Does an Inclusive Citizenship Law Promote Economic Development?

2.1 General Principles

The notion of citizenship has varied considerably throughout the evolution of mankind, though there are some common elements of citizenship principles. A general definition of citizenship is membership in a political society or group. With the birth of the nation state in the nineteenth century came the need to find a distinction between who belongs to the state and who does not, and therefore to create a legal distinction between nationals and foreigners.^[2] The former benefitted from rights such as voting, circulating within the county, or rights to work in the country for instance. They also had responsibilities, such as serving in the military, or paying taxes and voting (see Weil 2001).

Thus, most countries established in the nineteenth century, or at independence, a “code of nationality,” whose basic principles are often still intact today. This code, in most cases, defines who is a national, and if the nationality can be acquired, how it can occur (e.g., through marriage or through having lived for several years in the country). At the end of the 19th century, two visions of nationality laws were confronting each other. One vision, based on the declaration of human rights, is inclusive, and can extend, with certain conditions, the nationality to each type of citizen. The other view, more exclusive, defines a nation more as an ethnic community.

These two visions lead to three concepts of the citizenship laws.^[3] the inclusive version is reflected in the law of the soil (“Jus soli”); the principle being that a child born within a country’s territory acquires that country’s nationality—the norm in the “new world.” In this view, bonds of citizenship extend beyond basic kinship (and blood-ties) and encompass people of different genetic and geographical backgrounds. It describes the relation between a person and an overall political entity such as a nation and denotes membership in that entity. It is also generally characterized by some form of political participation, although the extent of such participation can vary considerably from minimal duties such as voting to active service in government. For males (and sometimes females), it is (was) often based on some form of military service or expectation of future military service. When there are many different ethnic and religious groups within a nation, citizenship may be the only real bond which unites everybody as equals without discrimination—it is a “broad bond”. In other words, jus soli citizenship links “a person with the state” and gives people a universal identity—as a legal member of a nation—besides their identity based on ethnic ties (see Turner and Hamilton 1994).

The exclusive version of the law of the blood (“Jus sanguinis”) promotes the principle that a child acquires the nationality from his parents (—and in some countries only from the father, not the mother), with the place of birth not being a factor.^[4] This is commonly the case in Europe, Asia and parts of Africa. In this philosophy, citizenship has legal aspects including rights (sometimes seen as a bundle of rights or a right to have rights). By definition, citizenship provided through jus sanguinis has an element of exclusion, in the sense that citizenship derives meaning, in part, by excluding non-citizens from basic rights and privileges. Citizenship is a powerful force to exclude persons, such as immigrants. In this sense, citizenship is not only about getting rights and entitlements, but it is a struggle to reject claims of entitlement for those residing outside the citizenry, such as migrants (Bertocchi and Strozzi 2010). Having said that, the integration of “outsiders” is determined by more than simply the jus soli/jus sanguinis dichotomy. For instance, naturalization is also crucial to the integration of immigrant or minority groups. If first-generation immigrants can be naturalized, then jus sanguinis does not necessarily cause the exclusion of immigrants (see discussion in the empirical part below).^[5]

The third option, becoming increasingly common, is a mixture of these two, creating a hybrid system. Whereas countries often started as jus soli or jus sanguinis principles, in recent times, reflecting the rising immigration levels, many countries have changed their policies to move towards the other’s philosophy. Germany, in 1999 significantly reformed its citizenship law which was based on the jus sanguinis principle, making it possible for foreigners residing in Germany for years, and particularly foreign children born in Germany, to acquire German citizenship. On the other hand, countries such as Britain have tightened the rules of jus soli law.

One consequence of the mixed regime is the modern emergence of “double nationalities.” Until the 19th century, the double allegiance was frowned upon, as it was viewed as a sign of potential traitor, or of a force of subversion, and hence prohibited in most countries. It also raised practical questions such as in which country should the person do his/her military service (some countries require military service from women). In case of war between the two countries, which camp would the person chose to fight for? While this attitude was understandable in a world ravaged by war, attitudes started changing in the second half of the twentieth century. These individuals would in fact increasingly be viewed as potential interlocutors in case of crisis. The United Kingdom was one if the first countries to recognize the dual citizenship in 1949, followed by France (1973) and others more recently. Some countries, such as Japan, Cameroon or the Republic of Central Africa, refuse the double nationality to this day (in the case of Japan, the child must choose at the age of 18), and can go as far as to deprive a child of its nationality if he/she seeks to be naturalized somewhere else. In many countries, formally colonized, such as in sub-Saharan Africa, this recognition of the dual-nationality started late, and only in the 1990s with the end of the cold war. This is in part a reflection of history. At independence, these countries wanted a clear break from their former masters and forced their people to choose their allegiance to the new country, or the former colonial power. This was especially an issue for colonialists who had been in the country for sometimes generations, and who didn’t want to choose between the two. The risk in many cases was that decedents of colonialists would pack their bags and take with them all their skills and knowledge. In addition, the other risk they faced was that the educated who had migrated overseas, may decide to acquire the nationality of the place where they reside, cutting the link between the country and its most experienced and educated citizens. This global phenomenon of increasing acceptance of the double nationality reflects in part the new reality emanating from the end of the cold war, with countries allowing for the double-nationality (e.g., Algeria, Burkina Faso, Mali) becoming the norm in sub-Saharan Africa, rather than the exception.

