You are on page 1of 14

Journal of Economic Literature 2015, 53(4), 961–974

http://dx.doi.org/10.1257/jel.53.4.961

Immigration and Globalization:


A Review Essay †
George J. Borjas*

This essay revisits the argument that the removal of worldwide immigration restric-
tions would induce a very large increase in world GDP. The recent books Exodus: How
Migration is Changing Our World by Paul Collier and The Price of Rights: Regulating
International Labor Migration by Martin Ruhs raise a number of questions about the
underlying economic model. The essay shows how these concerns can greatly attenu-
ate the predicted gains. ( JEL F22, F66, J11, J18, J61)

1.  Introduction increases global income and tends to equal-


ize prices and wages across countries.

A simple question lies at the core of any


examination of the link between immi-
gration and globalization: what exactly would
Decades of experience with various trade
liberalization policies, however, do not seem
to have had as much of an impact on global
a world without national borders—a world income or on international wage inequality
in which people could move freely from one as the proponents of free trade would have
country to another—look like? expected. This fact has motivated some
Economists have, in fact, devoted a lot of economists to consider yet another scenario:
effort to documenting how international dif- the removal of immigration restrictions that
ferences in economic conditions change as prevent the movement of people across
national governments lower the barriers that countries.
limit trade across countries. Much of interna- The modeling of economic adjustments
tional trade theory attempts to imagine what resulting from unrestricted international
happens to employment, prices, and incomes migration adds greatly to the complexity
when countries allow unrestricted flows of of describing what would happen in such
goods and capital across national boundaries. a world: How many persons would move?
One common theme in these models, which What would economic conditions in the new
has greatly influenced economic policy, is borderless world look like? What would hap-
that the removal of restrictions on such flows pen to the institutions and social norms that
govern economic exchanges in specific coun-
* Harvard Kennedy School. I am grateful to Kirk tries after the entry/exit of perhaps hundreds
Doran, Daniel Hamermesh, Gordon Hanson, and Fran- of millions of people? Would the institutions
cesc Ortega for comments on an earlier draft of this paper.

 Go to http://dx.doi.org/10.1257/jel.53.4.961 to visit the that presumably led to efficient exchanges in
article page and view author disclosure statement(s). the richer countries remain dominant and

961
962 Journal of Economic Literature, Vol. LIII (December 2015)

spread throughout the globe, or would these Ruhs’s The Price of Rights: Regulating
institutions be replaced by the political and International Labor Migration shows that
cultural inefficiencies that may have ham- practically all receiving countries walk by
pered growth in the poorer countries? These the trillion-dollar bills promised by the eco-
additional complexities arise from a simple nomic models, and instead set up a variety
fact embodied in Max Frisch’s profound quip of strict and sometimes draconian immi-
about the German guest worker program of gration restrictions. Put together, the two
the 1960s: “We wanted workers, but we got books suggest (at least to me) that perhaps
people instead.” it is time for a reappraisal of the economic
Although much of the economic litera- argument that unrestricted migration would
ture on immigration has typically focused on generate huge global gains.
estimating employment and fiscal impacts in
specific receiving or sending countries, there
2.  Basic Model
has been a parallel tradition that attempts
to examine the impact of international To get a better grasp of the issues at hand,
migration flows from a global perspective. it is best to begin with a description of the
Beginning with the seminal work of Hamilton basic model. The key question is straight-
and Whalley (1984), a number of studies forward: what types of gains or losses would
propose a variety of models that are then cal- accrue to the world’s population if countries
ibrated to describe what the economy would decided to remove all legal restraints to
look like if sovereign countries surrendered international migration and workers moved
their ability to restrict in- or out-migration to those countries that offered them the
flows.1 One common implication from these highest wages? To illustrate the source of the
simulations is that the global gains from the gains, consider the impact of migration flows
removal of immigration restrictions would in a two-region setting, the North (N) and
be huge, amounting to trillions of dollars the South (S). The North is an industrialized
annually. This finding has led to a popular region with relatively few workers, while the
metaphor that there are “trillion-dollar bills” South is a developing region with a large
lying on the sidewalk, ready for the taking, population. Initially, the wage in the North,​​
if only the receiving countries would remove w​ N​​,​ exceeds the wage in the South, w​ 
​​ S​​​.
the self-imposed migration barriers. As illustrated in figure 1, the two regions
The two books that form the basis for this have competitive labor markets with down-
review essay address some of these global ward-sloping labor demand curves that are
issues from very different perspectives. assumed to have the same slope. The two
Paul Collier’s Exodus: How Migration is demand curves, however, may have differ-
Changing Our World examines whether the ent intercepts. The intercept differential
available evidence suggests that the unre- measures the economic value of the dif-
stricted flows of labor will, in fact, generate ferent “infrastructures,” with the North’s
the sizable gains that are promised by the infrastructure allowing a specific worker to
generic study in the literature, while Martin have a higher value of marginal product in
the North than in the South. It is useful to
1 Representative studies include Benhabib and think of the infrastructure not only in terms
Jovanovic (2012), Clemens (2011), di Giovanni, Levchenko, of physical capital, but also as including the
and Ortega (2015), Docquier, Machado, and Sekkat (2012), value of the political, social, and cultural
Kennan (2013), Klein and Ventura (2007), Lundborg and
Segerstrom (2002), Moses and Letnes (2004), Pritchett institutions and organizations that regulate
(2010), and Walmsley and Winters (2005). social and economic interactions in the two
Borjas: Immigration and Globalization: A Review Essay 963

