Abstract
Allowing or restricting foreign movement is a crucial policy choice for leaders. We argue that freedom of foreign movement reduces the level of civil unrest under non-democratic regimes, but only in some circumstances. Our argument relies on the trade-offs inherent in exit and voice as distinct strategies for dealing with a corrupt and oppressive state. By permitting exit and thereby lowering its relative costs, authoritarians can make protest and other modes of expressing dissatisfaction less attractive for potential troublemakers. Liberalizing foreign movement can thus function as a safety valve for releasing domestic pressure. But the degree to which allowing emigration is an effective regime strategy is shaped by the economic opportunities offered by countries receiving immigrants. We find that freedom of foreign movement and the existence of economic opportunities abroad reduce civil unrest in non-democratic states. However, at high levels of unemployment in the developed world, greater freedom of foreign movement actually increases protest.
Introduction
One of the hallmarks of non-democratic regimes – particularly those we would term ‘dictatorships’ – is the central role of the removal of enemies (and potential enemies) in the survival strategies of leaders. ‘Unlike democracies, where politicians adjust policies to the median voter to be elected, brutal dictators adjust their constituency by eliminating citizens who are in opposition to the regime’ (Gregory, Schröder & Sonin, 2006: 2). While it is true that the removal of opposition is often extremely violent – as in the Soviet Gulag or the Cultural Revolution of Maoist China – there are other examples of the same ends being achieved by encouraging, or simply allowing, the option for potential troublemakers to leave (e.g. Pfaff, 2006). When this strategy works as intended, the people with grievances and the means to make them heard choose instead to find a better life elsewhere. As Pfaff (2006: 18) puts it, ‘[W]hen exit is easily available, it may tend to siphon off those alert and resourceful individuals who are the most ambitious and most discontented, subtracting from the reservoir of creativity and social capital remaining in the collectivity.’ By removing themselves, these individuals can leave behind a society less likely to challenge the regime.
This raises an interesting question for leaders of these autocratic regimes: should they ease restrictions on foreign movement if doing so might encourage troublemakers to leave? The straightforward answer seems to be ‘yes’. But if such leaders deal with grievances by encouraging exit without dealing with the underlying causes of the grievances, they expose themselves to real danger, if and when external conditions take away individuals’ hopes of realizing greater economic and political opportunities abroad. Now the troublemakers have to stay at home, remittance flows dry up, and the regime is likely left with no means of satisfying their domestic population. This can lead to unrest and protest, both peaceful and violent. Accordingly, allowing greater freedom of foreign movement might be a tempting strategy for leaders keen on reducing domestic unrest, but this decision carries with it trade-offs, as we show below, and can make the regime even more vulnerable than it otherwise would have been.
Over the following pages we draw out this logic using Hirschman’s (1970) model of exit and voice, demonstrating how the decision to either restrict or permit foreign movement has implications for instability within non-democratic countries, but that the effects of this policy choice are conditioned by the existence of economic and political opportunities abroad. Our results indicate that authoritarian regimes allowing greater freedom of foreign movement have lower levels of protest than those with more restrictions, but only while economic conditions in major receiving states remain strong. As global economic opportunities decline we see an increase in protest activity where liberal foreign movement policies are practiced, but little change in states where citizens’ right to foreign movement is restricted.
These findings have important substantive and theoretical implications. First, though other studies have applied Hirschman’s (1970) theory to explain social unrest in individual cases, such as East Germany (e.g. Hirschman, 1993; Gehlback, 2006), we apply this framework to a cross-national sample of states, shedding light on the generalizability of earlier work. 1 Second, we also contribute to the theoretical model underlying this body of research by explaining how the exit-voice dynamic is conditioned by exogenous factors – in this case, the availability of outside options. Third, our analysis indicates that the domestic effects of allowing greater freedom of foreign movement cannot be divorced from the broader global context. Liberalizing foreign movement, like economic liberalization, exposes states to an often volatile international environment, potentially generating disruptive blowback when foreign markets weaken. Finally, while the focus among scholars, pundits, and policymakers concerned with freedom of foreign movement (and migration specifically) has tended to be on the political effects in receiving countries (the countries to which migrants go), the theoretical mechanisms and results highlighted in this article suggest that more attention should be paid to the political consequences for sending countries (the states that migrants leave behind). Efforts encouraging authoritarian regimes to relax restrictions on foreign movement can, at times, actually strengthen authoritarians’ hold on power. By lowering the relative costs of the exit option, the corrupt and repressive practices that drive aggrieved citizens to action may, ironically, be made more likely to persist as a result.
