Open Trade, Closed Borders
Immigration Policy in the Era of Globalization
Margaret E. Peters∗
University of Wisconsin–Madison
November 9, 2011
Prepared for the 2011 Annual Conference of the International Political Economy Society
Please do not cite without the author’s permission
Abstract
This paper argues that immigration policy cannot be studied in the absence of trade and
capital policy. Trade is a substitute for immigration for both policymakers and firms. Opening
trade forces firms to become more productive, move overseas, or close their doors; all three
choices reduce firms’ incentives to push for open immigration and allows policymakers to move
immigration policy closer to the ideal policy of the rest of their constituents by restricting
immigration. Opening capital reinforces this effect by increasing the probability that firms will
move overseas in response to trade openness instead of remaining at home. The paper tests
this theory on a new dataset on the immigration policy of 19 states from the late 18th century
through the early 21st century. There is a powerful, robust and enduring relationship between
trade and immigration in the dataset. Trade openness leads to greater immigration restrictions
and this relationship is made stronger when we examine capital openness as well.
∗
Department of Political Science, University of Wisconsin–Madison, 110 North Hall, 1050 Bascom Mall, Madison,
WI 53706; [email protected]
Introduction
Explaining immigration policy has never been easy for political economists. Immigration policies
of states seem to vary in inexplicable ways: at times states are very open to immigrants while at
other times they are very closed. Moreover, policy is episodic. States seem to ignore the flow of
people for a generation and then, with seemingly no change in economic circumstance, close their
border. Moreover, policy is no more predictable when we look across nations as opposed to across
time. Democracies’ policies have varied as much as those of nations where citizens appear to have
less voice. While the conventional wisdom is that social forces are the key variable in explaining
the ebb and flow of border measures, these forces have only sometimes been met with action on
immigration policy. At times, including the late 1910s and early 1920s in the United States, cries
for immigration reform have been met with policy change. Recently, these same cries to “secure
our borders” have gone largely unheard by the US federal government. What, then, explains these
shifts in policy?
Unfortunately, our usual political economic explanation for this puzzle – the distributional costs
and benefits from openness – are inadequate. According to the simple two-factor, two-good StolperSamuelson model, openness through the movement of people, goods or capital affects prices and
wages in the same way, benefiting the abundant factor while hurting the scarce factor.1 Economic
theory, however, tells us little about which policy or policies the state should choose when it decides to pursue openness. Openness in any one of these policy areas, theoretically, should produce
the same effects as being open through any other policy or through any combination of policies.
Nonetheless, it appears that states rarely choose to be open on all three-policy areas. In the 19th
century, most labor-scarce states – those states that were most likely to face immigration pressures
– were open to immigration and capital but not to trade. In contrast, most of these same states have
been much more open to trade since the 1950s and to capital since the 1980s than to immigration.
Economic theory, however, does not seem to explain these choices between different policies.
Political analysts have been no better than economists at providing a clear explanation for the
choices that states make regarding immigration policy. When observing this puzzle, many scholars
(e.g Freeman 1995, Zolberg 1989, 2006) have turned to culture to explain the policy choices of
states, often invoking some type of prejudice as an explanation for why nations will close their
economies to immigrants. Yet, these theorists assume that prejudice is constant in most nations.
Since immigration policy varies, social theories of this sort may be descriptively true but not helpful
1
The Stolper-Samuelson model of trade builds on the Ricardian model of comparative advantage by allowing
countries to have different endowments. In the simple model, there are two countries, one relatively abundant in
labor and the other relatively abundant in capital. When these two countries trade with each other, they export the
good that uses the abundant factor intensively and import the good that uses scarce factor intensively. Owners of
the scarce factor suffers a real decrease in the returns to their factor as domestic production moves out of the good
that uses the scarce factor intensively and into the good that uses the abundant factor intensively. In contrast, the
abundant factor will see a real increase in the returns to their factor. Similarly, opening borders to the movement of
labor or capital will lead, respectively, labor to move from the labor-abundant country to the labor-scarce country
and capital to move from the capital-abundant country to the capital-scarce country. Movement of factors, therefore,
also increases the real returns to the abundant factor while decreasing the real returns to the scarce factor.
2
in predicting variation in policy.
What is missing from both the economic and political theories is a discussion of the structural constraints that policymakers face when choosing an immigration policy once they open their
economies to the free movement of goods and capital. I argue that the dynamics of immigration
policy formation differ depending on the openness of the economy to the movement of goods and
capital. When trade and capital are closed, immigration policy follows interest group politics. Firms
serve as the bulwark against anti-immigration groups, leading to a more open immigration policy.
When capital and goods are free to move, the dynamics become more complicated because openness
decreases firms’ need for immigrant labor while also increasing their ability to influence policy.
This paper argues that, in the long run, policymakers cannot simultaneously open their markets
to goods, capital and people because these flows are interdependent. Opening trade and capital
allows some firms to take their capital to labor rather than bring labor to capital, taking their
tax revenue and jobs with them and leads other firms to close by subjecting them to competitive
pressures, taking their tax revenues and jobs as well. Policymakers fear the consequences of losing
tax revenue and jobs as it could affect their chances of staying in office. They seek ways to keep these
firms at home. But the incentives that policymakers give firms to remain at home are politically
unpopular. The policymaker must balance her desire for tax revenues and jobs with her constituents’
preferences if she is to stay in office. This creates a dynamic where either firms are made less mobile
and more competitive, through capital and/or trade restrictions, while immigration is opened or
where firms are allowed to exit the state or close their doors and immigration policy is restricted.
By allowing firms to move overseas or close, policymakers remove the main bulwark — firms —
against anti-immigration groups and, therefore, face an incentive to close immigration policy. Thus,
in the long run, capital, trade and immigration cannot all be open.
One of the major obstacles to research on immigration has been the lack of longitudinal crossnaitonal data. While there have been a few studies that have coded immigration policy over time
(Albarracín 2003, Timmer and Williamson 1998), these studies have been focused on a limited
number of countries and years. In response to this lacuna, this paper assembles a new dataset of de
jure immigration policy across 19 countries over the last 200 years. After introducing the data, I
argue that the data suggest that states should be categorized into four categories — settler states,
European liberal democracies, export-oriented industrializers and autocracies — that vary in what
type of policies they use to regulate their borders and their openness to immigration. The data show
that the 19th century was a period of general openness; although, countries increasingly sought to
control immigration as the years went on. The Interwar period was a time of general closure, but
there were some side and back doors left open for some, mostly Northern European, immigrants.
After World War II, some states opened their doors to immigrants again, only to begin closing them
again at the end of of the 20th century.
What accounts for this increase in restrictions? I argue that trade policy can explain the pattern
of increased immigration policy restriction across all states. I test the relationship between trade
policy and immigration policy and find a robust and enduring negative relationship between these
3
two policies. Trade policy cannot be separated from immigration policy because trade policy is a
substitute for immigration policy. When trade is closed immigration is open and when trade is open
immigration is closed. The relationship is made stronger with the inclusion of capital openness. This
finding is supported if we examine policy in the 19th century or today, if we examine each state’s
trade policy and if we include a variety of control variables.
While the dataset was created to test the relationship between trade, capital and immigration
policy, it also provides the opportunity to test alternative hypotheses offered by scholars. I find that
most of the relationships hypothesized by these theories are not supported when tested using this
data. One theory, the fiscal argument, does help to explain the variation in immigration policy in
the post-World War II era in OECD states.
The paper continues as follows. I begin by further explicating the constraints that openness to
trade and capital place on the policymaker’s ability to use immigration policy. Next, I discuss the
alternative explanations in the literature. Third, I introduce the new dataset of the immigration
policy of 19 states from 1783-2010. Fourth, I test the relationship between trade, capital and
immigration policy and examine the evidence for the alternative explanations. Finally, I conclude.
The Constraints that Policymakers Face
In this section, I use a model of firm lobbying over immigration policy similar to the model developed
for trade policy by Grossman and Helpman (1994). While there are several groups in a polity that
could affect immigration policy — firms, labor, people who dislike immigrants for cultural reasons
(nativists), taxpayers and immigrants themselves — I examine the preferences of firms because
they are the main source of pro-immigrant sentiment in the polity. Native labor dislikes immigrants
because immigrants compete with them for jobs and push wages down (which is exactly why firms
like immigrants).2 Nativists dislike immigrants for the obvious reason: immigrants are different
from natives. Immigrants typically are for an open immigration policy as it allows them to be more
secure in their ability to stay in the state and/or bring in friends and family; however, immigrants
are not a powerful group. Until they gain the rights of citizens (and, sometimes, not even then),
immigrants can be expelled from the country, which limits the political power that immigrants
have. Moreover, as they are not citizens and therefore, lack the ability to punish the policymaker,
policymakers are less likely to listen to the demands of immigrants. Thus, firms are left as the only
powerful group that could be pro-immigration. If the firm is not for immigration then it is unclear
who would be for immigration. Given firms important role in immigration policy, I examine how
their preferences change due to changes in productivity, trade policy, and capital policy and how
policymakers respond to these changes.
2
Unions today have taken a more pro-immigration stance than they had in the past. In part this reflects the
realization that in many wealthy states, a segment of low-skill labor is comprised of illegal labor. Unions would
rather have these illegal workers brought out of the shadows so that all employers would have to respect minimum
wages and safety standards. As such, unions tend to be for a more open immigration policy with stricter enforcement.
Second, unions in many countries have been gaining immigrants as union members and therefore tend to represent
their interests as immigrants and not just their interests as workers.
4
I begin by assuming that the policymaker can only choose an immigration policy and a tax rate
for each firm; later the policymaker will be able to choose a trade and capital policy as well. In the
model, firms can lobby the policymaker for a preferred policy by paying a contribution; in general,
however, we could think about the contribution as the political capital that the firm is willing to
pay for the policy. In essence, by paying a contribution to the policymaker, the firm is paying an
extra tax, on top of the tax rate that the policymaker sets. To simplify the model, both the taxes
and the contribution are combined into one tax rate.
The game is as follows: firms offer the policymaker contributions schedules that lay out the tax
rate that the firm will pay for each potential immigration policy the policymaker could make. The
immigration policy is modeled as the total number of immigrants allowed into the state in a given
year, ρ. The policymaker observes the contribution schedules and chooses a level of immigration that
maximizes her utility function, G(·). Firms then maximize their profits, πi , given the immigration
policy. Firms may also be unable to make a profit given a tax rate and immigration policy and
choose to go out of business.
Following Grossman and Helpman, I examine Nash equilibrium with truthful contribution schedules, defined as a “contribution schedule that everywhere reflects the true preferences” of the firm
(1994, 840). A set of contribution schedules and a policy choice is a sub-game perfect Nash equilibrium if the firm does not contribute more for a policy than it would make under that policy; the
policymaker chooses the policy that gives her the greatest payoff given the contribution schedules,
and that the policy chosen also maximizes the firms’ payoffs given all the contribution schedules.3
The first criteria ensure that the firm can actually give the contribution it is offering for a given
policy and the second two criteria ensure that neither the policymaker or any of the firms want to
deviate from their choices in equilibrium.
The policymaker’s goal is to stay in office; to do so, she maximizes her utility function,
G(w, τ, ρ) = α1 w(ρ) + α2
X
τi (ρ) + α3 H(ρ)
(1)
i∈F
where w(ρ) is the wage for low-skill labor;
P
i∈F
τi is the total tax revenue, including contributions
that the policymaker receives, from each firm i in the set of all firms F and H(ρ) is a function that
measures the cultural benefits (costs) of immigrants. The policymaker places a weight αk∈[1,3] on
each element and the α’s must sum to 1. The first term, the wage level, measures how concerned
the policymaker must be about jobs and wages of low-skill workers. The second term measures
how much the policymaker cares about tax revenue. The policymaker could use the tax revenue
as a transfer to give her citizens, in a sense using it to buy votes, or she could keep the revenue
for herself, as we might expect an autocrat to do.4 Finally, the last term is used to capture all the
3
More formally, a policy choice is SPNE if (a) the contribution schedule for every firm is feasible, i.e. the firm
cannot contribute more than it would make, (b) the policymaker sets a policy to maximize her welfare given the
contribution schedules, (c) the policy chosen maximizes the joint welfare of the firm and the policymaker, given all
the other contributions and (d) there exists a policy that would elicit a contribution of zero from firm i, which the
policymaker likes as much as the equilibrium policy (Grossman and Helpman 1994, 839, 845).
4
I do not model the total welfare, as in Grossman and Helpman (1994), because I assume that constituents only
5
potential non-labor market costs and benefits of immigrants, including nativism (cultural costs),
cosmopolitanism (cultural benefits), the benefits to immigrants themselves of the policy and the
(supposed) fiscal costs of immigrants. If immigrants hurt citizens in the non-labor market, H is
negative; if they benefit citizens, H is positive.
The first order condition for the policymaker is
X ∂τi
dG
∂w
∂H
= α1
+ α2
+ α3
=0
dρ
∂ρ
∂ρ
∂ρ
i∈F
α2
X ∂τi
i∈F
∂ρ
= −α1
∂w
∂H
− α3
∂ρ
∂ρ
(2)
(3)
From equation 3, we see that, in the basic framework, the policymaker will need increased tax
revenue (contributions or political capital) in order to open up immigration further.5 Increased
immigration lowers the wage for low-skill workers, making the first term on the right hand side
of equation 3 positive and increases the non-labor market costs of migration, assuming that H is
decreasing in ρ, making the second term positive as well. As such, the total tax revenue she receives
through political capital, contributions or taxation will have to increase for immigration to increase.
Similarly, if contributions or political capital (tax revenue) decreases, the policymaker will restrict
immigration.
Firm Behavior and Immigration Policy under Closed Trade and Capital
I begin examining firm behavior when trade and capital are closed (henceforth, under autarky). I
assume that capital is immobile across industries but that labor is mobile across industries. The
profit function for each firm, i, in the set of all firms, F , is
πi = pi qi − ci (w(ρ), qi ) − τi (ρ)
(4)
where pi is the price of the good that firm i produces; qi is the quantity produced; ci (·) is the cost
function, which is assumed to be convex; w is the wage for low-skill workers and τi (·) is the tax
rate including the contribution schedule that the firm pays for a given immigration policy, ρ. The
costs of producing each good i is decreasing in the number of immigrants allowed into the country
because increased immigration lowers the wage for low skill workers.6 In contrast as we saw above,
care about the wage effects of immigration or similarly, that they do not realize that immigration has an effect on the
overall economic welfare of the state. I do not include prices because the overall effect of immigration on prices and
consumers utility is unclear. On the one hand, increased immigration will lead the costs, and therefore the prices,
of goods made with immigrant labor to drop; on the other hand, immigrants’ demand for goods will mean that the
price of some goods will rise. Without knowing the consumption bundle of consumers, it is impossible to know how
increased immigration will affect consumers’ and, therefore, the policymaker’s utility.
5
I assume that there are diminishing effects of immigration on both the wage and the non-labor market costs of
immigration, making both functions concave in ρ.
6
I assume that there is a numeraire good produced with only high-skill labor which sets the wage for high-skill
labor, without loss of generality, at 1. This assumption allows immigration to affect only the wage of low-skill labor,
which I assume to be the case, as the vast majority of immigrants over time have been low-skill workers. The wage
6
the taxes that the firm pays are increasing in the immigration policy to compensate the policymaker
for her loss in utility given an increase in immigration.7 I assume that no firm has market power;
instead, they are all price takers. Because the firms are price takers, in equilibrium, their marginal
cost will equal their marginal revenue, the price of good i, or their costs will exceed the price of
good i, which will force the firm to close. Assuming that the firm does not close,
∂c ∂w
∂τ
=−
∂w ∂ρ
∂ρ
where
∂c
∂w
is the change in costs due to a change in wages,
in immigration and
∂τ
∂ρ
(5)
∂w
∂ρ
is the change in wages due to a change
is the change in the tax that the firm pays for a given immigration policy.
The labor costs that the firm pays will decrease with an increase in immigration while the taxes
it pays will increase. Under autarky firms, therefore, will pay increased taxation for an additional
immigration until the additional benefit they receive from the immigrant in reduced costs equals
the marginal tax that they will have to pay. The decrease in costs due to increased immigration
will be lower for firms that are more productive or capital/high-skill intensive than for firms that
are less productive or low-skill intensive because these firms use less low-skill labor or
∂chigh−productivity
∂clow−productivity
<
∂w
∂w
(6)
Because firms are only willing to pay in taxes up to the marginal benefit they receive from another
immigrant, firms that need less labor — more productive or capital/high-skill intensive firms —
will be willing to pay less for each additional immigrant than firms that need more labor — the less
productive or low-skill intensive firms. As the proportion of more productive or capital/high-skill
intensive firms in the economy increases, the policymaker will receive less in taxation (contributions
or political capital) for the same immigration policy. With less tax revenue, the policymaker will
restrict immigration.
Further, if we allow for imperfect market competition (oligopoly), more productive and capital/highskill intensive firms in an industry have an incentive to spend political capital to restrict immigration. Firms now have market power and can compete for market share through price (or quantity)
competition. Increasing immigration, therefore, conveys an advantage to firms that use more labor
because immigration lowers the costs of these firms pay to a greater extent than for more productive
or capital/high-skill intensive firms. These less productive and low-skill intensive firms can lower
their price and collect a greater share of the market, reducing the profitability of more productive
and capital/high-skill intensive firms. More productive and high-skill intensive firms may want to
keep their competitive edge by limiting the availability of cheap labor for their less productive and
more low-skill intensive competition. Therefore, when the proportion of firms that are more profor low-skill workers will be less than 1; otherwise high-skill workers would take low-skill jobs.
