¡Descarga capitulo 6 economics of migration y más Transcripciones en PDF de Computación Gráfica y Animación solo en Docsity! 4 Selection in Immigration
In economic models of the migration decision, cross-country differences in economic conditions
play a central role in determining whether people move. The utility- or income-maximization
model and the gravity model predict that changes in relative economic conditions lead to
changes in migration flows. Chapters 2 and 3 focused on how changes in relative economic
conditions affect the number of people who migrate. This chapter focuses on how changes in
relative economic conditions affect the composition, or characteristics, of immigrants. Immi-
grants are unlikely to be perfectly representative of the population in the origin. They may
be predominately high skilled or predominately low skilled. They may be disproportionately
workers, students, children, stay-at-home spouses or retirees.
Who chooses to migrate is often as important to both origin and destination countries as the
number of migrants. Origin and destination countries can have many different perspectives on
what characteristics are “desirable” among migrants, These perspectives may or may not con-
flict. For example, origin countries may be concerned that out-migration of skilled workers
will slow economic growth, while destination countries may want to attract skilled workers in
order to boost economic growth. Origin countries may want students to go abroad to receive
an education but then return home to contribute to the economy. Destination countries may
want to attract tuition-paying foreign students and then have them leave so that they do not
compete with native-born workers. Alternatively, destination countries may not want to incur
the cost of educating foreign students unless they will stay and join the workforce.
The economics of immigration uses selection models to look at immigrants” characteristics.
Studies of immigrant selection have focused primarily on the relationship between immi-
grants” skill levels and returns to skill in the origin and destination. This chapter therefore
focuses on skill. However, the model presented in this chapter can be applied to other char-
acteristics as well. The key prediction of the model is that immigrants” characteristics depend
on the relative returns to those characteristics. The model can also be applied to return migra-
tion, as discussed later in this chapter.
The selection model can explain why not all migrants from an origin country go to the
same destination country. People in the same country typically share a culture and language,
and they are about the same distance from other countries. So why do some migrants go
to one place while others go elsewhere? The selection model developed below predicts that
people sort across potential destinations —including the origin—based on their skills and the
relative returns to those skills.
80 Immigrant Selection and Assimilation
The Roy model
The economist George Borjas (1987) developed an influential model of skill selection in
immigration. The model is based on a canonical model by Andrew Roy (1951) that examined
how self-selection into occupations affects the distribution of income. In Borjas's version of.
the Roy model, workers earn the average wage in their country plus a random term. The aver-
age wage differs across countries, as does the distribution of the random term. In the origin
country, the distribution of wages is
In(Wage) = 4 +€ (4.1)
where y is the average wage and € is a measure of how much wages vary across individuals rel-
ative to the average. (The model examines the natural log of wages because logarithmic func-
tions have nice properties that the model exploits.) People with positive values of € earn more
than the average, while people with negative values of £ earn less than the average. One way
to interpret € is as measuring the return to skill-——how much wages increase as skill increases.
The model assumes that € is distributed normally with a mean of O and a variance of O”.
(Borjas makes this assumption because normal distributions also have nice properties that the
model exploits.) The larger 0? is, or the bigger the variation in €, the higher the return to skills
is in the origin country and the greater income inequality is there. As the variation in income
increases, income inequality grows.
If everyone in the origin country migrates to the destination country, their distribution of
wages in the destination country is
In(ioge)=(1+6 (4.2)
where E is distributed normally with a mean of 0 and a variance of O? (terms with a tilde (=)
over them refer to the destination country, and terms without it refer to the origin country).
Everyone knows |, fl and the cost of migrating, and people know their own € and É.
The model assumes that € and É are correlated with, or related to, each other, with a
correlation coefficient equal to p. This correlation coefficient can range from 1 to —1. Bigger
numbers in absolute value mean a stronger relationship between earnings in the origin and the
destination. If P is positive, then people who have higher-than-average earnings in the origin
also have higher-than-average earnings in the destination, and people with lower-than-average
earnings in the origin have lower-than-average earnings in the destination. This would occur il
skills that are valuable in the origin are also valuable in the destination. For example, an engi-
neer in Kenya who has a PhD is likely to earn an above-average wage in Kenya and elsewhere,
while a street cleaner from India who has no formal education is likely to earn a below-average
wage in India and elsewhere. The more similar two countries are, the higher P is likely to be.