Panel A. North Panel B. South

SN SN* SS* SS

WN

A
Wage

Wage
W* W*
DN
C
WS WS

B DS

LN LN + M LS − M LS

Employment Employment

Figure 1. The Global Gains from Open Borders

regions. As is common in the literature, I however, prevent this flow. Suppose, how-
initially consider a short-run situation where ever, that all migration restrictions are sud-
the infrastructure is fixed in each region, so denly lifted and that migration from the
that the height of each labor demand curve is South to the North is costless. Workers will
fixed. Finally, assume that all labor is homo- flow from the low-wage to the high-wage
geneous and supplied inelastically in each region as long as any wage difference exists,
of the regions, with supply curves ​​S​ N​​​ and ​​SS​  ​​​, and the flow will stop only when the wage
respectively.2 ​​ ∗​​. This
in the two regions is equalized at w​​ 
The figure illustrates the initial labor mar- new equilibrium is reached when a total of M
ket equilibrium: the wage in the North is sub- workers have left the South and entered the
stantially higher than the wage in the South. North’s labor market.
Income-maximizing workers in the South The out-migration reduces the South’s
wish to move to the North to take advantage GDP by an amount equal to the area of the
of the higher wage. Immigration r­ estrictions, trapezoid ​C​, while the entry of M​ ​ workers
increases the North’s GDP by the area A ​  + B​.
2 Under some conditions, the homogenous labor The area of the trapezoid giving the North’s
assumption may be far less restrictive than it appears. It gain exceeds the area of the trapezoid giv-
is trivial to reformulate the model by interpreting employ- ing the South’s loss. In particular, note that
ment level L in figure 1 as the number of efficiency units
supplied in a given market, and the efficiency units are the South’s loss in GDP is also given by the
defined by an Armington (CES) aggregation of the num- (­
mirror-image) trapezoid B ​​ in the graph
ber of workers belonging to particular skill groups. Borjas illustrating conditions in the North’s labor
(2014, chapter 3) presents a technical discussion of the
implications of this approach to allowing for a heteroge- market. The global immigration surplus,
neous workforce. therefore, is given by the triangle ​A​. This
964 Journal of Economic Literature, Vol. LIII (December 2015)

s­urplus arises for two distinct reasons: The ­estimate of the ratio ​​α​N​​/​αS​ ​​​. In particular,
first is the relative imbalance of supply, with note that:
“too many” workers initially in the South. The η
​α​N​​ ___
​wS​   ​ ​​   = ​(​  ​αS​  ​​​  )​​​(​  ​LS​   ​​​  )​​​  ​​.
second, and more important one, is the fact ​wN​  ​​ ​LN​  ​​
R  = ​ ___
(2) ​ ___
that the Northern infrastructure increases
the marginal product of any worker.
In Borjas (2014), I proposed a parame- Equation (2) shows that it is trivial to calcu-
terization of this model that leads to trans- late the ratio ​​α​N​​​  /​​αS​ ​​​  once we know the geo-
parent and straightforward calculations and graphic distribution of the workforce, the
gives some sense of the magnitude of the value of the wage ratio R ​ ​, and the value of
key underlying effects. The transparency the elasticity ​η​.
is particularly useful because it exposes the World GDP prior to the relaxation of
robustness of the predictions to changes in immigration restrictions is given by the sum
the underlying assumptions. Specifically, of the relevant areas under the value of mar-
suppose that the inverse labor demand func- ginal product curves:
tions in the two regions are given by:
(3) ​​Y​ 0​​  = ​∫0 ​  ​​ ​αN
​ ​​ ​L​​  η​  dL  + ​∫0 ​  ​​ ​αS​ ​​ ​L​​  η​  dL 
​LN​  ​​ ​LS​  ​​

(1a)  ​log ​wN​  ​​  =  log ​αN


​ ​​ + η log ​LN​  ​​​,
​α​S​​ ___ ​α​ ​​
(1b) ​log ​wS​  ​​  =  log ​αS​ ​​ + η log L
​ S​  ​​​. = ​ _____
1 + η [
   ​ ​ ​  ​αN​  ​ ​​​  LN​  1+η
S
​  ​]​​,
​  ​ + ​LS​  1+η

These demand curves implicitly assume where ​​Y0​  ​​​gives the (known) value of world
that the aggregate technology in each of GDP in the initial equilibrium. Note that
the two regions can be represented by a the simultaneous solution of equations (2)
Cobb–Douglas production function. Fur­ and (3) uniquely solves for the values of the
ther, the demand curves incorporate the intercepts ​​α​N​​​ and ​​αS​ ​​​.
assumption that the factor price elasticity η​ ​ If there were no legal restrictions on
is the same in both regions (with η​   <  0​). In international migration and if migration
a linear homogeneous Cobb–Douglas world, were cost­less, the opening up of the borders
the factor price elasticity would equal (the equates wages between the two regions. This
negative of) capital’s share of income, so that equilibrium condition implies that the num-
the parameter η​ ​ will be set to −0.3 in the ber of movers is implicitly defined by:
simulation that follows.
A great deal of evidence suggests that ​α​N​​(​LN​  ​​  + M​)​​  η​
   ​​  = 1.
(4) ​​ ___________
  
the ratio ​​α​N​​/​αS​ ​​​ is greater than one. In fact, ​αS​ ​​(​LS​  ​​  − M​)​​  η​
the productivity of a specific worker could
easily triple or quadruple by moving him or Finally, the gains to world GDP (∆Y 
her from the South to the North. Clemens, = ​​Y​ 1​​​  − ​​Y0​  ​​​
) are given by the difference
Montenegro, and Pritchett (2008) report between the gains accruing to the North
that the adjusted wage ratio for a low-skill after the entry of ​M​ workers and the losses
worker between the United States and a suffered by the South after the exit of M ​ ​
large number of low-income countries is workers, or:

