No way out: travel restrictions and authoritarian regimes

Abstract

Why do some authoritarian regimes restrict their citizens’ freedom to travel abroad? This article posits that there is an inverse relationship between the freedom of foreign movement and authoritarian stability. Authoritarian leaders recognize this and hence try to control foreign travel by their citizens, for two reasons. First, civil rights increase significantly when travel policies are relaxed. Secondly, opportunities for contestation increase in the presence of civil rights. We test these propositions using panel data and various indicators of regime performance. Our results strongly suggest that by restricting travel, authoritarian regimes inhibit civil liberties. By suppressing civil liberties moreover, these regimes buy themselves stability.

Keywords:

• travel restrictions
• civil rights
• political liberalization

Acknowledgments

We are especially grateful to Melisa Patel for research assistance, Jonathan Moses for suggesting relevant literature, and to Abdeslam Marfouk for answering our data queries. For comments on various iterations of this paper, we would like to thank John Entelis, Justin Gest, Seva Gunitsky, Annika Hinze, Christian Houle, Carolyn Logan, Michael Miller, Margarita Mooney, Jonathon Moses, and particularly Jonathan Blake, Jim Jasper, Susan Woodward, participants at the CUNY Graduate Center workshop on Politics and Protest, and participants in ‘The International and Local Dimensions of Regime Change and Stability’ panel of the 2014 Midwest Political Science Association annual conference.

Notes

1. Recent work has addressed the effects of emigration on political development (Lodigiani & Salomone, 2012; O’Mahony, 2013; Pfutze, 2012; Spilimbergo, 2009). This literature is concerned, generally speaking, with the democratizing potential of outbound migration (Moses, 2011, 2012). These works notwithstanding, ‘one cannot really talk about a developed literature on sending-state emigration or practices in the way one can cite myriad works on receiving-state immigration policy [original emphasis]’ (Brand, 2006, p. 2).

2. Electoral autocracies can be divided into hegemonic and competitive authoritarian regimes. The latter differ from regimes that are more hegemonic in that ‘opposition forces use democratic institutions to contest seriously for executive power’ (Levitsky & Way, 2010, p. 16). Since our interest is in authoritarian regimes in general, we only distinguish between full autocracies and electoral autocracies, that is, between closed regimes and regimes that are ‘minimally pluralist,’ ‘minimally competitive,’ and ‘minimally open’ (Schedler, 2006, p. 3).

3. Restricting travel can also prove unpopular among those unable to exercise this right.

4. Fargues (2011, p. 5) observes in this regard that ‘North African governments long distrusted their migrant communities in Europe, suspecting them of being centers of political opposition, before coming to court them for their remittances’.

5. The coding is based on Cheibub et al. (2010) democracy/autocracy dichotomy, with autocracies in turn subdivided into full and electoral autocracies (Boix, Miller, and Rosato, 2013, p. 18). In full authoritarian regimes, political competition is suppressed. To be sure, regimes can allow participation even while suppressing contestation, as in communist one-party autocracies. Dahl (1971) labels these regimes inclusive hegemonies. In these countries, everyone votes but participation is meaningless since there is no alternative to the ruling party slate. These regimes are thus treated as full autocracies in the literature (e.g. Goldstone et al., 2010, p. 195). Electoral autocracies allow multiple parties to take part in electoral competition, but deny a level playing field to the opposition or violate the civil rights of their opponents (Levitsky & Way, 2010, p. 7).

6. As of 2009, 20 countries placed restrictions on the ability of women to legally obtain a passport or to travel out of their country, including Afghanistan, Burma, Iran, Jordan, Kuwait, Libya, Qatar, Saudi Arabia, Sudan, Swaziland, Syria, United Arab Emirates, and Yemen (United Nations Development Programme, 2009, p. 40). As of 2005, 11 countries required government permission or an exit visa for all citizens to travel: Belarus, Burma, Cuba, Ecuador, Equatorial Guinea, Iran, Laos, Libya, North Korea, Ukraine, and Uzbekistan (McKenzie, 2007).

7. On 14 January 2013, however, the Cuban government enacted the most comprehensive changes in its travel policies in over five decades. See http://www.ibtimes.com/cuba-travel-restrictions-lifted-us-wet-foot-dry-foot-policies-remain-same-1013514#.

8. See the article appearing in the New York Times on 22 February 2003 titled ‘No Exit: China Uses Passports as Political Cudgel’, available at http://www.nytimes.com/2013/02/23/world/asia/chinese-passports-seen-as-political-statement.html?pagewanted=all&_r=0.

9. Civil rights have been empirically related to this dimension in Dahl’s theoretical work. Of the two dimensions Dahl identified, contestation explains most of the variation in regime scores. See Coppedge, Alvarez, and Maldonado (2008).

10. Some also argue that restricting travel allows autocracies to prevent brain drain, that is, the flight abroad of qualified professionals on whom they have invested significant resources in training (Hirschman, 1978, p. 101). Breunig et al. (2012, p. 8) cite Cuba, Belarus and Turkmenistan in this regard. What is primarily an economic problem becomes political as well, since many autocracies are less-developed countries (LDCs) struggling to build viable nation-states in a difficult domestic and international environment. We assess this argument empirically but find it wanting in our analysis.

