Selection effects for migrants: some a priori possibilities

This post combines many different threads I’ve explored in earlier posts. Back in July 2013, I wrote a post arguing that it’s important to get a handle on both the quantity and the selectivity of migration. Recently, I wrote a series of blog posts laying out a detailed conceptual framework for the empirical analysis of migration (introductory post here, describes and links to other posts). While laying out this conceptual framework, I noted that, under any policy regime other than complete closed borders, there is likely to be both a selection effect and a treatment effect for migrants. Specifically, in part 3, I considered a situation where we assume for simplicity that the people who do not migrate are not affected by the act of migration. In that case, we can concentrate on selection and treatment effects for migrants and ignore the treatment effects on non-migrants. Our goal was to discuss the rank-ordering and quantitative comparison of the following four values (where X is the indicator of interest):

  1. Performance of natives of target country B on indicator X.
  2. Performance of natives of source country A (who would not move under either policy) on indicator X.
  3. Performance of potential migrants on indicator X if they were allowed to migrate (i.e., in the migration scenario).
  4. Performance of potential migrants on indicator X if they were not allowed to migrate (i.e., in the no-migration scenario).
  5. Ability to plan and execute a move.

Note that:

  • The difference between (2) and (4) measures the selectivity of migration relative to the source country.
  • The difference between (1) and (3) measures the selectivity of migrants relative to the target country, or equivalently, to their failure of assimilation (the assimilation may be “upward” or “downward” depending on how the migrants compare with the target country natives).
  • The difference between (3) and (4) refers to the treatment effect of migration on migrants.

The goal of this blog post is to come up with a priori arguments on how migrants might be selected on various parameters. We’re not concentrating on the treatment effect directly, except insofar as beliefs about the treatment effect affect the selection of migrants. In some cases, we cite empirical evidence to support the claim. But the goal is not to make concrete empirical predictions, but to lay out general considerations that would help make concrete predictions for specific migration policy regimes.

Because of the vagueness of our analysis, we don’t distinguish heavily between selectivity of migrants relative to source countries and selectivity relative to target countries. However, our arguments, as stated here, apply a priori far more to selection relative to source countries than to selection relative to target countries. This follows from the nature of the analysis: we’re trying to figure out who, from a given bunch of people in the same environment, would end up moving. Therefore, these are best thought of as arguments about emigrant selectivity (comparing (2) and (4)) than about immigrant selectivity (comparing (1) and (3)). At the end of the blog post, we’ll discuss what we can infer about immigrant selectivity from the information.

Also, for the most part, I restrict attention here to considerations that would be relevant even under open borders. There are some forms of selectivity that arise from fiat: migration policy dictates that migrants must satisfy a set of conditions in order to be allowed in. We discuss these only in passing here, and will return to explicit policy selection in a separate post (more remarks on this at the end of this post). Note that I ignore explicit policies in the post but I certainly consider them quite important. I am not an economic determinist.

Costs of moving

Migrants are moving to a new place. Even under an open borders regime, moving requires nontrivial fixed costs in terms of time, money, and emotional energy. The magnitude of the costs depends on the geographic distance moved, the cost differential (moving to a place with a higher cost of living means one’s savings are less use for covering the initial costs of setup, even if one expects to eventually recoup those costs through higher earnings), as well as the cultural and linguistic distance between the source and destination. Note that all these apply even under open borders. In a regime with migration restrictions, there are additional costs of time, money, and uncertainty in applying for permission to move. Depending on the feasibility of return migration, one may also need to dispose off assets before making the move. Those crossing borders illegally need to incur coyote fees and undertake time-consuming and dangerous journeys to reach their destination.