2.2 History of Citizenship Laws by Region

In continental Europe, jus soli law has historically been the dominant choice, a reflection of the feudal traditions linking a person to the lord, on whose land he was born (Bertocchi and Strozzi 2010). This explains why during the nineteenth century, most of the European continent drafted their citizenship laws on this model. Japan, which modeled its constitutional law on continental Europe, followed in the same tradition. One exception is France. With the French revolution, this link was broken, and the law became one of jus sanguinis. France, at the end of the nineteenth century, switched back from jus sanguinis to jus soli to grow demographically, following the lost war against Prussia, as well as to integrate foreign communities—this would it was thought permit for a strong military. By contrast, the British, kept the jus soli tradition, and spread it throughout the British Empire. Before WWI therefore, a pattern emerged where common law countries tended to have jus soli laws, while civil law countries tended to have jus sanguinis laws (see also Weil 2001).

The Western Hemisphere countries, such as the United States, chose the jus soli citizenship rule. Consistently with a country of immigrants, and with the specific purpose to protect the birthrights of black slaves, the 14th amendment in 1868 encoded the jus soli principle. The relatively limited benefits of U.S. citizenship versus US residency also means that the fiscal costs of providing citizenship to a newcomer were limited, with the potential upside of having an extra person working (with the cost of education falling on the migrants’ home country) (see Bertocchi and Strozzi 2010).^[6] Similarly, Canada a large and sparsely populated country, welcomed migrants into the country with jus soli citizenship law, as more people ensure that the country can expand and get the right skills.

In colonized countries, citizenship laws were initially typically transferred from the colonial power (Bertocchi and Strozzi 2010; Manby 2016).^[7] Countries with a strong, and long national sense, such as China, Egypt, or Japan, tend to make it hard to acquire the nationality, or to have a second passport. Others, as is typical in newer countries in the Western Hemisphere without necessarily a strong national sense of itself, make it generally easy to be naturalized, and often provide for automatic citizenship if a child is born on the soil of that country, the US being the best example. Many SSA countries, which were created artificially by the French, British and Portuguese, lacked a sense of national cohesion. At independence, there were major revisions in citizenship laws: Most former French colonies initially stuck with the principle of jus soli. On the other hand, many former British and Portuguese colonies switched to jus sanguinis citizenship laws, driven by ethnic considerations. Given that many countries were artificially created without consideration for diverse local ethnic groups, leading to conditions of political instability, jus sanguinis had the advantage, it was thought, of more easily creating a “national identity”. This was the case in Sierra Leone for instance, where the 1961 Constitution established that citizenship could only be transmitted by decent, and only to those whose father and grandfather were Sierra Leoneans and of Black-African descent (see Aleinikoff and Klusmeyer 2000).^[8] The problem with such a law in a heterogenous ethnic environment was that it excluded various ethnic and tribal groups, causing enormous problems, especially in the context of weak institutions. The Congolese Constitution of 1964 for instance, in an effort to exclude Rwandan immigrants, recognized citizenship only for persons whose parents were members of tribal groups established within the territory before 1908 (see Bertocchi and Strozzi 2010). Predictably, marginalizing certain groups, and in some cases creating de facto stateless people, who would in some circumstances rebel, was a consequence of such policies.

A look at Figure 1 illustrates the citizenship laws across the world. Citizenship laws, while marked by history, and changing only slowly, are not static. They evolved in some cases, often driven by economic or national considerations. As argued by Weil (2001), when the legal tradition fulfills the interest of a state in terms of migration, the national laws are maintained. Canada, a country of immigration, has maintained its jus soli laws. In continental Europe, countries often historically a source of emigration—to keep a link to immigrants—the authorities have kept the jus sanguinis citizenship laws in place, such as Germany. However, as Germany became increasingly a country of immigration, with large Turkish and other communities, it was felt that to integrate these communities better, the law had to be changed. In some ways, while there remain big variations across countries and regions, there has been, starting from a low level at the end of the 20th and beginning of the 21st century an increasing hybridization of jus sanguinis and jus soli laws, or a convergence. Jus sanguinis countries have increasingly accepted some form of naturalizations based on some tight criteria, while jus soli countries have tightened the requirements to citizenship, as was the case in the United Kingdom or Australia, for fear that migrants may burden too heavily the social welfare system for instance.

Figure 1:

Citizenship Laws, 2014

Source: Authors (compiled from various sources).