If p is negative, then people who have higher-than-average earnings in the origin have lower-
than-average earnings in the destination. For example, people who are relatively successful in
an agricultural economy that rewards physical strength and stamina might be relatively unsuc-
cessful in an industrialized economy that values education. The correlation coefficient may be
negative for immigrants moving from some developing countries to industrialized countries.
Selection in Immigration 83
inequality decreases in the origin relative to the destination, immigration increases
¡es less negatively selected. But if income inequality increases further in the origin
to the destination, immigration decreases and becomes even more negatively selected.
4.2 shows how the relative return to skill affects the direction of selection. Wages
'ed to increase linearly with skill, and skill is perfectly transferable across countries
= 1).The steepness of the skill-wage profile indicates the return to skill in a country-—
the line, the higher the return to skill. In Figure 4.2(a) on the left, the return to
higher in the destination than in the origin. The opposite is the case in Figure 4.2(b) on
the return to skill is higher in the destination than in the origin, people with low
earn more in the origin than in the destination, and people with high skill levels
in the destination than in the origin. In Figure 4.2(a), everyone whose skill level
s*, the skill level at which the lines cross, stays in the origin, while everyone with a
skill level migrates. In Figure 4.2(b), everyone whose skill level is below s* migrates,
ne with a higher skill level stays.
effect of changes in the relative return to skill can be seen by pivoting one of the skill—
in either figure while leaving the other line unchanged. For example, an increase
relative return to skill in the destination causes the dashed skill-wage lines to steepen,
4.3(a) shows. The skill threshold for migrating versus staying, s*, then changes, and
of immigrants changes as a result, If immigrants are positively selected and the
return to skill in the destination increases, the skill threshold for migrating falls. More
migrate, and migration becomes less selective as a result. Note that the direction of
changes only if the change in the relative return to skill is so large that the country
iously had the lower return to skill now has the higher return to skill, and vice versa.
effect ofa change in average income in the origin or the destination can be seen by shift-
of the skill -wage lines up or down in cither figure while again leaving the other line
(a) Positive selection
(b) Negative selection
Skill Ss Skill
Figure 4.2 Relative returns to skill and the direction of selection.
(2), positive selection occurs because the return to skill is higher in the destination than in the origin, as
icated by a more stecply sloped skill-wage curve in the destination (the dashed line) than in the origin
the solid line). In (b), negative selection occurs because the return to skill is higher in the origin than in the
tion, as indicated by a more stecply sloped skill -wage curve in the origin (the solid line) than in the
destination (the dashed line).
84 — Immigrant Selection and Assimilation
(a) Increase in the return to skill (b) Increase in average income
Wage Wage
O Skill e s Skill
Figure 4.3 The effect of changes in the relative return to skill or average income in the destinati
In (a), an increase in the relative return to skill in the destination decreases the skill threshold for mi
if there is already positive selection. More people migrate, and migration becomes less positively sel
In (b), an increase in the average wage in the destination decreases the skill threshold for migrating.
unchanged. An increase in the average wage in the destination causes the dashed skilla
line to shift up in Figure 4.3(b), for example. A change in average income does not affect
direction of selection, but it does change the magnitude of immigration and, by changing
the skill level of the marginal immigrant and the average immigrant. If there is already
tive selection, the skill threshold for migrating falls and more people migrate. Migration
becomes less selective in this case, as Figure 4.3(b) shows.
The effect of a change in migration costs can also be seen by shifting one of the lines.
example, an increase in migration costs is effectively a lower wage at all skill levels in the
tination, or a downwards shift in the skill-wage line for the destination.