ΔY  = ​∫​L N​  ​ ​​


around 4.1. ​LN​  ​​+M
The functional form assumptions in equa- (5) ​ ​​ ​αN
​ ​​ ​L​​  η​  dL 

− ​∫​L S​  ​​−  M​​ ​​ αS​ ​​ ​L​​  η​  dL.


tion (1) provide all the information that ​LS​  ​​
is needed to get a back-of-the-envelope
Borjas: Immigration and Globalization: A Review Essay 965

Table 1
The Simulated Global Gains from Unrestricted and Costless Migration

Value of R (​​wN​  ​​​/​​wS​  ​​​)


2 4 6
1. Net gain in world GDP (in trillions of dollars) 9.4 40.1 62.4
2. Number of migrant workers (in billions) 1.7 2.6 2.7
3. Number of migrants as percent of South’s workforce 62.3 94.8 98.6
4. Number of movers (in billions) 3.7 5.6 5.8
5. Percent wage change in the North −33.0 −39.3 −39.8
6. Percent wage change in the South 34.0 143.0 261.1
7. Change in income of capitalists (in trillions of dollars) 2.8 12.0 18.7
8. Percent change in income of capitalists 13.4 57.2 89.1

Notes: The simulations assume that world GDP is $70 trillion in the premigration regime; 600 million persons work
in the North; 2.7 billion persons work in the South; and the factor price elasticity is −0.3. The number of movers
equals the emigration rate of workers times the South’s population (assumed to be 5.9 billion).

​It is straightforward to simulate the model world GDP. Specifically, row 1 of the table
in equations (1)–(5). Table 1 reports the predicts that world GDP would increase by
results of the numerical exercise.3  The var- $40 t­rillion, almost a 60 percent increase.
ious columns of the table use alternative Moreover, these gains would accrue each
values of the wage ratio ​R,​ although I limit year after the migration occurs, so that the
the discussion to the case where ​R  =  4​. The present value of the gains nears one quadril-
simulation uses the 2011 World Bank esti- lion dollars! Not surprisingly, some econo-
mate that 600 million persons worked in the mists have latched onto variants of this model
high-income countries (a region with a pop- to engage in social engineering by proposing
ulation of 1.1 billion), and 2.7 billion persons that world poverty could be eliminated in
worked in the developing countries (where one fell swoop—if only countries would stop
the population was 5.9 billion). Finally, the being countries.
value of world GDP prior to the relaxation of A second important implication of the
immigration restrictions is $70 trillion. model is that there are going to be a lot of
As in the generic study in the literature, migrants. The simulation implies that 2.6 bil-
the removal of immigration restrictions lion workers, or 95 percent of the workforce
(combined with the assumption of costless in the South, will move. If these workers
mobility) would lead to a huge increase in bring along their families, the 95  percent
mobility rate implies that nearly 5.6 billion
3 The simulation assumes that all migration occurs
persons would move from the South to the
instantaneously. This assumption obviously simplifies the North.
mechanics of the exercise, but the costs and benefits will It is fair to say that this particular impli-
most likely be staggered over time. It is unclear how the cation of the model has not received nearly
results from a dynamic model (which would require addi-
tional assumptions about the evolution of the migration the same emphasis or attention as the fact
flow) would differ from the calculations reported in table 1. that world GDP would increase by tens of
966 Journal of Economic Literature, Vol. LIII (December 2015)

trillions of dollars. For example, the original North’s native workforce fall by almost
Hamilton and Whalley (1984) article spends 40  percent, and the earnings of Southern
a great deal of time poring over detailed workers increase by 143 percent.
estimates of the dollar gains, but curiously There is one final redistributive impact
neglects to report the number of movers that is worth documenting. Specifically,
required to achieve those gains at any point the income accruing to capitalists will also
in the study. change. Define the gains to “global capital-
The glossing-over of this particular impli- ists” as the excess income produced that is
cation may be the politically sensible thing not paid directly to workers. The increased
to do if one wishes to advocate these types returns to capital are given by:
of models in policy circles.4 However, it
is conceptually impossible to buy into the Δ Income of Capitalists 
(8) ​
argument that unrestricted immigration will
increase world GDP by $40 trillion without    = ​[​Y1​  ​​  − ​w​​  ∗​  (​LN​  ​​  + ​LS​  ​​)]​ 
simultaneously buying into the prediction
that this will entail the movement of billions − ​[​Y0​  ​​  − ​wN​  ​​ ​LN​  ​​  − ​wS​  ​​ ​LS​  ​​]​.​
of people from the South to the North.
These huge flows will necessarily imply As row 7 of table 1 shows, there will be a sub-
a substantial redistribution of wealth, and stantial increase in the wealth of global cap-
these distributional consequences also tend italists, amounting to about $12 trillion, or a
to be overlooked. Specifically, the ​​L​ N​​​ native 57 percent increase over their initial income.
workers in the North will be at the losing end In short, a world integrated by unrestricted
of the deal. After all, “factor price equaliza- migration flows creates large gains for some
tion” means precisely that: factor prices are groups, but also creates large losses for a
equalized, with initially high-wage workers group of workers who will vociferously fight
eventually earning less and initially low-wage the policy shift.
workers eventually earning more. As figure 1 It is the existence of this losing group of
shows, the influx of M ​ ​ workers reduces the workers that often leads to a degeneration
North’s wage from ​​w​ N​​​ to ​​w​​  ∗​​, and raises the of the immigration debate into a collec-
wage of all Southern workers (whether they tion of slurs and facile accusations of rac-
migrated or not) from w​  ​​ S​​​ to ​​w​​  ∗​​. The implied ism. Collier’s narrative, unfortunately, is not
percent wage changes are given by: immune: “A rabid collection of xenophobes
and racists who are hostile to immigrants
η
​w​​  ∗​  − ​wN​  ​​ ​L​  ​​  + M
( ​LN​  ​​ )
lose no opportunity to argue that migration
(6)  ​​ ________ ​​  
​wN​   ​   = ​​ _______
​  N  ​    ​​​  ​ − 1.​ is bad for indigenous populations” (p. 25).
The problem with such name-calling is that
η it downplays the fact that regardless of how
​w​​  ∗​  − ​wS​  ​​ ​ ​  ​​  − M
( )
L
(7) ​​ ________  ​  
​wS​  ​​  = ​​ _______
​  S  ​    ​​​  ​ − 1.​ the Northern workers actually feel about
​LS​  ​​ immigrants, their economic grievance is real
and will not go away.
Rows 5 and 6 of table 1 report these pre-
dicted wage effects. The earnings of the
3.  Productivity Spillovers