In recent decades, most of the conventions of the acquisition of a nationality have evolved. As an example, many countries have in the past regarded marriage as an important status changing event in people’s lives. For spouses, the most widespread practice among states, until the beginning of the twentieth century, was that a woman should obtain the nationality of her husband upon marriage. In other words, when marrying a foreigner, the wife would automatically acquire the nationality of her husband and lose her previous nationality. Even after the nationality of a married woman was no longer dependent on the nationality of her husband, legal provisions were still retained which automatically naturalized married women (and sometimes married men as well). This could lead to various problems, such as men becoming subject to military service obligations in another country—a major concern in times of international hostility. In more recent times, there has been a shift towards a principle that neither marriage nor dissolution of marriage automatically affects the nationality of either spouse, nor of a change of nationality by one spouse during marriage automatically affecting the nationality of their spouse. However, in many jurisdictions spouses can still obtain special and fast processing of applications for naturalization. Since the turn of the century, many countries have tightened the automaticity further, for fear of “marriage blanc”—a sham marriage—which permits the foreigner (often from a developing countries) to obtain the nationality of the spouse.

2.3 Jus Soli vs Jus Sanguinis: Different Impact on Economic Development?

How do citizenship rights impact economic development? A citizenship law can be thought of as a conflict resolution or conflict generating institution. If it is inclusive, it can provide positive social capital, raising trust, reducing transaction costs and reducing the probability and intensity of conflict for instance. This is especially the case when other conflict resolution institutions are weak (e.g., government is corrupt, court of laws is weak), as is the case in most developing countries.

As a result, our hypothesis is that citizenship laws impact economic development in lower-income countries, but not so much in countries with robust institutions. In practice, we hypothesize that jus sanguinis makes integration more difficult, and hence negatively impacts economic development. Several channels can be identified:

Distorting Investment. There are several channels through which investment may be negatively impacted, by having residents that do not have the nationality. First, such investors have short-time horizons—communities without citizenship in countries with weak institutions are vulnerable and have an incentive to store their money overseas. Thus, they would not necessarily maximize their exposures to the country, and would, particularly around elections, be cautious. In addition, their investment would be distorted: if property rights are not well protected because one lacks the local citizenship, one goes for investments with a quick payback (e.g., some agricultural sectors) or with limited capital investment needs (e.g., trading). The incentive is also to have a diversified business empire—rather than being too big in one sector—to avoid predation, and to diversify risk. This also means lack of specialization: investors will not go into assets that are high specialized (e.g., biotech firm), but rather into assets that are easily re-deployable. While not quite the same as not possessing the nationality, Montalvo and Reynal-Querol (2005) find, using a polarization index, that ethnic (religious) polarization has a large and negative effect on economic development through the reduction of investment.

Political Instability and Corruption. Marginalized minorities without nationalities are oftentimes at polar extremes, particularly in sub-Saharan Africa—either excluded from economic life or playing a disproportionally significant role in the local economy. Without citizenship, the marginalized minority cannot vote or impact public life through democratic means. One way for disenfranchised groups to draw attention to themselves is through protests and other forms of civil disobedience, or in other cases by a call to arms. As a result, governments may suppress minorities, spending more on military, etc., thereby weakening growth.

Conversely, when a non-national group plays a disproportionality significant role in the economy, their lack of protection by the state is a source of concern (given weak institutions and a potentially predatory state). Because of their vulnerabilities, influential minorities have an incentive to influence the political process through non-democratic means. They may have to use bribes, encouraging corruption, and therefore weakening institutions. In addition, investments must often be financed with local politicians, creating a patronage relationship. Thus, these communities don’t necessarily benefit from a strong state that could potentially expropriate them, or tax them significantly for that matter: a weak state that doesn’t tax (but also doesn’t provide public goods) may in fact be—in the short-run—in their interest. This also means that these minorities may have an incentive to weaken the state, through the policy of divide and rule, by bribing rival political groups (Thomson 1995). An inclusive citizenship law, on the other hand, is likely to align the interest of outsiders with those of the existing citizen, anchoring their longer-term interests on those of the nation.

One striking example if for instance the Lebanese diaspora, which is encountering different citizenship laws in the countries in which it is implanted. In Colombia, where a large Lebanese community exists, as jus soli applies, they have the citizenship and are well integrated: they are involved in local politics, and in businesses. Their long-term interest—in the rule of law, good infrastructure, good education of the local population—is aligned with their business interests. Even in case of political instability, they are unlikely to be expropriated because of their ethnicity. In Western Africa, however, the same business community is experiencing very different circumstances: In Sierra Leone for instance, the small Lebanese community, despite having been there for several generations as well, does not have the citizenship of the country due to the jus sanguinis law. Thus, it is not well integrated and not visibly present outside of businesses. Their longer-term interests are not necessarily aligned with those of the citizens of the country, as they know they can be expropriated any time. The result is that they may not, unlike in Colombia, benefit from the presence of a strong state. The risk of expropriation means that they are myopic in their investment decision, and will try to keep as much money out of the country as possible, rather than investing domestically. They are often involved in activities that don’t require much capital, such as import-export, or activities where cash flow start rapidly to accumulate (e.g., cinema, hotels). Socially, they tend to also be an endogamous community, marrying little with the local community.^[9]

