are Syria and North Korea, and I wish you much luck running your international healthcare recruiting operation in either one.

10. I could see the case for focusing on brain drain in, say, low-income and lower-middle income countries, and excluding upper-middle income countries (and I might do so if I had about a month to write this post). Look, your girl is tired.

11. This paper also attempts to estimate the causal effect of physician emigration on public health outcomes. Anyone who knows me knows I am a huge fan of Michael Clemens, but I don’t think these particular results are very robust. I think it is unlikely that the exclusion restriction holds for either instrument in the paper, and thus place relatively little weight on the results and do not discuss here.

12. There are approx. 316,000 Filipino nurses abroad, of 2.16 million overseas Filipino workers.

13. It’s possible that this is an underestimate - since 316,000 Filipino nurses remit $8B, the average Filipino nurse remits $25,000 a year. My guess is that most Filipino nurses go to the US, where salaries are higher; many African doctors go to the UK, where the average physician salary is not far off the median US nurse salary.

14. Some immigrant physicians will return to their home country - for instance, about one in three South African physicians have practiced abroad for at least some time. However, if they return to their home country, one can no longer consider the investment of their home country in their education a loss.

15. And that paper does not, in fact, do that.

16. Sorry philosophers.