Unformatted Attachment Preview
The Booth School of Business, University of Chicago
The Evolution of Citizenship: Economic and Institutional Determinants
Author(s): Graziella Bertocchi and Chiara Strozzi
Source: The Journal of Law & Economics, Vol. 53, No. 1 (February 2010), pp. 95-136
Published by: The University of Chicago Press for The Booth School of Business,
University of Chicago and The University of Chicago Law School
Stable URL: http://www.jstor.org/stable/10.1086/600080
Accessed: 11-03-2018 06:53 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
The University of Chicago Law School, The Booth School of Business, University of
Chicago, The University of Chicago Press are collaborating with JSTOR to digitize,
preserve and extend access to The Journal of Law & Economics
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
The Evolution of Citizenship: Economic
and Institutional Determinants
Graziella Bertocchi Università di Modena e Reggio Emilia
Chiara Strozzi Università di Modena e Reggio Emilia
Abstract
We investigate the evolution of the legal institution of citizenship from a political
economy perspective. We first present a median-voter model of the determination of citizenship laws. Next we test the implications of the model on a new
set of data on citizenship laws across countries. We show that citizenship laws
respond to economic and institutional determinants endogenously. When facing
increasing immigration, countries with a jus soli regime tend to restrict their
legislation, whereas countries with a jus sanguinis regime resist innovation. The
welfare burden does not prove to be an obstacle to jus soli legislation, but
demographic stagnation encourages it. A high degree of democracy promotes
the adoption of jus soli elements, whereas instability of state borders determined
by decolonization impedes it. Religion and ethnic diversity have no residual
effect.
1. Introduction
Every country of the world has established a complex system of rules that govern
the attribution of citizenship. As a consequence of the increasing pressure of
international migration, citizenship laws have moved to center stage on policy
agendas because citizenship laws not only affect the design of immigration policy
For helpful comments and suggestions, we thank the editor, Dennis W. Carlton; an anonymous
referee; and J.-P. Azam, G. Barlevy, D. J. DeVoretz, D. Fiaschi, N. Gennaioli, A. M. Mayda, D. Mitra,
L. Rocco, and K. Tatsiramos. We also thank the discussants and participants at the 2003 North
American summer meetings of the Econometric Society; the William Davidson Institute–Center for
Economic and Policy Research (CEPR) conference on Transition Economics; the Conference on
Economic Growth and Distribution; the CEPR conference on Understanding Productivity Differences
across Sectors, Firms and Countries; the 2004 Associazione Italiana Economisti del Lavoro conference;
the 2004 Economic Growth and Innovation in Multicultural Environments conference; the 2005
European Society for Population Economics conference; the CEPR conference on Institutions, Policies, and Economic Growth; the CEPR conference The Long-Run Growth and Development of the
World Economy; the 2005 winter meeting of the Econometric Society; the second Institute for the
Study of Labor (IZA) Annual Migration Meeting; the fourth Europaeum Economic Workshop; the
Consiglio Nazionale delle Ricerche workshop on International Economics and Development; and
seminars in Paris, Padua, Toulouse, Milan, Bologna, IZA, and Berlin. Financial support from the
Italian University Ministry and the European Commission is gratefully acknowledged.
[Journal of Law and Economics, vol. 53 (February 2010)]
䉷 2010 by The University of Chicago. All rights reserved. 0022-2186/2010/5301-0004$10.00
95
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
96
The Journal of LAW & ECONOMICS
but also interact with the workings of labor markets, affect welfare programs,
and influence demographic trends.
Citizenship is the legal institution that designates full membership in a state
and the associated rights and duties. It provides benefits such as the right to
vote, better employment opportunities, the ability to travel without restrictions,
legal protection in case of criminal charges, and the possibility of obtaining a
visa for a relative. There are also costs to citizenship, such as the military draft,
renunciation of one’s original citizenship, and the pecuniary and nonpecuniary
costs that may be required for naturalization and for recognition at the age of
majority. Examples of the latter costs are language and culture tests, waiting
periods, and a commitment to avoid activities leading to disqualification.
There are several ways to acquire citizenship: at birth, by naturalization, and
by marriage. The regulation of citizenship at birth, which determines citizenship
acquisition by second-generation immigrants, is rooted in the well-defined bodies
of common and civil law. The former traditionally applies the jus soli principle,
according to which citizenship is attributed by birthplace: this implies that the
child of an immigrant is a citizen as long as he is born in the country of
immigration. The latter applies the jus sanguinis principle, which attributes
citizenship by descent, so that a child inherits citizenship from his parents,
independent of where he is born. Despite being rooted in these principles, during
the twentieth century—and especially after World War II—citizenship laws in
many countries have gone through a process of continuous adaptation in conjunction with decolonization, the collapse of the socialist system, and the mounting pressure of international migration.
In the present article we investigate, from a political economy perspective, the
determinants and evolution of citizenship laws in the postwar period. To pursue
this goal, we assemble a new data set that codifies citizenship laws across countries, with a specific focus on the provisions that regulate access to citizenship
at birth. The data set is then used to study the dynamic adaptation of these laws
by relating the observed patterns to a number of potential determinants, including economic factors and other political and cultural factors that have been
found to be relevant in related research on institutions.
Modern sociopolitical theories have advanced several hypotheses concerning
the determinants of the dynamics of citizenship laws, on the basis of case studies
and nonquantitative cross-country comparisons. The legal tradition established
in a given country is generally believed to exert a persistent effect on current
legislation. The relevance of migration has also been investigated. In particular,
pressure from a large stock of migrants is perceived to be a factor that shapes
the attitude of a country toward citizenship policy. On the one hand, migration
could encourage legislation that allows granting automatic citizenship to all
newborns. On the other, it might also encourage restrictions of automatic citizenship in countries where it was originally applied. According to some sociopolitical theories, the combination of these forces tends to induce convergence
toward a mix of jus soli and jus sanguinis provisions (that is, a mixed regime)
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
97
for countries coming from different legal traditions (Weil 2001). For Europe,
Bauböck et al. (2006) instead stress the presence of divergent trends that lean
toward liberalization in some countries and toward restriction of access to citizenship in others. The influence of other economic forces is also recognized.
Because citizenship rights determine the ability to enjoy welfare benefits, the
shaping of nationality laws has been associated with the nature of the welfare
state, with a large government representing a potential obstacle to the retention
of jus soli (Joppke 1998). This argument, however, has to be weighed against
the potential gain resulting from the acquisition of relatively young new citizens
by countries with expensive pension systems and in the midst of a demographic
crisis. Political factors have also been found to be relevant. The presence of a
consolidated democracy is expected to lead to the adoption of jus soli, which is
viewed as treating aliens more equally. Stabilization of state borders should reduce
the pressure to preserve a national identity through jus sanguinis. Finally, an
additional factor that has been the subject of debate is the influence of national
character and culture. The theory advanced by Brubaker (1992) focuses on France
and Germany as having antagonistic kinds of nationhood, the former more
assimilationist and the latter more ethnocentric, which induce definitions of
citizenship.
In this article, we formalize the aforementioned hypotheses in a simple medianvoter model that guides our empirical investigation by generating testable implications and offering an interpretation of the resulting evidence. The model is
based on the assumption that the laws regulating the acquisition of citizenship
can be viewed as the outcome of a decision problem faced by a native median
voter, in a context where citizenship confers the right to vote on policy. In
making this decision, the native median voter takes into account the associated
benefits and costs, which depend on the share of migrants in the population
and the other factors suggested by the literature. The model indicates that the
share of migrants has a potentially ambiguous effect on the decision to grant
citizenship and voting rights to migrants because, for the native median voter,
it increases the loss of letting migrants vote but also increases the cost associated
with their exclusion. Moreover, the decision of natives can be positively influenced when migrants have relatively high income levels, the welfare state is small,
the native population is relatively old, and there is a high level of democracy, a
stable national border, and an inclusive national culture.
We test the aforementioned empirical implications and find that, in the postwar
period, citizenship laws responded to economic and institutional factors endogenously and systematically through a slow but steady process of adaptation.
Overall, our results suggest that migration moves national legislation in the
direction of jus sanguinis, not jus soli. In particular, when we take into account
the legal tradition governing citizenship, we find that countries with a jus soli
origin react to increasing migration by adding jus sanguinis elements. On the
other hand, the effect of migration has been negligible in countries with jus
sanguinis regimes. Therefore, the evidence does not support the hypothesis of
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
98
The Journal of LAW & ECONOMICS
convergence toward a mixed regime, because migration tends to induce restrictions but not extensions. Other economic factors also matter. Although the
welfare burden does not prove to be an obstacle for jus soli legislation, demographic stagnation encourages the adoption of mixed and jus soli regimes. Turning to institutional factors, we find that a high degree of democracy is significantly
associated with jus soli legislation, whereas border instability—in particular, after
decolonization—decreases its likelihood. Cultural characteristics determined by
religion and ethic fractionalization are not found to play a significant role.
Section 2 introduces the related literature. Section 3 reviews the historical and
legal background of the issues that we address. Section 4 presents our model of
determination of citizenship laws. Section 5 describes our data on citizenship
laws around the world. Section 6 empirically investigates the determinants of
current citizenship laws and presents our main results and a set of robustness
checks. Section 7 develops an alternative empirical strategy that highlights the
determinants of change in citizenship laws. Section 8 provides conclusions and
indicates directions for future research. Appendix A presents information about
the data used. Appendix B provides additional tables.
2. Related Literature
Our work is related to several branches of the economic literature. First, this
article adds to research on international migration and migration policy. Timmer
and Williamson (1998), Hatton and Williamson (2006), and Bertocchi and
Strozzi (2008) empirically analyze immigration policies enacted at the end of
the nineteenth century during the mass migration era, and O’Rourke and Sinnott
(2006) and Mayda (2006) estimate the attitudes of voters toward immigration
in the postwar period. The political economy of migration is modeled by, among
others, Benhabib (1996), Gradstein and Schiff (2006), and DeVoretz (2006). More
specifically, the role of citizenship policy is discussed by DeVoretz and Pivnenko
(2006), who investigate the economic costs and benefits derived from citizenship;
by Pritchett (2006), who evaluates citizenship policy within a broader discussion
of labor mobility and immigration policies; and by Chiswick and Miller (2009),
who evaluate the determinants of the decision to naturalize.