11. Restrictions in the Middle East may have had their origins in beliefs about gender differences. There is no reason to expect, however, that what is at issue in these countries is simply gender discrimination. Once restrictions are in place, they demobilize citizens politically. Consequently, unrestricted travel poses a threat to Middle Eastern governments. One indication of how threatening foreign travel remains to these governments is that, in some countries restrictions apply not only to married women, but also to unmarried ones.

12. Milošević’s Socialist Party regime consolidated power over its opposition thanks in part to the exodus from Serbia of many young educated Serbs (Moses, 2011, p. 12).

13. Examples include Cuba in 1965, 1980, and 1995 and East Germany in 1989.

14. We conducted a factor analysis to verify that the scores all reduce to one dimension or concept and hence can be used as an additive index. The results confirmed our expectation.

15. Freedom House also codes civil liberties worldwide. We opted not to use this index, however, since it also contains information on the independence of the judiciary and other social and economic liberties that in our view belong in a separate category of rights. For an overview of the Freedom House index, see http://freedomhouse.org/report/freedom-world-2013/checklist-questions-and-guidelines#.U1Uf1tJDuSp.

16. This is so as to avoid endogeneity, that is, a situation in which travel policy is not only or primarily affecting civil rights, but the reserve is also at work.

17. An alternative is an error correction model (ECM), which allow analysts to examine effects that can be permanent or transitory (DeBoef & Keele, 2008). The problem with these models is that they assume a long-run equilibrium between a dependent variable Y and an independent variable X. The goal is to estimate how quickly Y returns to equilibrium after a shock (i.e. a change in X). The problem with such a formulation is that travel restrictions may have precisely the effect an ECM is not equipped to evaluate, a change in the regime equilibrium.

18. Witness, for example, the color revolutions in the post-communist world (Diamond, 2008, p. 191) and the Arab Spring that began in North Africa in 2011 and is still being felt throughout the Middle East.

19. A Wooldridge test demonstrates the presence of serial correlation in the idiosyncratic error term.

20. One good example of this in our data is Jordan, a dictatorship with restrictions on travel and very high levels of social globalization. Jordan nonetheless appears stable politically.

21. See http://www.iab.de/en/daten/iab-brain-drain-data.aspx. While the calculations exclude tourists and asylum seekers, they include immigrants that entered these countries illegally intending to stay. The 20 countries are: Australia, Austria, Canada, Chile, Denmark, Finland, France, Germany, Greece, Ireland, Luxembourg, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, and United States.

22. When data is normally distributed, the coefficient of variation is given by the ratio of the standard deviation to the mean. When the mean equals the standard deviation, the coefficient of variation equals 1. This value is thus an anchor point between a low-variance and a high-variance series. See http://www.readyratios.com/reference/analysis/coefficient_of_variation.html. Our data is log-normally distributed. Consequently, the identity of the standard deviation and the mean will not provide the same intuitive interpretation. Nevertheless, the within-country coefficient of variation for emigration rate is only 0.54. This compares to the between- country coefficient, which amounts to 3.38.

23. The UN recognizes the problem. See its methodology for estimating the international migrant stock by country of origin, available at http://esa.un.org/unmigration/Documentation.pdf.

24. For a typology of the effects of outbound migration on home country politics, see Kapur (2014) and a special issue of the journal Studies in Comparative International Development (Meseguer & Burgess, 2014).

25. Examples include remittances and foreign aid.

26. A better approach in our view would have been to use public opinion data from sources such as the World Values Survey to establish empirically the existence of cultural proximity. We, however, chose not to pursue this approach for two reasons: first, personal contacts and information flows are much closer to what we think of as social globalization, leading to some extent to the cultural proximity among some nations that we observe. National cultures and any similarity among them, moreover, is also a function of important variables such as socio-economic development. We do observe that social globalization mostly depends on a country’s level of development, a finding confirmed by a recent study of visa restrictions placed on country nationals by destination countries. For more information, see Henly and Partners’ Visa Restrictions Index (available at https://www.henleyglobal.com/index.php). Whereas citizens of wealthy countries face few barriers to travel, citizens of poor, autocratic, or unstable countries are virtually trapped in their own countries by strict visa requirements, at least if they want to travel legally. Our data invariably confirms this, as the pairwise correlation between GDP per capita (in 2005 constant purchasing power parity dollars) and social globalization is high and statistically significant (r = 0.762; p < 0.000). We thus exclude economic development from our models. To retain the first two subcategories of social globalization while excluding the third, we rescaled the original index by multiplying it by their combined weight, 69%, or 0.69.

27. Brinks and Coppedge (2006) code 17 regions in their analysis, but several of their country classifications appear to be typos, so we corrected them using Wikipedia. The changes are as follows: Turkey from ‘Southern Europe’ to ‘Middle East’, Solomon Islands from ‘Sub-Saharan Africa’ to ‘Pacific’, Sao Tome and Principe from ‘Caribbean’ to ‘Sub-Saharan Africa’, Singapore from ‘East Asia’ to ‘Southeast Asia’, Maldives from ‘Sub-Saharan Africa’ to ‘South Asia’, Lesotho from ‘Sub-Saharan Africa’ to ‘Southern Africa,’ and Chad from ‘Northern Africa’ to ‘Sub-Saharan Africa’.