Exit, voice, and protest
Our argument is that adopting a liberal foreign movement policy creates, in certain contexts, a ‘safety valve’ for domestic political pressure. 2 At the same time, such a policy choice can foster greater pressure in non-democratic regimes when economic conditions abroad effectively foreclose foreign movement. Under these circumstances aggrieved citizens are likely to experience relative deprivation as remittance flows from friends and relatives living abroad decline, and, most importantly, because leaders who adopt policies allowing and encouraging foreign movement are tempted to view them as a substitute for actually redressing grievances in society (cf. Brownlee, 2007; Magaloni, 2008; Gandhi, 2010; Slater, 2010).
The logic of our argument is rooted in Hirschman’s (1970) treatment of the concepts of exit and voice, and their fundamental interrelationship. 3 For Hirschman, exit constitutes an individual’s decision to end a relationship with an organization with which they are dissatisfied, while voice is the expression of the individual’s dissatisfaction in the hope of ending and/or reversing an organization’s policies. While these are distinct strategies for dealing with a deteriorating organization, they are related. As Hirschman (1970: 37) writes, ‘the decision to exit will often be taken in the light of the prospects for the effective use of voice’ [emphasis in original]. Hirschman goes on to discuss the trade-offs inherent in the choice between voice and exit when he writes of ‘people … solving their problems through “physical flight” [rather than] through either resignation or through ameliorating and fighting in situ the particular conditions into which one has been “thrown”’ (1970: 107). Though Hirschman’s original articulation was in the context of an individual’s relationship with a firm, we can similarly conceive of politically dissatisfied citizens as having three options for dealing with their dissatisfaction with the state (Hirschman, 1978). First, if the government allows for freedom of foreign movement, the person may leave to seek opportunities elsewhere (i.e. exit). Second, the person may stay and attempt to redress their grievances by engaging in political action, through either informal political channels like protest, or formal channels such as elections (i.e. voice). Third, the person may choose to do nothing, seeing the costs of leaving and of engaging in political action as too high.
The feasibility and attractiveness of each option is shaped in large part by political factors, such as regime type and state policy. Autocratic governments, by their very design, do not typically provide opportunities for citizens to redress grievances through formal political channels. Particular forms of voice that are available to citizens in more democratic states (e.g. elections) are likely to be unavailable to citizens of authoritarian countries. Less institutionalized forms of voice are always possible, but given the exclusive and repressive nature of authoritarian regimes, they tend to carry significant costs. Citizens can, and often do, protest the policies or actions of authoritarian governments, but to be effective, citizens must possess the organizational capacity to pose a legitimate challenge to the regime, and they also must be willing to risk the costs of retaliation by the state, including imprisonment, torture, or even death. The difficulties inherent in collective action (Olson, 1965) combined with the threat of severe punishment pose a large obstacle to changing the status quo from below.
Given the often high costs of exercising voice in authoritarian countries, some citizens may prefer the exit option. Exit via emigration, though certainly costly, 4 can be a less costly means of achieving social, political, and economic aspirations than attempting societal transformation through political protest (Somin, 2008). Barring significant obstacles, disgruntled individuals may choose to live abroad, leaving behind a population that is more likely to remain quiescent in the face of a corrupt, or even abusive, regime. However, governments choosing to restrict citizens’ right to leave take this alternative off the table. Raising the costs of exit such that it approaches or even exceeds the costs associated with protest necessarily alters the calculus of those bent on change. 5 All else being equal, restrictions on foreign movement should increase the likelihood that protest is employed as a strategy.
While permitting exit may help secure a regime by directly removing potential troublemakers, it also has less direct stabilizing effects that are felt more broadly. Indeed, individuals who choose to leave may increase the opportunity costs of challenging the regime for those who remain behind. This is because migrants often remit some of their earnings to provide support for friends and family back home. These remittance flows have several sociopolitical and economic effects (e.g. Taylor, 1999; Adams & Page, 2005). Micro-transfers have been shown to mitigate unrest and reduce the likelihood of rebellion in the aggregate, in part because they raise individuals’ standard of living enough to dissuade them from engaging in collective action against the state (Frank, 2009). In other words, remittance flows are often sufficient to make loyalty (i.e. silence) more attractive than voice – even for potential agitators lacking the will or means to themselves choose exit.