Because our theory emphasizes that citizenship rights imply the right to vote,
our work is also related to Razin, Sadka, and Swagel (2002) and Dolmas and
Huffman (2004), who compare the effect of migration on the welfare state with
or without voting rights for the migrants. However, they do not examine the
determinants of these alternative regimes. On the other hand, the issue of franchise extension has recently received considerable attention within the literature,
with contributions by Acemoglu and Robinson (2000) and Bertocchi and Spagat
(2001), but with a focus on the conflict between rich and poor, whereas Bertocchi
(2007) concentrates on the conflict between men and women.
Viewed more broadly, our work contributes to research that focuses on the
historical determinants of institutions. Engerman and Sokoloff (2002) highlight
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
99
the relevance of wealth inequality and political factors in accounting for how
fundamental economic institutions develop over time. Acemoglu, Johnson, and
Robinson (2001) contribute to the understanding of how institutions evolve by
using historical variables as instruments for contemporary measures of the quality
of institutions.
This article also relates to the comparative legal approach initiated by La Porta
et al. (1998). The basic premise of this research line is the recognition that laws
in different countries are adopted or transplanted from a few legal traditions
and that the resulting legislative bodies reflect both the influence of legal origin
and the subsequent revision specific to individual countries. Balas et al. (2009)
specifically analyze the evolution of an index of formalism of legal procedure.
We add to this stream by focusing on the determinants of the dynamic adaptation
of nationality rules.
Finally, work by Alesina and Spolaore (1997) and Bolton and Roland (1997)
on the optimal determination of the size of nations, and thus of state borders,
is also relevant to our approach, both because country size in this literature is
the same as population size and is potentially influenced by migration and by
the legal status of immigrants and because borders play an important role in
the determination of citizenship rules.
3. Citizenship Laws in Historical Perspective
Citizenship policy can be viewed as part of broader migration policy. However,
contrary to other migration policy measures, such as quotas and visa requirements, that are typically adjusted to the business cycle and to current government
orientation, reforms of citizenship laws tend to be the outcome of long-term
processes of adaptation that often involve constitutional amendments.
In eighteenth-century Europe, jus soli was the dominant criterion, in accordance with feudal traditions that linked people to the lord who held the land
where they were born. The French Revolution broke with this heritage and, in
the Civil Code of 1804, reintroduced the ancient Roman custom of jus sanguinis.
Continental modern citizenship law was subsequently built on these premises.
During the nineteenth century, the jus sanguinis principle was adopted throughout Europe and then transplanted to its colonies. In imitation, Japan also adopted
jus sanguinis during this phase. On the other hand, the British preserved their
jus soli tradition and spread it through their own colonies, starting with the
United States, where it was later encoded in the U.S. Constitution. By the end
of the nineteenth century, the process of nation-state formation and the associated codification efforts were completed in continental Europe. At the same
time, the revolutionary period was over in countries that had been subject to
earlier colonization, and nineteenth-century colonization had extended the process of transplantation of legal tradition to the rest of the world. Therefore, most
countries had established specific provisions for the acquisition of citizenship
within a relatively well developed legal system, with jus soli being the norm in
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
100
The Journal of LAW & ECONOMICS
common-law countries and jus sanguinis regulating citizenship law in most civil
law countries, with important exceptions. For instance, civil law Latin America
had embraced jus soli early, whereas civil law France, with its colonies, had by
then already moved toward a mixed regime. However, the next century witnessed
a continuous transformation of citizenship laws across the world. Mentioned
below are some specific cases, drawn mostly from Joppke (1998), Aleinikoff and
Klusmeyer (2000, 2001), and Brubaker (1992).
3.1. The United States
Jus soli was encoded in the U.S. Constitution through the Fourteenth Amendment of 1868, with the specific purpose of protecting the birthrights of black
slaves. Given its history as a country of immigrants and its generally positive
attitude toward economic liberalism, the United States has an approach that is
still remarkably consistent with its original attitude in all aspects, ranging from
immigration policy to naturalization requirements. Debate about possible restrictions arose recently but never led to actual change.2 In particular, in the
1980s, jus soli came under attack regarding its applicability to the children of
illegal immigrants. A relatively young and thin welfare state contributes to the
fiscal sustainability of jus soli in this country.3
3.2. Australia
Current citizenship law in Australia differs considerably from that in the
United States, despite the nations’ common origin as countries of immigration.
Jus soli had also been introduced in Australia by the colonists. In the postwar
period, the country went through numerous legislative and administrative reforms. Jus soli survived until 1986. Since that time, to acquire citizenship, a
person born in Australia must have at least one parent who is either an Australian
citizen or a permanent resident.
3.3. Latin America
In the face of a civil law tradition transplanted by European powers, Latin
America has followed a rather peculiar pattern. At independence, most of the
incipient states chose jus soli regimes as a way to break with the colonial political
order and to prevent the metropoles from making legitimate claims on citizens
born in the new countries. Jus soli was encoded in the constitution of Brazil in
1824, in that of Venezuela in 1830, and in that of Argentina in 1853. Therefore,
most countries in Latin America already had a jus soli regime before the
2
In his analysis of Mexican immigration, Huntington (2004) criticizes current nationality regulations on the grounds that they represent a devaluation of citizenship. Pritchett (2006) discusses
the possible advantages of guest-worker programs that do not contemplate citizenship, from the
perspective of poor countries.
3
The relative thickness of the concept of citizenship, if compared with residency, is a related,
potentially relevant consideration: in the United States, for instance, citizenship is relatively thin in
the sense that it confers few additional benefits over residency.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
101
nineteenth-century waves of immigration began. Jus soli is still the prevalent
rule in the area, although it no longer attracts immigrants. Mexico is a special
case in that jus soli was adopted in the 1814 constitution but then was abandoned
in 1836, only to return to stay with a constitutional amendment in 1937.
3.4. The United Kingdom
British nationality law has been deeply affected by the imperial experience.
Because of its colonial history, until World War II the concept of nationality in
the United Kingdom was particularly extensive because all subjects of the British
Empire had equal access to British citizenship simply by establishing residence
in the United Kingdom. The British Nationality Act of 1848 created the status
of “citizen of the United Kingdom and colonies” for people with a close connection to the United Kingdom and its colonies. After a postwar wave of colonial
immigration, this open-door policy was progressively restricted, although special
status is still attributed to citizens of the British Commonwealth. Since the 1980s,
redefinitions of national citizenship have been effectively used as a form of
selective immigration policy. The 1984 British Nationality Act restricts jus soli
by establishing that a child born in the United Kingdom qualifies for British
citizenship only if at least one parent is a British citizen or resident.
3.5. France
The emergence of the nation-state in continental Europe was the main factor
that shaped citizenship law in this area. The revolutionary experience was particularly important for France, where jus sanguinis was first introduced with the
Civil Code of 1804 and was maintained for the entire course of the nineteenth
century, even though, early on, military considerations introduced elements of
jus soli. To ensure that children born to immigrants in France would be subject
to the draft, double jus soli became automatic in 1889, making the experience
of this country a unique one. After World War II, large-scale immigration,
especially from North Africa, raised concerns regarding assimilation. Citizenship
issues and the rights of immigrants became the subjects of heated debate in
French politics. In 1993, President Jacques Chirac introduced to the legislation
a restrictive revision that required a formal citizenship request from secondgeneration immigrants. With the Left regaining political power in 1997, however,
this restriction was revised considerably, with citizenship automatically assigned
at age 18 to immigrants’ children born in France who had neither requested nor
declined it. France is frequently compared with Germany. Brubaker (2002) influentially argues that the different paths followed by these countries have been
shaped by cultural difference, with France following its tradition as an assimilationist nation and Germany focusing on its ethnic identity.
3.6. Germany
The single most relevant event in the history of German citizenship law is
certainly the fall of the Berlin Wall, which paved the way for the achievement
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
102
The Journal of LAW & ECONOMICS
of stable national borders. Before that event, the massive immigration of guest
workers in the postwar period—mostly from Turkey but also from southern
Europe—had started to put under strong pressure, albeit without success, the
original Wilhelminian citizenship law of 1913, which established strong jus sanguinis ties with German emigrants overseas. With the foundation of the German
Democratic Republic and the consolidation of the Eastern Block, Germany found
itself in the paradoxical situation of having a large population of disenfranchised
foreigners born on its own soil and, at the same time, millions of ethnic Germans
living behind the Iron Curtain. Achieving border stability was a decisive factor
in moving Germany toward the long-delayed adoption of jus soli elements. A
first step in this direction was the Foreigner Law of 1990, which changed naturalization from a discretionary exception to the rule. A major overhaul of the
legislation, after intense political struggle, was finally approved in 1999. Jus soli
is now the norm in Germany (with the minor requirement that one parent has
lived in the country for 8 years). Other factors that may have delayed the introduction of jus soli in Germany are, as mentioned in the introduction, the
strong ethnic character of German national identity and the thick nature of the
German welfare state. The latter factor may have played a role in shaping the
evolution of citizenship policies in several other European countries, and especially in Scandinavian countries, where jus sanguinis was functional when
there were large emigration flows in the past but which have recently adapted
to the quickly changing conditions, especially in Sweden, where immigration
rates are high. As documented by Weil (2001), restricted forms of double jus
soli are now applied de facto in the vast majority of European countries, which
recently adapted their legislation to the globalization of international migration
and its increasing effect on Europe. In particular, in the European Union, with
the exception of Austria, Greece, and Luxembourg, access to citizenship by
second and third generations is facilitated.