Reducing Public Sector Efficiency. Studies have documented how divisions, whether ethnic, religious or linguistic ones, often negatively impact public sector performance, increasing patronage, lowering level of trust among the population and ultimately impacting negatively on economic development (see Alesina et al. 1999; Easterly and Levine 1997; Habyarimana et al. 2007; Miguel et al. 2004). The literature argues that the underdevelopment of sub-Saharan Africa is due some of the citizens not feeling fully integrated or full members of the state. In addition, the lack of inclusion of some segments of the population could impact government activities and the quality of institutions as well. While there is no study to our knowledge on the impact of nationality laws on public sector efficiency, there are clear parallels from the ethno-linguistic literature that would suggest an analogous effect. Ethnic groups were often split into different countries, meaning that there was often a lack of national identity (Englebert et al. 2002). As a result, this may make it more difficult to reach on public goods delivery, as would be the case in a more homogenous society. Mauro (1995), has discussed the impact of ethnic fragmentation on government activities and quality of institutions. Mauro (1995) was among the first to illustrate that a high level of ethnolinguistic diversity implies a lower level of investment. Easterly and Levine (1997) have shown that per capita GDP growth is inversely related to ethnolinguistic fractionalization in a large sample of countries. A large literature on US localities show that in more ethnically fragmented communities, public goods provision is less efficient, participation in social activities and trust is lower, and economic success, measured by growth of city sizes, is inferior. Evidence that trust does not travel well across racial lines is also supported by experimental evidence. Alesina et al. (2003) provide new measures of ethnic, linguistic and religious fractionalization for about 190 countries, and confirm most of these findings. These studies lead one to draw the conclusion that government spending may not be optimal in the presence of marginalized groups (see Habyarimana et al. 2009).^[10]

Distorting Labor market. Under jus sanguinis law, local minorities without citizenship may be excluded from parts of the labor market. In many countries, entire professions are not allowed to immigrants. For instance, in Thailand foreigners cannot become hairdressers, nor accountants. In Cambodia or Madagascar, foreigners are not allowed to purchase land. In a country like France, (non-EU) foreigners are not allowed to become directors of funeral companies. In these cases, a jus soli laws expands the labor market in a way that jus sanguinis law does not—providing a potentially larger pool of labor and creating a more efficient economy.

One caveat that needs to be highlighted is that citizenship laws might not have a uniform effect on economic development. As an example, one trend that has been observed in recent years is “Olympic citizenships” (jus talenti) (Jansen et al. 2018; Shashar 2011), whereby a “talent-for-citizenship exchange” is provided. According to this principle, a rising number of countries, typically wealthy ones like the US, or Gulf countries, increasingly provide a selective path to citizenship because these individuals may bring in prestige or wealth. In this context, citizenship itself is treated as a form of tradable asset, and not primarily with an aim to integrate individuals. This phenomenon is only accentuating an existing trend, which is to have a more calculated approach to citizenship in which a premium is placed on extraordinary ability and talent, rather than on direct ties to the country. Empirically, it’s difficult to test for this effect, given limited data, and because the focus in this paper is on developing countries, where the “Olympic citizenship” phenomenon is even less common than in advanced economies. While this phenomenon may develop further going forward, it’s impact on economic development is likely to be small (see also Shashar 2006). Indeed, conferring citizenship in this way is a rather recent phenomenon, and often done to a few select individuals, not on a large scale compared to other avenues of acquiring citizenship.

It is also worth noting that despite the economic advantages that jus soli law may provide, citizenship laws do not simply reflect pure economic considerations. They go to the heard of what it means to be a citizen, and hence involve emotional, identity and security considerations for instance (Shashar 2006). There are indeed good reasons for why countries may opt for jus sanguinis laws. First, jus sanguinis laws allows a country to maintain links with the descendants of its emigrants, which is viewed as an advantage for many developing countries, leading to the creation of large diasporas like Haitians for instance (Weil and Hansen 2001). Relatedly, under jus sanguinis, a country can deny citizenship to locally born children of citizens born abroad, under the logic that they lack a meaningful connection with the homeland’s culture (with the risk, though, that such a stringent system might exclude some of a country’s most productive potential citizens, especially in countries with large numbers of diasporas or expatriates). In regions with unstable borders, jus sanguinis might help preserve national identity (see Bertocchi and Strozzi 2010). This applies to Azerbaijan for instance. Likewise, the attempt to preserve an ethnic heritage through jus sanguinis has played a role in countries such as the former socialist in Eastern Europe and Asia, which have gone through a recent period of turmoil, or are going through a rapid demographic change. There are even economic reasons for why a country may reject jus soli laws, as they might incentivize migrants who lack the credentials (e.g., education, job market skills, etc.) to immigrate through lawful channels. Therefore, citizenship laws should not only be looked at through an economic lens, but other factors, such as identity and culture, as well as security considerations must be borne in mind.