4 Benhabib and Jovanovic (2012) and Kennan (2013) do


As noted above, the huge global gains typ-
report the estimated number of movers in these types of ically found in these types of numerical sim-
simulations. ulations have led a number of economists to
Borjas: Immigration and Globalization: A Review Essay 967

emphasize that the “gains from g­ lobalization” the entry of perhaps billions of new ­persons,
resulting from the decades-long effort to so that there is much ­ hand-wringing in
ease trade restrictions pale in comparison Collier’s discussion of social costs, and the
to the gains that are there for the taking if narrative depends far too much on refer-
countries simply removed all existing restric- ences to “mutual regard,” “trust,” “moral
tions on international migration. Clemens outrage,” and other equally hard-to-measure
(2011), for example, employs the metaphor concepts. I personally find it difficult to place
that there are trillion-dollar bills lying on much faith on the robustness of Collier’s
the sidewalk, ready to be easily picked up, if reported evidence when the heading of a
only policymakers in the industrialized world key empirical section in the chapter on social
would wisen up and remove all immigration consequences is titled “Some Illustrative
restrictions. Anecdotes” (p. 78), and focuses mainly on
Things that sound too good to be true, the experiences of the Afro-Caribbean com-
however, usually are. It is not surprising then munity in London. I have heard that the plu-
that the analysis of both Collier and Ruhs can ral of anecdote is data, but Collier’s specific
be interpreted as raising central questions anecdotes do not a data set make. As inter-
about the model that predicts the presence esting as the experiences of this particular
of these trillion-dollar bills on the sidewalk, immigrant community may be, I doubt they
as well as providing insights for understand- provide much information about what would
ing why nobody ever bothers to pick them happen if immigration restrictions were
up. The problem is easy to summarize: those removed and billions of persons moved to
bills are probably fake. the industrialized countries.
In particular, the simulation reported in There is also much to quibble about in
table 1 is a short-run, partial-equilibrium terms of the specific lessons that Collier
exercise, and its implications may have lit- draws from existing research studies. One
tle in common with what would happen in a telling example is Collier’s discussion of the
general equilibrium setting. Collier’s Exodus, wage effect of immigration on a receiving
in an important sense, marks a pivoting point country, a key parameter in any cost–benefit
in the literature by taking the l­ong-run con- exercise. Exodus is totally oblivious to the fact
sequences of migration flows much more that hundreds of thousands of words have
seriously than one sees in the stereotyp- been written on this topic in the American
ical study. Throughout the book, Collier context. The entire section discussing
emphasizes how the short-run impacts of these wage effects takes all of two pages
immigration can differ from what would be (pp. 112–13). Remarkably, not a single study
observed if the migration flow were to con- from the American literature is cited and the
tinue indefinitely: “Contrary to the preju- fact that there has been a lively debate over
dices of xenophobes, the evidence does not how to estimate these wage impacts and that
suggest that migration to date has had sig- there is still an ongoing debate about the
nificantly adverse effects. . . . Contrary to value of the factor price elasticity is ignored.
self-perceived ‘progressives,’ the evidence Similarly, Collier makes a crucial obser-
does suggest that without effective controls vation that is left dangling and unexplored:
migration would rapidly accelerate to the “desperate not to give succor to these groups
point at which additional migration would [i.e., the “xenophobes”], social scientists have
have adverse effects” (p. 245). strained every muscle to show that migration
Unfortunately, we know little (read: noth- is good for everyone” (p. 26). This is, in fact,
ing) about how host societies would adapt to quite a damning denunciation of both social
968 Journal of Economic Literature, Vol. LIII (December 2015)