This is an important point, as it implies that exit and voice do not have to be perfect substitutes at the individual level for the dampening effects to hold in the aggregate. In other words, not all would-be agitators have to utilize the exit option for freedom of foreign movement to ease domestic pressures: so long as some are using the option, and are also sending remittances home to supplement incomes, liberalizing foreign movement can reduce domestic political tensions. 6 This complements existing research showing how, ironically, the income benefits provided by remittances may actually perpetuate autocratic rule by reducing popular demand for public welfare, thereby allowing leaders to dedicate more state resources to patronage (Ahmed, 2012).
The immediate ‘safety valve’ effects of liberalizing foreign movement, combined with its broader economic implications, can change the calculations of potential agitators in ways that are favorable to the stability and survival of an authoritarian regime. Our logic suggests the following hypothesis:
H1: Freedom of foreign movement is negatively associated with antigovernment protest in non-democratic states.
This hypothesized effect, however, hinges on an important caveat. While greater freedom of foreign movement provides an alternative option for aggrieved individuals, the government implicitly ties the presence and degree of societal grievances to the economic and political conditions abroad. Though would-be agitators who have few opportunities for advancement in their home country may find greater expected utility in leaving (as opposed to remaining and fighting to advance change from within), their ability to do so is conditional on the presence of such opportunities elsewhere.
Indeed, scholars view migration as the result of both ‘push’ and ‘pull’ factors in the sending and receiving states, respectively (Money, 1999). The lack of economic and political opportunities, as well as repressive conditions in the individual’s home state, constitute some of the potential push factors. Pull factors, like the existence of political and economic opportunities abroad, must also factor into the individual’s decision to leave. Thus, even where autocratic regimes choose to allow greater freedom of foreign movement, exogenous conditions may limit citizens’ ability to take advantage of that policy. In particular, we suspect that the mitigating effect that freedom of foreign movement can have on domestic unrest is conditioned by the availability of economic opportunities in the developed democratic world. The advanced industrial democracies of Western Europe and North America have long been major destinations for those seeking greater economic and political opportunities. Western European states in particular, because of their colonial legacies, high concentrations of wealth, and need for labor, have been attractive destinations for individuals coming from less developed and politically repressive countries (Money, 1999). Though comprehensive cross-national time-series data on migration patterns are virtually non-existent, existing United Nations data provide some confirmation that such states host a large proportion of global migrants – in 2000 the G7 states alone hosted approximately 35.2% of global migrant stock (United Nations, 2004).
However, the attractiveness of such destinations is dynamic, not static. While many individuals may leave their home states in search of economic and political opportunities, periods of poor economic performance and high unemployment in the receiving states limit the availability of these opportunities, as there is lower demand and increased competition in the labor market. Many individuals rely on migrant networks when making the transition to the destination state (Money, 1999), and it is reasonable to expect that these networks will be less welcoming to newcomers who exacerbate these competitive pressures. Hard times also motivate political parties to focus on the needs of their core constituents and are frequently characterized by increasingly hostile attitudes toward immigrants and public pressure to restrict immigration (Hammar, 1985; Ogden, 1991; McLaren, 2003; Meguid, 2007; Islam, 2007). 7 Money (1999: 9), noting that periods of ‘economic recession [aggravate] competition and generate a rise of political pressures against immigration’, finds that higher unemployment rates in the developed industrial countries of Europe and North America are consistently and negatively associated with immigration, as illustrated by Figure 1. 8 Islam (2007) similarly finds that higher unemployment has a negative impact on immigration into Canada, specifically.
The effects of economic downturns in these states are not limited to only those who utilize the exit option (i.e. migrants). As previously noted, remittance payments by emigrants can also serve to change the cost–benefit analysis of challenging the regime for those who stay behind (Frank, 2009; Ahmed, 2012). Poor economic conditions in major host states increase the difficulty of sustaining these transfers, thereby eliminating their pacifying effect on individuals in the home country.