3.7. Decolonization
Postwar decolonization had a major influence on citizenship rules applied
around the world—not only through an indirect effect on the metropolitan
countries previously examined. The vast majority of African colonies that were
subject to civil law countries practicing jus sanguinis stuck to this principle after
attaining independence. On the other hand, many former British and Portuguese
colonies rejected the jus soli tradition and switched to an often strongly ethnically
tinged version of jus sanguinis. For instance, in Sierra Leone, the 1961 constitution established that citizenship is transmitted only by descent and only to
children whose father and a grandfather were Sierra Leoneans of African-Negro
descent. In situations in which instability was pushed to an extreme degree
because of the newness and the arbitrary borders of these countries and was
compounded by deep ethnic division, jus sanguinis tended to prevail as a way
to control more easily the formation of national entities. At the same time,
however, the associated exclusive notion of ethnic and tribal identity caused
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
103
enormous problems in countries where colonial rule had left shaky democratic
institutions. To the present time, ethnic conflict lies at the roots of the chronic
manipulation of citizenship rules in favor of one ethnic group over others. In
an effort to exclude Rwandan immigrants, the 1964 Congolese constitution recognized citizenship only for persons whose parents were members of one of the
tribes established within the territory by 1908. In 1981, Mobutu signed a new
law on nationality that required an ancestral connection to the population residing in the territory as far back as 1885. Marginalization and de facto statelessness of significant strata of the population are unavoidable outcomes of these
policies.
3.8. The Disintegration of the Union of Socialist Soviet Republics
Another major wave of citizenship law codification followed the disintegration
of the Union of Soviet Socialist Republics. The area had been sealed against
international migration; however, as for all empires, there had been considerable
internal migration. The Soviet Union occupied Estonia, Latvia, and Lithuania
in 1940. During the following decades, millions of Russians were encouraged to
settle in Latvia and Estonia (less so in Lithuania) to Russify them. Today, sizable
stateless, Russian-speaking minorities are still present. After independence, the
new citizenship laws of the three states reflected this heritage with an emphasis
on jus sanguinis as the basis for acquiring citizenship. The hostile attitude toward
ethnic Russians was especially strong in Latvia, whereas Lithuania, which was
less affected by Soviet immigration policy, had a more open approach. The issue
for these states was how to balance a need to reconstitute their national identity
around an ethnic model and a commitment to democratic values with respect
to the rights of minorities. Estonian and Latvian laws were sharply criticized by
international organizations on the grounds of human rights. In anticipation of
integration into the European Union, the Baltic states indeed addressed these
recommendations in more recent legislation, whereas discriminatory policies
persist in most other countries in the area. In contrast, the salient fact in shaping
the Russian Federation’s current citizenship policy is the perception that many
of its citizens are outside its borders, spread around the former regions of the
Union of Soviet Socialist Republics. Again, the perception of being a country of
emigrants is pushing to maintain jus sanguinis as the main principle, even though
small concessions to jus soli have been made.
4. The Model
This section develops a simple theoretical model that formalizes the hypotheses
that a widely interdisciplinary literature has advanced with respect to the determinants of citizenship laws. The aim of the model is also to generate testable
implications that can guide our interpretation of the empirical evidence.
We can view the laws regulating citizenship acquisition as the outcome of the
decision problem faced by the median voter in a context where citizenship confers
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
104
The Journal of LAW & ECONOMICS
the right to vote on policy. A native voter, when deciding whether to grant
citizenship, and thus the right to vote, to migrants is driven by the benefits and
costs associated with this decision.
A few warnings are in order before we present the details of the model. First,
although in practice acquisition of citizenship implies a set of rights larger than
the right to vote, and also implies some duties, our focus on the right to vote
as the main benefit is easily justified because political rights can be viewed as
an instrument through which migrants can achieve broader political, economic,
and social goals. Second, although the model concentrates on voting on a specific
policy—namely, a redistributive tax scheme that finances a public good—the
same approach can be extended to consider alternative agendas. Finally, in our
1-period framework, the distinction between different ways of acquiring citizenship becomes irrelevant; therefore, the predictions of the model can be applied
to laws concerning both citizenship at birth and naturalization.
We consider an economy in which a population of mass P consists of natives
with mass N and migrants with mass M, where M ⫹ N p P and M ! N. Migrants
are poorer than natives because they are relatively unskilled, so y N 1 y 1 y M,
where y N and y M denote average incomes for natives and migrants, respectively,
and y p (N/P) y N ⫹ (M/P) y M denotes the economy-wide average income. We
also assume that the income distribution is skewed to the right for each group—
namely, median income is lower than average income for both natives and
migrants and thus for the economy as a whole.
Both natives and migrants derive utility from consumption of a private good,
c i, and a public good, g, according to the equation
u i p c i ⫹ lg,
(1)
where l is a positive preference parameter. Both groups pay taxes in accordance
with a proportional income tax rate t such that 0 ! t ! 1 . Tax revenues are used
by the government to finance the public good according to the balanced budget
constraint
g p ty ⫺
t2
y,
2
(2)
where the second term denotes tax collection costs.
The tax rate is set by political choice under majority voting, as in Meltzer and
Richard (1981). Each enfranchised individual casts a vote on the tax rate. Assume
initially that only natives are citizens and therefore allowed to vote. Assume also
that society bears a cost k for the exclusion of migrants from citizenship. This
cost increases with the share of migrants in the population, reflecting the possibility that their disenfranchisement can lead to social unrest and even violence.
The cost is also affected by factors that determine the degree of inclusiveness of
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
105
the country’s culture as captured, for instance, by a jus soli tradition, according
to
kpK⫹h
M
,
P
(3)
where K denotes the degree of cultural inclusiveness and h 1 0. The cost enters
the individual budget constraint as follows:
c i ≤ (1 ⫺ t)y i ⫺ k,
(4)
i
where y denotes individual income. Equivalently, k could directly enter the utility
function. The expression for the indirect utility function of a native voter with
income y Ni is given by
( )
v Ni p (1 ⫺ t)y Ni ⫺ k ⫹ l t ⫺
t2
y,
2
(5)
which is single peaked with respect to the tax rate. We can therefore apply the
median-voter theorem, according to which the equilibrium tax rate is the preferred tax rate of the native median voter with income y N *, according to
t N* p 1 ⫺
1 y N*
.
l y
(6)
The level of the tax rate increases with the intensity of the preference for
public goods and with inequality, which is measured by the ratio of natives’
median income over the economy-wide average income, because migrants also
pay taxes. Note that, under our assumptions about income distribution, it is not
necessarily the case that (y N */y) ! 1, which implies that the tax rate is going to
be positive only if (y N */y) ! l.
The native median voter could avoid cost k by granting citizenship to the
migrants and thus accepting the tax rate that would prevail under universal
enfranchisement, which is given by
t* p 1 ⫺
1 y*
l y
,
(7)
where y* is the economy-wide median income. Because y N * 1 y* , it follows that
t N * ! t*—that is, the tax rate chosen by the native median voter is lower than
the tax that applies under universal suffrage. In particular, given our assumptions
about income distribution, the latter is certainly positive. The difference between
t N * and t* increases with the income gap between natives and migrants and
with the share of migrants in the population.
The native median voter faces a simple set of costs and benefits when considering the decision to grant citizenship. If migrants cannot vote, the native
median voter has to face cost k but can enjoy a smaller tax. If migrants vote,
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
106
The Journal of LAW & ECONOMICS
the native median voter has to pay more taxes, but k is avoided. It follows that
the native median voter decides to grant citizenship to migrants if and only if
(
(1 ⫺ t N *)y N * ⫺ k ⫹ l t N * ⫺
t N*
2
2
)
(
y ≤ (1 ⫺ t*)y N * ⫹ l t* ⫺
)
t*2
2
y,
(8)
where, following equation (5), we find on the left-hand side his indirect utility
function when migrants cannot vote and on the right-hand side his indirect
utility function when migrants can vote. Both the disenfranchisement cost and
the fiscal gain associated with the no-franchise status quo increase with the share
of migrants in the population. In addition, the fiscal gain also increases with the
income gap between natives and migrants, whereas the cost increases with the
degree of inclusiveness K of the country’s culture.
Even if the model is static, we can think of its dynamic implications in terms
of a sequence of repeated decisions. If, after a sequence of stationary decisions,
the economy is shocked by an increase in the migrant share because of a large
inflow, the median voter will respond, taking into consideration all the channels
involved, and this may result in an adaptation of the regulation. Note that a
given stock of migrants has a stronger effect on countries with relatively small
native populations because it is the share of migrants in the population that
matters.
The predictions obtained from the model to date indicate that a decision to
extend citizenship and associated voting rights is facilitated by a smaller income
gap between natives and migrants and by a larger degree of inclusiveness in the
country’s culture. If we interpret a jus sanguinis tradition as demonstrating a
low degree of inclusiveness and, thus, a low cost of exclusion, it follows that jus
sanguinis countries will be more reluctant to change. Finally, the effect of an
increase in the share of migrants is potentially ambiguous because a higher share
increases both the cost and the fiscal gain of disenfranchisement, generating a
trade-off. The net effect depends on which factor is stronger.
Although the present formulation of the model is designed to establish conditions for the extension of citizenship rights to migrants, it can also encompass
restriction. When the status quo is a jus soli regulation or, equivalently, a particularly generous naturalization policy, migrants who are already in the country
and have thus become citizens are simply considered to be natives themselves.
Together with ethnic natives, they decide whether to restrict the current regulation, taking into account the incoming waves of immigrants and following the
simple logic previously illustrated.
The basic model can be extended to consider several other potentially relevant
factors. First, the influence of the size of government on citizenship laws can be
captured by assuming that different countries exhibit different preference parameters toward government. Thus, having a relatively large government, as
captured by l, could make an open citizenship policy more costly by increasing
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
107
the tax differential. Empirically, we should therefore expect the size of government
to have a negative effect on the degree of inclusiveness in citizenship laws.
Second, demographic aspects can be considered by assuming that a younger
average age for migrants implies a larger ability to contribute to the welfare state
at any given level of income. Although our 1-period model cannot explicitly
reflect these aspects, we can interpret our tax as a lifelong contribution that is
higher for a migrant. This should facilitate the decision to grant migrants citizenship and implies that countries with relatively old native populations should
be particularly sensitive to these considerations and thus display a more open
attitude.