3.1 The Data and Sample

In the following section, we will assess empirically the impact of citizenship laws on economic development. How do we measure citizenship laws? We compile a data set of citizenship laws drawing on a series of sources, including the U.S. government report on citizenship laws of the World,^[11] the office of the United Nations High Commissioner for Refugees (UNHCR), constitutions and nationality laws released online by official country authorities, Manby (2016) and Bertocchi and Strozzi (2010). Using multiple sources enables us to document more accurately the features of different citizenship laws, avoid misinterpretations, and cover the largest sample of countries possible.

For this study, we collected data on the characteristics of the most recent citizenship laws (as of 2014), which for many countries have been in effect for decades as citizenship laws change little over time. We constructed a dummy variable taking one if a country has a jus soli regime or zero otherwise, as our main variable of interest. This variable is further refined subsequently, by distinguishing between a mixed regime and a jus sanguinis regime for countries where the jus soli dummy variable takes zero. We also compiled information on other features of citizenship laws, including the possibility or not to acquire citizenship by marriage, the existence of gender discrimination in the transmission of nationality, minimum requirements for naturalization such as period of residence, renunciation of other citizenship, and other requirements (e.g., language skills, income threshold).

Although the data are available for a worldwide sample, we restrict the main analysis in this paper to a sample of 105 developing countries for which data on citizenship laws and other variables are available (see Annex I). This is motivated by two factors. First, the literature section underscores how citizenship rights could harm economic development through exclusion, political instability, and lower private investment in countries with persistently weak institutions. These mechanisms are likely to more prevalent in developing economies that lack the necessary institutional checks and balances to tackle these issues. Second, on a technical ground, bundling developing and advanced economies raises the challenge of sample heterogeneity, which model specifications and econometric techniques can only partially address.

3.2 Model Specification

The basic idea is to assess to what extent citizenship laws explain differences in GDP per capita across developing countries, controlling for other factors. The underlying hypothesis we will test is that if a country has jus soli laws, it is presumably more inclusive in nature, and potentially more likely to be more prosperous, while a country with no jus soli laws is more likely to have disenfranchised groups that are not integrated in society and the economy, or that create tensions either because they are economically disenfranchised or are very successful compared to the dominant group.

Visualizing the data provides a vivid illustration of the striking difference in the average real GDP per capita in jus soli countries vs non-jus soli countries, with the gap having widened over the years (Figure 2). In 2014, income per capita in jus soli countries was 80% higher than in non-jus soli countries. Splitting the sample of non-jus soli countries in mixed regime and jus sanguinis countries confirms that jus soli countries are richer, but there is no clear pattern when comparing mixed regime to jus sanguinis countries. This may not be surprising as inclusiveness in many countries with a mixed regime might be similar to that of jus sanguinis countries, particularly in cases where the acquisition of citizenship by birth is subject to stringent requirements.

Figure 2:

Average Real GDP per Capita in Jus Soli and Non-Jus Soli Developing Countries, 1970–2014 (Constant US Dollars)

Sources: World development indicators, authors’ estimations.

The main purpose of the model will be to estimate how much of the real GDP per capita gap observed in Figure 2 is explained by the jus soli law. To do so, we build on the seminal paper of Acemoglu et al. (2001) which provides a starting point for our econometric model. The specification of the model is as follows:

As in Acemoglu et al. (2001), the measure of economic performance retained as a dependent variable in this paper is the level of GDP per capita at the end of the period considered in the study (1970–2014). The dependent variable y_i is explained by our main variable of interest, which is the dummy variable Jsoli capturing the presence of a jus soli law, and a set of control variables X_i. A positive ∂1, the coefficient of interest, would lend support to the hypothesis that a jus soli law is associated with higher economic performance. Consistent with Acemoglu et al. (2001) we include latitude and a measure of institutional quality (the index of rule of law) in the control variables. McArthur and Sachs (2001) argue that physical geography, which affects the disease environment and health characteristics of a country, matter for economic performance. Compared to the index of protection against expropriation used by Acemoglu et al. (2001), the rule of law indicator, compiled by the World Bank, is available for a larger sample of countries, therefore we decided to use it to avoid losing data points. Other control variables include factors that have been widely shown as key determinants of growth (hence economic performance), such as education (measured by the secondary school enrollment rate), trade openness (measured by the sum of imports and exports divided by GDP) and financial depth (measured by the ratio of credit to private sector over GDP). Although these variables are not considered in the original model of Acemoglu et al. (2001), we think it is important to include them given the potential impact of economic performance, and the added advantage of reducing the risk of omitted variable bias. ε_i is the error term of the country i.

One caveat to bear in mind is that the goal of this paper is not to looking at the growth literature per se (for an extensive survey, see Durlauf et al. 2005; Johnson and Papageorgiou (2020)). Therefore, the regression analysis only borrows a growth model that is one of the workhorses of the literature, and that is a priori the most suitable to the question addressed. We are mindful that issues of “model uncertainty” and the sensitivity of empirical results to small changes in specifications and samples (see Durlauf 2009; Henderson et al. 2012), which this paper tries to address through robustness checks.