9
Emigrants who find themselves out of work may even feel Migrant inflows and unemployment in four countries, 1980–2010
It may even be the case that the growing sense of relative deprivation among the populace actually serves to increase antigovernment protest beyond the levels seen in those states that more tightly restrict foreign movement, given the difference in expected fulfillment (Gurr, 1970). Relative deprivation theories focus on the difference between one’s expected well-being and one’s actual well-being (Gurr, 1970; Mason, 2004), and typically posit that, ‘when a prolonged period of objective economic and social development is followed by a short period of sharp reversal’, antigovernment activity is more likely (Davies, 1962: 6). Such ‘sharp reversals’ may occur in situations where citizens living under autocratic rule have received economic support from family members in foreign states, but suddenly have their economic support system dismantled by deteriorating economic conditions abroad. Similarly, in those states allowing freedom of foreign movement, individuals may be dissuaded from participating in antigovernment activity in the present if they have expectations or plans for future opportunities to leave in pursuit of economic and political opportunities abroad. However, economic downturns abroad can cause similar sharp reversals in citizens’ estimations of future prospects, making protest in the present a more attractive option. While an increase in relative deprivation may not be sufficient to cause a person to engage in full-scale rebellion against his or her government (Mason, 2004), it may serve to intensify grievances, thereby increasing the likelihood that one decides to participate in some form of antigovernment activity.
Theory and evidence thus suggest an important caveat for our initial argument: while autocrats can reduce the de jure costs of exit by liberalizing migration policy, the de facto costs of exit are not entirely under the government’s control and can still be high enough to effectively price out exit as a feasible option. This constraint seems most applicable when economic conditions in the major destination countries make them less amenable to receiving would-be emigrants. If an authoritarian regime permits foreign movement, but opportunities abroad taper, the consequences may be: (1) diminished ‘safety valve’ benefits from outward migration; (2) decreasing remittance inflows; and (3) a population of increasingly dissatisfied citizens with fewer options available to them. Consequently, a liberal migration policy may no longer provide its desired effect for autocratic rulers – indeed, it could even contribute to a more restive populace (e.g. Frank, 2009; Gurr, 1970). Thus, the autocrat who chooses to use freedom of foreign movement to reduce the number of aggrieved citizens in her country implicitly bets on the sustained success of the developed world.
H2: Economic hardship in developed democracies is positively associated with antigovernment protest in non-democratic states that permit greater foreign movement.
Research design
To test our argument we use a cross-national time-series dataset of 125 countries from 1981 to 2007. 10 As our outcome of interest is political unrest, our dependent variable is the total number of antigovernment protests per country-year, as identified by the Cross-National Time-Series Archive (Banks, 2010). 11 Previous work has found that the decision between nonviolent and violent tactics is often spurred by the government’s response to such activities, rather than an active commitment on the part of the movement itself to one form of protest or another (e.g. Heath et al., 2000; Lichbach, 1987; Mason, 2004; Regan & Norton, 2005). For example, if a government responds to protests with widespread and indiscriminate repression, it is more likely to be met with violent resistance (Lichbach, 1987; Moore, 1998; Regan & Norton, 2005). Thus, we make no assumptions about which specific form of protest will be made more or less likely, and measure our dependent variable as the sum of both violent and nonviolent antigovernment activities during a given year. As this variable is an event count, and given its overdispersion and the likelihood of positive contagion, we use negative binomial estimation (Cameron & Trivedi, 1998; Long, 1997).