7
For ease of solving the model, I do not allow the immigration policy to affect the price of the good by increasing
the demand for the good produced by firm i. If I did include this effect, firms would simply be willing to pay more
for each level of immigration as they receive a higher benefit from each immigrant.
7
ductive or capital/high-skill intensive in the economy increases, immigration policy should become
more restrictive under either perfect or imperfect competition. This leads to my first hypothesis.
Hypothesis 1 As labor productivity and/or capital/high-skill intensity of production increases,
immigration policy will become more restrictive.
The overall openness of immigration is also likely to be affected by the non-labor market costs
of immigrants (H(ρ)). If the non-labor market costs of immigration increase for all levels of immigration (shifting
∂H
∂ρ
downwards), the policymaker will need to be compensated with more tax
revenue for every possible immigration policy than she was before the increase in non-labor market
costs. Firms, therefore, will want fewer immigrants because immigration is more costly, leading to
a less open immigration policy. Similarly, if the non-labor market benefits of immigration increase
for every possible policy, immigration policy should be more open because firms have to pay less
for each immigrant. An increase in nativism or an increase in the welfare benefits that accrue to
migrants would shift the non-labor market benefits downwards. An increase in immigrants or their
political power, say for instance because they gained citizenship, would lead to an upward shift in
the non-labor market benefits of immigration. This leads to my next three hypotheses.
Hypothesis 2 As nativism increases, immigration policy will become more restrictive.
Hypothesis 3 As the number of immigrants in society increases and/or immigrant rights groups
gain more power, immigration policy will become more open.
Hypothesis 4 As the fiscal costs of immigrants increase (or are thought to increase), immigration
policy will become more restrictive.
Finally, the immigration policy will be affected by the weight that the policymaker places on each
item in her utility function. If the policymaker places more weight on the wage of low-skill workers
(α1 ), immigration policy should become more restricted. From equation 3, we see that an increase
in α1 increases the taxes that firms have to pay. By equation 5, we know that increasing the tax per
immigrant lower the total number of immigrants that the firm wants. By placing more weight on
wages, the policymaker makes increasing immigration more costly for the firm because the tax that
the firm has to pay for each immigrant increases. One way in which the weight placed on the wage
of labor would increase would be if the state democratized or if franchise expanded; both changes
in the political system make low-skill workers more important for reelection. Another way in which
the weight placed on wages could increase is if organized labor became more politically powerful.8
As organized labor gains power, they become more important for reelection, leading to a more
restrictive immigration policy. Finally, the weights could also measure the policymakers ideological
8
Here I assume that the power of labor organizations is in delivering votes rather than contributions and that
labor organizations dislike immigration due to the wage effects. If organized labor can provide contributions for
immigration restrictions, it is clear that immigration openness would be more expensive for firms and there would be
a more restrictive policy in equilibrium. If organized labor was pro-immigration, then it would be cheaper for firms
to “buy” additional immigrants and immigration policy should be more open.
8
position. Left leaning politicians are likely to place more weight on the wage of labor than right
leaning politicians. Conversely, right leaning politicians may place more weight on the non-labor
market costs of immigration because they want to appeal to typically conservative nativists, they
want to appeal to fiscal conservatives or both.9 This leads to my next four hypotheses.
Hypothesis 5 As the size of the franchise increases or as the state democratizes, immigration
policy will become more restrictive.
Hypothesis 6 As the power of labor increases, immigration policy will become more restrictive.
Hypothesis 7 When left leaning parties are in power, immigration should be more restrictive.
Hypothesis 8 When right leaning parties are in power, immigration should be more restrictive.
Immigration Policy under Open Trade and Closed Capital
Under autarky, firm preferences — and their willingness to spend political capital on immigration
— were driven by their productivity and their capital/high-skill intensity. Opening trade does not
affect these preferences. Instead it has two countervailing effects: it increases the influence that
firms have over immigration but also increases the proportion of firms that are more productive
and capital/high-skill intensive. In the short run which of these effects will prevail will depend on
the choices made by the policymaker; in the long run the changing composition of the economy will
predominate, leading to a more restrictive immigration policy.
I assume that trade openness is exogenous to both the firm and the policymaker, which I relax
later. Trade openness could affect firms in two ways depending on whether the trade is opened to
countries with different endowments (Ricardo-Viner model) or the same endowments (Melitz (2003)
model). I begin by examining what happens to firms that produce tradable goods when trade opens
under the Ricardo-Viner model. Firms that produce non-tradable goods are not, by definition,
affected by trade. These firms should continue to hold the same preferences as they held under
autarky.
Under the Ricardo-Viner model, trade openness is modeled as an increase in prices for the
tradable goods produced by capital/high-skill intensive firms and a decrease in prices for the tradable
goods produced by low-skill intensive firms. As prices for low-skill intensive goods decrease, these
firms have to decrease costs or close. There are three ways in which these firms could decrease costs:
increase their use of capital and/or high-skill labor, decrease the wage they pay or decrease their
tax bill. As we saw above, if firms choose to increase their use of capital and/or high-skill labor,
they will be less supportive of immigration and immigration will be more restrictive.
Assume there is only one sector i that faces lower profits under trade openness and all other
firms receive the same prices. If firms in sector i exit the market, they pay no taxes and provide no
9
We would expect that the policymaker would place more weight on nativists if they were more organized; for
example, if there was an anti-immigrant party which was a challenger to the policymaker or could be brought into her
coalition. This, of course, assumes that the anti-immigration party arose exogenously; otherwise, we would expect
that the policymaker would choose a policy that forestalled the emergence of the anti-immigrant party.
9
jobs, which decreases the wage for low-skill labor. The policymaker now must decide whether it is
better for her to give the firm incentives to stay in business or allow the firm to exit the market.
Without loss of generality, I assume that policymaker does not offer any tax breaks in addition to
offering increased immigration. The policymaker must increase immigration from the level that it
was under autarky to keep the firms in sector i in business without getting additional tax revenue.
The policymaker will offer increased immigration as an incentive if
α1 w(ρin ) + α2 τi (ρin ) − α3 H(ρin ) ≥ α1 w(ρout ) + α2 τi (ρout ) − α3 H(ρout )10
(7)
where ρin is the immigration policy needed to keep firm i in the market and ρout is the immigration
policy the policymaker would choose if firms in sector i exit the market. If firms in sector i exit the
market, the policymaker receives no tax revenue for those firms and equation 7 simplifies to
α1 w(ρin ) + α2 τi (ρin ) − α3 H(ρin ) ≥ α1 w(ρout ) − α3 H(ρout )
(8)
For the policymaker to be willing to increase immigration to keep firms in sector i in business,
wages and tax revenue must compensate her for increased immigration. If sector i is large and the
price decrease for good i with open trade is small, then the policymaker will be more likely to open
immigration to keep sector i in business. As a large sector if firms in sector i close, many jobs
will be lost and wages will fall sharply. If the change in prices due to open trade is small, then
only a small increase in immigration — and a small decrease in wages —- will be needed to keep
firms in sector i open. Conversely if sector i is small and the price change large, the loss of sector
i will result in only a slightly lower wage whereas the increased in immigration need to keep sector
i would be large, leading to a large decrease in wages and a large increase in the non-labor market
costs. Further, if firms in sector i are allowed to close, the policymaker could reduce immigration.
Wages would drop as the firms in sector i lay off their workers and policymaker could scale back
immigration without making these other firms worse off. Therefore, the non-labor market costs are
always higher if the policymaker chooses to subsidize sector i with increased immigration.
Although the policymaker may be willing to subsidize firms with immigration in the short run,
the combination of sector size, price change, and the non-labor market costs of immigration will lead
the policymaker, at some point, to realize that it is not in her interest to continue to subsidize the
sector i through immigration. Sectors that are small and face a large price decrease with openness
will be the first ones allowed to close. Further, the costs of keeping sectors in business are increasing
in openness because greater openness pushes prices down even more. As the state becomes more
open to trade, more sectors will follow suit, as it becomes increasingly expensive for the policymaker
to keep them open. Over the long run, then, as trade becomes more open, more sectors will be
allowed to close; the proportion of firms that are more productive and capital/high-skill intensive will
increase and immigration policy will become more restrictive. Figure 1 illustrates how immigration
10
To simplify the model I assume that the state is small; therefore, if it stops producing good i world prices will
not be affected.
10
Open
Closed
Immigration Policy
Figure 1: Immigration policy in response to trade openness if the policymaker uses immigration to
keep firms at home
Closed
Open
Trade Policy
policy might change in response to trade policy if the policymaker decided to use immigration to
subsidize sectors that were negatively affected by trade.
The policymaker could also decrease taxation instead of opening immigration to keep firms in
sector i in business. In this case immigration policy would remain the same as it was under autarky
but tax revenue would fall. As in the scenario above, if the policymaker allowed these firms to close,
she could restrict immigration without affecting other firms. If the sector was large and the price
change small, the policymaker could reduce taxes for the firms in sector i and be better off than
she would be if she allowed the firm to close. If, on the other hand, the sector was small and/or the
price change was large, the policymaker would be better off allowing the firm to close. As long as
trade does not open too much, the policymaker can subsidize firms that are hurt by trade with tax
breaks and keep immigration at the same level. Because the costs of subsidizing firms is increasing
in openness, at a certain point trade will open enough that the policymaker prefers to allow firms
to close, rather than to continue to subsidize them, and restrict immigration. As trade openness
continues, more firms will be allowed to close; the proportion of firms that are more productive and
capital/high-skill intensive will increase and immigration policy will become relatively restricted.
Figure 2 illustrates the effect of using tax policy to keep firms at home.
Opening trade, therefore, has two main effects which lead to changes in the immigration policy.
First, opening trade increases the influence that firms have over policy. Firms that are low-skill
intensive are more likely to close their doors because open trade leads the price of their goods to decrease, reducing profits. Because firm exit leads to losses of both jobs and tax revenue, policymakers
are more likely to place a greater weight on these firms’ preferences. In the short run, this could
lead to a more open immigration policy or a continuation of the current policy as policymakers use
open immigration or tax breaks to keep these vulnerable firms producing. In the long run, these
11
Open
Closed
Immigration Policy
Figure 2: Immigration policy in response to firm mobility/mortality if the policymaker uses tax
policy to keep firms at home
Closed
Open
Trade Policy
policies are unsustainable, leading to a more restrictive immigration policy. The increased immigration restrictions force low-skill intensive firms to increase their productivity or capital/high-skill
intensity of production or close their doors. Either choice will then lead to a second effect, namely
a change in the overall distribution of firms in the economy. The firms that continue to produce
will be more productive and capital/high-skill labor intensive.
These results do not depend on the assumptions of the Ricardo-Viner model; most importantly,
they do not depend on comparative advantage and interindustry trade. Instead, the results are
similar if I examine the effects of intraindustry trade under the Melitz (2003) model, which is an
extension of the Krugman (1980) model.11 Examining the effects of intraindustry trade is important
because a large proportion of trade is intraindustry rather than interindustry.
Under this model, countries with similar endowments open trade. However, exporting is not
costless; firms face some fixed costs that do not vary with the amount exported. Opening trade
leads the most productive firms to export and the least productive firms to close their doors. Due
to the costs of exporting, only the most productive firms can afford to export; these firms receive
higher returns than they did under autarky. As the highly productive firms export, they increase
the amount of labor they need and bid up the real wages, forcing the least productive firms to exit
(Melitz 2003, 1716). The policymaker under openness to intraindustry trade faces the same trade-off
as she did under interindustry trade: subsidize production of the least productive firms or allow them
11
The Krugman (1980) or Melitz (2003) model without costs to trade can also be thought to model increased
immigration under autarky. Increased immigration increases the country size; however increased country size has
no effect on firm level outcomes — the same number of firms produce the same output and earn the same profits
(Melitz 2003, 1706). Under this model, no firm would ever have an incentive to lobby for open immigration and
under autarky we would always see closed immigration. Empirically, this is not the case and therefore it is unlikely
that this model can explain immigration policy under autarky.
12
to close their doors.12 If the firm cannot survive with these lower taxes or increased immigration,
the firm must either also increase productivity or exit. Once the firm increases productivity or
exits, the policymaker, again, can decrease immigration. Therefore, we obtain the same results
under either model.13
Further, under the Melitz model if we relax the assumption of perfect competition, we see that,
again, the more productive firms may want to spend political capital to reduce immigration or limit
tax incentives. These firms can capture more market share if their less productive competition
closes. As such, they may want to hasten the demise of the less productive firms by ensuring that
less productive firms are not subsidized with open immigration or tax incentives.
I now relax the assumption that trade is exogenous to the policymaker. The policymaker now can
affect prices by lowering trade barriers. I assume that trade is opened reciprocally so that exportoriented firms have a reason to push for openness. Under the Ricardo-Viner model, prices on goods
that are made intensively with labor will drop and goods made intensively with capital/high-skill
labor will increase. The overall effect on consumers depends on their consumption bundle and
therefore is unclear. For the purposes of this model, I ignore the consumption effects. Under the
Melitz (2003) model, prices stay the same but more productive firms can earn higher returns under
trade than under autarky.14 Additionally, under both models total welfare increases. I assume that,
like immigration, trade policy is largely driven by firms in these states as they are capital abundant
and therefore capital should be the winner from open trade; the importance of labor can come into
the model through the α1 term.
Firms that “win” from trade openness are likely to pay contributions (increased taxation) for
trade openness; while firms that “lose” from trade openness are likely to pay contributions (increased
taxation) for closure. From Grossman and Helpman (1994, 846), each lobby will have to pay
according to the political strength of the rival. Therefore, if the export-oriented firms are politically
more powerful, the import-competing firms would have to pay more for trade protection than export12
To obtain the same results under tax incentives, we must make one more assumption; namely, that the most
productive firms do not want to “buy” more immigration as well. If the more productive firms wanted to buy
immigration as well, due to the increase in wages, I would expect that immigration would increase with trade openness.
However, more productive firms need labor that can work at a higher level of productivity. Most immigrants are
low-skill workers and are unlikely to possess the skills needed to preform at the desired level of productivity; therefore,
immigrants could not provide the labor that more highly productive firms need. Highly productive firms, therefore,
are indifferent to the choice of immigration policy.
13
These results, however, implicitly assume that production by either the more capital/high-skill intensive firms
under the Ricardo-Viner model or the more productive firms under the Melitz (2003) model does not increase so
much as to increase the economy-wide demand for low-skill labor. Since the firms expanding under trade in the
Ricardo-Viner model are more productive and/or capital/high-skill intensive and in the Melitz (2003) model are
more productive, the expanding firms are likely to want much more high-skill labor and only some low-skill labor.
The firms that exit the economy, on the other hand, are low-skill labor intensive or less productive. Therefore, when
they exit, they release much native labor. It is likely that the released native labor will more than meet the demand
for low-skill labor from the capital/high-skill intensive or more productive firms. Empirically, rising wage inequality
due to increases in productivity and trade seem to bear this out (see for example Feenstra and Hanson (1996) or
Kremer (2006)). Rising wage inequality is occurring precisely because low-skill jobs have not been replaced at a high
enough rate to keep low-skill wages at the same level. Therefore, I believe that this assumption is justifiable.
14
Prices stay the same because differentiated goods are traded in this model. In the Ricardo-Viner model, homogeneous goods are traded, leading to a price change.
13
oriented firms would for trade openness. Nonetheless, the import-competing firms have the threat of
closing if trade opens. This threat allows them to pay less as wages and tax revenue would decrease
if the firms closed. To obtain trade openness, the contributions from export-oriented firms must
outweigh the costs of subsidizing the losing firms or allowing them to exit the market. As noted
above, the policymaker can make the firms that benefit from trade equally well off if she subsidizes
the firms that face increased competitive pressures or allows these firms to exit. In equilibrium,
both trade and immigration can be open in the short term, but immigration will close in the long
term. This leads to my next hypothesis.
Hypothesis 9 Immigration policy may become more open, more restricted, or stay the same at
low to moderate levels of trade openness but will become more restrictive at high levels of
trade openness.
Immigration Policy under Trade and Capital Openness
I begin by assuming that capital is opened exogenously as well, which I relax later. If trade and
capital are open, some firms that are hurt by trade openness have one more option to lower their
costs — they can move production overseas where costs are lower. Moving production overseas
can occur in two ways: first, the firm could move their “factory” overseas through foreign direct
investment, or second, they could outsource part of their production process — the part that is
no longer productive to do at home — to a firm overseas. Firms cannot move production without
capital openness because they are legally unable to move their capital to either move their factory or
pay their outsourcer. Firms are unlikely to move production without trade openness because trade
protection would apply to the goods produced overseas, making them uncompetitive in the home
market. Not all firms will be able to move production; some firms, especially those that exploit
natural resources like agricultural or mining firms, simply will find moving their production process
prohibitively expensive.
To the policymaker, moving production overseas is the same as having the firm go out of business;
both the jobs and the tax revenue associated with that firm are lost.15 For firms that cannot make
a profit at home, moving overseas is likely to be a better option than closing. When firms move
overseas, rather than close, the owners continue to get their share of the profits into the future
rather than simply getting what little capital they can from selling the firm’s assets. As trade
and capital barriers decrease, moving overseas becomes a better option because the firm’s profits
from producing overseas increase. Although it is outside the scope of this paper to test, we should
expect that trade openness should be accompanied by capital openness as firms that are threatened
by trade openness should want the exit option. With the option to move production overseas,
the firm will be unwilling to pay as much in taxes for any given immigration policy because they
have an attractive exit option. Therefore, the policymaker must increase immigration more or she
must cut taxes further to keep the firm at home. Either of these options give the policymaker a
15
I assume that states cannot tax overseas production. This is not always the case, but it is clear that states tax
overseas earning at a lower rate, which would reduce their tax revenues.