scientists and the social science literature on economies without importing the “bad”
immigration, and is a clarion call for some organizations, social models, and culture
publication bias studies. It is certainly the that led to poor economic conditions in the
first time I have seen in print my long-held source countries in the first place. It seems
suspicion that many social scientists have inconceivable, however, that the North’s pro-
indeed engaged in vigorous and strenuous duction function remains unchanged after
workouts to ensure that they tipped the scale the admission of billions of new workers.
so that the published answer was “right.” As Echoing Max Frisch’s observation, Collier
a result, the safest way to digest academic bluntly states: “Uncomfortable as it may be
papers on immigration may be to follow . . . migrants bring their culture with them”
some Cold War advice from Ronald Reagan: (p. 68).
“Trust, but verify.” A preliminary approach to understanding
Regardless of these quibbles, Exodus how the “merging” of countries can affect the
makes an important conceptual contribution calculated gains is to do so explicitly within
by suggesting that because of the misguided the context of the parameterized model pre-
emphasis on showing that immigration “is sented earlier. As a first step, it seems sensible
good for everyone,” the literature has too to presume that the p ­ ostmigration intercept
often disregarded inconvenient facts, over- of the North’s inverse labor demand curve,​​
looked the potentially paradigm-changing α​  ∗N ​​,​  will lie somewhere between the orig-
general equilibrium effects, and proposed inal intercept and the South’s intercept.
the types of political upheavals that many Specifically:
observers would consider to be radical rear-
rangements of the social order. (9) ​​α​  ∗N ​ ​  = (1 − λ)​αN
​ ​​  + λ​αS​ ​​  ,​
Because there is no precise modeling and
measurement of the various costs and ben- where 0 ≤ λ ​ ​ ≤ 1. If λ
​ ​were equal to 0, the
efits, Collier does not provide a numerical intercept in the North is unaffected by the
estimate of how much accounting for the migration flow; the Northern infrastructure
general equilibrium concerns would reduce remains intact and the resulting estimates
the presumed global gains. In terms of the of the global gains are those summarized
model presented earlier, it is evident that the in table 1. If ​λ​were equal to 1, the immi-
problem with doing such a calculation is that grants “import” the entire set of institutions
we simply do not know what would happen and norms that led to the South’s poor eco-
to the shape of the North’s aggregate pro- nomic performance, totally overwhelming
duction function after the influx of billions the North’s infrastructure.5
of persons. Using this approach, it is straightforward
As Acemoglu and Robinson (2012) note, to conduct an alternative simulation of the
“nations fail” mainly because of differ- model. Suppose again that migration is
ences in political and economic institutions.
Analogously, Collier argues: “one reason
poor countries are poor is that they are short 5 I ignore the possibility that the intercept in the South’s
of effective organization” and “migrants are labor demand curve might also shift. Because there are
no labor flows from the North to the South, there is little
essentially escaping from countries with dys- reason to expect an improvement in Southern productiv-
functional social models” (pp. 33–34). For ity due to the importation of the efficient Northern insti-
immigration to generate substantial global tutions. It may be, however, that the self-selection of the
Southern out-migrants affects the stability and type of
gains, it must be the case that billions of institutional arrangements preferred by the Southerners
immigrants can move to the industrialized left behind.
Borjas: Immigration and Globalization: A Review Essay 969

Table 2
Accounting for Spillovers on Northern Infrastructure and Migration Costs

Intensity of externality

​λ = 0.0​ ​λ = 0.25​ ​λ = 0.5​ ​λ = 0.75​ ​λ = 1.0​


Panel A. Costless migration
1. Number of movers (in billions) 5.6 5.4 5.0 4.1 2.3
2. Dollar gain in world GDP (in trillions) 40.1 24.2 8.8 −5.5 −17.5
3. Dollar change in Northern income (in trillions) 12.7 5.3 −2.3 −10.2 −18.4

Panel B. Accounting for migration costs


1. Number of movers (in billions) 5.3 4.8 3.8 2.1 0.2
2. Dollar gain in world GDP net of migration costs (in trillions) 28.1 12.9 −0.9 −12.4 −20.0
3. Dollar change in Northern income (in trillions) 11.9 3.9 −4.5 −13.0 −20.0

Notes: The simulations assume that the wage ratio R ​ = 4​; world GDP is $70 trillion in the premigration regime;
600 million persons work in the North; 2.7 billion persons work in the South; and the factor price elasticity is −0.3.
The intercept of the inverse labor demand curve in the North in the postmigration period is a weighted average of
the premigration intercepts in the Northern and Southern regions, with the parameter λ ​ ​being the weight attached
to the Southern intercept; see equation (9). The “dollar change in Northern income” gives the change in the income
accruing to the Northern population after paying out the salaries of immigrants. The simulation in panel B assumes
that migration costs for a Southern worker equal ten times his initial salary, that it is costless to move all nonworking
dependents, and that the rate of discount is 5 percent.

c­ ostless so that persons can easily move back resulting from this set of assumptions is
and forth between the North and the South given by:
until wages are equalized across regions.
∫0 ​L​  ​ N​​+​M​​  ​​​ ​α​  N∗ ​ ​​  L​​  η​  dL 
This equilibrium is characterized by the ∗