The chief causal mechanism identified by our theoretical argument is the cost of exit, which is informed by two key factors: the stringency of state-imposed limitations on citizens’ ability to leave and the availability of economic opportunities in the major destination states that would receive them. To capture the first, we use the Freedom of Foreign Movement variable from the CIRI Human Rights Data Project (Cingranelli & Richards, 2008). It is coded 0 for states that have severely restricted freedom of foreign movement, 1 for states that have some restrictions on foreign movement, and 2 for states that have not restricted foreign movement at all. For the second, we use the average unemployment rate among the G7 countries (WDI, 2008). 12 While we acknowledge that this measure does not capture all of the possible alternatives for individuals seeking better conditions, these countries are among the most likely destinations for emigrants, accounting for 35.2% of global migration stock in 2000 (United Nations, 2004). 13 Given that our expectations here apply primarily to those states that actually allow their citizens to emigrate, we include an interaction term between the freedom of foreign movement variable and the G7 unemployment variable. 14
Although our theoretical argument does concern migrants, we do not include direct measures of emigration for a few reasons. First, migration is but one factor that is subsumed by our broader theoretical argument. Second, comprehensive and reliable cross-national data on emigration are simply not available. The United Nations and World Bank provide some statistics on emigration, but the data are severely limited both temporally and geographically (see United Nations, 2004; WDI, 2008). Furthermore, Bueno de Mesquita et al. (2003) note that there are serious problems with reporting bias, finding that as many as one-third of emigrants are not reported. These biases are most acute in the authoritarian states we are most concerned with here. Studies of emigration in East Germany have noted that the use of time-series migration data in that case is only viable because the East German government was unique in keeping such extensive records (Pfaff & Kim, 2003; Pfaff, 2006). Third, emigration flows themselves are likely endogenously related to our outcome variable, as exit is often motivated by political instability and uncertainty (e.g. Davenport, Moore & Poe, 2003; Moore & Shellman, 2004; Okamoto & Wilkes, 2008). In contrast, the two factors we isolate here – state-imposed restrictions on citizens’ ability to leave, and the availability of economic opportunities abroad – are less susceptible to this problem. Finally, and most importantly, our argument does not hinge on the number of emigrants exiting a country. Rather, it rests on how government decisions affect the cost–benefit calculations of dissatisfied citizens. Whether there are lower or higher barriers to exit has implications for the attractiveness of both exit and loyalty as alternatives to voice. As such, we believe that our approach is not only the most feasible, but also the most fitting for the specific arguments we seek to test.
While we are primarily interested in state policy on foreign movement and its consequences, it is likely that states that are more liberal on this front are also more open and liberal in other policy areas. Freedom of foreign movement could simply be a proxy for other factors, like integration into the global economy. We thus control for the degree to which the state is ‘globalized’ using the KOF Economic Globalization Index (Dreher, 2006; Dreher, Gaston & Martens, 2008). Since states that are more integrated into the global economy are likely more sensitive to fluctuations in the core economies, we also interact this variable with G7 unemployment. Similar to our variables of chief interest, we should expect that the benefits of economic integration have pacifying effects (e.g. Barbieri & Reuveny, 2005), but that domestic unrest in these more integrated states will increase when conditions in the developed world sour. Controlling directly for this alternative causal mechanism will give us greater confidence in any results found for the freedom of foreign movement variable and its interaction term.
We also control for other factors thought to affect the level of antigovernment protest within a country (e.g. Gurr, 1968; Schock, 1996; Murdie & Bhasin, 2011). Though we divide our analysis according to institutional and substantive elements of democracy, we include the Polity indicator of democracy in our primary models (Marshall & Jaggers, 2007). Domestic economic circumstances may be critical (MacCulloch, 2004). Economic development should correspond with lower levels of antigovernment protest, as participation likely requires that individuals give up more comfortable lifestyles for more dangerous ones. Similarly, access to wealth may give individuals more options for redressing grievances, while those in poorer states have fewer resources to employ (Muller & Weede, 1990). Alternatively, wealthier states may enjoy greater administrative capacity (e.g. more effective police and military forces), which can deter dissidents (e.g. Fearon & Laitin, 2003). Accordingly, we control for both GDP per capita (logged to account for skew) and economic growth. Both variables are taken from the World Bank’s World Development Indicators (2008).
State reliance on physical repression should also play a significant part in stimulating antigovernment activity (Gurr, 1968; Muller & Seligson, 1987; Regan & Norton, 2005; Carey, 2006). We thus use the CIRI Physical Integrity Rights Index (Cingranelli & Richards, 2008), which measures government respect for individuals’ rights to be free from torture, extrajudicial killing, political imprisonment, and disappearance. This variable ranges from 0 to 8, with higher values indicating generally better respect for these rights and lower values indicating more frequent use of repressive tactics by the government and its agents. Observance of other rights may also matter. State controls on movement within the country can make it difficult to coordinate and organize protest activities. We include a variable measuring freedom of internal movement, also from the CIRI Human Rights Dataset (Cingranelli & Richards, 2008).
Since protest events may be associated with broader unrest and violence (Murdie & Bhasin, 2011), we control for whether or not a state is experiencing an inter- or intrastate war using the UCDP/PRIO Armed Conflict Dataset (Gleditsch et al., 2002). Likewise, we control for population size, as larger populations present more opportunities for antigovernment action. We also account for the percentage of the population between ages 15 and 64 (WDI, 2008), as this is the subset of the populace that is most likely to be politically active (Murdie & Bhasin, 2011).