Another caveat is the assumption that jus soli laws, captured by a dummy variable, are more inclusive than other regimes. While integration of immigrant or minority groups is multidimensional, which a dummy variable might not fully capture, there is evidence from advanced economies that jus soli countries fare better in labor market mobility, access to education and health, political participation and access to nationality for immigrants (see Figure A1 in the appendix). Absent those data for developing countries, the citizenship dummy variable could be a reasonable proxy.

3.3 Econometric Strategy

At first sight, the ordinary least square estimator is an appropriate econometric approach to estimate Eq. (1) as the citizenship law is considered as a time-invariant variable during the study period spanning from 1970 to 2014. In this case, GDP per capita is the level in 2014 and the control variables are averaged over the period 1970–2014.

However, our OLS specification does not control for time-invariant country specific effects that allows one to capture unobservable country characteristics affecting economic performance. To address this issue, we also run panel regressions. The period of study is split in seven sub-periods of five years each, with GDP per capita being the level at the end of each sub-period and the control variables averaged over the 5-year subperiods. Since the citizenship law variable is time-invariant—we estimated that less than 5 percent of the countries in the sample might have changed their citizenship law during 1975–2014—the panel specification can only be estimated by the random effect estimator,^[12] which nevertheless relies on the strong assumption of the absence of correlation between the right-hand side variables and the country specific effects. We relax this assumption by using the Hausman-Taylor estimator, while resorting to the Mundlak test (Mundlak 1978) to identify which variables are likely to be correlated with the country specific effects.

Investigating the link between citizenship law and economic performance through the lens of the different econometric approaches discussed above allows us to compare how a country with a jus soli law performs on average relative to a country with alternative citizenship laws. Although quite informative, these approaches, however, do not consider the counterfactual outcome. Indeed, an interesting angle from which to reconsider the main research question of this paper, is to ask how a country with a jus soli law would compare to a situation where that country had adopted a different citizenship law. The counterfactual is obtained from a control group that has similar characteristics as the country of interest, but has not been “treated.” We exploit the fact that many countries had a jus soli law during the colonial period (by adopting the citizenship law of their colonizers),^[13] but at independence in the 1950s and 1960s, some countries decided to retain the jus soli law, while others chose to switch to jus sanguinis law. It is assumed that this change in citizenship law is exogeneous to the level of income per capita (and instead was mainly driven by political and ethnic considerations), and therefore provides an opportunity to tease out the effect of the citizenship law on economic performance.^[14] The synthetic control method of Abadie and Gardeazabal (2003), and Abadie et al. (2010) will be used to estimate the citizenship law’s treatment effect.

In sum, with the use of different economic estimators, different ways to slice the data and various model specifications, we aim to ensure that the impact of the citizenship law on economic performance observed in the empirical results is tangible and robust.

4.1 OLS Estimates

Table 1 reports the OLS regressions of the log of per capita real income in 2014 on the citizenship laws, plus the control variables in a sample of developing countries. First, we start by running a simple regression with latitude as the only control variable, and then we control for the rule of law, and subsequently for the other control variables, namely education, financial development and trade openness (column 1–6). Regardless of the specification, the results show that the jus soli law is strongly associated with higher income per capita. To get a sense of the magnitude of the marginal impact of the jus soli law on income level, we take the specification in column 7 of Table 1 which includes all control variables. The coefficient on the variable of interest indicates that a shift to from a jus sanguinis a jus soli law could yield on average a 34 log point increase in income per capita (about a 40½ percent increase), which represents close to three-fifth of the gap observed between the income per capita in jus soli and non-jus soli countries in Figure 2.

When controlling for the rule of law, the coefficient on the jus soli variable drops slightly (column 2) which tends to support the idea that the benefits of a jus soli law would be weaker in countries with strong institutions. We further tested this hypothesis by interacting the rule of law and jus soli variables (column 3). Although the coefficient on the interaction term has the expected sign, it lacks significance at conventional levels. Nevertheless, the coefficient on the jus soli variable drops further. These results suggest that the rule of law may play a role in how a jus soli law affects economic performance, but it is only a part of the story as the coefficient on the jus soli variable remains statistically significant. It is worth noting that in both specifications, the rule of law variable enters the regressions significantly, thus supporting the widely-shared view that institutional quality matters for growth (see for instance Acemoglu and Robinson 2010).^[15]

Turning to the other control variables, higher human capital and well development financial system are conducive to higher GDP per capita, as expected. However, we find little evidence for the geography to explain economic performance as the latitude variable is not statistically significant. Similarly, trade openness appears to have an ambiguous impact on income per capita.

In column 8 of Table 1, we substitute to the jus soli dummy variable a categorial variable taking the value of three for jus soli countries, two for mixed regime countries and one for jus sanguinis countries. The higher the value of the variable, the more inclusive the citizenship law. The coefficient of interest is positive and significant, supporting the above findings.