14
lower utility than the similar option in the case where only trade is open. Moreover, once firms
have moved overseas, they may find it in their interest to spend political capital (contributions)
to keep immigration restricted at home. If the firms move production to a state that sends many
migrants to the firm’s home state, closing (opening) immigration will lead to a lower (higher) wage
in the state where production is taking place. Firms producing in that state, therefore, have an
incentive to keep immigration restricted at home. This means that the policymaker will find that it
is better to allow the firm to leave sooner (or allow more firms to leave) than to try to keep them. I
expect, therefore, that in the short run immigration cannot open as much as it did under only trade
openness and that the short run should last for a shorter period of time as the policymaker chooses
to allow firms to move overseas sooner than she would allow them to close. This leads to the final
hypothesis.
Hypothesis 10 Immigration policy will not open as far, if at all, under trade and capital openness
as it did under trade openness alone. Further, immigration policy will be restricted at a lower
level of trade openness if capital is open as well than if only trade were opened.
Finally, we assume that capital policy is also endogenous. The policymaker now can choose a
trade, capital, immigration and tax policy. Firms choose contribution schedules over trade, capital
and immigration policy and after the policy bundle is chosen, the firm chooses to produce at
home with the same production function, increase productivity/capital/high-skill intensity, produce
overseas or close. Closing is the worst option and is only chosen if the firm will make negative profits
under the other three options. The policymaker knows that if she opens trade, she will be lobbied
to open capital as well. If trade and capital are open, the policymaker is worse off because she
receives less compensation for a given immigration policy or must cut taxes of the vulnerable firms
more. Therefore, to open trade, she must be compensated by the “winners" from open trade more
than she was when capital remained closed. Similarly, she must also be compensated by those who
want open capital. In equilibrium, therefore, we can have open trade and capital chosen by the
policymaker, but this policy bundle will only be achieved if those who want open trade and capital
are willing to pay for it. Under this scenario, all of the above observable implications still hold even
though trade and capital were chosen endogenously instead of exogenously.
I argue that firm preferences over immigration vary along two dimensions: productivity/lowskill labor intensity and their ability to move production locations. Firms that use production
technology that is low-skill labor intensive will need more labor than firms that use capital or highskill labor intensive production technologies. Similarly firms, regardless of production technologies,
will need less labor if the firm’s production process uses labor more productively. We expect,
therefore, that low-skill labor intensive and less productive firms will be pro-immigration whereas
capital or high-skill labor intensive and more productive firms may prefer to spend their political
capital on other issues or, in the case of oligopolistic competition, may prefer spend their political
capital on restricting immigration. The second dimension captures the ability of the firm to move
15
production overseas. Firms that can are relatively immobile will be more likely to need labor at
home, and therefore support immigration more, than firms that can move production. Once firms
move overseas, they will be indifferent to or potentially against immigration at home. Table 1
summarizes these two dimensions.
Table 1: Firm preferences for immigration along two dimensions
Mobility
Immobile/
Non-tradable
Mobile
Productivity & Skill/ Capital Intensity
Low
High
Pro-immigration
Indifferent/ Anti-immigration
Indifferent/ Anti-immigration
Indifferent/ Anti-immigration
Support for immigration is going to be driven in large part, then, by the low productivity and
low-skill intensive immobile firms. Under autarky, preferences will collapse to one dimension as trade
protection will make overseas production unprofitable and capital controls will limit the legal ability
of firms to move. As more firms become more productive or more capital or high-skill intensive,
support for immigration should decrease. With trade openness, the size of the low productivity and
low-skill intensive immobile firms shrinks, also reducing support for immigration. Low productivity
and low-skill intensive firms are the firms that are most likely to face increased competition from
overseas. These firms can choose either to become more productive or capital/high-skill intensive
or close; if capital is open, the firm can also choose to move overseas if it can. Any of the three
choices that the firm makes shrinks the size of the pro-immigration coalition.
We expect, that over the long run, policymakers will respond to the decrease in the size of the
pro-immigration coalition by closing immigration. Therefore, we expect that there will be a long
run negative relationship between trade and immigration, especially when capital is open. In the
short run, however, we think that policymakers may respond to firms’ threats to close or to move
overseas by offering them incentives, including increased immigration and tax breaks to stay open.
Yet we know that immigration policy, and other incentives like corporate tax breaks, tend to be
unpopular with the rest of the constituency. As such, these incentives cannot be maintained forever
and immigration will close in response to trade and capital openness.
Alternative Explanations for Immigration Policy
The extant literature on immigration has focused on two sets of variables to explain immigration
policy: societal-based variables and interest-based variables. For societal-based arguments, the key
variable is the “identity” of the state as it relates to immigration. Societies see themselves as either
immigrant or non-immigrant states and these identities affect the political discussion around border
controls. Interest-based arguments, on the other hand, look at the foundations of an individual’s
position on immigration, based on the labor market, cultural and fiscal effects of immigrants.
16
Societal-Based Variables
One set of explanations for immigration policy assumes that states have identities and these provide a
policy bias toward either open or closed immigration. Meyers characterizes this set of explanations
as arguing “the unique history of each country, its conception of citizenship and nationality, as
well as debates over national identity and social conflicts within it, shape its immigration policy”
(Meyers 2000, 1251). Scholars have used four different dichotomous variables to categorize the
“unique history” of each state: settler or ethnic states, homogeneous or heterogeneous states, jus
sanguinis citizenship or jus soli citizenship states and former colonial power or non-colonizers. Settler
states were built by immigrants and therefore, it is argued, have more open immigration policies
than do ethnic states. Freeman suggests that the “politics of immigration in the settler societies
was forged years ago and was institutionalized during eras when population movements were vital
to national development or even survival.” (1995, 877). The timing of these first immigration
experiences, he says, was critical in producing a receptive cultural context for further waves of
migrants (Freeman 1995, 877). In contrast, “large scale immigration (much of it non-European in
origin) posed serious dilemmas with respect to citizenship and nationality and emerged in countries
with little remembered history of such problems and slight inclination to accommodate the cultural
demands of new minorities” (Freeman 1995, 890).
Another version of the national identity argument is that ethnically heterogeneous and ethnically homogenous states have different immigration policies. Zolberg argues that homogenous
states will be wary of accepting immigrants in order to “preserve a functioning political community.
Massive immigration of any kind poses special problems. . . but the problems are compounded if the
newcomers are culturally very different people who have not been socialized into the same political
traditions” (1989, 411-412). Thus, states that currently have a more homogenous population are
likely to be more uncomfortable accepting immigrants and are likely to have a less liberal policy
(Money 1999).
A third way scholars have operationalized the national identity of states is through their citizenship policies. Societies that identify themselves as immigrant states tend to have more liberal
citizenship policies, based on jus soli (citizenship based on being born in the country). Societies
that identify themselves as non-immigrant states tend to have more restrictive citizenship policies,
based on jus sanguinis (citizenship based on the nationality of one’s parents). Those states with
more liberal citizenship policies, therefore, choose more open immigration policies because they are
“immigrant states” (Money 1999, 28).
Finally, due to their history as colonizers, some states have special relationships with their
former colonies that might also lead to more liberal immigration policies. For instance, prior to
decolonization, citizens of British colonies were treated as subjects of the British Crown and citizens
of France’s North African colonies were given French citizenship. At the time of decolonization,
these special privileges were not removed as the former colonies were seen as still belonging to the
colonizer’s sphere of influence. Migrants from these former colonies could enter as citizens, free from
restrictions. Non-colonizers, on the other hand, did not have these same special relationships and,
17
therefore, have been able to close their countries’ borders (Hansen 2002).
Each of the societal-based arguments hypothesize that immigration policies should vary by the
identity of the state as settler or ethnic state, its homogeneity, the state’s immigration policy, and
its history as a colonizer. Below, I test these propositions.
Interest-Based Variables
The second set of theories look at the role interest groups play in immigration policy formation.
Interest groups on immigration form due to the uneven distributional effects of immigration. These
theories argue that interest groups compete to open or close immigration based on their preferences.
Immigration policy will then reflect the interests of the most powerful organized group. The literature suggests that there are four groups, besides firms, that affect immigration policy: nativists,
immigrants themselves, labor and tax payers. We examine each in turn.
One effect of immigration is that it changes the national culture. This change is threatening to
some members of society, the nativists, who pressure the government for closure to immigration.
The set of arguments on nativism seeks, then, to explain when and why nativism arises. Some
theories argue that nativism arises as the result of the economic effects of immigrants (e.g., Ernst
1948, Olzak 1989) or as the result of the inability of some groups of immigrants to assimilate (e.g.,
Castro 1999, Hatton and Williamson 1998). While it is likely that nativists play a role in policy
formation, there has been no clear articulation of the conditions that allow nativists to affect policy.
Nativism is ubiquitous throughout history – for example, in the United States, there has been a
nativist reaction to the Irish and Germans in the 1840s, Asians in the late 19th century, Southern
and Eastern Europeans in the 20th century and Muslim immigrants today. Yet, as a group, nativists
are only episodically efficacious in altering policy.
In contrast to nativists, the foreign-born may benefit both culturally and economically from
open immigration. Open immigration allows the foreign-born to bring family members or friends
into the state, surrounding themselves with their culture and increasing the economic position of
their family. Immigrants typically want immigration to remain open. Tichenor (1994, 2002) argues
that it has largely been immigrant groups, run by immigrants, who have played a pivotal role
in the politics of immigration in the United States. Immigrants have been able to play this role
because left-wing parties consider immigrants to be a likely base of support (Faist 1994, Messina
2008, Money 1999). In addition to their role in electoral politics, Hollifield (2004a, b) argues that
immigrants use their constitutionally guaranteed rights and the courts to keep states from enacting
and enforcing anti-immigrant statutes.
There are two arguments that suggest that we focus on labor as the key interest group to
explain variation in immigration policy. Some scholars argue that we should focus on the power of
labor unions (Briggs 1984, 2001); while others focus on increasing income inequality (Timmer and
Williamson 1998). Unions oppose immigration because immigrants compete with native labor for
jobs.16 As unions gain strength, they force policymakers to increase the barriers to immigration.
16
Nonetheless, Avci and McDonald (2002) argue that unions have begun to support more open immigration policies
18
Timmer and Williamson (1998) argue that it was increasing inequality that led to the immigration
restrictions in the early 20th century through two mechanisms: as inequality increases due to
immigration, the median voter gets poorer and pushes for a more restrictive policy and/or citizens
vote for a more restrictions on immigration because they dislike increased inequality.
Finally, a number of other scholars have focused on the fiscal costs of immigrants as the variable
that explains immigration policy. Here, immigrants are typically portrayed as a drain on government
services, a fear that has had a long history. As early as 1738, the colony of South Carolina passed
a regulation forcing shipmasters to pay a bond on immigrants who were likely to become a public
charge (Neuman 1993, 1858). Most British Colonies followed suit, creating their own regulations
against those likely to become a public charge. Fiscal arguments suggest that citizens oppose
immigration because immigrants are thought to use more government services than they provide in
tax revenue (Hanson, Scheve and Slaughter 2007, Hatton and Williamson 2005a, b, Money 1999).
Rich tax payers oppose immigration because of the increased tax burden and poor tax payers —
those likely to be recipients of the welfare state — oppose immigration because immigrants crowd
them out (Hanson, Scheve and Slaughter 2007, Hatton and Williamson 2005a, b). Money (1999)
finds that when the fiscal burden of immigrants negatively affects swing areas, immigration is more
likely to be closed, as the policymaker caters to the preferences of these swing areas to stay in office.
Each of the interest group based arguments suggest that immigration policy should vary depending on which group has relatively more power: the nativists, the immigrants, labor, and tax
payers. Below, I examine the support for each of these arguments as well.
Cross-National Immigration Policy, 1783-2010
One of the major obstacles to research on immigration has been the lack of longitudinal crossnaitonal data. In response to this lacuna, this paper assembles a new dataset of de jure immigration
policy across 19 countries over the last 200 years. The 19 states were chosen due to their attractiveness to potential migrants. Previous research on migration suggests that potential migrants
choose locations where wages are high relative to the transaction costs of moving (Massey et al.
1993). States that are very wealthy, such as the United States, the United Kingdom, Germany and
Japan, are likely to attract migrants from all over the world; while states that are relatively wealthy
in comparison to their neighbors, such as Argentina, Brazil or South Africa, are likely to attract
migrants from their neighbors but not from countries far away. The states chosen, therefore, are all
wealthy states, in comparison to the rest of the world or to their neighbors.17 From the universe
of wealthy countries, the 19 states were selected to include those that have been much studied in
so that they can integrate immigrants who are in the country illegally into unions. Unions argue that illegal workers
do more to harm union interests, as they are unwilling to report employers for other illegal practices, than a more
open immigration policy would. A more open immigration policy, therefore, is a second best solution for unions
today; they prefer to have no immigration but would also prefer legal immigration to illegal immigration.
17
This criteria was operationalized as states with GDP per capita above 200% of the world average GDP per capita
in a given year or above 200% of the average GDP per capita for the geographic region in which the state is located.
The regional criteria captures states that are attractive to immigrants due to high wages combined with proximity.
These two criteria lead to the inclusion of 74 states over at least some part of the time period of 1800-2008.
19
the literature, such as the United States and European states, and those that have received less
attention, such as the Persian Gulf or East Asian states. The breadth of the dataset ensures that
previous arguments in the literature can be tested while providing external validity by examining
a wider range of immigrant attracting states. Table 2 lists the state included in the data set along
with the years of coverage for each state. Every state was coded through 2010, but they vary by
when they enter the dataset. States enter the dataset when they have control over their immigration
policy, either when they obtain responsible government, independence, or come into existence in
their current form.
Table 2: Countries included in the dataset and the dates of inclusion
Group
Country
Settler States
US (1790-2010)
Australia (1787-2010)
Canada (1783-2010)
New Zealand (1840-2010)
South Africa (1806-2010)
Argentina (1810-2010)
Brazil (1808-2010)
European liberal democracies
UK (1792-2010)
France (1793-2010)
Germany (1871-2010)
Netherlands (1815-2010)
Switzerland (1848-2010)
Export-oriented industrializers
Japan (1868-2010)
Hong Kong (1843-2010)
Singapore (1955-2010)
South Korea (1948-2010)
Taiwan (1949-2010)
Autocracies
Saudi Arabia (1950-2010)
Kuwait (1961-2010)
The dataset includes all laws on immigration and immigrant rights. For democracies, only
legislation passed was included; in the case of an autocracy, announced policy changes and legislation
passed in the autocracies that have legislatures were included.18 Immigration policy, rather than
flows, are coded because this paper is interested in why states have chosen the immigration policy
they have rather than how migrants have responded to these policies. To use flow data as a measure
of government policy, we would need to control for the demand to migrate, using a gravity model
as the trade literature does. However, there are great data limitations to using a gravity model
18
The use of only legislation or announced policy was due mostly to data availability reasons. Most administrative
action has not been documented in a systematic way. We know, for example, that during the Great Depression, the
US State Department was told to approve only a few visas. The State Department approved even fewer visas than
were available under the quota limit (Reisler 1976). We do not, however, have good documentation to know whether
the State Department had done this at other times or if other countries had used these administrative actions.
20
of immigration.19 The codings, therefore, are similar to using tariff levels for trade policy or IMF
capital controls measures for capital policy instead of using a measure that would capture non-tariff
or non-policy barriers to capital.
Immigration policy is an amalgam of several sets of policies, including policies that regulate who
gains entry to the state (border regulations), how the border is enforced (enforcement) and what
rights immigrants receive once they have entered the state (immigrant rights). Within each of these
three categories, states have used numerous policies to control their borders; these policies often
serve as substitutes for each other. Additionally, policies can belong to more than one category. For
example, family reunification policies are considered a right that immigrants may possess in Europe,
East Asia and the Persian Gulf and form a less important instrument in their immigration policies;
in contrast, family reunification is one of the main policy instruments in United States immigration
policy. Similarly, regulations on what positions immigrants can take and what sectors they can
work in has been a major policy tool to control entry in East Asia and the Persian Gulf; however,
the ability to work in any position is also seen as a right in liberal democracies. Refugee and asylum
policy are also hard to categorize; although these policies control entry into the state, they also are
seen as a human right. Table 3 lists the different dimensions of each category of immigration policy
and the basis of each dimension.20 Each dimension was coded from 1 to 5, with greater restrictions
taking lower values and more openness taking greater values.21 There are twelve total dimensions
— eight that regulate entrance to the state, of which four, work prohibitions, family reunification,
refugee and asylee policy, could also be considered rights; two that cover the rights that states give
immigrants and two that cover enforcement of the rules.
Border regulations are often what people think of as immigration policy: policies that determine
who gains entry to a state. Over the last 225 years, states have used many different types of border
control regulations to allow in the immigrants they want while keeping out those they do not. The
preferred method of control in the late 19th century was prohibitions by nationality, as seen by the
Chinese Exclusion Act in the United States or laws against Polish migration in Germany. In some
cases, exclusions against nationality could not be used due to concerns over diplomatic fallout of
19
There have been previous successful studies that use a gravity model of immigration flows (Leblang, Fitzgerald
and Teets 2009). However, even these studies have run up against data limitations. Most states have not collected
statistics on the origins of their immigrants over a long period of time and in the case of the Persian Gulf states these
data are simply not published. Furthermore, states often do not categorize the inflows of immigrants in the same
way and thus, it is hard to create comparable sets of immigration statistics. Collecting this data, especially before
the 1980s, therefore, is likely to be very difficult. Finally, if we are interested in a true flows measure, we would likely
want to know the number of illegal immigrants entering the state in a given year; however, this number is usually not
available given the nature of these flows. As such, I use a de jure immigration policy measure rather than a de facto
measure based on flows. In the coding scheme, I coded laws on enforcement to try to get around the issue of whether
illegal immigration is used as a backdoor open immigration policy by measuring whether states were enforcing the
laws they have on their books.