Summary statistics
Statistics based on sample from Model 2 shown in Table II.
Finally, we include a ‘peace-years’ counter to control for temporal trends in unrest and protest. This variable counts the number of years since the last observed occurrence of protest activity within a country (Banks, 2010). See Table I for summary statistics for these control variables.
As our theory applies primarily to those states that are non-democratic, we run several different models, shifting samples on the basis of authoritarianism. First, we run a series of base models on a sample of all non-G7 states. 16 We then limit the sample to those states that lack formal democratic institutions (states below 6 on the Polity IV scale) (Marshall & Jaggers, 2007). Next, we evaluate a sample of ‘substantively’ non-democratic states. This includes countries where the rights associated with political expression – the freedoms of speech, religion, and assembly – are most severely restricted. While there is some correlation between these two conceptualizations of non-democracy, there is also substantial variation across institutionally non-democratic states in terms of their willingness to permit the exercise of these rights on the ground. To determine our sample of substantively non-democratic states, we generate an index variable using measures of government respect for free speech, association, and religion, provided by the CIRI Human Rights Dataset (Cingranelli & Richards, 2008). 17 This variable ranges from 0 to 6, with higher values indicating better respect for these rights. As we want to focus on those states where these rights are highly restricted or absent entirely, we limit the sample to those countries that score only 0 or 1 on this index. Finally, we also limit our sample to states that are both institutionally non-democratic (those classified as non-democracies according to the Polity dataset) and substantively oppressive (those states classified using the CIRI index variable).
Negative binomial models: Freedom of foreign movement and antigovernment protest
Robust standard errors in parentheses. One-tailed significance tests used. * p < 0.05, ** p < 0.01.
Results and discussion
Models 1 and 2 in Table II are estimated on all non-G7 states for which the relevant data are available. We restrict the sample to non-democracies in Models 3–5.
The results of our base model (Model 1) demonstrate that the control variables behave largely as expected. Greater freedom of movement within a country positively affects protest activity, implying that collective action for popular movements is easier when there are fewer constraints on internal mobility. Likewise, liberal political institutions and larger populations both share a positive relationship with protest, consistent with previous studies (e.g. Murdie & Bhasin, 2011). Alternatively, governments that demonstrate greater respect for physical integrity rights may see less public unrest, as will those enjoying periods of economic growth and sustained periods of peace and stability. Economic development, the relative size of the working-age population, the presence of war, and colonial history demonstrate no consistent correlations here. These results persist across the subsequent models with some minor exceptions, which we discuss below.
Economic globalization and its corresponding interaction term perform as expected, and both are statistically significant. Greater integration into the global economy is negatively correlated with the likelihood of observing antigovernment activities. However, its interaction with G7 unemployment is positive and significant at the 5% level, suggesting the pacifying effects of liberalization are dampened when core markets weaken. These results imply that interdependent states are more sensitive to economic conditions in the developed Western democracies, and in a way that resembles our primary argument concerning the consequences of a liberal migration policy.
We introduce the variables of chief theoretical interest in Model 2. Even within this broad sample of states, we do find some support for our hypotheses. In the absence of free foreign movement, there is no evidence that economic conditions in the most-developed world – as indicated by the average unemployment rate among the G7 states – have any independent effect on the occurrence of antigovernment protest activity throughout the rest of the world. Greater respect for freedom of foreign movement, however, is estimated to have a statistically significant negative impact on the rate of antigovernment protests across non-G7 states, in accordance with Hypothesis 1. In contrast, the interaction between G7 unemployment and freedom of foreign movement is positive and significant at the 5% level, suggesting that states permitting freedom of foreign movement are at greater risk of experiencing antigovernment activity when economic conditions in the developed world worsen, lending support to our second hypothesis. By opening this channel to the outside world, states make themselves more vulnerable to blowback generated by forces over which they have no control – in this case, forces that obstruct entry for discontented nationals hoping to uproot in search of greater economic and political opportunities, or that suddenly choke off the flow of resources from friends or family who have already established themselves in freer, more prosperous countries.