restriction that: Δ ​Y​​  ∗​  = ​


(11) ​

​ ​  ∗N ​ ​​  (​LN​  ​​  + ​M​​  ∗​)​​  η​


α
 ​  = 1,​ + ​∫0 ​L​  ​ S​​−​M​​  ​​​ ​α​  S​​ ​L​​  η​  dL − ​Y​ 0​​  .​

(10) ​​ ______________
  
  
​α​  S​​ ​(​LS​  ​​  − ​M​​  ∗​)​​  η​

so that the equilibrium level of migra- Row 2 of the top panel of table 2 shows that
tion ​​M​​  ∗​​takes account of the externalities if ​λ​were equal to 0.5, the net gain falls from
that the migrants impose on the Northern $40 trillion to $8.8 trillion. In addition, if λ
​​
infrastructure.6 The gain in world GDP were equal to 0.75, the net gains become
negative because now the entire world’s
6 I assume that the size of the downward shift in the
workforce is largely operating under the
North’s labor demand curve (as measured by λ) is inde- inefficient organizations and institutions that
pendent from the level of migration, so that it essentially were previously isolated in the South but
occurs immediately after the relaxation of immigration have now spilled over to the North.
restriction and a single person moves. A generalization of
the model would allow for the value of λ to depend on the Let me stress that this is only a simula-
size of the migration flow. tion—and one should put as much faith in
970 Journal of Economic Literature, Vol. LIII (December 2015)

these numbers as one puts on the promise worker’s salary. Similarly, Artuc, Chaudhuri,
that trillion-dollar bills lie strewn all over the and McLaren (2010, p. 1021) estimate aver-
sidewalk. The exercise, however, teaches a age moving costs that are around ten times
lesson that has far too often been ignored: the annual wage for workers who move from
the gains from unrestricted immigration one sector to another in response to trade
depend largely on how the infrastructure in shocks in specific industries.8
the receiving economies adjusts to the influx To easily illustrate the attenuating effect
of perhaps billions of persons. Although we of migration costs, suppose that these costs
have no idea about how this adjustment will are constant in the working population and
pan out, there will be an adjustment. equal to ​π​times a worker’s initial salary in
In fact, even these estimates are proba- the South. Assume further that nonworking
bly too optimistic, because I have assumed dependents tag along with the “householder”
that migration is costless. Migration costs, and migrate for free. The equilibrium con-
however, are real, sizable, and will further dition that equates the present value of the
reduce the global gains. Consider, for exam- gains from migration with the costs and that
ple, the wage differences between Puerto implicitly defines the number of migrants is
Rico and the United States. In 2010, the then given by:
mean annual earnings of a construction
worker in his t­hirties was $23,000 in Puerto ​α​  ∗N ​ ​  (​LN​  ​​  + ​M​​  ∗​​)​​  η​ − ​αS​ ​​  ( ​LS​  ​​  − ​M​​  ∗​​)​​  η​
Rico and $43,000 in the United States.7 The    
(12) ​​ ___________________________ r ​
    
annual income of a young Puerto Rican con-
struction worker, therefore, would increase     =  π (​α​S​​ ​LS​  η​  ​) ,​
by $20,000 annually if he or she were to
migrate, implying a lifetime present value where r is the rate of discount (assumed to
of around $400,000 (if the rate of discount be 5 percent).9 The bottom panel of table 2
is 5 percent). A Puerto Rican nonmover— reports the results of the simulation assum-
and two-thirds of Puerto Ricans have chosen ing that ​π​  =  10. Although there are now
not to move—is leaving almost a half-million obviously fewer movers, there are still a
dollar fortune unclaimed. This fact is consis- lot of them. In the case where there are no
tent with the canonical income maximization spillovers (​λ  =  0​), the number of movers
model of migration only if the costs of migra- falls only from 5.6 to 5.3 billion. However,
tion are at least that high for the many peo- the gains from migration fall substan-
ple who choose not to move. tially because the calculation of the gains
Although this calculation may seem con- must now account for the cost of moving
trived, studies that rely on structural mod- over 5.3  billion people. The annualized
els of labor flows often provide similarly
large estimates of migration costs. Bertoli,
Fernández-Huertas, and Ortega (2013, p. 89) 8 Kennan and Walker’s (2011) study of interstate migra-
calculate that migration costs for the aver- tion for a sample of young workers distinguishes between
age low-educated Ecuadorian immigrant in the costs incurred by the small group of actual movers and
the United States are almost nine times the the cost of a potential move by a randomly chosen person.
While the former can be negative, the cost of “hypothetical
moves to arbitrary locations” is over $300,000 (Kennan and
Walker 2011, p. 232).
7 These wage differences do not adjust for price dif- 9 The equilibrium condition in equation (12) assumes
ferences. The Penn World Table reports that in 2010, the that potential migrants in the South internalize the exter-
PPP-adjusted per capita GDP in the United States was nality they impose on the North’s production technology so
almost twice as large as that in Puerto Rico. that ​​M​​  ∗​​gives the optimal number of immigrants.
Borjas: Immigration and Globalization: A Review Essay 971

global gain—net of migration costs—is such characteristics as quotas, nationality