What attributes does the high cost of moving select for? It’s hard to say, but here are some guesses:

  • Money: People who have more money can afford the costs of moving more easily.
  • Strong future orientation (i.e., lower discount rate): People who think of life a few years ahead are more likely to be willing to migrate than people who engage in hyperbolic discounting.
  • Willingness to break ties: People who are heavily attached to their family and home culture would find the move more difficult, whereas people who define themselves less by their present relationships can move more easily.
  • Adventurousness, openness to experience, and willingness to take risks

Opportunity costs of migration

When people migrate, they leave behind their home, family connections, and a culture that they are more familiar with and may be attached to. What sort of people are willing to leave that behind? Here are some guesses:

  • People who have little to lose by leaving are most likely to do so. This could be because they are at high risk of being victimized by violence, are heavily discriminated against by people where they live, are cultural misfits, or cannot find any use of their job skills where they currently live.
  • People who have skills or assets that cannot be transported easily and can be leveraged most in the homeland are least likely to leave. For instance, people who are good-looking by the standards of their culture may have the best prospects in their homeland (however, if a huge diaspora from the country already exists, they might be able to marry a member of the diaspora settled elsewhere through a family connection or other introduction). People who inherit a big family business that they can continue running, but aren’t particularly entrepreneurial, may just prefer to stay where they are to keep running the business. People with skills in politics have the best shot at politics in their home country, given that voters everywhere are likely to discriminate in favor of people who were born in the country and fit in culturally and linguistically. People who have completed expensive location-specific qualifications (such as in law) may prefer to stay in the home country because they’d need to re-qualify to practice law in a new country.

Note that there’s a contradiction of sorts between the little to lose criterion (which suggests that poor people may be more keen to migrate) and the observation that wealthier people can more easily afford to migrate. We’ll talk more about this later.

Benefits of migration

Migration generates huge benefits for some people, and scant benefits for others. The main benefit of migration from lower-productivity regions to higher-productivity regions is the place premium: one can earn more with the same skills simply by migrating to a new country.

The following attributes predict the benefits of migration:

  • Larger absolute wage gains predict migration. Highly skilled individuals, who command high incomes in general, are likely to have large absolute wage gains.
  • Larger proportional wage gains predict migration. Increasing one’s income from $1000/year to $10,000/year looks a lot more attractive than increasing one’s income from $30,000/year to $40,000/year.
  • Greater knowledge of, affinity for, or ability to learn, the language, customs, and culture of the new place predicts more migration.
  • Greater ideological or political affinity with the place they’re moving to (see Ilya Somin’s blog post on the subject).
  • Presence of diasporas from the source country in the target country can make migration more attractive. This is part of the diaspora dynamics model developed by Paul Collier.
  • Other geographical and health-related considerations could play a role. For instance, I’ve been told that in the 19th/20th century US there was a wave of migrants with lung diseases to the southwest to benefit from the dry air there. This was intranational migration, but presumably there could be international migration for similar reasons under open borders. The selection effects here are unclear, for instance, it may be that emigration of the unhealthy makes it such that the people who stay on in inhospitable climates are unusually healthy and fit.

Migration for one’s children

In my blog post on whether there might be too much or too little migration, I talked about how the costs of migration are borne by the migrant, but the benefits are shared by the descendants. I had noted at the time that, because migrants may not fully take into account the benefits to their descendants (even though they care somewhat about their descendants) this might lead to too little migration.

I want to bring up the same point, but with a focus on selectivity rather than raw quantity. People who strongly care for the future of their children and later descendants (including unborn descendants) are more likely to be willing to migrate. This probably selects for two things:

  • Strong future orientation (we already talked about this in the context of overcoming the costs of migration, but it takes on added importance if one is thinking of one’s children or grandchildren, particularly the unborn ones).
  • Greater love or concern for one’s future family. Note that this is in some tension with the fact that migrants are generally more willing to break ties with their existing families in order to migrate.

I’m planning to do a post on how migration can be considered a sacrifice for future generations, where I’ll explore this in more detail.

Selectivity of migration on income and wealth

The a priori considerations provided above paint a mixed picture of the role that income and wealth play. The following emerge:

  • Higher wealth allows people to fund their move more easily.
  • On the other hand, less wealth means people have less to lose and are more desperate to migrate.
  • Huge wage gains attract more migration. But huge wage gains in absolute terms are linked to higher incomes, whereas huge wage gains in proportional terms are linked to lower incomes.
  • Higher wealth may be correlated with other traits that predict greater or lesser ability to migrate. This is particularly the case for self-acquired wealth, but might also apply for inherited wealth to the extent that parental wealth correlates with parental attributes and via that with the person’s attributes.