To address potential endogeneity issues given the potential feedback effect from income per capita to the control variables, we rerun the regressions in Table 1 using the average values in the 1990s (1990–99) for rule of law, education, financial development and trade openness. The regressions shown in the appendix (Table A1) yield comparable results.^[16]

4.2 Panel Estimates

The main purpose is to see if OLS results hold using different econometric techniques and augmenting the model with country specific and time dummies. We start with a pooling regression, and then use a random effect estimator (column 2 and 3, Table 2) as the jus soli variable does not change over time.^[17] The jus soli variable continues to be highly significant in all specifications, thus providing support to the idea that jus soli law countries tend to be more developed than countries with alternative citizenship laws. As the coefficients of the random effect model may be biased if there is a correlation between the explanatory variables and the country specific effects, we overcome this challenge by performing the Mundlak test to select the control variables that are likely to violate the assumption of the random effect model. The Mundlak test suggests that the rule of law, education and financial development variables are likely to be correlated with the country specific effects.^[18] As shown in column 4 of Table 2, rerunning the regression using the Hausman-Taylor estimator does not alter the results; the jus soli variable still comes out statistically significant.

This result is also robust when the control variables ranging from rule of law to private credit ratio are lagged by one period (column 5), or when their initial values are used (column 6).

Finally, we rerun the regressions in Table 2 replacing the jus soli variable with the categorial variable taking the value of 1 for jus sanguinis countries, 2 for mixed regime countries and 3 for jus soli countries. The results presented in Table 3 remain in line with the above findings. Moreover, the results reveal that the mixed regime appears not to be associated with higher income per capita relative to jus sanguinis countries (column 5), consistent with Figure 2 which shows no clear evidence of consistently higher income per capita for mixed regime countries.^[19] This may not be surprising as in many mixed regimes, jus sanguinis remains the prevalent mode of acquisition of citizenship and getting citizenship through birth on the territory may be subject to numerous conditions. These include typically: (i) the requirement that one of both parents should be born in the country, (ii) the visa status of the parents, and (iii) a minimum period of residence for the parents and/or the child.

4.3 Additional Specifications

To test the robustness of our main result, we add three additional control variables, namely government consumption as a percent of GDP, an index of ethno-linguistic fractionalization and a measure of protection of property rights, in line with previous studies such as Alesina et al. (1999), Easterly and Levine (1997), and Montalvo and Reynal-Querol (2005). Interesting results emerged and are displayed in Table 4. Higher government consumption is associated with lower GDP per capita, as is a high ethno-linguistic fractionalization, whereas better property rights stimulate economic development. The jus soli variable retains its sign and significance, confirming our previous findings (Table 4, column 1 to 4).

Since in addition of the law of the blood or the law of the soil, a country’s citizenship can also be acquired by marriage and naturalization, we constructed a composite index to measure the inclusiveness of the citizenship law along these dimensions. The index is obtained from the principal component analysis of several sub-indicators, namely: a dummy variable taking 1 if citizenship can be obtained through marriage and 0 otherwise; a dummy variable taking 1 if there are no restrictions for women with regard to the right to pass their nationality to a spouse of a child, and 0 otherwise; the number of years of residency required before being eligible to apply for naturalization; a dummy variable taking 1 when a country’s naturalization law does not require renunciation of other citizenship, and finally the number of other conditions required for naturalization (including language skills and income threshold). For this composite index, the higher the score, the less restrictive is the acquisition of nationality by marriage and naturalization. As expected, and consistent with the previous results, this composite index enters the regression with a positive and significant coefficient (Table 4, column 5). Moreover, the interaction term between this composite index and the jus soli dummy variable is negative and significant, pointing to some substitutability among the different dimensions of a citizenship law as far as the economic effects are concerned. This implies that for a jus sanguinis country, the income loss relative to a jus soli country, could be reduced if the former has more favorable legal provisions on access to citizenship through marriage and naturalization, everything else being equal.^[20]

4.4 Accounting for the counterfactual using the Synthetic Control Method

As underlined in Bertocchi and Strozzi (2010), decolonization had a major influence on citizenship law. For instance, many Latin American countries switched to a jus soli law at the independence, a departure from the civil law of European settlers where acquisition of citizenship by blood is most common. Bertocchi and Strozzi (2010) noted that this move was a way to break with the colonial political order and to prevent the former colonial power from making legitimate claims on citizens born in the newly independent countries. Many former British and Portuguese colonies moved the other way at independence, by rejecting the jus soli law of their colonizers in favor of the jus sanguinis law. On the other hand, most African colonies that were subject to civil law countries kept jus sanguinis principles in the post-colonization period.