20
A more detailed explanation of the dimensions and their coding is listed in Appendix A.
21
Theoretically, immigration policy has no bounds; states could always pay people more and more money to come
to their country; for example in the 1960s, South Africa not only paid for the cost of passage for European workers,
but also established them with bonuses and other monetary goods. At the other end of the spectrum, states could
also denaturalize part of their population and force them to leave, as the Germans did to the Jews under the Nazi
regime or the South Africans did to Africans under Apartheid. These examples, however, are very rare. Most states
fall in between these two extremes the vast majority of the time.
21
Table 3: Dimensions of immigration policy
Category
Dimension
Coding criteria
Border regulations
Nationality
Skill
Quotas
Recruitment
Work prohibitions
Family reunification
Refugee policy
Asylum policy
Number of nationalities restricted
Restrictions based on skill or wealth
Numerical limits on entry
Policies aimed at recruiting immigrants
Restrictions on industries or positions held
Distance of relatives allowed special entry
Entrance policies for refugees outside the state
Entrance policies for those claiming
refugee status at the border
Immigrant rights
Citizenship
Other rights
Who can be a member of the state
Other rights immigrants possess
Enforcement
Deportation
Other enforcement
Who can be deported and how
Other enforcement measures in place
the exclusions. In these cases, skill requirements were used to achieve the same ends; for example,
Great Britain did not want her dominions passing prohibitions against Indian immigrants due to
protests from within India. After World War II, when nationality as a basis for immigrant laws was
delegitimized, skill requirements replaced nationality requirements in many states as a way to keep
out the same immigrants.22
States also have used measures which seek to both regulate the entrance of migrants and their
participation in the labor market. Recruitment measures captures the laws that regulate the recruitment of immigrants by both governments and the private sector. Work prohibitions capture
which sectors immigrants can work in and how many immigrant workers a firm may hire.
Family reunification policies and quotas, in contrast to nationality regulations, skill regulations,
recruitment measures and work prohibitions, have rarely been used to control immigration. Family
reunification policies, in contrast to the policies examined thus far, regulate one form of potentially
22
I code skill requirements as immigration restrictions because most potential immigrants over the last 225 years
have been low-skill. Prior to the transportation revolutions that allowed the low-skill laborer to travel on his own,
most low-skill laborers were brought as slaves, indentured servants, or convicts (Hatton and Williamson 1998). It
was not until the 1840s that free European migration to the Americas flowed at the same rate as coerced African
migration (Hatton and Williamson 1998, 7). Thus even at a time when most white migration was high-skill, most
migration in total was still in unskilled labor. Finally, while there has always been migration of high-skill labor, the
movement of this labor has not been politicized in the receiving states nor has it been seen as much of a threat to
the national interest (however, the flow of highly skill workers has been greatly politicized due to the effect on the
sending country, see Bhagwati and Hamada (1974), Clemens (2007), Kuhn and McAusland (2006), Stark (2004)).
The lack of politicization of high-skill immigration is likely due to the small numbers of these immigrants relative to
their population in receiving states (although they can be a large percentage of skill workers in the sending state)
and their ability to assimilate (Hatton and Williamson 1998, 140)). For instance in the United States in 2005, the
number of green cards for high-skill workers were approximately 10% of all green cards, similar to the number of
refugees offered green cards (Office of Immigration Statistics 2006, 18). Additionally, in 2005 H1B visa holders made
up less than two percent of all temporary visa holders (Office of Immigration Statistics 2006, 60). In other OECD
states the percentage of high-skill workers has been larger than in the United States, but their skill level should be
viewed in light of policies in these states that select high-skill workers. The skill composition of legal immigrants in
these states is part of the equilibrium outcome and not due to a change in the composition of potential immigrants.
22
non-economic migrants, family member of citizens and residents of the state. Quotas are also
different from the other policies in that they used to regulate the total intake of immigration rather
than regulate the type of immigrant. Most states did not separate out family immigration from
other forms of immigration until the 20th century. Since then, most states have allowed at least
some family reunification, usually allowing spouses (in some cases only wives), fiancees and children
of citizens and immigrants entrance to the state. Similar to family migration, many governments in
the 19th century simply admitted as many people as were willing to come to their states and could
pass the standards for admission; in the 20th century, a few states have placed numerical quotas on
immigration.
The final two dimensions of border regulations also look at potentially non-economic migrants:
refugees and asylees.23 The advent of modern refugee policies was the aftermath of World War
II when most of the Allies outside of Europe took in European refugees. After the World War
II refugee crisis ended, most states continued to accept refugees on an ad hoc basis. Only after
the Vietnamese refugee crisis did the United States, Australia, Canada and New Zealand create
refugee policies to resettle a small number of refugees per year. The United States, by far, has
the most generous refugee resettlement program, taking in about 80,000 refugees per year; Canada
and Australia which have the next most generous programs take about 11,000 and 15,000 refugees
per year respectively. More recently, Brazil, Argentina, Germany, the Netherlands and Japan have
agreed to take in refugees and South Korea has always had a policy of resettling refugees from North
Korea.
In contrast to refugee policy, most states have created asylee policies to address those seeking
asylum in their country. The most generous policy was passed in Germany after World War II, where
the right to asylum was enshrined in the constitution. These asylum policies became contentious
after these states restricted employment-based immigration programs in the 1960s and 1970s. At this
point, the political discourse on immigration shifted to asylum seekers who were seen as economic
migrants in disguise. The states in Europe began enacting safe-third country restrictions, quicker
determination of asylum cases, safe country of origin determinations and restricting asylees’ access
to the social welfare system.24 In 1992, Germany ended its constitutionally based asylum policy
and brought its policy in line with the rest of the EU. Similarly, most other states have sought to
curtail asylum rights in recent years and the Persian Gulf states and Singapore have never enacted
them.
States also vary in the legal rights granted to their immigrants. The most important right that
states can offer immigrants is citizenship; citizenship allows the immigrant to stay in the state and
23
A refugee is defined as a person fleeing from their country who is outside the state they are trying to enter. Asylee
is defined as someone who is at the state’s borders or inside the country who wants to claim refugee status. Refugee
and asylee policies are categorized as border regulation policies, rather than rights, because refugees and asylees are
potential labor market participants. Additionally, it is often thought that asylees are economic migrants who cannot
enter in another way (see for example Kay and Miles (1988) and Kaye (1994)).
24
Safe third country restrictions allow the asylum claim to be automatically classified as manifestly unfounded if
the asylee travelled through a country that the receiving state has deemed as “safe.” Safe countries of origin are
states that are deemed by the government to be safe and therefore, asylum claims are found to be unfounded without
a hearing.
23
have the same rights as natives. There are several other rights that the state could give immigrants
as well. For example, above, we examined the right to work in any industry and the right to bring in
family members; these two rights are also important regulatory measures for controlling the border.
Other rights include property rights, access to the welfare system and the right to vote. All of these
rights likely affect the number of immigrants in the state. Citizenship given to adults is usually given
to their children, increasing the total size of the population. The right to stay in the country after
losing one’s job, to join a union and anti-discrimination laws will affect the total amount of money
the immigrant can make and therefore, affect his choice of if and where to migrate. Additionally,
the right to invest in land or a business will affect the calculation of whether to enter or to stay in a
state. Access to the welfare system is also thought to affect the number of immigrants in the state;
previous studies have argued that immigrants choose to move to states with a large social welfare
system (however Zavodny (1997) finds no effect of welfare policies on immigration in the United
States). Finally, other political rights, such as the right to assembly, the right to vote, freedom of
religion and anti-discrimination laws may positively affect migrants choice of where to settle.
There is no definitive proof that rights affect migrants decision of where to move; however, there
is evidence that states act strategically when giving immigrant these rights. For example, in the
19th century, liberal citizenship policies and the right to set up ethnic-based state-supported schools
were used by both Canada and Brazil to attract immigrants. Brazil would even pay for ministers to
attract German and Swiss settlers. More recently, the rights debate has been focused on deterring
rather than encouraging immigrants. In the United States, the debate over immigration in the
1990s often focused on limiting immigrants’ access to the social welfare system as a way to deter
immigrants. Similarly in Europe, limiting asylum seekers rights to the social welfare system has
been seen as a way to deter them. Finally, the treatment of immigrants has affected the sendingcountries’ willingness to allow their citizens to migrate. In the 1920s, India passed legislation to
limit the recruitment of their workers due to their mistreatment. Similarly, the Philippines has
prohibited workers from going to various Persian Gulf states and other countries over disputes
about the treatment of Filipino workers. Immigrant rights, therefore, are likely to affect the flow of
immigrants to a state.
The final two dimensions of immigration policy focus on enforcement of immigration rules. A
very restrictive immigration policy that is not enforced is not that different from an open policy,
as many people can still enter the state and work illegally. States use deportation most often as an
enforcement mechanism. The coding of deportation measures how easy it is to get rid of unwanted
immigrants. Increased safe guards, such as access to courts, make deportation more difficult. On
the other hand, increasing the number of deportable offenses to include illegal immigration, losing
one’s job, going on welfare or being part of a political or social group makes deportation easier.
Paying immigrants to leave the country also is clearly a sign that the government wants to decrease
immigration. Other than deportation, states use a variety of measures to enforce their borders.
These policies include placing additional penalties on illegal immigration, sanctions on transportation companies that bring illegal immigrants, sanctions on employer that hire illegal immigrants and
24
measures to police the border. States also often give amnesty to those in the country illegally; in
some cases the illegal immigrant is allowed to stay while in others he is forced to leave but does not
have to pay any of the penalties for being in the country illegally. The dimension, other enforcement,
seeks to a measure these other enforcement mechanisms.
The policy dimensions can be integrated in a general measure of immigration policy for each
country-year. The final goal of a state’s immigration policy is to attract and allow in a certain
number of immigrants each year. There is not, however, a consensus in the literature about how
these different dimensions affect the flow of immigrants; although, it is clear that not all dimensions
affect the number of immigrants equally.25 I use principal components analysis on all twelve the
dimensions of immigration policy to recover an immigration policy score.
Principle components analysis reveals that these twelve dimensions combine to create two different factors: immigration policy and rights of immigrants.26 The first factor, immigration policy,
places more weight on nationality, skill, recruitment, quotas, enforcement and deportation policies
than the second, rights of immigrants, which places more weight on family reunification, refugee,
asylee, citizenship, rights and work prohibition policies; hence, the names for the two factors. The
first factor also correlates highly (0.95) with a standardized average of the nationality, skill, quota,
recruitment, work prohibitions, deportation and enforcement scores. Figure 3 illustrate the change
in the immigration policy score cross-nationally over the last 225 years.27
The data suggest that states should be categorized into four categories — settler states, European
liberal democracies, export-oriented industrializers and autocracies — that vary in what type of
policies they use to regulate their borders and their openness to immigration. The settler states
consist of the land abundant states of the New World: the United States, Argentina, Australia,
Brazil, Canada, New Zealand and South Africa. The settler states were open throughout most of
the 19th century. The United States was the first state to begin restricting immigration to any
great degree in the 1880s when it imposed a ban on Chinese immigration and began to restrict
various undesirable classes (the poor, the sick and the less skilled). Soon the other settler states
followed suit with their own restrictionist policies. Most of these policies remained in force and
were strengthened during the Interwar period, especially with onset of the Great Depression. After
25
There has been only one study thus far (Leblang, Fitzgerald and Teets 2009) that has examined how immigration
policies have attracted immigrants. In that study, however, only citizenship policy and a simple coding of immigration
policy were included.
26
There are four eigenvalues greater than 1, but the other two are 1.14 and 1.05 and neither of these explain much
of the variation.
27
For robustness, I created a second measure of immigration policy coding refugee, asylee and family reunification
policy differently. As we have seen, in the 19th century most states did not have special exemptions in their policies
for refugees, asylees and family members; however, they did not need these exemptions since other immigration
policies did not keep out these groups of people. In the coding used to create figure 3, I coded these policies in these
years as a 1, as there was no policy in place. For the second coding, I coded these policies as a 5 for these years and
as a 1 once there was a policy that would exclude a refugee, asylee, or family reunification immigrant. This coding
produced just one factor, which correlates with the coding of immigration policy at 0.9. The first four eigenvalues
for this analysis are 4.73, 1.46, 1.13 and 1.01; given the large size of the first factor in comparison to the others, it is
likely the case that there is only one factor. Additionally, the data for 8 states, the United States, Canada, Australia,
New Zealand, South Africa, the United Kingdom, France and Germany were recoded by a second coder; the data
between the two codings correlate at 0.9.
25
Immigration policy
2
1
0
-1
-2
2
1
0
-1
-2
2
1
0
-1
-2
2
1
0
-1
-2
26
1800 1850 1900 1950 2000
Singapore
Taiwan
1800 1850 1900 1950 2000
Switzerland
South Africa
New Zealand
Netherlands
Brazil
US
Year
1800 1850 1900 1950 2000
Saudi Arabia
Japan
UK
Argentina
Kuwait
Hong Kong
France
Australia
1800 1850 1900 1950 2000
Figure 3: Immigration policy for each state
1800 1850 1900 1950 2000
South Korea
Germany
Canada
World War II, each of the settler states reopened their policy to some extent before closing the door
again. Nonetheless, immigration policy was never opened to the same extent that it was prior to
World War I. Interestingly, the United States has had a more restrictive policy than the other four
settler states throughout the majority of the time period.
The European liberal democracies consist of capital abundant states — the United Kingdom,
France, Germany, the Netherlands and Switzerland. The European liberal democracies exhibit
similar patterns as the settler states. With the exception of Germany, they tended to be relatively
open in the 19th century, closed throughout the interwar period, opened some after World War II
and then closed again since the 1970s. In contrast to much of the immigration literature, which
argues that European states have had more restrictive immigration policies due to their history as
“ethnic states,” I find that while these states were slightly more restrictive during the 19th century,
they were still relatively open to immigration. As such, many of these states had large inflows of
people from other European states, for example, workers from Southern Europe and refugees from
religious persecution. In the 20th century, Europe again does not look vastly more restrictive than
the settler states; in fact, many of the European states were more open to immigrants throughout
this period than the United States.
The third group — the export-oriented industrializers — consists of states that industrialized
by orienting their markets to export goods to the major markets: Japan, South Korea, Hong Kong,
Taiwan and Singapore. After World War II, these states had a relatively restrictive policy. They
relaxed these restrictions in the 1970s and 1980s but have restricted immigration since then. These
states have tended to regulate their borders with sectors specific prohibitions on immigrant labor
and strict regulation of the recruitment of workers.
Finally, there are the natural resource dependent autocracies of Saudi Arabia and Kuwait. These
states were relatively open directly after World War II and have been restricting their borders since
the 1980s. They have tended to have few restrictions on the skill or nationality of workers who can
enter, but have, like the export-oriented states, regulated the sectors in which immigrants can work
as well as adopting draconian enforcement measures to deport immigrants when they are no longer
needed.
Not every state fits into each group on all dimensions. For example, the United States, while
in some respects acting as the other settler states, is often an outlier. The United Kingdom often
adopts similar policies as the other European liberal democracies but occasionally adopts policies
similar to her Dominions. Finally, Singapore often adopts immigrant rights and enforcement policies
more similar to the policies used by Saudi Arabia and Kuwait than the policies used by the other
export-oriented industrializers. Nonetheless, among the members of each group there are many
similarities in both policies used and in their overall immigration policy. Additionally, there is great
similarity in the overall immigration policies among the four groups.28
28
The policy coding may seem to be at first glance at odds with immigrant stocks and flows. For example, by
measures of stocks, Saudi Arabia and Kuwait, with 25% and 62% foreign born respectively, should be very open; in
some cases, more open than many states were in 19th century (Ratha and Xu 2008). Similarly, by examining the
stock of foreign born, the United States looks as open today as it did 100 years ago (Office of Immigration Statistics
27
Examining the Evidence
The data show that, even though these groups of states have used different immigration policies,
all states are today more restrictive of immigration than in previous years. What accounts for this
increase in restrictions? I show below that trade policy, in combination with capital policy, can
explain the pattern of increased immigration policy restriction across all states.
The Relationship between Trade, Capital and Immigration
The historical data on immigration, trade and capital policy show trade in the 19th century was
relatively more restricted while immigration was more open. In the 20th century, however, this
relationship reversed. Capital policy, comparatively, has varied more over time; money moved
easily across the globe throughout the 19th century and since the end of Bretton Woods. Markets
were closed in the Interwar years and in the early post-World War II period. Figure 4 shows the
relationship between trade, capital and immigration policy for each state.29
As can be seen in figure 4, in the 19th century, most of the settler states had relatively high
tariffs, no capital controls and open immigration. To some extent, the settler states in the 19th
century were forced to have this constellation of policies — closed trade and open immigration and
capital — if they wanted to develop their domestic markets: each had few workers and needed
protection from European goods in order to foster indigenous industrialization. This arrangement
changed, however, in the early years of the 20th century. Settler states began to remove their
tariff barriers (some which were reinstated with the Great Depression) and closed their immigration
policies. Throughout the 20th century, these states continued to open their borders to the free
movement of goods while closing their borders to people. There have been moments of reopening
of immigration, although only in periods of capital controls when businesses could not freely move
production overseas.
The European liberal democracies had lower tariff barriers in the 19th century than did the
settler states and were not as open to immigration. On the heels of the Great Depression, Great
2006). However, examining stocks and flows alone hides the demand for immigration. Demand for immigration
has risen as transportation costs have decreased greatly over the last 200 years. Further, increases in information
technology has lowered the opportunity costs of migration as migrants have greater knowledge of the employment
situation in the receiving state and spend less time looking for employment. Moreover Sassen (1988), for example,
argues globalization has actually increased the size of potential migrants through disrupting the traditional economy
and has led to “step” migration as migrants first leave their villages for cities in their home state and then leave
the cities for another country. Therefore, examining stocks or flows without controlling for demand would lead to a
biased picture of immigration policy.