Third, the level of democracy can influence the outcome because it implies a
constraint on the political rights of the natives, which can be modeled with an
income franchise requirement. If only rich natives are allowed to vote, the associated tax rate will be higher than otherwise, thus amplifying the tax cost that
follows the decision to allow migrants to vote. The testable implication is that
the decision to grant citizenship is positively influenced by the domestic level
of democracy.
Fourth, border instability can be captured in a version of the model in which
the size of the native population, and thus the population share of migrants, is
subject to uncertainty. Indeed, when the national border is unstable, the main
problem is to establish who can be considered a native. Under standard riskaversion behavior, introducing a random component into the model should affect
the voting decision by reducing the tax rate.
In summary, the share of migrants has a potentially ambiguous effect on the
decision of the natives to extend citizenship and voting rights to migrants because,
for the native median voter, it increases the tax disadvantage of letting migrants
vote while increasing the cost associated with their exclusion. Moreover, we expect
that the natives’ decision will be positively influenced by an inclusive national
culture, a relatively high income level for migrants, a small welfare state, a
relatively old native population, a high level of democracy, stable national borders,
and a large native population.
5. The Data: Citizenship Laws of the World
We compile a data set of postwar citizenship laws across the countries of the
world. The principal source for the information that we codify is U.S. Office of
Personnel Management (2001), a directory that provides synopses of citizenship
laws currently in force in 190 countries. The sources for the directory are embassies, the Library of Congress, and the U.S. Department of State. We supplement this information with Central Intelligence Agency (2002), United Nations
High Commissioner for Refugees (2003), and the survey in Weil (2001). The
principal focus of our codification is acquisition of citizenship at birth, but we
also collect information about naturalization requirements.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
108
The Journal of LAW & ECONOMICS
5.1. Citizenship at Birth
We attribute to each country an appropriate code for citizenship laws in 2001,
in 1975, and at the beginning of the postwar period. We use 1948 as the starting
point although there were nearly no reforms in citizenship laws during the first
half of the century, so most of the legislation in place in 1948 had been developed
much earlier.4 As in the analysis of legal origins in La Porta et al. (1998), we
treat as predetermined the specific legal provisions regulating access to citizenship
in 1948, at least relative to the subsequent developments that are the focus of
the present investigation. Indeed, although the nineteenth century witnessed a
first wave of adaptation of citizenship legislation from either the civil law tradition
or the common-law tradition, there was no further evolution in the subsequent
half-century, despite major historical events such as World War I. By coding
citizenship laws in 1975, we divide the postwar period into two subperiods of
equal length.
In our classification, we focus on the presence of jus soli elements in the
legislation of a country. This approach is justified by our primary interest in the
potential effect of citizenship laws on immigrants rather than emigrants. We
divide countries into three groups for 1948, 1975, and 2001: countries subject
to jus sanguinis without any jus soli elements (group 1), those that apply a mixed
regime (group 2), and those subject to full jus soli (group 3). A mixed regime
includes elements of both jus soli and jus sanguinis regimes.5 Our data set
includes those 162 countries for which we were able to collect information on
both original and current citizenship laws and for which postwar migration data
were available.6
The differential patterns of evolution of citizenship laws in 1948, 1975, and
2001 are summarized by the transition matrices in Tables 1–3, which reveal
considerable variation both across countries and over time. Tables 1 and 2 show
that, in 1948, jus soli was the rule in approximately 47 percent (76 of 162) of
the countries, whereas jus sanguinis dominated in 41 percent (67 of 162 countries), and the mixed regime was adopted in the remaining 12 percent (19 of
162 countries). Among the countries that were under a jus soli regime in 1948
were the United States, Canada, all the Oceanian countries, most of Latin America, the British and Portuguese colonies in Africa and Asia, the United Kingdom,
Ireland, and Portugal. In 1948, jus sanguinis predominated in most of Europe,
including its eastern part. As explained in Section 3, France was unique in its
early choice of a mixed regime. Because we treat colonial territories as being
subject to the regime of the metropolitan countries until independence, the group
with the mixed regime in 1948 includes France and its colonies.
4
By treating 1948 as the initial year, we include the postwar decolonization period, with the
exception of the Middle East, which gained independence from the British and French administration
in 1943–48.
5
For details on our classification criteria, see Section A1.
6
For details on migration data, see Section A3.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
109
Table 1
Evolution of Citizenship Laws: Transition Matrix, 1948 versus 2001
Citizenship Laws in 2001
Citizenship Laws in 1948
Jus sanguinis
Mixed
Jus soli
Total
Jus Sanguinis
Mixed
Jus Soli
Total
46
11
31
88
20
6
9
35
1
2
36
39
67
19
76
162
Table 2
Evolution of Citizenship Laws: Transition Matrix, 1948 versus 1975
Citizenship Laws in 1975
Citizenship Laws in 1948
Jus sanguinis
Mixed
Jus soli
Total
Jus Sanguinis
Mixed
Jus Soli
Total
63
10
28
101
3
7
1
11
1
2
47
50
67
19
76
162
Table 3
Evolution of Citizenship Laws: Transition Matrix, 1975 versus 2001
Citizenship Laws in 2001
Citizenship Laws in 1975
Jus sanguinis
Mixed
Jus soli
Total
Jus Sanguinis
Mixed
Jus Soli
Total
81
2
5
88
20
9
6
35
0
0
39
39
101
11
50
162
By 1975, 31 percent (50 of 162) of the countries had jus soli, 62 percent (101
countries) had jus sanguinis, and 7 percent (11 countries) had mixed regimes.
The main event justifying this evolution is decolonization, with many former
colonies switching to jus sanguinis from jus soli when the United Kingdom and
Portugal were the metropolitan countries and from the mixed regime in the case
of France (see Section 3). As of 2001, 24 percent (39 of 162) of the countries
had jus soli, 54 percent (88 countries) jus sanguinis, and 22 percent (35 countries)
mixed regimes (see Table 3). It has mostly been the adaptation of the legislation
of many European countries, relaxing pure jus sanguinis in favor of a mixed
regime, that explains the pattern observed for the second subperiod.
Among the countries that still adhered to the jus soli principle in 2001 were
the United States, Canada, New Zealand, and Ireland (the latter, however, recently
introduced restrictions to jus soli with a June 2004 referendum). The United
Kingdom and Australia, on the contrary, no longer adhere to jus soli and have
adopted a mixed regime. Overall, jus sanguinis is currently the most common
regime, with 69 percent of the countries in Africa, 83 percent of those in Asia,
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
110
The Journal of LAW & ECONOMICS
and 41 percent (down from 88 percent) of those in Europe. The growing group
of countries in which a mix of provisions is applied is particularly well represented
in Europe, where 56 percent of European countries, including the United Kingdom, which formerly had a jus soli regime. On the other hand, jus soli predominates in the Americas, where it is followed by 89 percent of the countries
in Latin America and by all of North America (namely, the United States and
Canada).
The matrices present three different patterns of transitional dynamics: stability,
switch, and convergence. Stable countries lie along the diagonal. Looking at Table
1, we see that a large fraction (28 percent, or 46 of 162 countries) started and
ended as jus sanguinis countries. In other words, 69 percent (46 of 67) of the
countries that originally followed jus sanguinis principles have continued to do
so. By contrast, 22 percent (36 of 162 countries) steadily follow jus soli principles:
this means that only 47 percent (36 of 76) of the countries that originally followed
jus soli principles did not change their policies. Off the diagonal, there is a sizable
proportion of countries (19 percent, or 31 of 162) that switched from jus soli
to jus sanguinis by completely eliminating birthplace as a criterion: most of
them, as mentioned, are former African colonies of the United Kingdom and
Portugal that made this radical choice at independence. Looking at the two
subperiods, we see that most of these switches occurred between 1948 and 1975.
Finally, there is evidence of a process of convergence to a mix of jus soli and
jus sanguinis, which affects 18 percent of the countries (29 of 162, of which 20
converged from jus sanguinis by adding jus soli elements, whereas nine converged
from jus soli by restricting it), and evidence that it intensified between 1975 and
2001.
In Tables 4–6 we present additional information on the evolution of citizenship
laws. Here we report changes in the laws. During 1948–2001, a total of 74
countries (46 percent) experienced a change in the laws. Of these countries, 51
changed to jus sanguinis and 23 to jus soli; 45 changes occurred in the first
subperiod and 33 in the second.7 In particular, in the first subperiod, the majority
of the countries that experienced a change (29, or 64 percent) were originally
jus soli. As mentioned, this pattern is determined largely by the behavior of
former colonies. In the second subperiod (Table 6), the majority of the countries
that experienced a change (20, or 61 percent) were originally jus sanguinis, with
most adopting more open legislation.
The above discussion suggests a relevant role for border stability. To investigate
this issue, we introduce a set of dummies that capture a country’s history of
border changes. In particular, we distinguish three causes of border instability:
decolonization, the Berlin Wall, and other border changes.8 If we compare the
transitional dynamics of the full sample with those of the countries that did not
7
8
A few countries experienced more than one change.
Section A2 describes how the three border change dummies are constructed.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
111
Table 4
Changes in Citizenship Laws, 1948 to 2001
Citizenship Laws in 2001
Citizenship Laws in 1948
Jus sanguinis
Mixed
Jus soli
Total
No Change
Toward Jus Sanguinis
Toward Jus Soli
Total
46
6
36
88
0
11
40
51
21
2
0
23
67
19
76
162
Table 5
Changes in Citizenship Laws, 1948 to 1975
Citizenship Laws in 1975
Citizenship Laws in 1948
Jus sanguinis
Mixed
Jus soli
Total
No Change
Toward Jus Sanguinis
Toward Jus Soli
Total
63
7
47
117
0
10
29
39
4
2
0
6
67
19
76
162
Table 6
Changes in Citizenship Laws, 1975 to 2001
Citizenship Laws in 2001
Citizenship Laws in 1975
Jus sanguinis
Mixed
Jus soli
Total
No Change
Toward Jus Sanguinis
Toward Jus Soli
Total
81
9
39
129
0
2
11
13
20
0
0
20
101
11
50
162
experience a border change, we count for the latter a much smaller proportion
of switches from jus soli to jus sanguinis. This pattern confirms the relevance
of border changes, especially those caused by decolonization.