As noted above, the nature of the income and wealth pattern may matter more than the amount. People whose income and wealth is heavily tied to their current location are likely to stay, whereas those whose income and wealth are tied to transportable skills or assets are more likely to move to places where their skills and assets can be best used to earn more.

One example of a “wealthy with little to lose” combination is (relatively) wealthy members of minority groups that are forcefully evicted as path of ethnic cleansing, or anticipate that this will happen. This was the case with Indians in Uganda, businessmen who found themselves on the wrong side of the border in the runup to the Partition of India, Tamils in Sri Lanka, and many others. Market-dominant minorities may in general fear hostile political environments and may be eager to leave when populist political parties or opinions are ascendant.

What does empirical evidence suggest about the relation between income/wealth and emigration? This blog post by Michael Clemens reviews the evidence and concludes that for countries below something like $6,000–8,000 GDP per capita (at US prices), countries that get richer have more emigration. The plot of emigration flow in terms of GDP per capita peaks at this income range, as does the plot of emigrant stock in terms of GDP per capita. Clemens writes:

Social scientists have six theories for this “mobility transition”. I review these theories and the evidence for them in the paper. Briefly: 1) Development is usually accompanied by a demographic transition that favors a corresponding mobility transition, 2) development means that more people can afford to emigrate, 3) development means that more people can access the information they need to emigrate, 4) development tends to disrupt economic structures that keep people immobile, 5) development shapes domestic inequality in ways that foster migration, and 6) development in country A means that people in country B are more likely to give visas to migrants from A.

The Zelinksy model of mobility transition is also relevant.

Selectivity of migration on criminality

The following are some considerations:

  1. The strong future orientation needed to migrate suggests that migrants will be less criminal, because crime generally involves short-term benefits and long-term costs, and criminals generally discount the future heavily. In addition to future orientation, the ability to execute the move might also filter for other relevant positive traits that predict lower criminality.
  2. The fact that migrants are likely to have more money (in order to fund their moves) and the fact that richer people commit fewer violent and property crimes, argues in favor of migrants being less criminal.
  3. The fact that migrants often need to cheat and lie in their visa applications in order to be able to migrate, or that they cross borders illegally, might lead to migrants being selected for higher levels of criminality.

In a later post series on crime and open borders, we’ll weigh these considerations against one another. The general bottom line will be that emigrants have substantially lower crime rates than natives of their source countires, and this is attributable in large part to selection.

Selectivity of migration on enterprisingness

People who move have strong future orientation, adventurousness, openness to experience, and willingness to take risks. This suggests that they are more likely to be enterprising in the general sense. In some cases, this translates to being more entrepreneurial (see here for more on existing research). The following are some other considerations:

  • To the extent that regulations on migrants make it easier for them to stay in standard, steady jobs, they are less likely to engage in entrepreneurship. This is a major issue for migration to the US: it’s much easier for high-skilled migrants to get a work visa working at a big company than to start a company. Note that this effect could operate at both a selection and a treatment level: entrepreneurial people may shy away from migrating to a place where it’s not that easy to start a business, and people who’ve already migrated may prefer to continue in an existing company than start a business.
  • To the extent that regulations (or societal discrimination) inhibit work in the formal sector, migrants are more likely to start their own small businesses. For instance, it may be easier for families to start a restaurant and have family members work at it so that younger members can contribute and they can circumvent labor laws. Note that this type of entrepreneurship isn’t the “create a billion-dollar business” type, and has lower value per entrepreneur, but it is still important to society. At the same time, artificial restrictions on formal sector employment may lead to too many family businesses and a more inefficient economy overall because family businesses cannot avail of the economies of scale.
  • The amount of wealth that migrants have affects whether they can afford to experiment with entrepreneurial ventures. As we saw, the relationship between migration and wealth is unclear.