defined by: restrictions, and the rights granted to immi-
grant groups. Presumably, each receiving
(​∫0 ​ 
​LN​  ​​+​M​​  ∗​
Δ Y′  = ​
(13) ​ ​​ ​α​  ∗N ​ ​​  L​​  η​ dL  country enacts an immigration policy that
maximizes the specific country’s national
+ ​∫0 ​  ​​ ​αS​ ​​ ​L​​  η​ dL − ​Y0​  ​​)​ 
​LS​  ​​−​M​​  ∗​ interest (however defined). Sometimes,
these decisions lead to policies that encour-
age the admission of skilled workers (as in the
− r(π ​α​S​​ ​LS​  η​  ​ ) ​M​​  ∗​.​ Canadian point system), or favor the entry
of relatives of earlier immigrants (as in the
Table 2 shows that these net gains fall from United States), or restrict the rights granted
$40 to $28 trillion when there are no exter- to specific types of migrants after arrival,
nalities, and from a positive gain of $9 ­trillion including the right to vote, the right to move
to a loss of almost a trillion dollars when internally within the country, or the right to
λ  = 0.5. become a naturalized citizen. As an example
Of course, we have no idea what the costs of how specific some of these restrictions can
of migration will actually be if migration be, Ruhs (p. 119) notes Singapore’s require-
restrictions were to be removed and billions ments that a “foreign employee shall not
of people from poor countries were on the go through any form of marriage. . . with a
move. The formation of social networks Singapore Citizen or Permanent Resident,”
among migrants could substantially lower and that a “foreign employee shall not
the costs of migration for the second or third become pregnant or deliver any child in
billionth mover. But congestion costs in the Singapore.”
receiving countries could also increase expo- Ruhs calculates various cross-country cor-
nentially, making it harder to resettle that relations to determine which set of country
marginal migrant. Regardless, the global characteristics are associated with specific
gains from unrestricted migration need to be types of immigration restrictions. Many of
contrasted with the costs of moving billions these correlations are inherently interest-
of people if the exercise is to be taken seri- ing—for example, high-income receiving
ously. After all, it seems that migration costs countries are more welcoming to high-skill
do not need to be all that high to make those than to low-skill immigrants. Due to the
trillion-dollar bills disintegrate even faster. nature of the cross-country empirical exer-
In contrast to Collier’s Exodus, where the cise, however, it is far from clear what under-
focus is on the gains and losses accruing to lying mechanism leads to the enactment of
specific groups in the population, Ruhs’s specific policies.
The Price of Rights attempts to explain Despite this indeterminacy, Ruhs’s
why receiving countries enact policies that detailed documentation of the existence and
restrict the number and types of migrants variation in immigration policies has a cru-
that are allowed to enter. As a result of the cial implication for the models that predict
different focus, Ruhs’s analysis raises con- huge gains from unrestricted migration. To
cerns about the simulated gains from global- put it bluntly, why exactly are the receiving
ization that are conceptually different from countries being so stupid? Why do policy-
those of Collier, but equally important. makers in these countries not buy into the
Ruhs examines an exhaustive data set that models and enact policies that would sub-
summarizes various restrictions on immigra- stantially increase national income? Why do
tion policy for forty-six countries, including countries like Canada and Australia, which
972 Journal of Economic Literature, Vol. LIII (December 2015)

offer both very high wages that would keep loss. In short, the general equilibrium effects
attracting immigrants and vast geographic can easily turn a receiving country’s expected
regions waiting to be filled, keep strolling windfall from unrestricted migration into an
on that mythical sidewalk, keep seeing those economic debacle.11
trillion-dollar bills, and just keep walking My inference from Ruhs’s The Price of
right on by? Rights is that receiving countries endoge-
This point can be quantified in the context nously choose those policies that are most
of the simulation. Suppose that the capital beneficial for them. And those countries’
stock in each region is owned by the capi- revealed preference—the fact that they
talists in that region. It is then straightfor- repeatedly keep ignoring the advice of the
ward to calculate the change in the income social engineers—contains valuable infor-
that accrues to all Northerners after they pay mation. If the trillion-dollar benefits were
out the immigrants’ salaries. This quantity is really there for the taking, would not some
given by: receiving country have already chosen to go
down that path? The fact that these coun-
ΔNet Income of North 
(14) ​ tries instead keep enacting immigration
restrictions hints at the possibility that per-
    = ​∫0 ​  N
​L​  ​​+​M​​  ∗​
​​ ​α​  ∗N ​ ​​  L​​  η​ dL  haps those trillion-dollar bills are not real. I
know that an easy retort to this interpretation
− ​∫0 ​  ​​ ​αN
of the evidence is that the policymakers and
​ ​​ ​L​​  η​ dL − ​w​ N∗ ​ ​​  M​​  ∗​,​
​LN​  ​​
populations of the receiving countries form a
“rabid collection of xenophobes and racists.”
where ​​w​ N∗ ​​​  is the wage paid in the Northern But another interpretation, which may be
labor market after the relaxation of immigra- just as valid, is that perhaps those policymak-
tion restrictions.10 Row 3 of table 2 reports ers and native populations know something
the income change accruing exclusively to that the social engineers ignore: there are
Northerners. If there were no spillovers few gains to be had after accounting for the
and if migration were costless, this income adverse spillovers.
would increase by around $13 trillion as a The striking variation in the types of restric-
result of unrestricted immigration. It would tions that different receiving countries impose
seem, therefore, that receiving countries on specific types of international migrants sug-
have a huge incentive to remove immigra- gests a promising avenue for research. After
tion restrictions; the size of the national all, the variation may provide a great deal of
economic pie increases and the country information about how receiving countries
could, in theory, redistribute some of its perceive and quantify the potential externali-
additional wealth so that all natives in the ties that would arise if the country were hit by
receiving country are better off. The simu- very large supply shocks from specific places.
lation also shows, however, that the increase These different choices may be amenable
in Northern income quickly dissipates and to empirical study by carefully examining
turns negative if there are significant spill- how the adopted policies reflect preexisting
overs. If the parameter λ ​ ​= 0.5, for example, local conditions, i­ncluding the geographic,
the $13 trillion gain turns into a $2 trillion
11 In an unpublished (and prescient) working paper,
10 Note that the wages paid in the North and South in Davis and Weinstein (2002) present a general equilibrium
the postimmigration period will not be the same if migra- framework that implies substantial losses for receiving
tion costs are positive. countries.
Borjas: Immigration and Globalization: A Review Essay 973