Summary statistics for our data set of citizenship laws are reported in Table
7. The correlation between citizenship laws in 1948 and 2001 is .42, which points
to some persistence, as confirmed by the even higher correlations noted between
the 1948 and 1975 laws (.60) and the 1975 and 2001 laws (.81).
5.2. Citizenship by Naturalization and the Citizenship Policy Index
Naturalization policies are also relevant to the issues at hand. Indeed, facilitation of naturalization for immigrant parents may represent a substitute mechanism for conferring citizenship on children born in jus sanguinis countries. In
addition, the general attitude revealed by a country’s regulation of citizenship
at birth may be reflected in its naturalization laws, with countries with jus soli
regimes traditionally making naturalization much easier, at least for resident
aliens. In jus sanguinis countries, naturalization requirements again tend to be
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
112
The Journal of LAW & ECONOMICS
Table 7
Citizenship Laws Data Set: Summary Statistics
Citizenship laws in 2001
Citizenship laws in 1975
Citizenship laws in 1948
Changes in citizenship laws, 1948 to 1975
Changes in citizenship laws, 1975 to 2001
Changes in citizenship laws, 1948 to 2001
Naturalization in 2001
Citizenship policy index in 2001
Observations
Mean
SD
Min
Max
162
162
162
162
162
162
142
142
1.698
1.685
2.056
.043
⫺.204
⫺.173
2.458
.415
.835
.916
.941
.451
.488
.655
.920
.299
1
1
1
⫺1
⫺1
⫺1
1
0
3
3
3
1
1
1
4
1
correlated with the provisions introduced for citizenship at birth. Basic rules for
naturalization may include a period of residence, renunciation of other citizenship, familiarity with the language and customs of the country, and availability
of adequate means of support.
We code naturalization only for 2001 on the basis of the available information
for 142 of our 162 countries. We classify countries on the basis of the number of
years of residence required for naturalization by constructing four classes (114
years, 6–14 years, 5 years, and X4 years).9 In our data set, 62 countries (44 percent)
require 5 years of residence, which can be considered a relatively open attitude,
whereas 46 percent require more time, and only 10 percent are more open.
We then combine the information that we collected on citizenship at birth and
naturalization within a single measure, although data on naturalization are available
only for the year 2001 and for a subset of countries. We construct an index of
citizenship policy defined on a 0–1 interval. To construct the index, we treat
citizenship laws in 2001 as an ordinal variable by associating jus soli elements with
lower number of years of residence required for naturalization. The corrected
Cronbach’s a of the indicator is .54, which can be explained by the fact that the
correlation between citizenship laws and naturalization is only .37. Alternative
ways to define naturalization classes yield similar conclusions.10 Table 7 also reports
summary statistics for naturalization and the citizenship policy index.11
6. Determinants of Citizenship Laws
6.1. Empirical Specification
In bringing our theoretical model to the data, to investigate the determinants
of the evolution of citizenship laws in the postwar period, we select the following
9
We do not consider naturalization by marriage because it is heavily dependent on family law.
Dual-citizenship provisions constitute another potentially relevant aspect of citizenship policy,
but we do not use this variable because of limited information.
11
For European countries only, the British Council has compiled an index of civic citizenship and
inclusion (British Council Brussels 2005); Bauböck et al. (2006) collect statistics on the acquisition
of nationality.
10
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
113
variables. The dependent variable is Current Citizenship Laws, which is categorical and can take three values: 1 if the country has a jus sanguinis regime, 2
if the country has a mixed regime, and 3 if the country has a jus soli regime.
Among regressors, we employ measures of migration, the size of government,
and the age structure of the population, although we cannot gauge the effect of
the income gap between natives and migrants because of a lack of data. We
interpret the presence of a jus soli tradition as an indicator of the degree of
inclusiveness of the national culture. To control for other cultural characteristics,
we also consider religion and ethnic diversity. Border stability is measured by
our border change dummies. We also control for countries with particularly
small populations.
We use a multinomial logit specification to achieve maximum generality. In
fact, this choice does not impose any ex ante ordering among the three regimes—
for example, an ordering by increasing inclusiveness toward immigrants, with
jus soli being associated with maximal inclusiveness and jus sanguinis with minimal inclusiveness. Moreover, in principle, those regressors that affect the probability of being a jus soli country may not always have the opposite effect on
the probability of being a jus sanguinis country or may not have any effect at
all (and vice versa). (Further discussion on this point is presented in Section
6.3.)
We use a panel sample that includes information on two cross sections of 162
countries: the first cross section refers to the 1950–75 subperiod, and the second
cross section refers to the 1976–2000 subperiod. In the full specification that we
present, the multinomial logit model that we run has the form
L it p a ⫹ bM it ⫹ cSit ⫹ dM itSit ⫹ eTt ⫹ Z it f ⫹ it,
(9)
with i p 1, . . . , 162 and t p 1, 2, where t p 1 denotes the 1950–75 subperiod and t p 2 denotes the 1976–2000 subperiod.12
In specification (9), L it denotes citizenship laws in country i at the end of
period t, a is a constant term, M it is migration stock as a percentage of the
population in country i at the beginning of period t, Sit is a dummy for the
presence of jus sanguinis in country i at the beginning of each subperiod,
M itSit is the interaction between the previous two variables, Tt is a period dummy,
Z it is a vector of additional explanatory variables, and it is an error term.13
The set of explanatory variables Z it can be divided into two groups. The first
group includes dummies capturing the geopolitical position of a country. Within
this group, we consider the border change dummies and dummies for Latin
American, southern European, and small countries. The second group of explanatory variables includes the size of government as a share of the gross
12
For comparison purposes, we also run multinomial logit models for two types of more parsimonious specifications.
13
In particular, Li1 and Li2 are citizenship laws in country i in 1975 and 2001, respectively, Mi1 is
migration stock in country i in 1960, Mi2 is migration stock in country i in 1980, Si1 p 1 if country
i has jus sanguinis in 1948, Si2 p 1 if country i has jus sanguinis in 1975, T1 p 0, and T2 p 1.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
114
The Journal of LAW & ECONOMICS
Table 8
Summary Statistics
Current Citizenship Laws
Changes in Citizenship Laws
Jus Sanguinis Origin
Civil Law
Migration Stock
Net Migration Flow
Decolonization
Berlin Wall
Other Border Changes
Latin America
Southern Europe
Small Country
Sub-Saharan Africa
British or Portuguese Colony
Socialist
Oil
Political Rights
Catholic Share
Ethnolinguistic Fractionalization
Government Consumption
Share of Young
Log GDP Per Capita
Gini Index
Observations
Mean
SD
Min
Max
324
324
324
324
300
318
324
324
324
324
324
324
324
324
324
324
276
324
272
263
324
263
155
1.691
⫺.080
.519
.673
5.646
⫺.116
.250
.052
.022
.173
.043
.160
.265
.333
.173
.086
3.730
31.535
.349
19.617
34.54
7.488
40.425
.875
.485
.500
.470
9.713
8.551
.434
.223
.146
.379
.204
.368
.442
.472
.379
.281
2.040
35.572
.299
11.342
3.47
1.249
9.966
1
⫺1
0
0
.011
⫺47.95
0
0
0
0
0
0
0
0
0
0
1
0
0
2.492
27.01
4.979
20.495
3
1
1
1
70.673
63.35
1
1
1
1
1
1
1
1
1
1
7
97.3
.890
72.233
53.36
10.060
63.180
Note. The sample includes two cross sections of 162 countries. The reference period is 1950–2000. The
first cross section includes the 1950–75 subperiod, and the second cross section includes the 1976–2000
subperiod. GDP p gross domestic product.
domestic product, the share of young people in the population, proxies for
cultural characteristics such as religious affiliation and ethnolinguistic fractionalization, and a measure of democracy.
Appendix A presents information on the definitions and sources of our variables. Table 8 presents their summary statistics.
We can now suggest a number of specific hypotheses, consistent with the
model previously outlined, regarding the potential role of the aforementioned
factors, starting with the variables that we consider to be focal to our analysis—
namely, immigration and legal tradition. Throughout the remainder of this article, we organize our comments in terms of the effects of each of our regressors
on the probability of adopting either a mixed regime or a jus soli regime instead
of a jus sanguinis regime. This organization follows the specification of our
multinomial logit estimates, where we consider jus sanguinis to be the reference
category.
Migration (M it) is measured by the stock of migrants as a percentage of the
population at the beginning of each subperiod. For the first subperiod, the
available data refer to the stock in 1960,14 whereas for the second subperiod they
14
Earlier data are not available. See Section A3.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
115
refer to the stock in 1980. By considering the migrant stock near the beginning
of each period, we avoid any potential endogeneity problem of migration with
respect to citizenship laws, because it is unlikely that stocks evaluated at the
beginning of the period can be affected by subsequent changes in citizenship
laws.15 A positive coefficient for the mixed regime would indicate that high rates
of migration incline countries toward it rather than toward jus sanguinis, with
a similar interpretation for the jus soli regime.
A crucial control in our regressions, as previously discussed, is legal tradition
in the matter of citizenship. We measure it with a dummy for countries that
apply jus sanguinis at the beginning of each subperiod (Sit). We select this dummy
because jus sanguinis is the most persistent of the three regimes, which thus
suggests a particularly significant role for this initial legislation. A negative value
for the coefficient of the dummy for a mixed and jus soli regime would imply
that jus sanguinis countries are less likely to end up in the mixed and jus soli
groups, thus confirming the persistence of the original laws.
The interaction between the jus sanguinis dummy and migration should reveal
additional information: if positive, its coefficients would indicate that jus sanguinis countries facing large migration tend to add jus soli elements. In particular,
a positive coefficient for the mixed regime would suggest convergence toward
the intermediate group. On the other hand, the coefficients of the interaction
could also turn out to be negative because, in the presence of a large stock of
migrants, the reaction of natives could be conservative. To assess the total effect
of migration on jus sanguinis countries, we also need to take into account the
coefficients of the interaction.
