A few months ago, someone pointed me to François Gemenne who last year gave several interviews to German-language newspapers on migration. As the Swiss “Tagesanzeiger” phrased his claim (my translation from the German):
“The migration researcher François Gemenne states that whether borders are open or closed has no effect on the extent of refugee flows. But on the number of those who die“.
While I agreed with a lot of the arguments that Gemenne made, I was baffled by this assertion. It was not the first time I had come across such a claim. As far as I can tell, also the Dutch migration researcher Hein de Haas holds a similar view although I am not sure (cf. his inaugural lecture and especially his second assertion).
My intuition is wildly different. I would expect very many immigrants if borders were actually open, even so many that a gradual transition might be warranted. And it could not be a narrow definition of “refugee” that makes the difference. As far as I understand Gemenne and de Haas, they are referring to migrants (including refugees in a narrow sense). That got me thinking whether I missed something.
Still I did not change my mind. According to Gallup polls, there are literally hundreds of millions worldwide who would like to migrate to a developed country. Maybe it is easier to express your desire in a poll than realizing it if it were possible. But to get from hundreds of millions to practically no one, you would have to assume that almost all respondents entertain totally unrealistic plans. And even if the polls were unreliable, there would still be strong economic incentives to migrate. Earnings differentials are huge, so it makes a lot of sense for very many people to migrate. What explains the vastly differing views?
It took my some time to realize that maybe both claims can be right, and that is so because we are simply talking about two different things when we use the shorthand “open borders.” On the one hand, “open borders” can mean that you scrap border controls or reduce them to a minimum. Whether de facto or de jure, there is hardly any impediment to crossing the border. I think that is the first intuition most people have. It also ties in with a vision of tearing down walls and fences.
On the other hand, “open borders” can mean that foreigners have the same or at least very similar rights as domestic citizens: they are allowed to settle and work in the country and they also have associated rights, e.g. to open a bank account, register a car, travel back and forth, etc. Maybe “open borders” is not the correct term here, and “freedom of movement” comes closer or “free labor mobility.”
The two senses of “open borders” are independent. In principle you could have “open borders” in one sense without having “open borders” in the other. On the one hand, you can imagine a country with a very high wall at the border that does not allow anyone in without some official procedure. But at the same time, the country could grant foreigners the right to settle and work. Having such border controls would seem superfluous, but they might be in place for security reasons or because of some other rationale. And on the other hand, there could be a country that does not control who enters the country, but treats all foreigners who settle and work in the country as in breach of the law and persecutes them accordingly.
I think this distinction can explain the different views. If foreigners are allowed to settle and work in the country, there is in principle no limit to how many could come. There might be some temporary constraints, e.g. how many new jobs can be created over the short term, or how many apartments can be built. But over time, there is no apparent upper bound. If there is a very high level of demand for immigration, as indicated by international polls or as a consequence of high earnings differentials, there should be a lot of immigration. The main point here is that immigrants are allowed to immigrate legally and have a secure and permanent status.
Now in the other case, things are very different. Everyone could enter the country because there is no real impediment to crossing the border. However, they would not have a legal status. And that means lots of things. E.g. they can be sent back if they are apprehended, maybe fined and imprisoned. Or their belongings might be confiscated. Another consequence is that they lack the security of legal immigrants and cannot make dependable plans over the long term. They would also find it hard to bring in their families because they are in a continually precarious situation.
The differences do not stop there. So far the problem for illegal immigrants is that they are in the crosshairs of the government. But enforcement can also be directed against domestic citizens, e.g. landlords and employers who deal with illegal immigrants. Throw in many other handicaps: you cannot get insurance, you cannot register a car, you cannot get an official degree or a license, you cannot buy a ticket from an airline and travel back and forth, etc., and it adds up to a very difficult situation for an illegal immigrant—no matter whether border controls are strict or non-existent. Enforcement will never be perfect, so there should still be a niche for illegal immigrants.
Why a niche? Suppose there were a lot of illegal immigrants in the country. Maybe it is hard to find a few percent in the population (e.g. for the US it is about 3%, for Germany it might be less than 2% or even 1%). It could also be hard to track down their landlords and employers. However, the more illegal immigrants there are, the easier it is to find them. If every second person were an illegal immigrant, even Inspector Clouseau could find one.
The basic relationship should be something like this: If S is the share of illegal immigrants in the population, then it takes a certain number of checks to find one, and this number should be proportional to 1 / S, a hyperbolic function that goes to infinity as you approach 0.
It would be extremely costly to find the last illegal immigrant in the country. But costs come down considerably as the share of illegal immigrants rises. At some point, you would not even have to search for illegal immigrants in the first place because you stumble over them as you do other things, e.g. when the government tries to collect taxes. So the real constraint might come from enforcement within the country, not from enforcement at the border, and it might be enforcement against domestic citizens and not so much against the illegal immigrants themselves.
My conjecture would be that the main constraint is the size of the illegal labor market. You cannot hire illegal immigrants in any line of business. E.g. if they are very visible for the public, it won’t work. Or in a large company with specialized functions (separate HR department, different divisions), a lot of people would be involved, and that would make it difficult to hide illegal employees from the government. Or in a business that creates paper trails for everything and where money flows are rather transparent, employing illegal immigrants should also be a major challenge.
