Ancestry into Opportunity: How Global Inequality Drives Demand for Long-distance European Union Citizenship

Abstract

This paper investigates the relationship between global inequality and dual citizenship by analysing citizenship acquisition from abroad in the European Union (EU). Most EU countries now offer facilitated naturalisation to descendants of emigrants and co-ethnics abroad, which requires neither residence nor renunciation of former citizenship. Since the 1990s, over 3.5 million people have used this opening to obtain dual citizenship from a European country to which they often have little if any connection. I analyse this phenomenon using a data-set that I constructed from previously unanalysed administrative statistics. The data were used to test an original theory that explains patterns of demand for dual citizenship in the context of a global hierarchy of citizenship worth. The analysis demonstrated that demand was much higher in Latin America and Eastern Europe than in North America and Western Europe. Non-Western applicants were drawn to the practical benefits of EU citizenship, and their level of demand varied in response to economic conditions like unemployment. In contrast, Western applicants displayed lower demand for citizenship and were unresponsive to economic incentives. The paper contributes to the literature by demonstrating the relationship between citizenship and global stratification as well as highlighting a widespread instrumental approach to dual citizenship.

Keywords:

• Citizenship
• Globalisation
• Transnationalism
• Migration
• European Union

Acknowledgements

I have received valuable comments from Andreas Wimmer, Yinon Cohen, Alejandro Portes, Thomas Espenshade and Edward Telles, as well as from Alfredo Garcia, Maria Abascal and the participants of Princeton's Center for Migration and Development workshop.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

[1] In this paper, I use the term ‘citizenship’ in a dual sense, to refer to both the formal status of membership in a state (which is synonymous with ‘nationality’) and the array of practices and beliefs that are included in an individual's relation to her state of citizenship.

[2] This figure includes statistics for Romania, Bulgaria, Germany, Switzerland and Finland listed in Appendix 1, as well several other sources for which year-specific data were unavailable: Hungary (622,531 acquisitions in 2011–2014, from Bálint 2014), Croatia (1,150,000 acquisitions in 1991–2006—the vast majority of them facilitated reacquisitions—from Štiks 2010) and Spain (491,101 acquisitions in 2009–2011, from Izquierdo and Chao 2015). It also uses 1998–2010 statistics for Italy (1,003,403 acquisitions, from Tintori 2012) instead of the 1998–2007 statistics used in the data-set. If we also add countries for which data were unavailable, such as France and Portugal, the total figure should be even higher.

[3] I use the term ‘long-distance’ in a political, not geographical sense (cf. Anderson 1992). Therefore, it includes dual citizenship in a neighbouring country. Furthermore, I do not deny that some people lead truly transnational lives and have strong ties to two countries.

[4] A detailed discussion of facilitated naturalisation schemes in EU countries can be found in Dumbrava (2014), Pogonyi, Kovács, and Körtvélyesi (2010) and the detailed country reports produced by the EUDO on Citizenship (http://eudo-citizenship.eu).

[5] Milanovic decomposed the contribution of aggregated within-country and between-country income variances to income inequality that was calculated for the entire world population, based on estimates of household income. The 1/3–2/3 breakdowns reported here refer to the part of the global variance in income that were explained by the combination of location and class—about 65% in 1870 and 80% in 2000 (Milanovic 2012).

[6] This pattern emerges from the democracy and conflict indices listed under ‘Independent variables’, as well the State Fragility Index (http://www.systemicpeace.org) and the Economist Democracy Index (http://www.eiu.com) and the Henley and Partners Index of International Visa Restrictions (http://www.henleyglobal.com).

[7] I include Israel in the non-Western category in spite of its high income level, because Israeli citizens do not enjoy Western levels of security and stability.

[8] I tested an additional model in which unemployment, growth, conflict and democracy were operationalised as a relation between residence-country and granting-country levels. The key findings were unchanged.

[9] It might seem at first that the depletion effect should be expressed as a negative correlation. Such a correlation, however, would correspond to a graph that zigzags up and down, with increases followed by decreases and vice versa. None of the acquisition dyads followed that pattern. A much more common pattern was an inverted-U shape, with increases followed by increases and decreases by decreases. Presumably, the depletion effect is initially offset by the diffusion effect, but gains momentum as applications accumulate. This assumption is reinforced by the fact this it was positive and highly significant in all the models. When including this variable, the Z-score and logged models produced near-identical results; when omitting it, they were very different.

[10] While we can be sure that almost all those who acquired citizenship became dual citizens, the data does not allow us to determine with certainty where citizenship applicants lived, since most granting countries only provided statistics on applicants’ citizenship. Undoubtedly, some applicants were living outside their first country of citizenship (see Mateos 2013). Nonetheless, the term ‘residence country’ accurately captures the first citizenship for at least 90% of applicants, since even high-emigration residence countries had no more than 10–12% of their population living abroad.

[11] I can read German and Italian. I used translation software for the Romanian and Finnish texts and confirmed the translations with the individuals who provided the information.

[12] It would be extremely difficult to calculate the eligible population for even one dyad, let alone all of them. For example, 18 million Americans declared Italian ancestry in the census (United States Census 2011), but this is not a reliable basis for calculating the population that is potentially eligible for Italian citizenship, since anyone with any degree of Italian ancestry is potentially eligible. In most other cases, no eligibility data were available at all.

[13] The use of aggregate data creates a potential for ecological fallacy, i.e., an unjustified inference from group correlations to individual correlations. No cross-national statistics were available on the individual characteristics of citizenship applicants. In light of these limitations, the arguments and findings do not pertain to individuals, but rather describe the effect of aggregate conditions on aggregate levels of demand. Thus, this study tries to explain operate at the level of the eligible population rather than that of the individual agent.

[14] Bulgaria and Romania restricted long-distance naturalisation in the years leading up to their EU accession. Once accepted in 2007, they began handling long backlogs, focusing on Macedonia and Moldova (Iordachi 2010; Smilov and Jileva 2013). Therefore, some of the increase in those two residence countries was caused by changes in supply, not demand. However, post-accession increases were observed in almost all residence countries, suggesting that at least part of the increase is the result of changes in demand.

[15] I set the delay at one year on the basis of an estimate that the mean processing time was between 1 and 1.5 years. I also ran the models with a two-year lag and the main findings held.

[16] The main findings were robust to running the analysis without the dyad dummies.

[17] The sample size in Table 3 is smaller than in Table 2 because the first year of observation in each dyad was omitted when the diffusion/depletion variable was included.

[18] I also ran separate regressions for each granting country and found no systematic differences between them.