We exploit the change in citizenship rules at independence to assess from a different perspective how citizenship law affects economic development. Table 5 presents the dynamics of citizenship laws for the countries in the sample from the period before independence to 2014.^[21] Countries along the diagonal, accounting for half of the sample, have not changed their citizenship law; a high number, reflecting the persistence of citizenship laws. It is also noteworthy that inclusiveness of citizenship laws has regressed over time. While only 6% of jus sanguinis countries (1 country) moved to a jus soli law after the independence, and 40% (seven countries) of jus sanguinis countries added jus soli elements to their citizenship laws, 33% (six countries) of mixed regime countries and 37% (19 countries) of jus soli countries tightened their citizenship laws by moving to a jus sanguinis law. Also, among the countries that have followed jus sanguinis principles in 2014, half of them used to have a jus soli law before the independence.

Ideally, the synthetic control method (SCM) allows to analysis the different regime changes but there are several considerations to account for. First, as we want to be selective by focusing on bold changes in citizenship laws, we leave out the transition to and from mixed regimes, allowing to reduce the number of different combinations of regime changes from six to two. Second, since only one country^[22] in our sample moved from a jus sanguinis law to a jus soli law, analyzing this regime change presents a major downside: the findings cannot be generalized to a large group of countries. Third, the SCM requires that the dependent variable and the covariates have no missing data points, which presents a challenge for a sample of developing countries over a long-time period. For this reason, it is important to focus on the regime change with the largest sample. Considering all these constraints, we narrow down the analysis to the pool of countries that had jus soli law in the colonial period, giving us a sample of 26 countries after dropping countries with insufficient data (see Annex 1). Then, we observe the most recent citizenship law to identify countries that have maintained the jus soli law (15 out of the 26 countries) and the rest that shifted to jus sanguinis law. The basic premise is that if a more inclusive citizenship law favors economic development as posited in this paper, one should expect a country that moved away from a jus soli law to be poorer than a comparable country which has maintained the jus soli law.

The synthetic control method (SCM) developed by Abadie and Gardeazabal (2003) and extended in Abadie et al. (2010) provides a framework to construct the counterfactual of a country that moved to jus sanguinis law, drawing from the donor pool consisting of countries that did not change their citizenship rules after the independence. The underlying idea of the SCM is that, even though the counterfactual of the treated unit cannot be observed, a combination of the realized outcomes in the control units (units that do not undergo treatment over the sample period) may be able to produce a reliable estimate of that counterfactual. This assumes that the optimal weights are selected in such a way that the synthetic control unit matches certain pre-treatment characteristics of the treated unit as closely as possible.^[23]Cavallo et al. (2013) further extend the SCM to allow for multiple treated units and at possibly different times.

We adopt the Cavallo et al. (2013) pooled SCM to look at the evolution of income per capita during the period 1960–2014 in the selected sample of countries.^[24] For the sake of simplicity, we assume that the change in citizenship law occurred in 1970 for all treated units.^[25] To match the treated units with the synthetic control, the covariates in Eq. (1) are good candidates for the pre-intervention variables, but longer time series data are lacking for most of them. As a result, we used latitude, an index of human capital provided the Penn World Tables,^[26] population size, real capital stock and the level of real income per capita in the pre-intervention period as covariates.^[27]

Figure 3 depicts the evolution of real income per capita in the median treated country and that of the counterfactual as estimated by the pooled SCM.^[28] Using the median smooths out noises in the estimates and ensures that the results are not driven by outliers. Consistent with the results from the standard regression techniques, countries that have switched to a jus sanguinis law ended up with a lower income per capita than where they would have been had they kept the jus soli law after they became independent. It is striking that the magnitude of the income gap is very close to the OLS estimate (Table 1, column 7). Indeed, the treatment effect of the median country estimated by the SCM points to a 37.8 log point loss in income per capita by 2014 (about a 46% decline) compared to a difference of a 34-log point found with the OLS regressions. This provides evidence of the robustness of the results across a range of econometric techniques.

Figure 3:

Synthetic Control Method: Treatment Effect of Moving from a Jus Soli Law to a Jus Sanguinis Law.

Source: Authors’ calculations.

While Figure 3 provides a useful overview of the magnitude of the treatment effect, delving into country-specific results is also informative. Figure A2 shows some heterogeneity among the treated units, with the treatment effect of eight out of 11 having the expected sign as their income per capita is lower than that of the counterfactual. For those 8 countries, the income loss from having shifted away from a more inclusive citizenship law ranges from a 137-log point (The Gambia) to a 17-log point (Nigeria) in 2014. However, surprisingly 3 treated countries outperformed their counterfactual, probably because either the model fails to account for idiosyncratic factors in the post-treatment period or the matching during the pre-treatment period is weak.^[29]

We also run placebo tests to assess the statistical significance of the effect. The tests consist in running the same model on each country in the donor pool, assuming that unit was treated at the same time, while excluding the actual treated unit from the synthetic control. Figure A3 provides evidence of a lack of a strong treatment effect in placebo cases,^[30] suggesting that the observed difference in income per capita between the treated and control unit is due to the actual treatment. To ensure that the synthetic control makes a good counterfactual, we exclude from the donor pool units for which the pretreatment root mean squared predictive error, is larger than that of treated unit. The results remain largely unchanged.