So, typical businesses that employ illegal immigrants have to be rather small and/or intransparent (e.g. restaurants, construction, agriculture). Or employers are not even businesses, but private households (e.g. for nannies, gardeners). That alone severely limits the size of the labor market for illegal immigrants relative to the total economy. And it also pushes illegal immigrants into jobs with relatively low productivity because the relevant businesses cannot participate as openly and widely in exchange, or else risk making themselves vulnerable. A constraint on their size also means a constraint on the sophistication of their production processes.
One indication that there is some rather stringent bound on the share of illegal immigrants in the population is that estimates are in the lower single digits for all developed countries. There are different estimates, but none of them ever go beyond a few percent. Obviously, the level of border enforcement is quite different. If that were the main reason for keeping the share of illegal immigrants low, I would expect not only more variation, but also some countries with very high shares of illegal immigrants.
A further indication is that the same holds for US states, though with more dispersion. Nevada was the state with the most illegal immigrants, 10% in 1990 and 7.3% in 2010. But most states were in the lower single digits, with some that had almost no illegal immigrants. Still, most US states had a share of illegal immigrants above 1% or 2% and many even above 3%. Now if the main constraint were border enforcement and not internal enforcement, I’d expect illegal immigrants to settle right across the border. What’s the use of branching out to other states? It only means more costs, being farther away from home, and having to find your way when you could stay where others have already blazed a trail.
However, if internal enforcement were more imporant, such behavior would make perfect sense. The more illegal immigrants there are, the easier many things get, but also the more conspicuous you are and the more likely detection becomes. One guess would be that less diverse parts of the country are also less congenial to illegal immigrants than more diverse parts of the country where you can slip through more easily. Actually, there is a rather strong connection between diversity in a state and the share of illegal immigrants (although admittedly there could be other explanations). Here is what I got when I ran diversity in different states (probability that two random people have the same ethnicity) against the share of immigrants:
There are two ways internal enforcement can work against illegal immigrants: Living in the country per se could be the problem (e.g. renting accomodation, controls by the police, etc.) or finding a job to stay afloat. In both cases, illegal immigrants have to rely on illegal markets which are part of the shadow economy. And that introduces a constraint on the share of illegal immigrants because for developed countries with a certain level of enforcement, there is an upper bound on how large the shadow economy can be relative to the official economy. By its very nature, estimates for the size of the shadow economy are not easy to come by. According to one approach by Friedrich Schneider, estimates for 2013 were in a range from 6.6% for the US to some 15% for countries in Northern Europe and 20% in Southern Europe.
However, these estimates have been challenged by Thiessen and Kholidilin for what I think are good reasons. One observation was that a shadow economy of some 15% of GDP would mean the government is missing out on a lot of taxes it could collect. As Ulrich Thiessen explained in an interview, that was exactly what the German government thought: 70 billion euros. So in 2004 a task force, 6,000 to 7,000 strong, was created to fight illegal employment (not primarily by illegal immigrants). The whole affair turned out rather disappointing for the German government, though. While the task force cost about 600 million euros, it raked in a measly 30 to 40 million euros of the supposedly 70 billion floating around, and the Federal Court of Auditors (Bundesrechnungshof) politely demanded an explanation from the government what they were doing.
The results by Thiessen and Kholidilin would indicate a much smaller shadow economy of 1.1% for the US, less than 3% for Northern Europe, and mostly less than 5% for Southern Europe. No matter what the correct estimate is, there has to be a bound on the shadow economy if there is some level of enforcement. And the labor market for illegal immigrants has to be at most as large as the shadow economy, and hence it has to stay rather limited.
But why would the labor market be the bottleneck? It could also be controls by the police going from door to door. My explanation would be that incentives for the government to go after illegal employment and, by implication, also illegal immigrants are very strong. As the example above shows, governments get very active when they sense an opportunity to collect extra taxes. And it makes more sense to go after those who could pay fines and have the money: employers.
By comparison, going after illegal immigrants has rather little political potential. Yes, the public always wants a crackdown in the naïve belief that illegal immigration could be lowered to zero. But in reality, even if the government were somewhat successful in this regard and could shrink the stock of illegal immigrants, it would be hard to even prove it happened with such obscure estimates. Figures for the stock of illegal immigrants in Germany range from 100,000 to well over 1 million (cf. Worbs/Wolf/Schimany: Illegalität von Migranten in Deutschland, page 7).
Of course, this is no proof that the main bottleneck for illegal immigrants has to be the labor market. But let me refer you to three papers that would more or less support my conclusion:
- Camacho/Mariani/Pensieroso: Illegal Immigration and the Shadow Economy. – The authors propose a model that connects illegal immigration and the shadow economy. They then derive general relationships. Border enforcement appears less effective than internal enforcement on the labor market. And if the two compete (cf. extended model in the appendix), more border enforcement can paradoxically lead to more illegal immigration because the shadow economy expands.