29
Trade policy is measured as the percent of imports that come into the country without duties and is from Clemens
and Williamson (2004) and updated for this paper by the author. Capital policy is measured as an indicator variable
taking the value of 1 when capital controls are in place and 0 otherwise and is from Bardo et al. (2001) and was
also updated. I use this trade measure because it is the best measure that captures trade policy over a long period
of time. The only other variable available over a similarly long period of time is exports plus imports over GDP.
This measure conflates trade policy and other aspects of the world economy, such as GDP growth, crises and the
like. Other measures that contain non-tariff barriers have only become available more recently. I use the indicator
variable for capital controls due to its availability over a long time period as well. Newer measures, such as Quinn
and Toyoda (2006) or Chinn and Ito (2008) measures, are only available for the years since World War II.
28
Trade Policy
1
.5
0
1
.5
0
Switzerland
Netherlands
1800 1850 1900 1950 2000
Year
1800 1850 1900 1950 2000
Saudi Arabia
Japan
UK
Argentina
Capital Controls
Immigration Policy
1800 1850 1900 1950 2000
Singapore
South Africa
New Zealand
Taiwan
Brazil
US
South Korea
Germany
Canada
1800 1850 1900 1950 2000
Trade Policy
1800 1850 1900 1950 2000
Kuwait
Hong Kong
France
Australia
Figure 4: Trade, capital and immigration policy
2
1
-2 -1 0
2
1
-2 -1 0
2
1
-2 -1 0
1
.5
0
1
.5
0
2
1
-2 -1 0
29
Immigration Policy
Britain, France and Germany all greatly increased their tariff barriers and enacted capital controls.
After the war, these states slowly dismantled their tariff barriers but kept their barriers to the free
movement of capital. It was in this immediate post-war rebuilding period that these states signed
bilateral labor migration treaties, which opened their economies to some low-skilled workers. After
the demise of the Bretton Woods exchange rate system and the rise of the European Monetary
Union, European states dismantled their barriers to the free movement of capital but increased
their restrictions on immigration.
Export-oriented states and the autocracies have differed considerably in their trade, capital
and immigration policies. The export-oriented states and the autocracies have tended to have
relatively low barriers to goods but high barriers to people and to money. Immediately after World
War II, export-oriented states limited immigration to highly skilled workers, such as executives for
multinational corporations, and most of these states also enacted capital controls. After removal of
capital controls, these states opened their economies slightly to immigrants; however, this opening
did not last in the face of firms migrating to locations, such as China and Malaysia, that had cheaper
labor forces. The autocracies, in contrast, opened their borders to workers of all skill levels and
opened their borders to capital. Saudi Arabia and Kuwait have recently shifted to a more restrictive
immigration policy in line with their development strategies to create high-skilled service industries.
As seen in figure 4 immigration and trade tend to be negatively correlated. Table 4 examines
the relationship between trade, capital and immigration more formally by regressing immigration
on trade and capital.30 The models contain country and year fixed effects, a linear time trend for
each country and robust standard errors. Models 1 and 2 include all years of data. Trade squared
and trade squared interacted with capital policy are not included because they are not significant
and do not greatly improve the fit of the model. Over all the data, there is a negative relationship
between trade and immigration and this relationship is stronger when capital is open than when it
is closed. The relationship between capital openness and immigration is positive. This is largely
driven by the fact that states used very few capital controls during the 19th century — in fact
Bardo et al. (2001) codes all states as having no capital controls prior to the Interwar period —
when immigration was also very open.
Next, I examine the 19th century and the Interwar period. Model 3 examines the 19th century
and 20th century prior to the start of World War I. As noted above, capital was not restricted
during this time period, so capital is not included in this model. The trade squared term does not
improve the fit of the model either and so is not included. The relationship between trade and
immigration is less strong during this time period. During the interwar period, the relationship is
stronger; when states opened their doors to both trade and capital, they were more likely to close
immigration. Again, a term for trade squared and trade squared interacted with capital are not
30
As I argued above, the rational policymaker chooses a bundle of trade, capital, immigration and tax policy in
order to maximize the best interests of her firms. Firms choose whether or not to produce, what technology to
use and where to produce based on the policymaker’s decision on the policy bundle. The policies examined below
cannot be thought of as exogenously occurring. Therefore, the analysis is restricted to determining if there is a robust
correlation between these choices in the data.
30
31
1786
0.83
-0.01***
(0.00)
4.39***
(0.68)
-2.91**
(0.83)
(1)
All Year
1356
0.78
-0.01***
(0.00)
2.01*
(0.91)
2.34+
(1.21)
-2.65+
(1.31)
-0.81
(0.89)
(2)
All Years
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade2 *Capital
Trade*Capital
Capital policy
Trade policy2
Trade policy
DV: Immigration policy
545
0.61
-0.01
(0.00)
3.31+
(1.63)
-1.85
(1.75)
(3)
Pre-1914
201
0.64
-0.04**
(0.01)
5.61**
(1.35)
6.45*
(2.73)
-7.10*
(2.76)
-1.51
(1.33)
(4)
Interwar
506
0.68
-0.03*
(0.01)
2.29
(1.35)
3.43+
(1.61)
-4.47*
(1.81)
0.63
(1.71)
(5)
Pre-1946
850
0.44
1.08
(4.92)
-3.26
(7.58)
-104.05
(60.19)
221.84
(127.73)
-118.32+
(67.70)
-0.00
(0.00)
-0.90
(3.31)
(6)
Post-WWII
291
0.33
-8.24*
(2.85)
11.77*
(4.74)
-173.42***
(38.58)
366.74***
(83.37)
-193.88***
(45.08)
-0.01
(0.01)
5.73*
(2.55)
(7)
Bretton Woods
Table 4: Immigration policy regressed on trade and capital policy
559
0.35
20.45***
(5.05)
-27.91**
(7.26)
-29.33
(48.87)
60.37
(102.89)
-31.00
(54.15)
-0.01
(0.00)
-13.60**
(3.58)
(8)
Post-Bretton Woods
included because they are not significant and do not improve the fit of the model. When capital
was closed, there is not a statistically significant relationship between trade and immigration. This
relationship is likely driven by two opposing factors. Capital was closed in this time period in
response to economic crises, especially the Great Depression. Trade and immigration during this
time period were often both closed in response to crises as well, which result in a positive relationship
between trade and immigration. After the worst years of the Depression, trade was opened in some
states while immigration remain closed, which resulted in a negative relationship between trade
and immigration. Over the entire period prior to the end of World War II, there is a negative
relationship between trade and immigration and again, this relationship is stronger when capital
was open.
Finally, I examine the post-World War II era. In this time period, both a squared trade term and
trade squared interacted with capital are included as they improve the models’ fit. Overall, there is
an inverse U shaped relationship between trade and immigration when capital is open. During the
Bretton Woods era, this relationship is stronger. For states that kept their capital markets closed,
increasing trade openness led them to open immigration. States that had open capital markets
had more restrictive immigration policies and had an inverse U shaped relationship between trade
and immigration. States that kept their capital markets closed were willing to offer firms increased
immigration as they face increased competitive pressure from international competition. States that
opened their capital markets were willing to open immigration at first, but closed immigration as
trade opened further. After the end of the Bretton Woods era, only those states that continued to
keep capital markets closed were willing to offer immigration openness as a way to subsidize firms
threatened by open trade and even these states could not continue to offer these subsidies as trade
open further. For states which had open capital markets, there is no relationship between trade
and immigration. This result may be driven by the lack of variation in trade policy for these states.
Trade policy is very open during all of these state-years; 95% of state-years had tariff rates of less
than 10% and 75% had rates of less than 3.5%. It is less likely that the small changes in trade
policy in these states had much effect on their firms; instead it is likely that most of the firms that
would use immigrant labor and would be affected by trade had responded to previous rounds of
trade opening by becoming more productive, moving overseas or closing their doors.
Over all the data, immigration policy is negatively affected by trade policy in the long run, but
how immigration policy responds to trade policy varies by time period. Prior to the end of World
War II, policymakers respond to opening trade by closing immigration. This pattern is not only
observed in the overall data but also in individual country data (see tables 7 - 23 in Appendix B).
As a robustness check, I examined the relationship between trade and immigration in each state.
There is a negative relationship between trade and immigration in Argentina, Brazil, Canada, New
Zealand, Germany, the UK, and the US for at least some of the period prior to World War I. After
World War II, we see that policymakers are willing to open immigration in the short run in response
to trade openness but closed in the long run. This is seen in many states for at least part of the
period: Australia, Brazil, Canada, New Zealand, France, Germany, the Netherlands, Singapore,
32
South Korea, Taiwan and Kuwait.
Table 5 examines why some states in some time periods use immigration as a subsidy to keep
firms at home instead of using tax incentives or allowing firms to close or move overseas. As argued
above, policymakers have two ways to subsidize firms: they can offer them increased immigration
or decreased taxation. States that are more reliant on business taxation will be more likely to use
immigration to keep firms at home. I include the importance of corporate tax revenue, operationalized as business taxes as a percentage of GDP from Cusack (2000b) and interact it with trade policy
and squared trade policy as a way to measure this. Increasing reliance on corporate taxation should
result in states being more likely to use immigration policy as a way to keep firms at home. As a
second way in determine preferences over which tool to use, I include party ideology and interact
it with trade policy. We think that left parties — the excluded category — would be more willing
to use immigration as a subsidy because they want to continue collecting corporate tax revenue to
pay for social services. Therefore, right parties should be less likely to increase immigration in the
face of increased trade and capital openness.
The ideology of the party in power should also help shed light on hypotheses 7 and 8 as well
as the importance of nativists and labor. Given the lack of consensus on what activates nativist
sentiment, I cannot explicitly test this argument. As a proxy, I examine the ideology of the party in
power; right parties are typically the parties that nativist join. Therefore, if the nativist hypothesis
is correct, immigration should be more restrictive under right parties than left parties. If the labor
hypothesis is correct, immigration should be more restrictive under left parties. The data on parties
in power for the OECD states are from Cusack (2000a) and the data on parties in Brazil and
Argentina are from Coppedge (1997). For each state, an indicator variable was created to capture
whether the party in power was in the left, center, or right for that country.
Table 5 examines the importance of corporate taxation. Models 1 and 2 include the corporate
tax measure and, therefore, examine data from the years 1951-1995 for the states in the OECD
and models 3 and 4 include the party in power measure and therefore, examine 1947-1997 for the
states in the OECD and Argentina and Brazil. In order to not include triple interaction terms,
models 1 and 3 examine the relationships when capital is closed and models 2 and 4 examine the
relationships when capital is open. As can be seen by the insignificant interaction terms, neither
the inclusion of corporate taxation nor party in power interacted with trade affect the relationship
between trade and immigration. Instead, it seems that most states and most parties were willing
to use immigration to keep firms from facing closure due to trade pressure in the early stages of
openness — as seen by the positive correlation between trade and immigration under closed capital
during the Bretton Woods era — but were not as willing to use immigration to keep firms from
moving overseas when capital was open or to keep firms from closing when world trade was opened
greatly — as seen by the inverse U relationship on states with open trade and capital during the
Bretton Woods era and on states with only open trade in the Post-Bretton Woods era. This is also
evidence that nativism and organized labor cannot explain immigration policy nor the relationship
between trade and immigration.
33
Table 5: Immigration policy regressed on preference for taxation, party in power and trade and
capital policy
DV: Immigration policy
Trade policy
Trade policy2
Corporate taxation
Trade*Corporate taxes
Trade2 *Corporate taxes
(1)
Capital Controls
(2)
No Capital Controls
(3)
Capital Controls
(4)
No Capital Controls
15.82
(58.05)
-28.28
(73.01)
4.95
(7.73)
-7.71
(11.31)
12.16
(14.76)
250.69
(192.23)
-291.00
(218.40)
35.12
(40.04)
-49.06
(55.86)
56.64
(64.05)
16.69
(12.13)
-28.40
(17.70)
125.60+
(57.79)
-147.22*
(62.73)
23.28
(63.10)
-17.24
(62.96)
-31.36
(87.43)
32.20
(98.20)
25.82
(86.98)
-35.43
(96.93)
0.00
(0.00)
-91.15+
(41.57)
215
0.80
Center Party
-0.01
(0.01)
-8.42
(40.94)
0.00
(0.01)
-180.60
(138.00)
7.21
(13.58)
19.63
(11.13)
-11.40
(19.57)
17.25
(24.58)
-30.33
(16.89)
46.33+
(24.74)
0.01***
(0.00)
-12.21
(8.02)
162
0.81
198
0.81
247
0.45
Right Party
Trade*Center
Trade2 *Center
Trade*Right
Trade2 *Right
Linear time trend
Constant
Observations
R2
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
34
Test of Additional Hypotheses and Alternative Explanations
Now, I turn to examining the additional hypotheses from the model as well as the alternative
hypotheses. The first set of alternative hypotheses — the societal-based variables — focus on
the identity of states: settler or ethnic states, homogenous or heterogeneous states, jus sanguinis
citizenship or jus soli citizenship states and former colonial power or non-colonizers. Given these
differences in immigration histories, the first version of the theory argues, settler states should
have more open policies today than ethnic states. As we saw above, the data do not support this
argurment. First, the assumptions on which this theory is based are incorrect. In the 19th century,
both the settler states and the ethnic states — the European liberal democracies and Japan — had
relatively open immigration policies. Any difference in societal views of immigration, therefore, was
not based on the different policies. Additionally, if scholars had only examined 19th century and
early 20th century immigration policy, they would have come to the opposite conclusion as settler
states tended to restrict immigration more during the Interwar period than did the ethnic states.
Finally, today, the immigration policies of the settler states and the ethnic states look remarkably
similar, suggesting little residual difference based on their past histories.
The other societal-based theories offered by the literature are similarly inconsistent with the data.
The second version, based on the ethnic heterogeneity of states, can be dismissed by examining the
immigration policies of heterogenous states, such as the United States, Brazil and Singapore, against
homogeneous states, such as South Korea, Taiwan, Saudi Arabia and Kuwait. We see that recently
all seven states have had very similar immigration policies even though they vary in their level of
heterogeneity. Similarly, there is little difference in the immigration policies today of states that
have jus soli citizenship — as most of the settler states do — and states that have jus sanguinis
citizenship policies — as most of the rest of the world has. Again, if we examine immigration in
the Interwar period, we would have come to the opposite conclusion. Finally, there appears to be
few differences in policy among colonizers (the United States, United Kingdom, France, Germany,
Netherlands and Japan), non-colonizers (Switzerland)and former colonies (Argentina, Australia,
Brazil, Canada, Hong Kong, Kuwait, New Zealand, Saudi Arabia, South Africa, South Korea and
Taiwan). Instead, these states, regardless of colonial status, have had similar immigration policies
over time.
Table 6 examines interest group based hypotheses as well as the other hypotheses from the
model. Hypothesis 1 argued that increasing productivity should lead to decreased support for open
immigration by firms. I included three measures of changing productivity in models 1 through
3. One problem with studying productivity cross-nationally is that there is very limited data
on productivity, especially for earlier periods. Following Comin and Hobijn (2009), I use various
measures of technology adoption. Previous work shows that technology adoption is correlated
with increases in labor productivity and thus, with a reduced need for labor. Therefore, as the
need for labor decreases, immigration policy should become more restrictive. I use three different
measures, one from agriculture and two from industry, of technology adoption from the CHAT
dataset (Comin and Hobijn 2009) to test the effect of variation in productivity on immigration:
35
fertilizer use (standardize by the percentage of the economy in agriculture (the World Bank 2011),
railway mileage, and the amount of steel produced using a basic Bessemer.31 Due to data availability
and the use of these technologies, I use them to examine productivity in different time periods.
Railroads were a greater technological advancement in the 19th and early 20th centuries than they
were later and therefore, are used to measure technology adoption in the 19th and early 20th
centuries. Steel made from the basic Bessemer process is used to examine the role of productivity
in the Bretton Woods era. During this time period, the Bessemer process was becoming obsolete
due to newer technologies32 Unfortunately, we do not have good data on the adoption of these
other technologies during this time period. Instead, decreasing use of the Bessemer process should
correlate with increasing use of newer technologies after World War II until its obsolescence in the
late 1960s and early 1970s. Steel made with the Bessemer process during this time period should,
therefore, be positively correlated with immigration policy as it signals the use of a less productive
technology. Finally, data on fertilizer is only available from the early 1960s through 2000. The use of
fertilizer and other chemical inputs in agriculture reduce the amount of labor needed in agriculture
and should be correlated with productivity. As we see in models 5 through 7, increasing productivity
leads to a more restrictive immigration policy. Additionally, the inclusion of productivity does not
affect the results on trade and capital.
I also include net union density from Golden, Lange and Wallerstein (2009) in models 4 through
6 to measure the power that labor unions have to shed light on hypothesis 6 as well as alternative
arguments based on the power of organized labor (for example Briggs 1984, 2001). Net union
density is only available for OECD states; therefore, these models only examine the effect of the
variables on these states. The coefficient on net union density is signed as hypothesized, but is
not statistically significant. The results are similar if we examine only the Bretton Woods or the
post-Bretton Woods era as well.
The focus on labor is also found in arguments that use income inequality as a predictor of
immigration policy, such as Timmer and Williamson (1998). To test the role of inequality I examine
inequality using the Gini coefficient in model 3. There is not data on the Gini coefficient prior to
the 1970s. As a robustness check I have also included the top 10% income share and found similar
results.33 As can be seen in model 3, the coefficient on Gini is not statistically significant nor is
does it have the hypothesized sign. If rising inequality leads to increased support for immigrant
restrictions, as hypothesized by Timmer and Williamson (1998), we would expect the sign to be
negative.