Refugees and selection
Borjas discusses a third type of selection, when immigrants are from the low end of the
distribution in the origin but are in the upper end of the wage distribution in the desti
country. This is particularly likely to happen when P is negative. Borjas terms this “refugee
ing” or “inverse sorting” and argues that this case is particularly likely for migration from
tries that adopt a Communist or Socialist government that confiscates privately owned
When such transitions occurred in Eastern Europe, Cuba and parts of Latin America during:
twentieth century, many highly skilled people whose prospects at home worsened mi
Those migrants typically were successful in the market-oriented economies of their desti
However, not all refugees are successful. Many have low skill levels and do poorly in
the origin and the destination. If they are doing well in the origin, they have little re:
flee. Refugees may be negatively selected relative to the destination if they would not mi
absent some adverse event, such as a natural disaster or a civil war. Because their migrati
not motivated by potential economic gains, they are unlikely to be selected on characteri
that are valued in the destination. Refugees also may have little choice of destination co
Selection in Immigration 85
but simply flee to the nearest safe country or to a country that will take them, not necessarily
country in which they will be the most successful. Ultimately, the direction of selection
nong refugees depends on the nature of the refugee-producing event and other idiosyncratic
ctors (Chin and Cortes, 2014).
mediate selection
migration costs depend on skill, intermediate selection may occur instead of positive or
gative selection. Intermediate selection is when immigrants are from the middle of the skill
d wage) distribution. Suppose the return to skill is higher in the origin than in the destina-
as in Figure 4.2(b). If all immigrants have the same migration costs, negative selection
curs. But if migration costs decrease as skill increases, intermediate selection may occur
cad. Low-skilled people have bigger gains from migrating but also higher migration costs
people with more skill. Low-skilled people therefore may not benefit from migrating.
aple with intermediate skill levels may benefit from migrating from a country with rela-
ely high inequality because their migration costs are lower than for low-skilled people.
here are several reasons why migration costs might decrease as skill increases. People with
skill levels and hence low incomes may face “cash-in-advance” or liquidity constraints,
tions where they cannot save or borrow enough to pay the costs of migrating. Such con-
is may not be binding for people with higher skill levels and hence higher incomes.
gration policies that favor skilled migrants may make migration costs lower for skilled
s than for unskilled migrants. In addition, migrant networks may increase with skill.
ration costs then would decrease as skill increases since having a bigger migrant network
ces migration costs, all else equal.
gration costs instead increase with skill, intermediate selection may occur from coun-
with relatively low inequality. Intermediate selection might occur if an origin country
relative low inequality imposes highly progressive taxes—tax rates that increase with
on emigrants, for example. In that case, high-skilled, high-income people may stay
er to avoid paying high taxes if they emigrate.
selection model applies to selection on both observable characteristics and unobservable
. Education and income are examples of observable characteristics. Unobserv-
eristics are, by definition, characteristics that datasets do not include since they are
able or easily quantified by researchers. Ambition and willingness to work hard are
es of unobservable characteristics. Economists usually measure unobservable character-
estimating wage models. The residual, or error term, after controlling for observable
ics serves as a measure of unobservable characteristics. Selection on observable and
ble characteristics is usually in the same direction, although not always.
ng up the model
Hlection model has two key insights. First, the number of immigrants depends not only
ive incomes in the origin and the destination but also on the variances of those incomes,
relative return to skill. Second, not just the number of immigrants but also their com-
heir skill level—depend on the relative return to skill. In essence, less-skilled
88 Immigrant Selection and Assimilation
5 + Guyana
Haiti q Barbados,
e .
»
Zimbabwe * a * E
e».
Papua New Guine4 + .t e
Log odds of emigration for tertiary educated
.
«é e sor. ys
1 . osio UI e lreland
ARAN Mexico
a: $ 3 > E * Slovenia
e ¿eye anal
runs e ose +. *Z *Australia
pana. Le *+ Kazakhstan
..
54 0. + United States
+ *Russia
Turkmenistan
7
3
9 7 5 3 E] 1 3 5
Log odds of emigration for primary educated
Figure 4.4 Emigration rates for adults by source country and education, 2010.
Source: Emigration rate from Briicker, H., Capuano, S., Marfoulk, A. (2013) “Education, gender
international migration: Insights roma panel-datasct 1980-2010 "Available at http: / /www.iab.de/en/d:
iab-brain-drain-data.aspx [12 December 2013]. The data include adults age 25 and older.
Like the United States, Russia and Turkmenistan have higher returns to skill than most
the twenty OECD destination countries included in the data used here and are below
45” line in Figure 4.4. Mexico is also below the 45” line, although closer to it than Rus
Turkmenistan and the United States.