e­ conomic, religious, linguistic, and historical further than the different systems of distribu-
linkages among the various countries. tive justice proposed by Nozick and Rawls for
There is, in fact, a related and underap- evidence that assumptions drive conclusions.
preciated inference that can be drawn from Abstracting from these ethical issues, there
Ruhs’s exhaustive accounting of immigration is a clear message for anyone examining the
restrictions. The existing research on the link between immigration and globalization:
economic impact of immigration typically beware of social engineers who promise the
treats the policy parameters that regulate existence of trillion-dollar bills on a mythi-
immigration flows into a receiving country as cal sidewalk at the end of the rainbow; those
exogenously determined, and then exploits promises are often based on flimsy modeling
the policy-induced variation in the size and and inadequate evidence.
composition of these flows to measure the
various economic effects. The observed pol- References
icy, however, is endogenous. This endoge-
Acemoglu, Daron, and James A. Robinson. 2012. Why
neity suggests that the effects observed in a Nations Fail: The Origins of Power, ­ Prosperity,
particular context may provide little insight and Poverty. New York: Random House, Crown
into the economic impact that similar sup- Publishing.
Artuc, Erhan, Shubham Chaudhuri, and John
ply shocks would have in other places and McLaren. 2010. “Trade Shocks and Labor Adjust-
at other times. In fact, it seems likely that ment: A Structural Empirical Approach.” American
a particular policy is chosen because that Economic Review 100 (3): 1008–45.
Benhabib, Jess, and Boyan Jovanovic. 2012. “Optimal
choice leads to the greatest benefits and/or Migration: A World Perspective.” International Eco-
smallest costs in that place and at that time. nomic Review 53 (2): 321–48.
The application of that specific policy in any Bertoli, Simone, Jesús Fernández-Huertas, and Fran-
cesc Ortega. 2013. “Crossing the Border: Self-Selec-
other context would likely lead to a dimi- tion, Earnings, and Individual Migration Decisions.”
nution of the benefits and/or an increase Journal of Development Economics 101: 75–91.
in the costs. A little humility about what Borjas, George J. 2014. Immigration Economics. Cam-
bridge, MA: Harvard University Press.
we actually know would seem to be a prereq- Clemens, Michael A. 2011. “Economics and Emigra-
uisite before anyone proposes a breath­taking tion: Trillion-Dollar Bills on the Sidewalk?” Journal
rearrangement of the world order. of Economic Perspectives 25 (3): 83–106.
Clemens, Michael A., Claudio E. Montenegro, and
Before concluding, let me point out that Lant Pritchett. 2008. “The Place Premium: Wage
I have assiduously avoided the ethical issues Differences for Identical Workers across the U.S.
surrounding the relaxation of immigration Border.” Center for Global Development Working
Paper 148.
restrictions throughout the essay. I am cer- Davis, Donald R., and David E. Weinstein. 2002.
tainly not qualified to comment on the moral- “Technological Superiority and the Losses from
ity of the restrictions that countries enact to Migration.” National Bureau of Economic Research
Working Paper 8971.
restrict population flows across international di Giovanni, Julian, Andrei A. Levchenko, and Fran-
borders. Although these ethical issues are cesc Ortega. 2015. “A Global View of Cross-Border
often alluded to (both Collier and Ruhs offer Migration.” Journal of the European Economic Asso-
ciation 13 (1): 168–202.
lengthy discussions of these issues), the moral Docquier, Frédéric, Joël Machado, and Khalid Sekkat.
argument is often far too ideological and too 2012. “Efficiency Gains from Liberalizing Labor
steeped in an author’s value system to be very Mobility.” Universite catholique de Louvain Institut
de Recherches Economiques et Sociales Discussion
convincing. Moreover, I suspect that the axi- Paper 2012–23.
oms one postulates about the foundations Hamilton, Bob, and John Whalley. 1984. “Efficiency
of a just society are very likely to influence and Distributional Implications of Global Restric-
tions on Labour Mobility: Calculations and Policy
the ending point regarding the morality of Implications.” Journal of Development Economics 14
immigration restrictions—one need look no (1): 61–75.
974 Journal of Economic Literature, Vol. LIII (December 2015)

Kennan, John. 2013. “Open Borders.” Review of Eco- Moses, Jonathon W., and Bjørn Letnes. 2004. “The
nomic Dynamics 16 (2): L1–13. Economic Costs to International Labor Restrictions:
Kennan, John, and James R. Walker. 2011. “The Effect Revisiting the Empirical Discussion.” World Devel-
of Expected Income on Individual Migration Deci- opment 32 (10): 1609–26.
sions.” Econometrica 79 (1): 211–51. Pritchett, Lant. 2010. “The Cliff at the Border.” In
Klein, Paul, and Gustavo J. Ventura. 2007. “TFP Differ- Equity and Growth in a Globalizing World, edited
ences and the Aggregate Effects of Labor Mobility in by Ravi Kanbur and Michael Spence, 263–86.
the Long Run.” B. E. Journal of Macroeconomics 7 (1). ­Washington, DC: World Bank.
Lundborg, Per, and Paul S. Segerstrom. 2002. “The Walmsley, Terrie L., and L. Alan Winters. 2005. “Relax-
Growth and Welfare Effects of International Mass ing the Restrictions on the Temporary Movement of
Migration.” Journal of International Economics 56 Natural Persons: A Simulation Analysis.” Journal of
(1): 177–204. Economic Integration 20 (4): 688–726.

You might also like