To examine this hypothesis 4 and arguments about the role of the fiscal system, I use the Cusack
(2000b) measure of taxation for social spending in models 4 through 6 as this data is only available
for OECD states as well. Scholars argue that as the welfare state has increased in size, immigrants
31
There are several other measures on technology adoption in the CHAT dataset, but most of them have very
limited data coverage and are not used for that reason.
32
It was replaced by the blast oxygen process and electric arc furnace, both of which began to be used after World
War II.
33
The Gini coefficient is available for more states than the top 10% income share.
36
37
450
0.70
Observations
R2
71
0.96
-0.01
(0.01)
-0.29+
(0.09)
0.01
(0.02)
-2.91*
(0.68)
-469.26+
(137.58)
970.20+
(286.48)
5.60*
(1.24)
-501.05+
(149.04)
-0.03**
(0.00)
5.88*
(0.65)
0.03+
(0.01)
(2)
Bretton Woods
All States
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Constant
-0.02*
(0.01)
1.71
(1.44)
0.01
(0.01)
0.62
(0.48)
0.27
(0.22)
0.98
(1.74)
3.90*
(1.18)
-5.11**
(1.29)
-0.03**
(0.01)
Linear time trend
Trade2 *Capital
Trade policy2
Trade*Capital
Capital policy
Trade policy
War
GDP growth
Polity
Citizenship
Net union density
Welfare taxation
Gini
Fertilizer Used (Technology)
Steel produced (Technology)
Railroad km (Technology)
DV:Immigration policy
(1)
Pre-1946
All States
310
0.50
23.17*
(8.30)
-24.38
(53.78)
47.61
(112.84)
-32.97*
(11.61)
-23.04
(59.16)
-0.00
(0.01)
-15.91*
(6.02)
0.00
(0.00)
-0.47
(0.61)
-0.02+
(0.01)
0.01
(0.00)
(3)
Post-Bretton Woods
All States
360
0.80
-0.01
(0.02)
-0.89
(1.45)
-0.00
(0.08)
-10.75
(6.09)
16.63
(49.89)
-42.67
(106.70)
13.64
(8.73)
26.56
(56.98)
-0.02**
(0.01)
10.35*
(3.99)
-0.08**
(0.02)
-2.08
(1.14)
(4)
Post-1946
OECD
164
0.47
0.04*
(0.02)
-0.26
(0.73)
0.01
(0.09)
-4.69
(5.65)
-68.79***
(12.79)
137.26***
(26.80)
3.40
(7.96)
-68.16**
(14.35)
0.01
(0.01)
1.60
(2.39)
-0.03
(0.02)
-1.69
(2.34)
(5)
Bretton Woods
OECD
196
0.64
-0.22**
(0.05)
0.57
(1.20)
-0.05
(0.08)
-16.38
(61.06)
-110.68
(102.96)
230.20
(219.62)
10.70
(70.98)
-119.53
(116.98)
-0.02+
(0.01)
18.85
(43.45)
-0.03
(0.02)
-1.21
(0.97)
(6)
Post-Bretton Woods
OECD
Table 6: Immigration policy regressed on additional and alternative hypotheses and trade and capital policy
1786
0.87
3.10***
(0.74)
-2.10**
(0.70)
0.23**
(0.07)
(7)
All Years
All States
become more expensive as they are thought to use the social welfare system more than natives. As
immigrants place a greater burden on the fiscal system of the state, they create a backlash against
open immigration, leading to a more restrictive immigration policy. Therefore increased taxation for
social spending should lead to increased restrictions on immigration. Over the entire the post-World
War II period, increasing the size of the welfare state is correlated with restricting immigration at a
high level of significance. The inclusion of welfare spending makes the results on trade stronger as
well. For the OECD states, there is little correlation between trade and immigration if we do not
control for the size of the welfare state.
I also included Polity (Marshall, Gurr and Jaggers 2011) to measure regime type and the size
of the franchise to test hypothesis 5. I find little effect of regime type on immigration policy. To
control for economic crises, I include GDP growth and I also control for wars. Neither of these
variables affect immigration policy either.
Finally, model 7 examines the role that immigrant groups play in the policymaking process by
examining a slightly different dependent variable for all years of the data (similar to model 1 in
table 4). Immigrants are thought to want to keep immigration open so that their family members
or co-ethnics can enter the state and/or so that their place within society will be more secure. If
this is true, then when immigrants are more powerful, immigration policy should be more open.
Unfortunately, as with nativism, it is hard to explicitly test the strength of immigrants argument
as there is little comparative cross national data on immigrant strength. Nonetheless, we use
citizenship as a proxy and expect that immigrants should have relatively more influence in states
where citizenship is more easily obtained. Immigrants who can become citizens are likely to be
more important to policymakers than immigrants who cannot easily become citizens. To study the
relationship between immigration policy and citizenship, I recoded the immigration policy variable
to exclude citizenship and then regressed it on citizenship. As we can see in model 7, there is a
positive effect of citizenship policy on the over all immigration policy, as it is recoded, and there
is still an effect of trade and capital openness on immigration policy. This relationship could be a
product of the strength of immigrants due to their ability to become constituents of the policymaker;
on the other hand, it could be the case that policymakers, who are more supportive of immigration,
are also more supportive of liberal citizenship policies and vise versa. As a robustness check, I also
examined the time periods separately and found that there is little effect of citizenship after World
War II (table 24 in Appendix B). The relationship between trade, capital, and immigration remains
the same.
This section examined whether the additional hypotheses from the model and existing theories
from the literature on the causes of immigration policy are consistent with our data. We also
examined whether the inclusion of these variables elucidated puzzles we found in the relationship
between trade and immigration. There was some support for the additional hypotheses from the
model. Hypothesis 1 argued that increased use of labor saving technology leads firms to need less
immigrant labor and change their preferences over immigration; the data support this hypothesis.
Additionally, we found that as the non-labor market costs increase due to an increase in the size of
38
the social welfare system, immigration became more restricted. The inclusion of welfare spending
also made the relationship between trade and immigration stronger. Most of the examined theories
from the literature, however, were not supported by the data. I found that societal-based variables
received little support. I also found little support for interest group arguments based on the strength
of nativism, immigrants themselves, or labor. Further, these interest-based theories provided no help
in explaining the relationship between trade and immigration.
Conclusion
States’ immigration policies do not conform to the typical patterns that political analysts expect:
democracies have had more open and more restrictive immigration policies than autocracies; majoritarian governments have policies that are sometimes similar to and sometimes different from
consensus democracies; sometimes immigration policy responds to economic conditions; sometimes
immigration policy responds to increased nativist sentiment and so on. All these expectations rely
on variation in domestic factors to explain immigration policy, ignoring the international context in
which policy is made. This paper argues that international factors cannot be ignored: immigration
policy cannot be understood without considering the interactive effects of trade and capital policy
on a nation’s economy.
Trade, immigration and capital openness should be viewed as policy substitutes. Policymakers
can choose to open their economies by opening to people, money or goods; however, there is likely to
be a hierarchy among these choices. Open trade and capital is preferred in most situations to open
immigration since opening to foreign goods and capital is likely to engender a lower level of nativism
than opening to foreign people. Furthermore, people use the social welfare system, whereas goods
do not, increasing either the actual or perceived fiscal costs of immigration. Policymakers are likely
to close immigration when they can and use some other tool to open their economies.
If this is correct, why do we ever see any immigration openness? Open immigration has benefits
to firms. Business interests are typically powerful; therefore, when they demand openness, we
should expect policymakers to capitulate. This paper argues that opening trade and capital as
well as the use of labor saving technology, however, change firm preferences, making them less
supportive of open immigration. By opening trade, policymakers subject firms that use immigrant
labor to increased competition from nations with lower labor costs or from firms with greater
productivity. Once these vulnerable firms increase their productivity, move overseas or close their
doors, they no longer care about open immigration at home. In the long run, the policymaker loses
the support of these firms for open immigration and now finds that she can restrict immigration
without encouraging the wrath of firms that use immigrant labor.
In the short run, the effects of trade and capital openness are less clear due to cross-cutting
effects. Firms that are threatened by trade openness gain increased influence because policymakers
are concerned that they might move overseas or close, leading to fewer jobs and lower tax revenues.
In the short run, immigration could open because policymakers are willing to provide firms cheap
39
labor to keep them in business. Immigration could stay the same if the policymaker chose to provide
firms with tax cuts instead of increased immigration to keep these firms profitable. Immigration
would close, however, if the policymaker chose to let these firms close or move overseas instead of
subsidizing them. In the long run, however, immigration is unpopular and tax cuts are unsustainable. At this point, the mobile firms will move overseas and the immobile firms must become more
productive or close their doors. Regardless, the overall demand for immigration will drop, leading
anti-immigrant groups to have relatively more influence and immigration policy to become more
restrictive.
In order to test the relationship between trade, capital and immigration across nations, this
paper created a dataset that includes information on border regulations, enforcement policies and
immigrant rights laws in 19 representative countries. These laws were categorized into 12 dimensions
to capture the different ways in which states can control immigration. Using these different dimensions, this paper offered a variable that captures immigration policy in all of its forms, comparable
across countries and time.
The 19 states that were examined were categorized into four groupings: settler states, European
liberal democracies, export-oriented states and the autocracies. The states were grouped because of
their use of similar policies to control their borders. Even though these groups have used different
discrete policies to control their borders, over time, we see remarkable similarity in the composite
immigration policy. The 19th and early 20th centuries were a period in which states opened their
borders to a degree that has not been seen since.
How can we explain this pattern of immigration policy across states? I showed that this pattern
can be explained by examining these states’ trade and capital policies as well as the use of labor
saving technology. The data show a robust, enduring and powerful negative relationship between
trade and immigration. Trade and immigration are substitutes both economically and politically.
In the 19th century, when moving goods was relatively expensive given transportation costs and
when states protected their borders with tariffs, states opened their borders to the movement of
people. At the end of the 19th century and beginning of the 20th, when transportation costs fell
and the movement of goods increased, immigration policy became more restrictive. After World
War II, many states closed their capital markets, maintained some trade protection and reopened
their borders to migrants to a small degree. The level of protection against foreign goods during
this time was not as high as it had been in the 19th century and neither was immigration as open as
it was in the 19th century. As the world globalized goods and capital markets opened again in the
second half of the 20th century and the early 21st century, the restrictions on immigration increased
as well. In addition to increased globalization of world goods and capital markets, the 19th, 20th,
and 21st centuries have seen great increases in the use of labor saving technology. The data shows
that increased use of technology has allowed firms in these states to use less labor and spend less
political capital on immigration, leading to a greater restrictions on immigration as well.
In addition to examining the role that trade and capital have played in states immigration
policymaking, this paper examines existing theories of immigration policy. Using the 19 country
40
dataset, I find the size of the welfare state affects immigration policy, as predicted. Most of the
dominant theories of immigration policy, however, are not supported by the data.
In sum, this paper increases our understanding of immigration policy by examining the role
that international factors play in constructing domestic policy. Economists have long argued that
the movement of goods, people and money are substitutes; yet, political analysts have ignored
this argument when they have studied immigration policy. Instead, these scholars have focused
on domestic factors and have therefore missed the role that trade has played in the formation of
immigration policy. This paper brought these international factors back into focus by arguing that
trade and capital openness affect firm preferences over immigration and thus affect the way that
policymakers respond to firms. In addition, it provided new data on which to test both this argument
and alternative hypotheses that have rarely been tested using longitudinal cross-sectional data. I
found that there is a powerful, robust and enduring relationship between trade and immigration
policy in the cross-national data. This relationship is made even stronger with the inclusion of
capital policy. Thus, I conclude that we cannot examine immigration policy without considering
trade and capital policy.
41
References
Albarracín, J. 2003. Explaining Immigration Policies in Argentina during the 1990s: European
Immigration, ‘a Marriage in Sickness and in Health’. In XXIV Congreso Internacional de LASA.
LASA.
Avci, G. and C. McDonald. 2002. “Chipping away at the fortress: Unions, immigration and the
transnational labour market.” International Migration 38(2):191–213.
Bardo, M., B. Eichengreen, D. Klingebiel and M. Soledad Martinez-Peria. 2001. “Is the Crisis
Problem Growing More Severe?” Economic Policy 16(32).
Bhagwati, J. N. and K. Hamada. 1974. “The Brain Drain, International Integration of Markets for
Professional and Unemployment: A Theoretical Analysis.” Journal of Development Economics
1:19–42.
Briggs, V. 2001. Immigration and American Unionism. Cornell, NY: Cornell University Press.
Briggs, V. M. 1984. Immigration policy and the American labor force. Baltimore: Johns Hopkins
University Press.
Castro, M. J. 1999. Toward a new nativism? The immigration debate in the United States and
its implications for Latim America and the Caribbean. In Free Markets, Open Societies, Closed
Borders? Trends in International Migration and Immigraiton Policy in the Americas. Miami:
North-South Center Press.
Chinn, M. D and H. Ito. 2008. “A new measure of financial openness.” Journal of Comparative
Policy Analysis: Research and Practice 10(3):309–322.
Clemens, M. 2007. “Do Visas Kill? Health Effects of African Health Professional Emigration.”
Center for Global Development Working Paper No. 114.
Clemens, M. A. and J. G. Williamson. 2004. “Why did the Tariff-Growth Correlation Change after
1950?” Journal of Economic Growth 9(1):5–46.
Comin, D. A. and B. Hobijn. 2009. The CHAT Dataset. Working Paper 15319 National Bureau of
Economic Research.
Coppedge, M. 1997. “A Classification of Latin American Political Parties.” Kellogg Institute Working
Paper 244.
Cusack, T. R. 2000a. “Center of Political Gravity Dataset.”.
URL: http://www2000.wzb.eu/alt/ism/people/misc/cusack/ds ets.en.htm
Cusack, T. R. 2000b. “General Government Taxation Dataset.”.
URL: http://www2000.wzb.eu/alt/ism/people/misc/cusack/ds ets.en.htm
42
Ernst, R. 1948. “Economic Nativism in New York City During the 1840s.” New York History
46:170–186.
Faist, T. 1994. “How to define a foreigner? The symbolic politics of immigration in German partisan
discourse, 1978-1992.” West European Politics 17(2):50–71.
Feenstra, R. C. and G. H. Hanson. 1996. “Globalization, Outsourcing, and Wage Inequality.” The
American Economic Review 86(2):240–245.
Freeman, G. P. 1995. “Modes of Immigration Politics in Liberal Democratic States.” International
Migration Review 29(4):881–902.
Golden, M., P. Lange and M. Wallerstein. 2009. “Union Centralization among Advanced Industrial
Societies: An Empirical Study.”.
URL: http://hdl.handle.net/1902.1/10193 UNF:3:RvCaQbChZc7ffAtCuUlj0g== Miriam Golden
[Distributor]
Grossman, G. M. and E. Helpman. 1994. “Protection for Sale.” The American Economic Review
84(4):833–850.
Hansen, R. 2002. “Globalization, Embedded Realism, and Path Dependence: The Other Immigrants
to Europe.” Comparative Political Studies 35(3):259–283.
Hanson, G. H., K. Scheve and M. J. Slaughter. 2007. “Public Finance and Individual Preferences
Over Globalization Strategies.” Economics & Politics 19(1):1–33.
Hatton, T. J. and J. G. Williamson. 1998. The Age of Mass Migration: Causes and Economic
Impact. New York: Oxford University Press.
Hatton, T. J. and J. G. Williamson. 2005a. “A Dual Policy Paradox: Why Have Trade and Immigration Policies Always Differed in Labor-Scarce Economies?” National Bureau of Economic
Research Working Paper 11866.
Hatton, T. J. and J. G. Williamson. 2005b. Global Migration and the World Economy. Cambridge,
MA: The MIT Press.
Hollifield, J. F. 2004a. “The Emerging Migration State.” International Migration Review 38(3):885–
912.
Hollifield, J. F. 2004b. France: Republicanism and the Limits of Immigration Control. In Controlling
Immigration: A Global Perspective. Stanford, CA: Stanford University Press pp. 183–214.
Kay, D. and R. Miles. 1988. “Refugees or Migrant Workers? The Case of the European Volunteer
Workers in Britain (1946-1951).” Journal of Refugee Studies 1(3-4):214.
Kaye, R. 1994. “Defining the agenda: British refugee policy and the role of parties.” Journal of
Refugee Studies 7(2-3):144.
43
Kremer, M. 2006. “Globalization of Labor Markets and Inequality.” Brookings Trade Forum pp. 211–
228.
Krugman, P. 1980. “Scale economies, product differentiation, and the pattern of trade.” The American Economic Review 70(5):950–959.
Kuhn, P. J. and C. McAusland. 2006. “The International Migration of Knowledge Workers: When
is Brain Drain Beneficial?” National Bureau of Economic Research Working Paper 12761.
Leblang, D. A., J. Fitzgerald and J. Teets. 2009. “Defying the law of gravity: The political economy
of international migration.” University of Virginia.
Marshall, M. G., T. R. Gurr and K. Jaggers. 2011. “Polity IV Project: Political Regime Characteristics and Transistions, 1800-2009.”.
URL: www.systemicpeace.org/polity/polity4.htm
Massey, D. S., J. Arango, G. Hugo, A. Kouaouci, A. Pellegrino and J. E. Taylor. 1993. “Theories of International Migration: A Review and Appraisal.” Population and Development Review
19(3):431–466.
Melitz, M. J. 2003. “The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry
Productivity.” Econometrica 71(6):1695–1725.
Messina, A. M. 2008. “The logics and politics of post-WWII migration to Western Europe.” West
European Politics 31(5):1096–1097.