Turning to our geopolitical dummies, if border stability really counts as a
prerequisite for the introduction of automatic birthrights for immigrants, as
suggested by some of the political theories mentioned in Section 1, we should
expect negative signs for the coefficients for our border change dummies. We
introduce the Latin America dummy to capture the peculiarity of this region’s
experience. As explained in Section 3, most of Latin America adopted jus soli
long before our sample period, so its current regimes are not determined by
postwar developments and, in particular, by its postwar migration experience.
If indeed the behavior of Latin America differs significantly from the rest of the
sample in being associated with a higher probability of adopting jus soli, it should
exhibit a positive coefficient for this kind of legislation. For southern Europe,
we should expect a positive coefficient for the mixed regime because these countries experienced quickly increasing migration during the second subperiod, with
most of the revision of the legislation toward mixed regimes occurring in the
past 15 years or so. Finally, because migration data reveal that countries with
15
Even taking into account the anticipation of future changes in citizenship laws in making
migration decisions, the endogeneity of our migration measure is ruled out by the fact that such
changes were extremely rare during the first half of the twentieth century. The potential endogeneity
of migration is further addressed in Section 6.3.1, where alternative measures of migration are
introduced.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
116
The Journal of LAW & ECONOMICS
small populations tend to have large and erratic migration figures, with a disproportionately small effect on their legislation, we should expect negative signs
for the coefficients of this dummy.
The size of government is meant to proxy for the nature of the welfare state:
if a thicker, more expensive, and more redistributive structure represents an
obstacle to automatically granting citizenship to the children of relatively poor
immigrants, we should find negative coefficients. On the other hand, if young
immigrants offer a solution to domestic demographic imbalances, we should
find that countries with a higher share of young people in the population are
less prone to adopt jus soli elements, thus resulting in negative coefficients.
In an effort to capture additional dimensions of cultural differences, other
than legal tradition, we include the share of Catholics in the total population
and an index of ethnolinguistic fractionalization.
The establishment of a consolidated democracy—measured by Political
Rights—should exert a positive effect on the probability of jus soli legislation,
even though hostility toward the assimilation of outsiders may persist for a
protracted period even in a democratic country.
Pairwise correlations among our dependent and independent variables are
presented in Table 9. Current Citizenship Laws, the dependent variable, is highly
correlated with initial citizenship laws, as identified by Jus Sanguinis Origin
(⫺.64), whereas its correlation with Civil Law is much lower (⫺.15). Citizenship
laws are also significantly correlated with Migration Stock, Small Country, Political Rights, Catholic Share, and Ethnolinguistic Fractionalization. Pairwise correlations between all our independent variables are not reported, for reasons of
brevity, and can be summarized as follows. Jus Sanguinis Origin is negatively
correlated with Decolonization (⫺.31) and Latin America (⫺.43). The Latin
America dummy is positively associated with Catholic Share (.52). Levels of
political rights tend to be low in countries with high ethnolinguistic fractionalization (⫺.36). The share of young individuals in a population is positively
associated with migration stocks (.35). Overall, these stylized facts are in line
with previous research and economic intuition. It is also clear that several of
our independent variables are closely interrelated and that it may be difficult to
disentangle their specific effects on the evolution of citizenship laws.
6.2. Results
The results of our multinomial logit models are in Table 10, which presents
three different specifications. Multinomial logit (1) is the core specification and
includes only the core variables—Migration Stock and Jus Sanguinis Origin plus
the period dummy. Multinomial logit (2) is an expanded specification that adds
to specification (1) the dummies that we discussed above. Finally, multinomial
logit (3) is our full specification, which adds to specification (2) the other potentially relevant economic and institutional regressors. Jus sanguinis is the reference category. Hence, the results in Table 10 report the effect of the explanatory
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
117
Table 9
Pairwise Correlations among Dependent and Independent Variables
Variables
Migration Stock
Net Migration Flow
Jus Sanguinis Origin
Civil Law
Decolonization
Berlin Wall
Other Border Changes
Latin America
Southern Europe
Small Country
Sub-Saharan Africa
British or Portuguese Colony
Socialist
Oil
Political Rights
Catholic Share
Ethnolinguistic Fractionalization
Government Consumption
Share of Young
Log GDP Per Capita
Gini Index
Current
Citizenship Laws
Change in
Citizenship Laws
⫺.12*
⫺.11⫹
⫺.64**
⫺.15**
⫺.09
⫺.08
⫺.07
.60**
⫺.05
.14*
⫺.12*
.02
⫺.25**
⫺.13*
.29**
.39**
⫺.10⫹
⫺.03
⫺.03
.05
.22**
⫺.12*
⫺.06
.48**
.33**
⫺.42**
.24**
⫺.02
.04
.10⫹
⫺.12*
⫺.19**
⫺.26**
.14*
⫺.06
.18**
.10⫹
⫺.20**
.06
⫺.20**
.34**
⫺.24**
Note. The sample includes two cross sections of 162 countries. The reference period is
1950–2000. The first cross section includes the 1950–1975 subperiod, and the second cross
section includes the 1976–2000 subperiod. GDP p gross domestic product.
⫹
P ! .10.
* P ! .05.
** P ! .01.
variable on the probability of choosing either the mixed or the jus soli regime
relative to the jus sanguinis regime.
Starting with specification (1), we find that the core variables are all significant.
In particular, Migration Stock and Jus Sanguinis Origin display negative coefficients for both the mixed and jus soli regimes, whereas both coefficients are
positive for Period. This means that large migration and a jus sanguinis origin
decrease the probability of adopting mixed or jus soli legislation, rather than jus
sanguinis, and that the probability of adopting mixed or jus soli legislation
increases in the second subperiod.
In specification (2), which includes the dummies, Jus Sanguinis Origin still
exerts a negative effect on the probability of adopting a mixed or jus soli regime,
whereas Migration Stock remains significantly negative only for the probability
of adopting a mixed regime. The period dummy is significantly positive only
for the probability of a mixed regime. As expected, Decolonization displays two
negative coefficients; in other words, having experienced a border change related
to decolonization negatively affects the probability of adopting either a mixed
or a jus soli regime. The southern Europe dummy has a positive coefficient for
the mixed regime, confirming that the regimes of southern European countries
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Table 10
Determinants of Citizenship Laws: Multinomial Logit Estimates
Specification (1)
(N p 300)
Mixed
Migration Stock
Jus Sanguinis Origin
Period
Decolonization
Southern Europe
Latin America
Small Country
Jus Soli
Specification (2)
(N p 300)
Mixed
Jus Soli
Government Consumption
Share of Young
Catholic Share
Ethnolinguistic
Fractionalization
Political Rights
Log likelihood
Maximum-likelihood R2
McFadden’s R2
McFadden’s adjusted R2
Cragg and Uhler’s R2
Count R2
Adjusted count R2
Mixed
Jus Soli
⫺.051⫹
⫺.055**
⫺.054⫹
⫺.032
⫺.167*
⫺.045
(⫺1.86)
(⫺2.81)
(⫺1.79)
(⫺1.61)
(⫺2.30) (⫺1.53)
⫺2.044** ⫺6.048** ⫺2.712** ⫺6.117** ⫺4.568** ⫺7.59**
(⫺4.39)
(⫺5.43)
(⫺4.99)
(⫺4.90)
(⫺5.24) (⫺4.38)
1.887**
1.193**
1.409*
⫺.272
2.385**
.211
(4.28)
(3.02)
(2.28)
(⫺.47)
(2.65)
(.27)
⫺1.296⫹ ⫺2.211**
⫺.084
⫺1.523
(⫺1.87)
(⫺3.41)
(⫺.09)
(⫺1.55)
⫹
1.224*
⫺.153
1.234
1.187
(2.03)
(⫺.14)
(1.69)
(.87)
⫺.799
2.866**
⫺.621
4.495**
(⫺.70)
(3.73)
(⫺.51)
(2.93)
⫺.804
.415
⫺2.566⫹
.523
(⫺1.23)
(.69)
(⫺1.88)
(.64)
Jus Sanguinis Origin #
Migration Stock
Constant
Specification (3)
(N p 224)
⫺1.092**
(⫺2.73)
⫺195.77
.45
.32
.29
.53
.71
.32
.626*
(2.48)
⫺.024
(⫺.03)
⫺155.94
.58
.46
.4
.68
.81
.55
1.353*
(2.43)
.156⫹
(1.81)
.035
(1.07)
⫺.193⫹
(⫺1.75)
.012
(1.46)
.084⫹
(1.75)
.063**
(2.68)
⫺.397*
(⫺2.43)
.01
(.70)
.023
(.02)
.548**
(2.84)
3.716
(1.05)
⫺95.24
.68
.57
.44
.79
.83
.64
1.277
(1.02)
.426*
(2.46)
11.165*
(2.09)
Note. Jus sanguinis is the reference category. The reference period is 1950–2000. Robust z-statistics clustered
at the country level are in parentheses.
⫹
P ! .10.
* P ! .05.
** P ! .01.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
119
have a higher probability of becoming mixed. The Latin America dummy has
a positive coefficient for jus soli because countries in this region have a higher
probability of adopting this regime. The dummy for small countries is not
significant.