It is difficult to use Figure 4.4 to draw clear conclusions about the validity of the selectis
model. On the one hand, the fact that most immigrants are positively selected seems at
with the selection model. We should observe a mix, with countries with relatively low retu
to skill above the 45” line and countries with relatively high returns to skill below the 45" li
On the other hand, the data combine 20 destination countries, some of which have hi
returns to skill than many origin countries and some of which have lower returns to
Looking at a single destination country may therefore be more useful.
Figure 4.5 looks at selection among immigrants to a single country, the United Sta
The horizontal axis in the figure is the ratio of earnings of workers with a tertiary educ
tion to workers with only a primary education in 31 countries. This ratio is a proxy for
return to skill. The vertical axis is the natural log of the ratio of the emigration rate to
United States among adults with a tertiary education to the emigration rate among ad:
with a primary education for those 31 origin countries. The selection model predicts
the figure should show a negative relationship—+the more high-skilled workers earn relati
to low-skilled workers in the origin country, the lower the emigration rate should be am:
Selection in Immigration 89
>
3 E + Turkey
Pas + Switzerland Al
3 3 Darpiad + United Kingdom
3 «Finland
New Zealand,
+ Franco
25 "+ Sweden A
E ó + Norway *Belgium +Chile
> pe Israel
2 .s «Slovakia
, E 15 Spain "Netheriands - + Luxembourg
ñ «Ireland
4 8 *Korea
3 «Germany
«Italy + Portugal
ESO Australia + Hungary
0.5 + Greece
E + Canada
Eze,
1 15 2 25 3 3.5 4 45 5
Earnings of tertiary-educated workers/earnings of primary-educated workers
Figure 4.5 Relative carnings and emigration rates to United States for tertiary- and primary-
educated adults, by origin country.
Source: Emigration rate from Briicker, H., Capuano, S. and Marfoulk, A. (2013) “Education, gender and inter-
national migration: Insights from a panel-dataset 1980-2010.” Available at http: / /www.iab.de/en/daten/iab-
brain-drain-data.aspx [12 December 2013]; carnings data from table A6.1 of OECD (2013) Education at
a Glance 2013. Available at: http://www.oecd.org/edu/cag2013%20(eng)--FINAL%2020%20June%20
2013.pdf [19 March 2014].
illed workers relative to low-skilled workers. However, this is not the case. Statisti-
there is not a relationship between the return to skill and the emigration rate among
origin countries.
'e early research finds results that appear consistent with the selection model. For
le, Borjas (1987) analyzes data on male immigrants from 41 countries to the United
He concludes that male immigrants” wages are negatively related to income inequality
ir origin country, as measured by the 90/20 income share. This finding is consistent with
selection model. However, the finding is sensitive to what other variables are included in
lel. Deborah Cobb-Clark (1993) applies Borjas's approach to female immigrants from
tries to the United States. She similarly finds that immigrants” wages tend to be neg-
related to income inequality, as measured by the 95/20 income share. She also finds
immigrants” wages may be negatively related to the return to schooling in their origin
; as measured by the return to completing secondary education (high school). These
are again consistent with the selection model but sensitive to what other variables are
in the model.
studies of the selection model were hampered by data limitations. Testing the selec-
model requires data on the returns to skill in the origin and in the destination and data on
ion rates by skill level. Early studies had proxies for the rates of return to skill in the
and destination and data on the characteristics of immigrants, but they lacked data on
90 — Immigrant Selection and Assimilation
people who did not migrate. Early studies therefore examined the relationship between ti
average skill level of immigrants from a country and the relative rate of return to skill.
relationship is not a true test of the selection model because it says nothing about the skills
immigrants relative to the population of the origin country. For example, an immigrant
developing country might look low skilled compared with a U.S. native or an immigrant
an industrialized country but might be from the top portion of the developing country's ski
distribution. Early studies were also limited to examining only one or a handful of destinati
countries.
In recent years, the Internet and faster computers have allowed researchers to use bet
data. Frédéric Docquier, Abdeslam Marfouk and colleagues led the way by creating esti
of emigration rates by education from a large number of origin countries to OECD desti
tion countries. Such estimates allow researchers to better test the validity of the selecti
model, The results provide support for the model, but only under certain conditions.