Meyers, E. 2000. “Theories of International Immigration Policy - A Comparative Analysis.” International Migration Review 34(4):1245–1282.
Money, J. 1999. Fences and Neighbors: The Political Geography of Immigration Control. Ithaca:
Cornell University Press.
Neuman, G. L. 1993. “The Lost Century of American Immigration Law (1776-1875).” Columbia
Law Review 93(8):1833–1901.
Office of Immigration Statistics. 2006. 2005 Yearbook of Immigration Statistcs. Washington DC:
Department of Homeland Security.
Olzak, S. 1989. “Labor Unrest, Immigration, and Ethnic Conflict in Urban America, 1880-1914.”
The American Journal of Sociology 94(6):1303–1333.
Quinn, D. and A. M. Toyoda. 2006. “Does Capital Account Liberalization Lead to Growth?” Review
of Financial Studies 21(3):1403–1449.
Ratha, Dilip and Zhimei Xu. 2008. Migration and Remittances Factbook 2008. Washington, D.C.:
The World Bank.
44
Reisler, M. 1976. By the sweat of their brow: Mexican immigrant labor in the United States, 19001940. Westport, CT: Greenwood, Press.
Sassen, S. 1988. The Mobility of Labor and Capital: A study in international investment and labor
flow. Cambridge: Cambridge University Press.
Stark, O. 2004. “Rethinking the Brain Drain.” World Development 32(1):15–22.
Tichenor, D. J. 1994. “The Politics of Immigration Reform in the United States, 1981-1990.” Polity
26(3):333–362.
Tichenor, D. J. 2002. Dividing lines: The politics of immigration control in America. Princeton,
NJ: Princeton University Press.
Timmer, A. S. and J. G. Williamson. 1998. “Immigration Policy Prior to the 1930s: Labor Markets,
Policy Interactions, and Globalization Backlash.” Population and Development Review 24(4):739–
771.
World Bank. 2011. “World Development Indicators.”.
URL: http://data.worldbank.org/
Zavodny, M. 1997. “Welfare and the locational choices of new immigrants.” Economic Review-Federal
Reserve Bank of Dallas pp. 2–10.
Zolberg, A. R. 1989. “The Next Waves: Migration Theory for a Changing World.” International
Migration Review 23(3):403–430.
Zolberg, A. R. 2006. A Nation by Design: Immigration Policy in the Fashioning of America.
Cambridge, MA: Harvard University Press.
45
Appendix A: Coding Scheme for Cross-National Data
Federal states: All federal states are coded according to the policy of the most open member of
the state until the time when the federal government takes sole responsibility for immigration policy.
This coding scheme is used because most federal states allow the free movement of persons among
the members of the federation. Therefore, an immigrant who can come to one of the members can
then have access to all of the members of the federation. Due to this coding, Switzerland is not
included in the dataset prior to 1848 because citizens of the different cantons did not have the right
to live in another canton. Among EU members, the immigration policies of the most liberal country
is not coded as the policy for these countries because freedom of movement does not extend to third
country nationals. When EU policy does affect some or all of the nations — either because they
implement the regulations with national legislation, such as the Blue Card program for high-skill
workers, or have delegated control to the EU as in Schengen — the EU policy is coded as the policy
for each nation it affects.
Universality by Nationality: How selective is the state about letting immigrants in based on
their national origin? Does nationality matter at all? Are there few national groups or many allowed
in? A score of 1 represents that few or no nationalities are allowed in. A score of 5 represents that
all nationalities are treated equally. One issue that this brings up is that in the late 20th and early
21st centuries, states often gave some groups preferential access to their labor market while having
an overall policy of equality. For example, New Zealand uses a point system with no national origin
criteria but also has a special program with the island nations of the South Pacific for seasonal
workers. In this case, these preferential access programs - because they are almost always for lowskill workers - are coded in the universality by skill category. This coding rule is used because the
policy is to increase, and not deny, access to the state.
1 Few or no nationalities allowed in.
2 More than a few but not many. Example: if a European country only allowed immigrants from
other EU countries.
3 Many nationalities allowed in but not all. Example: Between 1924-1965, the US quota system
allowed in many Northern Europeans, some Southern and Eastern Europeans, anyone from
the Western Hemisphere and no one from Asia.
4 Almost all nationalities allowed in. Example: In the late 19th century, only Chinese were excluded
from the US. Additionally, numerical limits by country but not differentiated by country.
Example: Current US law restricts migration from each country to 20,000.
5 No exclusions based on nationality
Universality by Skill or Income: Does the state restrict by the skills or income an immigrant
possesses? Does it use a point system with points given for education or special skills? Are people excluded based on profession (i.e., no prostitutes), illness (e.g., no epileptics), or likelihood of
46
becoming a public charge? A score of 5 on this scale indicates that the country has no restrictions
by skill and a score of 1 means only the very highest skilled workers (executives, high-level intracompany transfers) are allowed in. Again, when states exempt one group from these restrictions either by nationality in the case of the New Zealand seasonal workers program for Polynesians or
by a general category like the seasonal agricultural workers program in Britain - the score increases.
1 Only highly educated, high income earners allowed in; many excludable classes.
2 Mostly high educated, high earners, but some allowances for low-skilled workers; some excludable
classes.
3 Preference for high-skill workers but many opportunities for low-skilled workers; some excludable
classes.
4 Few slots reserved for high-skill workers (i.e like the H1B visa in the US); most visas open for
anyone; few excludable classes (e.g., only criminals, those likely to become a public charge).
5 No skill restrictions for any visas; no excludable classes.
Citizenship: How easy is it to obtain citizenship? What determines citizenship for children born in
the country (jus sanguinis, jus soli, double jus soli)? Are there racial discriminations in citizenship?
How easy is it for the government to denaturalize citizens? A score of 1 denotes states where
citizenship is only given through birth through one parent (usually the father). A score of 5 denotes
jus soli citizenship (citizenship given to all children born in the state) and an easy naturalization
process. Racial discrimination in citizenship policies leads to a lower score as well.
1 Only by birth from a native father or mother.
2 Only by birth through either native parent and/or grandparent.
3 Very difficult process to obtain citizenship (language requirements, difficult test) and/or many
years to citizenship (more than 10 years) and/or children receive citizenship through either
parent or grandparent.
4 Moderately difficult process (relatively easy language requirements and/or an easy test) and/or
moderate time to citizenship (more than 5 but less than ten years) and/or children born in
state automatically get citizenship.
5 Fairly easy process (e.g., no language requirements) and short time to citizenship (5 or less years)
and children born in state automatically get citizenship.
Immigrant Rights: What rights do immigrants have once in the state? Are there racial/national
origin discriminations? Does the government try to integrate immigrants or does it just expect
them to assimilate? How easy is it to get permanent residency? A score of 1 indicates few legal
rights: immigrants had to be registered; they had to go through invasive health checks; they do not
47
have the right to marry nationals; they could only live in specific locations; they could only work
for specific employers; they have no access to the welfare state; they cannot own land; they are
discriminated against and they cannot gain permanent residency. In states coded as 1, immigrants
can basically only work the job in which they were hired for and cannot leave the housing provided
for them by their employer. A score of 5 indicates parity to citizens: complete access to the welfare
state; voting rights; no restrictions in where they can live or work; no restrictions in property rights
and a robust anti-discrimination program.
1 Almost no legal rights; immigrants must leave state if they leave their job; cannot own property;
cannot access the welfare state; they have to register, no freedom of religion, no permanent
residency, etc.
2 Some rights but land ownership and ownership of companies restricted; limited access to the
welfare state.
3 Ability to change jobs freely, some ownership of real property or companies; some access to the
welfare state, some racial discrimination in laws.
4 Access to most welfare policies; few restrictions on ownership of property or firms.
5 Total access to welfare state, voting rights without citizenship, no restrictions in property ownership, integration policies, no racial discrimination, few years to permanent residency.
Refugee: Does the state have a resettlement policy or does it just resettle refugees on an ad hoc
basis? How selective is their refugee policy? Do they let in many refugees? Are refugees only defined
as those who meet the 1951 Convention or 1967 Protocol or is there are more expansive definition?
Refugee policy is coded as a 1 if the country has no special policy and a 5 if the country is willing to
resettle large numbers of refugees without taking into consideration the refugees’ qualifications. This
last criteria is to distinguish the more generous refugee policies of the current day with those after
World War II when most receiving countries placed occupational restrictions on refugees, selecting
for higher skilled migrants. Ad hoc refugee programs for one group during the crises are coded as
relaxing refugee restrictions and the magnitude of the change is based on the number of refugees
the state was willing to allow in. The change in coding only lasts as long as the refugee program
was in place; for example, when New Zealand took in Ugandan refugees in 1973, but no other years,
the increase in the refugee score is only calculated for 1973.
1 Almost no refugees allowed in; those that are allowed in must follow normal immigration procedures.
2 Some refugees allowed in; special refugee visas but refugees chosen by some sort of preference or
must be able to pass tests that non-refugee immigrants take; few reasons for being a refugee
or ad hoc policy.
48
3 Special refugee visa, preference system but not overly burdensome; moderate number of refugees
allowed in; must follow some of the requirements that a non-refugee immigrant would have to
pass; the UN definition of a refugee is followed.
4 Large number of refugees allowed in; no preference system or very weak system; easy to obtain
refugee visa; exemption from requirements of non-refugee immigrant; at least the UN definition
of a refugee is followed.
5 Large number of refugees; no preference system or requirements; very easy to obtain refugee visa;
many categories of refugees included not just the UN definition.
Refugee Provisions: Coded 0 before first mention of refugee; 1 after.
Asylum: How easy is it to gain asylum? What rights do asylum seekers and asylees have? Are
they kept in detention centers? Are they repatriated? Is there only one asylum status or is there
temporary protected status as well? What are the procedures and are there legal safeguards?
1 No asylum.
2 Extremely difficult process; asylum granted only in a few cases; little ability to work or access to
welfare state while awaiting determination; little recourse if not granted asylum; no temporary
protected status; limited access for political refugees.
3 Difficult process; asylum granted for more cases; some access to the welfare state or labor market,
more recourse including ability to access courts if denied; some temporary protected status
allowed.
4 Fairly easy process; asylum granted to many groups; access to labor market and welfare system;
access to courts and other procedures if denied; temporary protected status given to many
groups.
5 Easy process; asylum granted for most cases; access to labor markets and welfare state; constitutionally protected procedure; no need for temporary protected status because almost everyone
gets asylum.
Asylum Provisions: Coded 0 before first mention of asylum; 1 after.
Recruitment: Are there special visas or procedures to recruit labor or settlers? To recruit workers,
do employers have to advertise first or otherwise seek approval from a government ministry? How
many industries can recruit? Do firms have to pay levies or other taxes for foreign workers? Does
the government pay for passage or give settlers or workers other benefits to induce them to come?
A score of 1 denotes that all workers have to follow the same requirements as all other immigrants
and that firms cannot recruit from overseas. An example of this is the US Contract Labor Law. A
5 denotes that the government will pay for passage of any immigrant and will give the immigrant
money, land, or other goods to help him to settled.
49
1 No special procedure or visa, come in under the same system of regulation as everyone else; labor
recruitment prohibited.
2 Small set of visas for special groups of workers (i.e. agricultural workers); trigger to reduce
numbers based on employment data; employers are not allowed to pay for moving expenses;
many restrictions including no unemployed natives in the industry.
3 Moderate number of visas for all groups or many groups obtain visas; employers allowed to pay
for moving expenses; some procedures for recruiting workers.
4 Few or no restrictions on visas for any type of worker, employers are allowed to pay moving
expenses; few restrictions or procedures for obtaining work visas.
5 Government program to recruit workers or settlers, government pays for the workersÕ transportation cost and helps pay for firms or government officials to recruit workers.
Labor prohibitions: How many occupations can the immigrant work in? Are there requirements
to have a certain number of native workers in an occupation/firm or that foreign workers can only
make up a certain percentage of workers? How many occupations do the rules cover? All? Just
certain industries? Are there racially based policies? A score of 1 means that immigrants are not
allowed to work in any industry. This is not the case for any of the states in this sample. A score
of 5 means that there are no restrictions or in modern times, that the only restrictions are in highly
sensitive national security positions.
1 Immigrants completely blocked from the labor market.
2 Immigrants restricted from many occupations; less than 30% of the workers in a given occupation/firm can be immigrants (covering most or all of occupations).
3 Immigrants restricted from some occupations; 30-50% of workers in given occupation/firm can
be immigrants (covers some occupations).
4 Immigrants cannot hold public sector positions; 50% or more of the workers in a given occupation/firm can be immigrants (covers some occupations).
5 Immigrants can hold any position (except for highly sensitive national security positions); no
restrictions on the number of immigrant workers in a given occupation/firm.
Deportation: How easy is it to deport an immigrant? What safeguards exist? Does the state
engage in mass expulsions or pay people to leave the country? A score of 1 denotes that there are
many deportable offenses, including losing one’s job and there are few administrative or judicial
safeguards. A score of 5 is given if there are few deportable offenses (usually deportation is limited
to criminals) and/or clear judicial checks.
1 No appeals process; many deportable offenses, including losing oneÕs job.
50
2 Administrative process with few checks; fewer deportable offenses.
3 More checks on the process and even fewer deportable offenses.
4 Judicial checks on process including going to the highest court in the land and/or very few
deportable offenses.
5 Almost no deportable offenses (conviction for a criminal offense, but not for an immigration
offense) and clear judicial checks.
Enforcement: How strongly does the state enforce its borders? Are there employer sanctions,
fines or prison time for illegal immigrants? Are there amnesties? During an amnesty are immigrants
allowed to stay or just leave without paying a fine? A score of 1 denotes a high spending country,
with severe employer sanctions, sanctions on those who are in the country legally including fines
and prison time, bonds to ensure that immigrants leave and identification papers that are hard to
forge. A score of 5 denotes no enforcement beyond basic police enforcement.
1 High spending, employer raids or hard to forge national work idÕs, strong employer sanctions,
bonds placed by employers to ensure that migrants go home, large number of enforcement
officials.
2 Slightly less spending, fewer raids or easier to forge national work id, border enforcement is strong
but not impossible to over come.
3 Even less money, no raids, easy to forge idÕs, some border enforcement.
4 Very little enforcement, screening at points of entry, little enforcement on employers.
5 Basically no enforcement.
Family: Do family members get special treatment? Can they immigrate more easily than others?
Are there racial or skill distinctions? A score of 1 indicates that no family members are given special
treatment and a score of 5 indicates that many family members are given special treatment. Most
states fall somewhere between a 2 - special treatment for wives and minor children only and a 4 wives and minor children and sometimes parents can enter without difficulty and all other relatives
can be sponsored with some occupational or skill requirements. One issue with family migration is
that states did not seem to consider it a necessary policy to have when there were few restrictions
by nationality or skill. Family reunification policies only came into being once other restrictions
were put in place. Given that the states have no policy on family migration during these times,
these years are scored as a 1.
1 No special provisions for family reunification; family members must enter under the same procedures as others.
51
2 Only wives and minor children of citizens or legal permanent residents can be sponsored, but are
free from other controls.
3 Increased number of relatives can be sponsored but only by citizens and/or relatives (except minor
children and wives) need to possess same characteristics as non-family immigration (i.e., if
there is a literacy test, relatives must pass the test); relative in the country has to pay bond
or otherwise be responsible.
4 Many categories of relatives can be sponsored by citizens or residents but still must possess same
characteristics as non-family immigrants (except minor children and wives); relative in the
country has to be responsible for immigrant.
5 Many categories of relatives can be sponsored by citizens or residents and they do not need to
possess the characteristics of non-family immigrants (exemption from literacy exams, etc.);
no bond required or responsibility for relative in the country.
Family Provisions: Coded 0 before first mention of special provisions for families; 1 after.
Quota: Is there a quota and how restrictive is it? Quotas are only coded when the quota is
a numerical limit on a large portion of immigrants, not when it is a target for the number of
immigrants. Targets, like policy statements or development plans, are not coded because they are
not changes in legislation but usually administrative policies. The quota does not need to be binding
on all immigrants. This is because it is rare to have a quota that binds on all immigrants. Usually
at least wives and minor children of citizens are allowed in above the quota; this policy is denoted in
the family immigration policy coding. Sometimes, the quota is only on one class of immigrants, such
as the Hong Kong quota on Chinese immigrants, but this class makes up the majority of immigrants
entering the country. Again, high-skill workers from other countries could enter above the quota;
although, interestingly, wives and minor children of Hong Kong belongers (equivalent to citizens)
cannot. This is denoted in the other categories.