Finally, in specification (3), the effects of migration, a jus sanguinis origin,
and the period dummy are confirmed.16 The interaction term between Migration
Stock and Jus Sanguinis Origin is significant and positive for both the mixed
and jus soli regimes, uncovering a tendency for countries with a jus sanguinis
origin that experience large migration to add jus soli elements. However, the
strength of this tendency is questioned by the fact that, if one evaluates the
coefficients of Migration Stock and the interaction together, the total effect of
migration becomes negligible. In this extended version, the coefficient for Decolonization loses significance. The role of the Latin America and southern
Europe dummies is confirmed, whereas Small Country now has a negative effect
on the probability of adopting a mixed regime. The size of government has a
positive and significant coefficient for the probability of adopting a jus soli
regime, whereas Share of Young exerts a negative effect, which supports the
hypothesis that countries with relatively old populations are more likely to choose
mixed and jus soli regimes. Catholic Share and Ethnolinguistic Fractionalization
are both insignificant, and a high degree of democracy positively affects the
probability of adopting either a mixed or a jus soli regime.17 For specification
(3), inspection of the marginal coefficients in Table 11 confirms that migration
increases the probability of jus sanguinis and decreases that of a mixed regime
and that the interaction between jus sanguinis and migration is negative, consistent with the positive coefficients for jus soli and a mixed regime in Table
10.18 Moreover, the marginal effect of Decolonization reveals that this regressor
retains a significantly negative influence on the probability of jus soli. The marginal effects also allow us to quantify the effect of our regressors. In particular,
having experienced a decolonization border change decreases the probability of
being a jus soli country by approximately 14 percent, whereas an increase in
migration of 1 percentage point increases the probability of being a jus sanguinis
country by approximately 2.5 percent and decreases the probability of having a
mixed regime by 2.3 percent. In addition, an increase in migration of 1 percentage
point for a jus sanguinis country decreases the probability of being jus sanguinis
by approximately 2.7 percent, in combination with the 2.5 percent increase due
to the direct effect of migration.
We also consider additional covariates that have often been found to be significant in related research on the determinants of institutions. Quantitative and
16
For specifications (1) and (2), we obtain the same results using a balanced sample composed
of the 224 countries that constitute the reference sample for estimation of our full specification.
17
For all three specifications, we verified that our results are not driven by outliers.
18
The estimated marginal effects are calculated by holding the independent variables at their means.
For dichotomous independent variables, the marginal effect is the change from zero to one, holding
all other variables at their means.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
120
The Journal of LAW & ECONOMICS
Table 11
Determinants of Citizenship Laws: Marginal Effects
(Multinomial Logit, Full Specification)
Migration Stock
Jus Sanguinis Origin
Period
Decolonization
Southern Europe
Latin America
Small Country
Jus Sanguinis Origin # Migration Stock
Government Consumption
Share of Young
Catholic Share
Ethnolinguistic Fractionalization
Political Rights
Jus Sanguinis
Mixed
Jus Soli
.025*
(2.49)
.875**
(18.7)
⫺.303⫹
(⫺1.95)
.123
(.80)
⫺.291
(⫺1.51)
⫺.611**
(⫺4.15)
.088
(.59)
⫺.027*
(⫺2.36)
⫺.010*
(⫺1.98)
.061**
(2.65)
⫺.002
(⫺1.29)
⫺.124
(⫺.56)
⫺.107**
(⫺3.50)
⫺.023*
(⫺2.09)
⫺.193*
(⫺2.56)
.336**
(2.86)
.016
(.13)
.170
(1.03)
⫺.194**
(⫺2.85)
⫺.203**
(⫺3.16)
.021
(1.50)
.004
(.76)
⫺.018
(⫺1.18)
.001
(1.14)
⫺.028
(⫺.16)
.069**
(2.61)
⫺.001
(⫺.34)
⫺.681**
(⫺7.92)
⫺.034
(⫺.44)
⫺.139⫹
(⫺1.77)
.121
(.51)
.805**
(5.75)
.114
(.86)
.006
(.92)
.007*
(2.01)
⫺.043*
(⫺2.15)
.001
(.50)
.152
(.96)
.037⫹
(1.90)
Note. The marginal effects shown above denote the multinomial logit estimates of our full specification
(specification [3]) in Table 10). The reference period is 1950–2000. Robust z-statistics clustered at the
country level are in parentheses. N p 224.
⫹
P ! .10.
* P ! .05.
** P ! .01.
qualitative development indicators, such as income per capita and inequality,
could reveal whether a richer, more equal country is more prone to adopting
jus soli elements. However, both per capita gross domestic product and the Gini
index of inequality tend to be associated with migration and with democracy
and fractionalization; therefore, they are unlikely to add independent explanatory
power to a regression. In fact, they fail to add any further significance to the
previous results. Moreover, a dummy for oil countries could account for the
fact that most of them have been experiencing huge immigration, which has
had no effect on their still very restrictive legislation (often based on Islamic
family law). A dummy for socialist countries might work as an alternative to
our Berlin Wall dummy.19 However, we cannot include the socialist, oil, Berlin
Wall, and other border change dummies in the regressions because the countries
19
The correlation between the Berlin Wall dummy and socialist country dummy is .48.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
121
identified by them do not exhibit enough variability with respect to the dependent
variable.20
Overall, our results indicate that migration, the original laws, and our geopolitical dummies exert a significant effect on current citizenship laws and that
other factors, such as government size, demographics, and democracy, also contribute to their determination. In particular, we show that legal tradition tends
to affect the current legislation persistently, even though the process of transplantation can be discontinuous in the case of former colonies. In addition, legal
tradition interacts in a complex fashion with the way countries react to migration.
Overall, our results suggest that migration inclines national legislations in the
direction of jus sanguinis, not jus soli. However, this correlation conceals a more
complex pattern, which is revealed once the interaction between migration and
legal tradition concerning citizenship is considered. Countries with a jus soli
origin react to increasing migration by adding jus sanguinis elements. But countries with a jus sanguinis origin that have experienced more immigration tend
to add jus soli provisions so that, on balance, in jus sanguinis countries the effect
of migration turns out to be negligible. Therefore, the evidence does not support
the hypothesis of convergence toward a mix of provisions suggested by some
political theories (see Section 1) because liberal countries tend to restrict legislation while restrictive countries tend to resist innovation. The presence of
countervailing forces highlighted by the model is therefore confirmed by our
findings. Contrary to the implications of the model, the welfare burden proves
not to be an obstacle for jus soli legislation. However, this could be because
many of the countries with extended welfare systems may favor immigration
because of their demographic crises. Indeed, as predicted by the model, demographic stagnation encourages the adoption of mixed and jus soli regimes.
Moreover, the effect that we observe for the size of government could be explained by the fact that it proxies for European-style, relatively open social
democracies. Finally, the evidence confirms that a higher degree of democracy
is associated with more jus soli elements, whereas cultural traits captured by
religious affiliation and ethnolinguistic fractionalization appear to be irrelevant.
In sum, our empirical findings match the theoretical insights of the model
and provide a deeper understanding of the forces shaping citizenship laws, in
particular as far as the effect of migration is concerned.
6.3. Robustness
In this section, we present a number of alternatives to our benchmark regressions to investigate whether they are robust to different specifications, samples, and estimation techniques.
20
For example, countries affected by the Berlin Wall always adopted jus sanguinis.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
122
The Journal of LAW & ECONOMICS
6.3.1. Alternative Specifications
We experiment with our multinomial logit specifications by using alternative
covariates. First, we replace our measure of migration with a range of alternative
measures. As outlined above, our migration measure was chosen to minimize
potential endogeneity bias. When we replace our migration stocks with average
migration flows (computed with reference to each subperiod),21 the coefficients
for migration are insignificant in all three specifications, which could suggest
that migration flows are endogenous with respect to citizenship laws. Additional
tests involve alternative measures of migration stocks.22 In Table B1, specification
(1) is the full specification with the average migration stocks for 1960–70 and
1980–90. Some differences emerge in the coefficients for migration and the
interaction of migration with jus sanguinis, with the latter losing significance.
Because, in the postwar period, migration has been highly regulated by policy
in most receiving countries, and because citizenship laws can be viewed as part
of migration policy, the simultaneous determination of citizenship laws and
migration represents a concern when we enter within-period data on migration
instead of beginning-of-period data. In summary, our beginning-of-period migration stocks prove to be the most adequate measures of the role of migration.23
Because jus sanguinis and jus soli are, in principle, closely associated with the
civil and common-law systems, respectively, the influence of legal tradition can
also be analyzed through a dummy for the presence of a civil law tradition.24 If
the coefficients of the two alternative dummies are the same, one could conclude
that our detailed codification of the original citizenship laws does not add much
to what we can already learn from the broader legal tradition of a country.25
When we replace Jus Sanguinis Origin with Civil Law (Table B1, specification
[2]), the latter turns out to be substantially less significant and reduces the
significance of most other regressors, which suggests that civil law is a much
weaker predictor of current citizenship laws than original citizenship laws. We
also replace our decolonization dummy with a dummy for British or Portuguese
colonies (identifying those countries that were characterized by jus soli legislation
during the colonial period) and with the sub-Saharan Africa dummy,26 but both
have insignificant coefficients. When we replace the southern Europe dummy
with migration flows—because the quickly increasing immigration in the second
subperiod determines the peculiar behavior of this region—we find that its
21
Migration stocks and flows have a correlation of .45.
As possible alternatives for the first and the second subperiods, we select 1970 and 1990, respectively, and the 1960–70 and 1980–90 averages, respectively.
23
We also experiment with a specification using the migrant stock in 1960 for both subperiods.
Results from this instrumented specification are similar to those presented in Table 10.
24
The dummy is equal to one if a country belongs to a civil law tradition and to zero if a country
belongs to a common-law tradition. The correlation between civil law and jus sanguinis is .35.
25
The correlation between the two dummies is .35.
26
Decolonization is significantly correlated with British or Portuguese Colony (.30) and SubSaharan Africa (.31).
22
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
123
coefficient is insignificant, possibly because flows are multicollinear with respect
to stocks and endogenous with respect to the dependent variable.
6.3.2. Alternative Estimation Techniques
Alternative estimation techniques broadly confirm the same results shown in
Table 10. In particular, an (unreported) alternative multinomial probit model
delivers the same qualitative results. We also run ordered logit regressions in
which current citizenship laws are explicitly treated as an ordinal variable, which
we assume here to be ordered by increasing inclusiveness—namely, we assume
that jus sanguinis corresponds to minimal inclusiveness and jus soli to maximal
inclusiveness. These results are presented in Tables B2 and B3. Our previous
conclusions are confirmed, with Migration Stock and Jus Sanguinis Origin exerting a negative effect on the adoption of jus soli.27
6.3.3. Alternative Sample Criteria
We also run multinomial logit regressions on a cross-sectional sample composed of country averages over the 1950–2000 period.28 The results for this cross
section reveal a much lower level of significance for several covariates, in particular for migration and its interaction with the initial laws.29 The same finding
applies to alternative variants with a cross section over each subperiod (1950–75
or 1976–2000). For the same cross section, an alternative ordered logit regression
also achieves much weaker results than in our panel,30 whereas ordinary least
squares regressions using our citizenship policy index in 2001 and an indexed
version of 2001 citizenship laws as alternative dependent variables both have
insignificant coefficients for migration (see Table B4).31 This may be because a
single cross section of countries includes less information than our panel.