For example, Michele Belot and Timothy Hatton (2012) examine the education levels
immigrants from 70 source countries to 21 OECD countries. They find that immigrants tend
be more positively selected in terms of education as the difference in wages between high-
low-educated workers, which proxies for the relative return to skill, widens between the des
nation and the source country. In other words, as the return to skill increases in the destil
relative to the origin, selection becomes more positive. However, they only find this res
when they control for the poverty rate in the source country. Belot and Hatton hypothesize
poverty prevents low-skilled people from migrating from countries with high returns to ski
Jeffrey Grogger and Gordon Hanson (2011) apply the selection model to data on the ed
cation levels of immigrants from more than 100 source countries to 15 high-income OEG
destination countries. They note that immigrants to the OECD countries they examine
positively selected even though in many cases the model predicts, based on the relative retu
to skill, that they should be negatively selected. Grogger and Hanson find that bigger dife
ences in the relative return to skill between the destination and the origin decrease immig
selectivity, the opposite of the model's prediction.? They obtain this result using a variety
measures of the relative return to skill. They argue that liquidity constraints—the inability
save or borrow enough to migrate-—may explain why low-skilled people do not move
though they would experience the biggest proportional gains from moving. Other reseas
also finds evidence at odds with the Roy model and attributes it to migration costs or ba
to immigration by the low skilled (e.g., Briicker and Defoort, 2009).
There are several additional limitations that apply to both earlier and more recent studies
One is that most available data combine all types of immigrants. Ideally, researchers would ha
separate data on economic immigrants, refugees and family-based immigrants. The model
a different prediction for refugees than for economic immigrants, and it does not apply well
family-based immigrants who do not plan to work. Data that combine all types of immigran
may appear to not support the model simply because the data combine groups with differe
predicted effects. Another limitation is that researchers typically do not know immigra
characteristics prior to migration. Immigrants” earnings in the origin before migrating
usually not known, nor are their education levels prior to migrating. If immigrants acq
education in the destination, it may not be a surprise that immigrants have more educatio
than people who do not migrate.*
Selection in Immigration 93
ed relative to the destination country skill distribution. However, more evidence on
ge growth among refugees relative to natives, not just other immigrants, in various desti-
tion countries is needed.
One way to examine the role of immigration policy in selection is to look at how immi-
nts are selected in the absence of immigration policy. During the late 1800s and early
300s, the United States imposed few barriers to immigration from Europe. Virtually
one in good health with the funds to migrate could enter the United States and stay
e. Ran Abramitzky, Leah Platt Boustan and Katherine Eriksson (2012, 2013) examine
migrants from Norway to the United States during that time. One-quarter of Norway's
pulation eventually migrated to the United States. Norway had a more unequal income
tribution than the United States at the time, so the Roy model predicts negative selec-
n. Historical records indicate that immigrants were indeed negatively selected in terms
family wealth. Norwegians who, because of their birth order or sibling composition,
uld expect to inherit land were less likely to migrate. In addition, migrants had poorer
thers than non-migrants did.
Today, there are no barriers to immigration to the United States by people born in U.S. ter-
tories, such as Puerto Rico. Earnings inequality is higher in Puerto Rico than in the United
jates, suggesting that Puerto Rico—U.S. migration should be negatively selected. Research
s that this is indeed the case. As of 2000, almost 45 percent of working-age men born in
erto Rico who had not completed high school had moved to the United States, compared
th 30 percent of college graduates (Borjas, 2008). Interestingly, Puerto Rico also experi-
s inflows from the United States, typically by descendants of earlier Puerto Rico-U.S.
grants. These U.S.—Puerto Rico migrants are positively selected on education, as the Roy
odel predicts.
The United States also allows unrestricted immigration by citizens of the Federated States
Micronesia (FSM), a set of islands in the western Pacific Ocean. Research shows that FSM—
LS. immigrants have higher education levels than non-migrants (Akee, 2010). Immigrants
tend to be positively selected on earnings. Earnings inequality is higher in FSM than in
aii and Guam (the main places FSM migrants go), so this positive selection is counter to
Roy model's prediction. The relative return to skill is thus not the only factor that deter-
ines the composition of migration flows,
Zable 4.2 Distribution of Mexican residents and immigrants by education in 2000
Years of education Men Women
Mexican residents Mexican immigrants Mexican residents Mexican immigrants
0-9 69.4 60.1 72.5 62.0
10-11 45 55 4.0 9
12 10.1 212 44,2 20.4
1315 4.7 3.3 Hed 198
16+ ES 5.0 8.0 4.8
Source: Table 2 of Chiquiar, D. and Hanson, G. H, (2005) “International Migration, Self-Selection, and the Distri-
bution of Wages: Evidence from Mexico and the United States” Journal of Political Economy 113(2), pp. 239-281.