- Boswell/Straubhaar: The Illegal Employment of Foreign Workers: an Overview. – The conclusion of this article is: “Efforts to limit the supply of irregular labour through entry controls or internal checks, meanwhile, are socially and economically costly. The best strategy would be to develop better ways of tackling the problem at the demand side: increasing the costs and probability of apprehension for employers, and lowering the costs of hiring regular workers.”
- Leerkes/Engbersen/van der Leun: Crime among irregular immigrants and the influence of internal border control. – The authors show for the Netherlands how internal enforcement is effective in driving illegal immigrants out of the labor market (and as an unintended consequence into outright criminal activities). There is also evidence that the Dutch police—quite reasonably in my view—do not put a lot of effort into chasing illegal immigrants, but focus on fighting crime, no matter what status the criminals have.
If all this is correct, the situation for illegal immigrants on the labor market could look like this stylized graph (my apologies to everybody who draws the diagram with quantity horizontally, but in this way I can make my point more easily):
There are very many potential immigrants who are interested in a job (red), but the number of jobs available (blue) is capped by a rather hard constraint. Such a situation might explain the observations de Haas and Gemenne have in mind. If the number of jobs available goes down with the general state of the economy, there should be corresponding movements in the stock and then in the flow of illegal immigrants. And the response is rather strong.
Enforcement at the border leads to a fixed sum an illegal immigrant has to pay upfront. Typical estimates would suggest a few thousand dollars. This alone shows that border enforcement is effective to some extent. Actual travel expenses for legal travel are usually much lower (e.g. less than 100 euros for crossing the Mediterranean). The relationship between the size of the flow and the cost of detection should also be hyperbolic: If millions tried to cross the border it would be very easy to clamp down on the flow. But only to a point. It gets harder and harder, the fewer people you try to detect (I looked at an extreme effort in my post: “How to Seal a Border” for the border regime of the GDR).
Still, this does not have to mean that border enforcement is the main bottleneck for illegal immigrants. With an average stay of two or four years, border enforcement works in the sense that it inflicts a cost on immigrants and throttles large flows. But it is more like a “tax” of 100 or 200 dollars per month. Illegal immigrants might also face credit constraints, so it could be hard to get the initial sum together. However, there are ways around this, at least to a certain extent, e.g. support by previous immigrants, families, loans that have to be paid back afterwards, etc.
Now, if enforcement at the border is stepped up, the fixed sum goes up and hence also the “tax.” This means the red curve shifts to the right. But if the number of jobs (blue line) is more or less capped, the stock of illegal immigrants will not change a lot. Likewise, with less enforcement at the border: The change could be small. I am not sure I understand the argument why there would not be any change, but it might be that the effect of increased border enforcement stays rather limited. Apart from that, border enforcement is directed at the flow. If illegal immigrants stay longer to make up for higher up-front costs, even with lower turnover, the stock of illegal immigrants could stay steady.
If this reasoning is correct, then there are two conclusions:
You can only make the argument on the effect of border enforcement if you assume that enforcement within the country stays the same and immigrants do not obtain legal status. If you are willing to bite this bullet, I might agree, but I am not impressed. Surely, it would be an improvement if illegal immigrants could enter the country in safer ways. However, what if the real wall is the one around the labor market? And if you advocate granting immigrants also legal status you are in a different case and it is no more obvious why your argument is correct. It is probably false.
There is also a problem for the opposite position. For years, restrictionists in Europe have been making the argument that there are hundreds of millions who would like to immigrate, and all that’s keeping them back are strict border controls. So you would see an exodus once you relaxed enforcement. That cannot be true either. As long as you have enforcement within the country and there is no legal status, numbers have to stay rather modest for illegal immigrants (note: this does not have to apply to those who enter the country perhaps illegally, but who have a chance of obtaining legal status like asylum seekers or refugees in a narrow sense).
For legal immigration, the regime at the border is unimportant. And there is no such constraint on the supply of jobs for immigrants (or only in the short run). Hence the conclusion from low numbers of illegal immigrants and little sensitivity to the level of enforcement at the border does not carry over to the number of legal immigrants if other policies change. Granting more people legal status could mean very high numbers. Confounding the two cases is just a category mistake.
To sum up:
- If you understand “open borders” as no enforcement at the border, but no legal status, and enforcement within the country at a level as for developed countries now, then there should be a rather low bound for the percentage of illegal immigrants. Empirical evidence would suggest a range in the lower single digits.
- The main constraint is probably the labor market for illegal immigrants, which has to be a part of the shadow economy. And for developed countries, there appears to be a constraint on the size of the shadow economy, perhaps in the lower single digits as a percentage of the official economy. This implies a rather stringent bound on the number of jobs available for illegal immigrants. Combined with very many potential immigrants, changes in the border regime could make only a small difference for the stock of illegal immigrants, all else equal. The stock should react strongly to the state of the economy, though.
- If you understand “open borders” as granting legal status and the right to settle and work in the country as for domestic citizens, then the regime at the border is unimportant. There is no apparent bound for the number of immigrants, and the numbers in global polls could well materialize. Lumping illegal and legal immigration together is not warranted. The two cases are quite distinct.