7. An Alternative Approach: The Determinants of
Change in Citizenship Laws
In this section, we study the evolution of citizenship laws with an alternative
approach that is able to provide additional insights. Although specification (9)
focuses on current citizenship laws as the dependent variable, we develop an
27
For all specifications, we run a test for the parallel regression assumption. The test is an approximate likelihood-ratio test of proportionality of odds across response categories. In all cases, the
test provides evidence that the parallel regression assumption has been violated. Our multinomial
logit specification is hence superior to an ordered logit specification.
28
Here the dependent variable is citizenship laws in 2001; Migration Stock refers to 1960, and Jus
Sanguinis Origin in 1948 is the initial law.
29
We also experiment with migration stocks in 1970 and 1980 and with average migration flows,
with similar results.
30
We apply Cook’s distance method to the results to show that they are not driven by outliers.
31
The same qualitative results are found in an unreported regression with naturalization in 2001
as a dependent variable.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
124
The Journal of LAW & ECONOMICS
alternative specification designed to identify more specifically the determinants
of a change in laws.
7.1. Empirical Specification
In the alternative specification, the dependent variable is categorical and can
take three values: ⫺1 if the country changes its laws toward jus sanguinis, 0 if
no change occurs, and 1 if the country changes its laws toward jus soli. For our
panel, we run a multinomial logit model of the following form for its full
specification:
Vit p a ⫹ bM it ⫹ gTt ⫹ Z it d ⫹ hit,
(10)
where a is a constant term and hit is an error term. The vector Z it contains all
the variables previously described, except for those that present zero-cell problems—namely, Jus Sanguinis Origin, its interaction with Migration Stock, and
Latin America. Pairwise correlations among our new dependent variable, changes
in citizenship laws, and the independent variables are presented in Table 9,
showing that migration is again negatively correlated with the dependent variable.
7.2. Results
Regression results for the multinomial logit model are presented in Table 12,
where the reference category is no change. Hence, the results represent the
influence of the explanatory variable on the probability of choosing either restriction toward jus sanguinis or expansion toward jus soli. Migration has a
positive effect on the probability of restriction and a nonsignificant effect on the
probability of expansion. The period dummy indicates that the second subperiod
witnesses an increase in the probability of expanding. Decolonization and Southern Europe exert a positive effect on change in both directions, more significantly
so for restriction for the case of Decolonization, whereas the opposite holds for
Southern Europe. The negative coefficient for restriction confirms that the size
of government actually prevents it, whereas, again, the presence of a relatively
young population provokes resistance to extension, meaning that countries with
a relatively old population are more likely to liberalize their legislation. Ethnic
diversity emerges as a significant factor of change toward jus sanguinis.
Table 13 reports the marginal coefficients for the regressions in Table 12 and
confirms the restrictive effect of migration emerging from Table 12.32 Moreover,
the implications of Table 13 are in line with those of Table 11, which reports
the marginal effects for the multinomial logit specification. For instance, in Table
11, Decolonization exerts a significantly negative effect on the probability of
having jus soli legislation, which is consistent with the fact that, in Table 13, it
emerges as a factor that facilitates change toward jus sanguinis.
As for specification (9), we perform a full set of robustness checks for specification (10). The Gini index once again fails to add any significance, whereas
32
The estimated marginal effects are calculated as explained in footnote 18.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
125
Table 12
Determinants of Change in Citizenship Laws: Multinomial Logit Estimates
Toward Jus Sanguinis
Migration Stock
Period
Decolonization
Southern Europe
Small Country
Government Consumption
Share of Young
Catholic Share
Ethnolinguistic Fractionalization
Political Rights
Constant
Log likelihood
Maximum-likelihood R2
McFadden’s R2
McFadden’s adjusted R2
Cragg and Uhler’s R2
Count R2
Adjusted count R2
⫹
.039
(1.90)
.652
(.65)
2.898**
(2.82)
2.2⫹
(1.94)
⫺.341
(.50)
⫺.05⫹
(1.95)
.02
(.15)
⫺.006
(.70)
1.981*
(2.23)
.104
(.84)
⫺4.364
(.96)
⫺107.58
.33
.3
.15
.45
.81
.16
Toward Jus Soli
⫺.006
(.12)
1.736⫹
(1.73)
2.058⫹
(1.68)
2.155**
(3.44)
⫺.593
(.60)
.032
(1.12)
⫺.493**
(2.82)
.001
(.12)
2.508
(1.52)
.234
(1.08)
9.771*
(2.04)
Note. The reference category is no change in citizenship laws. The reference period is 1950–2000.
Robust z-statistics clustered at the country level are in parentheses. N p 224.
⫹
P ! .10.
* P ! .05.
** P ! .01.
per capita gross domestic product has a significant effect on expansion. Alternative measures of migration confirm an irrelevant effect on change in the laws.
When we replace Decolonization with British or Portuguese Colony, the alternative dummy is again significant because of the removal of the Latin America
dummy, which comprises several former Portuguese colonies. The dummy for
sub-Saharan Africa is not significant when used instead of Decolonization.
Overall, these results complement those of Section 6 because they highlight
which factors induced the observed evolution of the legislation for each possible
direction of change. In particular, migration is confirmed as a factor that favors
restriction.
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
126
The Journal of LAW & ECONOMICS
Table 13
Determinants of Change in Citizenship Laws: Marginal Effects (Multinomial Logit)
Migration Stock
Period
Decolonization
Southern Europe
Small Country
Government Consumption
Share of Young
Catholic Share
Ethnolinguistic Fractionalization
Political Rights
Toward Jus Sanguinis
No Change
Toward Jus Soli
.003*
(2.01)
.044
(.58)
.374*
(2.00)
.299
(1.10)
⫺.022
(⫺.53)
⫺.004*
(⫺2.02)
.003
(.29)
⫺.000
(⫺.71)
.145*
(2.12)
.007
(.75)
⫺.003
(⫺1.26)
⫺.102
(⫺1.31)
⫺.442*
(⫺2.42)
⫺.415*
(⫺2.33)
.038
(.74)
.003
(1.27)
.013
(1.18)
.000
(.60)
⫺.223**
(⫺2.39)
⫺.015
(⫺1.10)
⫺.000
(⫺.19)
.058*
(2.05)
.069
(1.16)
.116
(1.06)
⫺.015
(⫺.65)
.001
(1.28)
⫺.016**
(⫺2.74)
.000
(.19)
.077⫹
(1.89)
.007
(1.12)
Note. The marginal effects presented are the multinomial logit estimates shown in Table 8. The reference
period is 1950–2000. Robust z-statistics clustered at the country level are in parentheses. N p 224.
⫹
P ! .10.
* P ! .05.
** P ! .01.
8. Conclusion
We studied the theoretical and empirical determinants of the legal institution
of citizenship in the postwar period using a new set of data that we compiled.
We developed a simple median-voter model in which natives grant citizenship
rights to migrants after taking into account the associated benefits and costs.
The model predicts that migration has a potentially ambiguous effect on the
legislation and that this effect is also influenced by cultural factors, including
the degree of inclusiveness in a country as reflected by its original citizenship
legislation. In bringing the model to the data, we found that citizenship laws
indeed respond endogenously and systematically to a number of economic and
institutional factors. Our investigation reveals that migration has an overall negative effect on liberalization of citizenship legislation and adoption of jus soli
elements. Moreover, a country’s legal tradition affects the way that it responds
to migration. In particular, jus soli countries react to increasing migration
through restriction, whereas the effect of migration is negligible in jus sanguinis
countries. Therefore, the evidence does not support the hypothesis of convergence toward a mixed regime that includes both jus soli and jus sanguinis
elements. Border instability emerges as a decisive factor in shaping citizenship
laws, particularly in connection with decolonization, reflecting discontinuities
This content downloaded from 198.91.32.141 on Sun, 11 Mar 2018 06:53:43 UTC
All use subject to http://about.jstor.org/terms
Evolution of Citizenship
127
for the transplantation of legal institutions. Countries with larger welfare systems,
older populations, and more extensive political rights tend to be associated with
more diffused elements of jus soli.
More generally, a clear implication of our investigation is that institutions
should not be presumed to be exogenous, because they do adapt to both economic and noneconomic factors. The endogeneity of institutions to economic
factors represents a challenge for research aimed at demonstrating that institutions are crucial determinants of economic performances. By showing that
citizenship laws are shaped not only by the broader legal origins but also by
other institutions, such as the internal system of political rights and the international system of relations as reflected by state borders, we also establish that
different institutions are interrelated. This finding represents another challenge
for further research on the formation of legal rules and the effect of institutions
on economic outcomes.
Citizenship laws are still changing. Further research will study the future
evolution of citizenship policy by using projections of international migration
patterns in combination with available predictions about the future course of
democratization and border changes. Finally, citizenship laws can be viewed as
a link within a legal system between the public and the private spheres of influence. Many issues that fall within the former sphere of influence—such as
commercial law, labor regulation, and government activities—have already been
investigated. Our methodology can be extended to the study of other evolving
bodies of law such as family law, rules of inheritance, and women’s rights.
Appendix A
Data
A1. Citizenship-at-Birth Classification
Group 1: Countries with a Jus Sanguinis Regime
In this group, we include countries where citizenship is passed to a child on
the basis of at least one of his or her parents being a citizen of that country,
regardless of the actual country of birth of the child, and where citizenship is
not granted as a result of birth within the country. In the application of jus
sanguinis, countries may differ with regard to some factors—for example, the
right of the father versus the right of the mother to transmit citizenship by
descent, the requirement of citizenship for one or both parents, and the relevance
of the parents’ marital status. Most of these factors depend on the interaction
between local family law and citizenship law. A common exception to the general
principle of jus sanguinis is automat...