94 Immigrant Selection and Assimilation
Selection among Mexico—U.S. immigrants
Economists have devoted considerable attention to the direction of selection among Mexico—
U.S. immigrants because of the sizable magnitude of this migration stream. The Roy model pre-
dicts that immigrants from Mexico will be negatively selected since Mexico has greater earnings
inequality and higher returns to schooling than the United States. However, several studies show
that Mexican immigrants tend to be more educated than non-migrants (e.g., Chiquiar and Han-
son, 2005; Orrenius and Zavodny, 2005; Caponi, 201 ye Table 4.2 shows the distribution of
adult Mexicans living in Mexico and in the United States in 2000. The proportions who have at
most nine years of education or 16 or more years of education are lower among immigrants than
among non-migrants. The opposite is true for intermediate levels of education. The proportion
with 12 years of education is about twice as high among immigrants as among non-migrants.
Importantly, these data do not include people who migrated before age 21. This pattern is |
therefore unlikely to be due to Mexican immigrants acquiring education in the United States.
In addition, research shows that Mexico—U.S. immigrants” skill levels would place them in
the middle or upper end of Mexico's earnings distribution (Chiquiar and Hanson, 2005). This
again suggests that the direction of selection in Mexico—U.S. migration is intermediate or pos-
itive, not negative. Despite this evidence of intermediate selection on education, Mexico-U.S.
immigrants appear to be negatively selected from Mexico's earnings distribution—they
are from the bottom of the earnings distribution there. This suggests that workers who earn
less in Mexico than they should given their skill level—workers with a relatively low return
to skill—are more likely to migrate (Ambrosini and Peri, 2012; Kaestner and Malamud,
2014). Together, these patterns are consistent with Mexico—U.S. immigrants being negatively
selected on unobservable characteristics, whereas intermediate selection occurs on observable
characteristics like education.
Barriers to migration by the least skilled may explain why the direction of selection in
Mexico—U.S. immigration does not match the model's prediction with regard to observable
characteristics. Such barriers include limited networks and liquidity constraints. Research
shows that Mexico-U.S. migration is negatively selected from communities with sizable
migrant networks but positively selected from communities with small migrant networks
(Ibarraran and Lubotsky, 2007; McKenzie and Rapoport, 2010). Less-skilled migrants may
rely on friends or family in the United States to help them find jobs and housing there. Skilled
migrants, in contrast, may be better able to navigate U.S. job and housing markets on their
own. As Mexican migrant networks have grown over time in the United States, immigrants
from Mexico are likely to have become more negatively selected. Research indicates this |
appears to be the case (e.g., Campos-Vazquez and Lara, 2012).
Bigger networks also help potential immigrants borrow from people who have already
migrated. This relaxes liquidity constraints that may prevent the poor from migrating, Selec-
tion from rural Mexico tends to be more positive than from urban Mexico, in part because
of differences in the return to skill but also because of differences in networks and wealth
(Fernández-Huertas Moraga, 2013). An anti-poverty program in Mexico named Oportuni-
dades that gives money to low-income households leads to more migration by those house-
holds, consistent with liquidity constraints (Angelucci, 2013). The additional migration
Selection in Immigration 95
ers the average skill level among migrants, suggesting that the direction of selection is
sitive or intermediate.
U.S. border enforcement appears to affect selection among unauthorized Mexican immi-
rants. Selection becomes more positive when U.S. border enforcement, as measured by
tal hours worked along the U.S.—Mexico border by U.S. Border Patrol agents, is stricter
enius and Zavodny, 2005). Stricter enforcement raises migration costs because unautho-
ed immigrants are more likely to need to hire a coyote-—a smuggler—to help them enter
United States. In terms of Figure 4.2(a), the destination wage line effectively shifts down
en migration costs go up. The degree of positive selection then increases, and the number
immigrants falls.
lection on health
economic research on immigrant selection has focused on immigrants” education or
ings, but a few studies have examined selection on health. Studies typically find that immi-
rants are positively selected on health relative to both non-migrants in the origin and natives
the destination (e. 2+, Kennedy et al., 2006). This hypothesis is known as the “healthy immi-
nt effect.”