Selectivity of migration on political attitudes

While there are many migrants who leave because of political persecution, this political persecution often has more to do with ethnic identity and religious beliefs than with specific political beliefs. (There is some relation between religious beliefs and political beliefs, but it’s very tenuous). The following are some general remarks:

  • The very fact that migrants left their home country suggests that they are not overly attached to the institutional or policy framework of that country. This doesn’t mean they actively dislike it. This creates a prior against migrants replicating the policies of their home countries. Empirically, there is little evidence of home country policy replication: people from communist countries aren’t noticeably in favor of communism and don’t seem to want to impose communism on other countries. At any rate, they haven’t been successful doing so. On a related note, see Ilya Somin’s blog post on immigration and political freedom.
  • People who have a strong aptitude or interest in politics (in the sense that they want to become political activists or politicians) are likely to stay in their home countries, because it’s easier to make headway in politics as a native.

Remarks about the distinction between selectivity with respect to source and target countries

The arguments above concentrate on what we expect regarding selectivity relative to the migrants’ source countries, because we’re trying to answer the question: of a given set of people in a given environment, who’d be most willing and able to leave? But people in the receiving countries are more interested in comparing immigrants to natives, in order to figure out how immigration affects the overall societal composition.

To what extent can the above arguments make predictions about how immigrants compare with natives? We need to know both how much the countries differ (the (1) versus (2) gap) and how large the treatment effect of migration is (the (3) versus (4) gap).

For instance, let’s say we have a low-productivity poor country A and a high-productivity rich country B. By our general arguments, we think that the people who migrate from A to B are likely to be more enterprising, more future-oriented, and more adventurous than those who stay behind in country A. How do they compare with country B? The conclusion we draw depends on what we think of the relative levels of these traits in the two countries.

One school of thought is that the distribution of traits in the populations of both countries is similar, so that the (conjectured) fact that emigrants do better on these traits than natives also implies that immigrants will do better on these traits. For instance, one might argue that there isn’t any difference between the levels of future orientation in China and Taiwan. Therefore, immigrants from (low-income) China to (high-income) Taiwan, who are selected relative to their source country with respect to future orientation, are probably also selected relative to their target country.

Another school of thought is that the reason country B is richer and has higher productivity is that the people there are more enterprising, more future-oriented, more adventurous, etc. For instance, one might argue that the United States is more entrepeneurial than the United Kingdom, and this accounts for the difference in their per capita levels of income and wealth. In this case, even though emigrants from country A score higher on these traits than natives of country A, it’s unclear how they compare with natives of country B. There are two separate issues to consider to figure this out:

  • How strong is the selection effect of migrants relative to their source countries, in comparison with the difference between source and target countries? Even if the US is more entrepreneurial than the UK, that difference on average might be much smaller than the selection effect for migrating.
  • How much of a treatment effect is there on migrants? The strength of the treatment effect arguably depends on the age of migration. Those who migrate as young children, and do not grow up in an isolated culture, are likely to be exposed to similar cultural influences as the natives of the target country, though they still experience a different home culture and prenatal environments, and are genetically close to country A. Note that though treatment effects are stronger for young people, selection effects may be weaker, because young people are often dragged along by their parents rather than being active participants in the decision to move.

Remarks on differences in selectivity of different migration policy regimes

The extent of selectivity depends heavily on the nature of the migration policy regime. Thus, the level of selection under open borders is likely to be quite different (and in general, much weaker) than the level of selectivity under the status quo.

The majority of the considerations outlined in this post apply to migration even under open borders. The main difference is that rigid legal constraints on whether one can migrate, and the amount of bureaucratic red tape one has to go through to migrate, both reduce under open borders. If we are trying to quantitatively ballpark the level of selectivity, we need to keep in mind its sensitivity to the policy regime. In a future post, I’ll explore ways that governments can (and do) affect the selectivity of migration through explicit migration policy.

3 thoughts on “Selection effects for migrants: some a priori possibilities”

Leave a Reply