1 Less than 0.25% of population can enter annually
2 0.25-0.5% of population can enter annually
3 0.5-1% of the population can enter annually
4 Over 1% of population can enter annually
5 No quota
52
Appendix B: Additional Regression Tables
Table 7: Immigration policy regressed on trade policy for South Africa
DV: Immigration
policy
Trade policy
Linear time trend
Linear term only
Square term only
Both terms
During Apartheid
After Apartheid
-3.50***
(0.58)
0.02***
(0.00)
0.38
(4.10)
-0.04***
(0.01)
1.68
(1.18)
0.00
(0.00)
-4.17*
(2.01)
-0.93*
(0.39)
-10.23
(31.96)
0.00
(0.00)
7.70
(37.78)
6.45
(23.14)
-1.06*
(0.45)
6.76
(5.47)
56
0.08
56
0.08
56
0.08
41
0.72
15
0.74
-3.61*
(1.66)
0.00
(0.00)
Trade policy2
Constant
Observations
R2
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
53
54
136
0.79
-0.02***
(0.00)
3.72***
(0.39)
-2.19***
(0.60)
136
0.79
-0.02***
(0.00)
2.28***
(0.11)
-2.90***
(0.82)
Square term only
136
0.79
-0.02***
(0.00)
7.36**
(2.35)
-7.72*
(3.64)
7.97
(5.03)
Both terms
119
0.91
6.32***
(0.61)
-6.42***
(0.71)
-0.00
(0.00)
-0.05
(0.27)
-0.60+
(0.31)
Interacted with Capital
136
0.79
-0.02***
(0.00)
3.72***
(0.39)
-2.19***
(0.60)
Pre-1914
Table 8: Immigration policy regressed on trade policy for Argentina
Linear term only
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade*Capital
Capital policy
Trade policy2
Trade policy
DV: Immigration policy
41
0.87
-0.05***
(0.00)
4.28***
(0.48)
1.54**
(0.50)
1914-1954
31
0.43
0.00+
(0.00)
-0.96+
(0.50)
-0.64*
(0.29)
1955-1985
20
0.37
0.01
(0.01)
-4.02**
(1.18)
1.54
(1.29)
Post-1985
55
144
0.91
-0.02***
(0.00)
-0.14
(0.20)
3.77***
(0.24)
144
0.90
-0.02***
(0.00)
2.42***
(0.07)
5.42***
(0.33)
Square term only
144
0.91
-0.02***
(0.00)
-1.66
(2.14)
6.01+
(3.15)
-3.28
(4.54)
Both terms
116
0.88
-0.02***
(0.00)
0.13
(0.50)
-1.25*
(0.53)
1.27*
(0.59)
4.02***
(0.63)
Interacted with Capital
116
0.89
20.15***
(4.92)
-21.47**
(6.51)
9.69
(10.79)
-23.25
(25.10)
13.72
(14.48)
-0.02***
(0.00)
-11.04**
(3.36)
Interacted with capital
Table 9: Immigration policy regressed on trade policy for Australia
Linear term only
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade2 *Capital
Trade*Capital
Capital policy
Trade policy2
Trade policy
DV: Immigration policy
73
0.92
-0.03***
(0.00)
3.20***
(0.78)
1.04
(0.66)
Pre-1938
71
0.62
-0.01**
(0.00)
-17.53***
(3.42)
27.95***
(4.74)
-36.59***
(6.96)
1938-2009
56
131
0.69
-0.01***
(0.00)
4.29***
(0.25)
-3.68***
(0.61)
131
0.69
-0.01***
(0.00)
1.97***
(0.19)
-7.22***
(1.33)
Square term only
131
0.69
-0.01***
(0.00)
3.75**
(1.13)
-2.88
(1.81)
-1.85
(3.95)
Both terms
116
0.89
1.50
(1.15)
0.87
(1.31)
0.01***
(0.00)
-0.33
(1.11)
-2.30+
(1.32)
Interacted with Capital
131
0.69
-0.01***
(0.00)
4.29***
(0.25)
-3.68***
(0.61)
Pre-1914
Table 10: Immigration policy regressed on trade policy for Brazil
Linear term only
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade*Capital
Capital policy
Trade policy2
Trade policy
DV: Immigration policy
31
0.83
-0.12***
(0.01)
16.15***
(1.87)
-2.67+
(1.57)
1914-1944
28
0.50
-0.01*
(0.00)
-5.59***
(1.45)
6.63***
(1.29)
1945-1973
28
0.90
-0.01***
(0.00)
-23.78***
(6.16)
36.80***
(8.63)
-47.17***
(10.71)
Post-1973
57
142
0.78
-0.02***
(0.00)
2.33*
(0.92)
0.98
(1.28)
Linear term only
142
0.78
2.60
(1.70)
-0.02***
(0.00)
3.01***
(0.09)
Square term only
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade policy2
Trade policy
DV: Immigration policy
142
0.81
-42.65***
(7.78)
56.91***
(10.44)
-0.03***
(0.00)
32.66***
(5.42)
Both terms
35
0.71
-0.01***
(0.00)
1.53***
(0.26)
0.04
(0.25)
Pre-1900
13
0.96
-0.05***
(0.00)
14.41**
(3.19)
-8.40*
(3.46)
1900-1914
Table 11: Immigration policy regressed on trade policy for Canada
15
0.89
-0.07***
(0.01)
8.26***
(1.43)
1.33
(1.94)
1914-1930
79
0.53
116.04***
(15.13)
-158.22***
(19.09)
0.03***
(0.01)
-85.43***
(10.92)
Post-1929
58
140
0.84
-0.01***
(0.00)
1.91***
(0.30)
-0.54
(0.41)
140
0.84
-0.01***
(0.00)
1.53***
(0.04)
-0.50
(0.56)
Square term only
140
0.85
-0.01***
(0.00)
7.73***
(1.67)
-9.11***
(2.49)
11.94**
(3.59)
Both terms
58
0.84
0.31
(2.30)
-0.65
(2.38)
-0.00
(0.00)
2.05***
(0.38)
-2.01**
(0.67)
Interacted with Capital
81
0.94
-0.02***
(0.00)
3.99***
(0.21)
-2.45***
(0.26)
Pre-1950
Table 12: Immigration policy regressed on trade policy for New Zealand
Linear term only
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade*Capital
Capital policy
Trade policy2
Trade policy
DV: Immigration policy
35
0.74
-0.00
(0.00)
-21.22***
(5.63)
31.26***
(7.92)
-42.38***
(10.35)
1950-1986
24
0.21
0.01+
(0.00)
1.91
(2.37)
-4.07
(2.64)
Post-1986
59
220
0.89
-0.02***
(0.00)
3.14***
(0.24)
-1.75***
(0.37)
Linear term only
220
0.89
-3.29***
(0.71)
-0.02***
(0.00)
2.04***
(0.06)
Square term only
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade policy2
Trade policy
DV: Immigration policy
220
0.89
-1.31
(0.91)
-0.95
(1.76)
-0.02***
(0.00)
2.86***
(0.58)
Both terms
124
0.58
-0.01***
(0.00)
2.13***
(0.26)
-0.60+
(0.35)
Pre-1914
139
0.70
-0.02***
(0.00)
3.01***
(0.30)
-1.63***
(0.40)
Pre-1929
Table 13: Immigration policy regressed on trade policy for the US
19
0.31
-0.00
(0.00)
-2.97***
(0.66)
1.91**
(0.65)
1929-1948
62
0.68
0.00
(0.00)
9.10***
(2.02)
-11.44***
(2.72)
Post-1948
60
145
0.92
-0.02***
(0.00)
2.35***
(0.37)
0.11
(0.41)
145
0.92
-0.02***
(0.00)
2.45***
(0.09)
-0.01
(0.51)
Square term only
145
0.93
-0.02***
(0.00)
-18.03*
(7.38)
28.94**
(10.36)
-35.30**
(12.36)
Both terms
118
0.92
-2.50*
(1.19)
2.44+
(1.25)
-0.02***
(0.00)
3.18***
(0.74)
-0.19
(0.82)
Interacted with Capital
73
0.95
-0.02***
(0.00)
-18.55***
(3.74)
29.71***
(5.25)
-35.55***
(6.39)
Pre-1938
Table 14: Immigration policy regressed on trade policy for France
Linear term only
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade*Capital
Capital policy
Trade policy2
Trade policy
DV: Immigration policy
12
0.47
0.01***
(0.00)
-1.74**
(0.46)
-0.78**
(0.19)
1938-1949
43
0.87
-0.04***
(0.00)
-64.24**
(19.54)
98.61***
(27.37)
-117.33**
(33.62)
1950-1992
17
0.51
0.04*
(0.02)
-471.82*
(190.12)
627.63*
(255.28)
-667.85*
(265.14)
1993-2010
61
124
0.44
-0.01***
(0.00)
2.84***
(0.50)
-3.59***
(0.59)
124
0.48
-0.01***
(0.00)
0.56***
(0.14)
-5.89***
(0.89)
Square term only
124
0.63
-0.00
(0.00)
-19.89***
(3.07)
30.81***
(4.67)
-50.84***
(7.16)
Both terms
113
0.72
20.35***
(3.14)
-21.64***
(3.31)
0.00+
(0.00)
0.71*
(0.31)
-1.92***
(0.39)
Interacted with Capital
19
0.38
-0.13
(0.09)
28.50
(21.96)
-30.26
(22.32)
Pre-1890
Table 15: Immigration policy regressed on trade policy for Germany
Linear term only
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade*Capital
Capital policy
Trade policy2
Trade policy
DV: Immigration policy
31
0.36
-0.01***
(0.00)
2.74+
(1.46)
-3.03+
(1.60)
1890-1931
13
0.78
-0.04***
(0.00)
2.36***
(0.22)
-0.46
(0.34)
1931-1945
41
0.68
-0.02+
(0.01)
-186.61+
(97.17)
258.70+
(134.70)
-285.44+
(155.76)
1945-1990
20
0.79
-0.02
(0.02)
10.37
(7.60)
-10.23
(9.62)
1991-2010
62
57
0.92
-0.03***
(0.00)
-14.67***
(3.07)
19.61***
(3.23)
Linear term only
57
0.92
-0.03***
(0.00)
-0.20
(0.67)
20.64***
(3.22)
Square term only
57
0.92
-0.03***
(0.00)
56.69
(104.99)
-77.04
(141.88)
101.57
(148.19)
Both terms
57
0.97
18.63***
(4.80)
-19.80***
(5.00)
-0.01*
(0.00)
-17.22***
(2.01)
18.72***
(1.59)
Interacted with Capital
Table 16: Immigration policy regressed on trade policy for Netherlands
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade*Capital
Capital policy
Trade policy2
Trade policy
DV: Immigration policy
21
0.85
-0.02***
(0.00)
-629.98*
(248.11)
859.99*
(339.63)
-910.85*
(366.84)
1953-1973
36
0.66
0.00
(0.01)
9.79*
(3.86)
-11.47*
(4.87)
1974-2009
63
59
0.63
-0.01***
(0.00)
0.07
(0.44)
0.15
(0.71)
Linear term only
59
0.63
-0.01***
(0.00)
0.16
(0.12)
-0.25
(0.81)
Square term only
59
0.79
-0.01*
(0.00)
-47.43***
(8.24)
65.99***
(11.36)
-75.82***
(13.29)
Both terms
32
0.97
-4.33**
(1.48)
4.70**
(1.64)
-0.01**
(0.00)
4.13*
(1.76)
-4.63*
(1.80)
Interacted with Capital
Table 17: Immigration policy regressed on trade policy for Switzerland
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade*Capital
Capital policy
Trade policy2
Trade policy
DV: Immigration policy
15
0.29
0.00+
(0.00)
-1.52
(1.01)
0.30
(0.99)
1951-1966
44
0.83
-0.01***
(0.00)
-50.20***
(8.81)
70.39***
(12.14)
-78.16***
(13.85)
1966-2009
64
142
0.90
-0.01***
(0.00)
1.39***
(0.23)
0.80**
(0.25)
142
0.90
-0.01***
(0.00)
1.89***
(0.07)
1.41***
(0.36)
Square term only
142
0.93
-0.02***
(0.00)
9.43***
(1.16)
-11.34***
(1.71)
18.37***
(2.43)
Both terms
115
0.93
-7.01***
(0.70)
7.36***
(0.73)
-0.02***
(0.00)
2.31***
(0.24)
0.29
(0.27)
With Capital
40
0.73
-0.00
(0.00)
4.84***
(0.60)
-4.09***
(0.66)
Pre-1905
14
0.78
-0.04***
(0.00)
-0.90
(3.45)
6.38+
(3.29)
1905-1918
Table 18: Immigration policy regressed on trade policy for the UK
Linear term only
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade*Capital
Capital policy
Trade policy2
Trade policy
DV: Immigration policy
19
0.83
-0.01***
(0.00)
1.81**
(0.50)
-0.11
(0.14)
1919-1938
38
0.53
-0.03***
(0.01)
3.10***
(0.59)
0.88+
(0.52)
1938-1979
31
0.20
-0.01*
(0.00)
-9.23*
(3.72)
10.78*
(4.56)
1979-2010
31
0.33
-0.00
(0.00)
353.84*
(158.74)
-478.39*
(214.07)
500.21*
(219.50)
1979-2010
65
130
0.79
-0.01***
(0.00)
-0.60
(1.12)
0.89
(1.21)
130
0.79
-0.01***
(0.00)
0.05
(0.25)
0.94
(1.37)
Square term only
130
0.80
-0.01***
(0.00)
-41.62
(26.25)
57.72
(36.46)
-63.96
(41.36)
Both terms
112
0.81
-9.65**
(3.08)
10.12**
(3.25)
-0.02***
(0.00)
4.85
(3.05)
-4.60
(3.24)
With Capital
79
0.37
-0.00***
(0.00)
-2.96***
(0.79)
3.12***
(0.83)
Pre-1952
51
0.15
0.00**
(0.00)
-0.01
(0.59)
-1.80*
(0.70)
Post-1952
Table 19: Immigration policy regressed on trade policy for Japan
Linear term only
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade*Capital
Capital policy
Trade policy2
Trade policy
DV: Immigration policy
61
0.40
-2.38*
(1.15)
2.37+
(1.21)
-0.01***
(0.00)
-1.32+
(0.67)
1.61*
(0.72)
With capital (pre-1952)
51
0.38
-4.32
(3.23)
4.28
(3.37)
0.01*
(0.00)
-1.04
(0.88)
-0.92
(0.82)
With capital(post-1952)
Table 20: Immigration policy regressed on trade policy for Singapore
DV: Immigration
policy
Linear term only
Trade policy
Square term only
Both terms
Interacted with capital
1978-2010
12.31**
(3.58)
48.52***
(7.20)
1132.15
(815.29)
-1122.00
(840.57)
1662.37+
(884.30)
-1665.92+
(896.30)
47.02***
(7.03)
Trade policy2
Capital policy
-0.03***
(0.00)
-46.56***
(6.84)
-0.03***
(0.00)
-11.65***
(1.61)
-0.02**
(0.01)
-851.65
(605.44)
-9.12
(5.94)
9.64
(6.02)
-0.03***
(0.00)
-12.58**
(3.50)
34
0.64
34
0.64
34
0.66
34
0.94
Trade*Capital
Linear time trend
Constant
Observations
R2
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Table 21: Immigration policy regressed on trade policy for South Korea
DV: Immigration policy
Trade policy
Linear term only
Square term only
Both terms
-0.01***
(0.00)
-0.17
(0.37)
-0.50
(0.61)
-0.01***
(0.00)
-0.26*
(0.10)
10.52**
(3.16)
-14.50**
(4.44)
-0.00*
(0.00)
-7.50***
(2.12)
53
0.33
53
0.34
53
0.40
-0.18
(0.42)
Trade policy2
Linear time trend
Constant
Observations
R2
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
66
-0.03***
(0.00)
-1245.81+
(660.29)
27
0.95
Table 22: Immigration policy regressed on trade policy for Taiwan
DV: Immigration policy
Trade policy
Linear term only
Square term only
Both terms
-0.02*
(0.01)
-4.81
(3.03)
4.31
(4.32)
-0.02
(0.01)
-0.92*
(0.38)
128.73**
(35.58)
-141.34**
(41.12)
-0.01
(0.01)
-94.11**
(25.68)
30
0.39
30
0.38
30
0.60
5.15
(3.69)
Trade policy2
Linear time trend
Constant
Observations
R2
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Table 23: Immigration policy regressed on trade policy for Kuwait
DV: Immigration policy
Trade policy
Linear term only
Square term only
Both terms
-0.01***
(0.00)
0.56
(2.38)
-1.07
(2.62)
-0.01***
(0.00)
-0.10
(0.58)
236.51***
(62.77)
-252.29***
(67.61)
-0.01***
(0.00)
-174.00***
(45.95)
36
0.57
36
0.57
36
0.69
-0.93
(2.46)
Trade policy2
Linear time trend
Constant
Observations
R2
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
67
68
1786
0.87
1356
0.82
-0.01***
(0.00)
1.20
(0.83)
0.21*
(0.09)
-0.48
(0.80)
1.57
(0.93)
-1.87+
(1.00)
(2)
All Years
Standard errors in parentheses
+ p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
Observations
R2
Constant
Linear time trend
Trade2 *Capital
Trade policy2
Trade*Capital
Capital policy
-0.01**
(0.00)
2.90***
(0.68)
0.23**
(0.07)
-2.10**
(0.70)
Citizenship
Trade policy
(1)
All Year
DV: Immigration policy
without citizenship
545
0.64
-0.00
(0.00)
2.16
(1.36)
0.22
(0.16)
-1.63
(1.28)
(3)
Pre-1914
201
0.76
-0.04***
(0.01)
3.04*
(1.25)
0.48+
(0.26)
-0.29
(0.62)
4.91*
(1.86)
-5.78*
(1.90)
(4)
Interwar
506
0.81
-0.03***
(0.01)
0.61
(0.57)
0.42***
(0.07)
0.89
(0.61)
2.12**
(0.62)
-3.05**
(0.64)
(5)
Pre-1946
850
0.46
0.05
(0.19)
2.40
(5.95)
-95.63
(60.86)
203.40
(129.14)
-4.96
(8.81)
-108.21
(68.43)
-0.00
(0.01)
-1.67
(4.18)
(6)
Post-WWII
291
0.35
0.14
(0.11)
-8.04*
(2.86)
-187.55***
(39.74)
397.02***
(86.19)
11.43*
(4.74)
-210.08***
(46.74)
-0.01
(0.01)
5.63*
(2.44)
(7)
Bretton Woods
559
0.40
-0.11
(0.14)
20.33**
(5.67)
-22.55
(47.74)
45.97
(100.42)
-27.96**
(7.99)
-23.38
(52.82)
-0.01
(0.00)
-13.04**
(4.19)
(8)
Post-Bretton Woods
Table 24: Immigration policy without citizenship regressed on citizenship, trade and capital policy