- There are several potential explanations for the healthy immigrant effect. First, health tends
be positively related to education and income. Immigrants tend to be positively selected
education relative to non-migrants and must have enough income to bear migration costs,
it is not surprising that immigrants tend to be healthier than non-migrants in the origin.
ít immigrants also tend to be healthier than natives in the destination, even in destination
ountries where immigrants have less education and lower incomes, on average, than natives.
migration policies that screen immigrants on health, income or skill may contribute to the
salthy immigrant effect.
Self-selection may also play a role. Characteristics that increase the likelihood a person
“comes an immigrant, such as having a low discount rate, may be associated with healthy
iors. Such behaviors include exercising, eating a healthy diet and not smoking. Immi-
ts also tend to be relatively young, so adverse health conditions may have not yet man-
fested for many immigrants, The healthy immigrant effect tends to decrease as duration of
esidence in the destination increases, as discussed in the next chapter.
Irish immigrants to England are an interesting example of selection and health. As Liam
Delaney, Alan Fernihough and James Smith (201 3) show, Irish immigrants to England who were
before 1920 or after 1960 tend to be healthier than their counterparts who remained in
and and than the English. However, the opposite is true for those who were born between
920 and 1960. That group of migrants tended to be negatively selected on education as well,
nd they had experienced relatively high rates of child abuse. This created high psychic costs of
faying in Ireland for many of them. Although studies typically emphasize the psychic costs of
igrating, it is important to note that some people may face greater psychic costs of staying, The
dy notes that although these migrants have relatively poor physical and mental health, they
obably benefited from the fact that health care quality tended to be higher in England than in
land when they migrated.
98 — Immigrant Selection and Assimilation
Never migrate Migrate and stay
in destination
Skill
Figure 4.6 The direction of selection in permanent and return migration if the relative return
skill is higher in the destination.
given in equation 4.11 migrate and never return. People in the middle region migrate and
return. Both cutoffs shift to the left as the average wage in the destination, Él, increases and
the right as the average wage in the origin, a, increases. An increase in the value in the origi
of skills acquired abroad, k, widens the intermediate range by causing the lower cutoff to
to the left and the higher cutoff to shift to the right. More people will migrate and then re?
because skills acquired abroad are more valuable in the origin. An increase in $, the relati
return to skill in the destination, shrinks the intermediate range by causing the cutoffs to mi
closer together. Fewer people migrate, but those who do are less likely to return migrate.
1 Ó is positive and less than one, the return to skill is lower in the destination than in
origin, The Roy model then predicts that immigration is negatively selected. The inequaliti
in equations 4.11 and 4.12 switch direction, The most-skilled people never migrate. Amo
migrants, the least skilled remain in the destination while those in the middle region ret
migrate.
In essence, the model predicts that return migration reinforces the direction of selection k
migration. If migration is positively selected, it becomes more positively selected over ti
as a consequence of return migration. If migration is negatively selected, it becomes mo)
negatively selected over time as a consequence of return migration.
The model also predicts the optimal duration of migration, T*, which is
. _f-u+(n-1)e+k (4.13
2k
T
Equation 4.13 yields several predictions.*The optimal duration of migration increases as
average wage in the destination (H) rises and decreases as the average wage in the origin (M)
rises. If the return to skill is higher in the destination than in the origin (8 > 1), the opti
Selection in Immigration 99
tion of migration increases as skill, £, increases. The opposite is true if the return to skill
er in the origin.
ical evidence on selection in return migration
and Bratsberg examine the validity of the model using average wages and out-migration
És among U.S. immigrants by country of origin. They find that immigrants” average wage
reases as the out-migration rate increases if the return to skill —measured using the 90/20
e share-—is lower in the origin than in the United States. In this case, the model predicts
selection in initial migration is positive, and selection in return migration is negative.
ersely, average wages decrease as the out-migration rate increases if the return to skill
her in the origin than in the United States. In this case, the model predicts that selection
¡al migration is negative, and selection in return migration is positive. The relation-
between average wages and out-migration rates that Borjas and Bratsberg find match
nodel's prediction that out-migration intensifies the direction of selection. They also find
out-migration rates from the United States increase as origin country GDP rises, which
nsistent with the model.
related research, Bratsberg (1995) finds support for the model among foreign students
in the United States. The lower the return to skill or income inequality in their home
, the more likely students are to stay in the United States after finishing their studies.
ents tend to return if they receive a high return in the origin on their studies. Ifincome
ty is lower in the origin than in the United States, the average wage among highly
cated immigrants from that country falls as more foreign students stay in the United
es after completing their studies. Conversely, the average wage among highly educated
'ants increases as more foreign students stay in the United States after completing their
es ifincome inequality is higher in the origin than in the United States. In other words,
least skilled leave if income inequality is lower in the origin, pushing average wages up
ng those who remain. The most skilled leave if income inequality is higher in the origin,
ing down average wages among those who remain. This result is consistent with the
Out-migration from the United States generally appears to be negatively selected. One
finds that out-migrants are negatively selected among foreign-born scientists and engi-
s in the United States (Borjas, 1989). Another study likewise finds that out-migration
n the United States is negatively selected among immigrants as a whole (Lubotsky, 2007).
pattern is different for Sweden and Finland. Migration to those countries is negatively
ected on education, while return migration from those countries is positively selected
poth and Saarela, 2007). Sweden and Finland have much less income inequality and lower
s to skill than the United States, which may explain this pattern.
The direction of return migration to Mexico from the United States has received consid-
ble attention. Return migration to Mexico from the United States is high-—an estimated
percent of people who left Mexico for the United States during 2005 to 2010 returned
Mexico by 2010, for example (Campos-Vazquez and Lara, 2012). Researchers have com-
ed return migrants with people who have never migrated and with migrants who have
ined in the United States. One study finds positive selection among return migrants
100 Immigrant Selection and Assimilation
relative to never-migrants in 1990 but negative selection for men and intermediate selection
for women in 2010 (Campos-Vazquez and Lara, 2012). The change could be due to a decrease
in the return to skill in Mexico relative to the United States over time, a decrease in migration
costs or an increase in networks that enables less-skilled people to migrate but also leads to
them returning. Return migrants appear to be positively selected among the pool of Mexican:
migrants (Ambrosini and Peri, 2012; Biavaschi, 2012).
Final thoughts on selection
Selection can be measured relative to either the origin or the destination. Origin countries
are mainly interested in the characteristics of the people who leave relative to those who stay.
Destination countries are mainly interested in the characteristics of immigrants relative to
natives and earlier immigrants, not relative to the origin population. Both origin and des-
tination countries care about whether return migrants are positively or negatively selected
among immigrants in the destination. The next chapter discusses another way of thinki
about immigrant selection: assimilation, or how well immigrants do in the destination, both.
initially and over time.
Problems and discussion questions
1 Explain how a decrease in the return to skill in the origin affects the direction of selection
and the number of immigrants if (a) immigration is positively selected, or (b) immigra-
tion is negatively selected. Assume that the return to skill in the destination and migration
costs do not change.
2 Explain how a decrease in the average wage in the destination affects selectivity and the:
number of immigrants if (a) immigration is positively selected, or (b) immigration is neg-
atively selected. Assume that the return to skill in the destination and migration costs do.
not change.
3 Suppose a destination country adopts a policy that guarantees all residents, including.
immigrants, a minimum income well above the current lowest income in the origin.
How will this affect the direction of selection and the number of immigrants if (a) immi-
gration is positively selected, or (b) immigration is negatively selected? Assume that the:
earnings distribution in the origin and migration costs do not change.
4 Cana destination country experience both positive and negative selection in immigration
at the same time? Why or why not?
5 Can a destination country experience both immigration from an origin country and
return migration to that country at the same time? If so, what does that imply about the
directions of selection for immigration and return migration?
6 Explain what happens to the number of immigrants, the number of return migrants and
selectivity in initial and return migration if the value of skills acquired in the destination:
declines in the origin country. What happens if instead the return to skill falls in the des-
tination relative to the origin?
7 Income taxes can affect immigrant selection. Suppose the destination country has a more
progressive tax structure than the origin country. How would this affect selection?