While this initial evidence is encouraging, we have argued that the theoretical linkage between emigration and public protest will be strongest for the subset of states that most severely restrict other modes of expression. Accordingly, we limit the sample in Model 3 to countries that lack formal democratic political institutions. 18 However, there can be considerable variation among even those states that are institutionally non-democratic. Some that have implemented liberal institutional reforms may still resist the kinds of substantive reforms that allow people to speak or associate freely; others that maintain exclusive political processes may nevertheless extend these more accessible means of expression to their citizens (Brownlee, 2007; Magaloni, 2008; Gandhi, 2010; Slater, 2010). We alternatively limit the sample in Model 4 to states that repress freedoms of speech, association, and religion. 19 For Model 5, we combine these rules to limit the sample to only countries that are both institutionally and substantively non-democratic.
When conceptualizing non-democracies in strictly institutional terms (Model 3), the results change little. The estimated coefficient on freedom of foreign movement is negative and significant at the 5% level. However, when we conceptualize non-democracies in substantive terms, we find a larger effect. As the results from Model 4 demonstrate, substantively non-democratic states that impose fewer constraints on foreign movement experience less protest activity, on average, than those that constrain foreign movement. This evidence supports our first hypothesis, and we take it to imply that, when domestic discontents can freely move to more favorable locales, they may be less compelled to take their grievances to the streets. Similarly, those that benefit indirectly from foreign movement – through remittances or otherwise – might reconsider the costs of exercising voice relative to simply staying silent.
However, this pacifying effect is conditional on the availability of opportunities abroad. Across all models, the interaction between G7 unemployment and freedom of foreign movement is positive and significant. The results indicate that poor economic conditions in major receiving states can undermine the effectiveness of a liberal migration policy as a mechanism for domestic stability. By our logic, this reflects the backlash we may expect when exit becomes a less viable option.
The fifth column presents our results from the most limited sample – institutionally and substantively non-democratic states. As expected, we find that the substantive impact of our primary independent variables is strongest among these most authoritarian regimes. Figure 2 illustrates the marginal effect of freedom of foreign movement across the range of unemployment observed in the estimation sample. At the low end of the x-axis we see that increasing the freedom of foreign movement variable from 0 to 2 leads to a decrease of approximately 0.65 in the expected number of protest events. 20 However, this effect only holds when economic conditions in the destination states are at the best levels found in the estimation sample (i.e. when the average unemployment rate is near 6%). As the G7 unemployment rate increases the effect of increasing freedom of foreign movement becomes statistically insignificant. As economic conditions in the G7 states deteriorate, greater freedom of foreign movement can even make an authoritarian state worse off. At the highest average G7 unemployment rate found in the sample (approximately 8.8%) the marginal effect of increasing foreign movement from its most restrictive to its most permissive is associated with an increase of approximately one protest event. In other words, if for no other reason than providing a means for the development of higher expectations, greater freedom of foreign movement may contribute to greater dissatisfaction when those expectations are dashed (Gurr, 1970).

Marginal effect of freedom of foreign movement on predicted count of antigovernment protests

Effect of G7 unemployment on predicted probability of protest
Though these predicted differences seem small, they are actually fairly substantial given our selection of states. Among these most authoritarian countries, no protest was observed in 62% of the cases; and of the cases for which protest activity did occur, 91% experienced four events or fewer within the given year. Since even one protest event may be considered a rare and unwelcome event for such repressive regimes, we believe the predicted differences associated with freedom of foreign movement indicate a substantively meaningful finding.
Figure 3 shows the predicted probability of observing any antigovernment protest activity as a function of respect for freedom of foreign movement and the concurrent unemployment rate among the G7 countries. In accordance with our expectations, states that maintain stricter controls over foreign movement are at a higher baseline risk of protest – a predicted probability of about 0.4 here – but this baseline risk is not particularly sensitive to fluctuating economic conditions in the developed world, and actually seems to fall to about 0.35 at higher values of G7 unemployment. Alternatively, authoritarian states that permit some freedom of foreign movement are better off when G7 unemployment is low, predicted to have a 0.29 probability of experiencing protest. However, they face an increasing risk of protest as economic conditions in the core deteriorate, reaching a predicted probability of 0.45 when G7 unemployment approaches 8.8%. Finally, those states that impose the fewest barriers to foreign movement demonstrate the greatest sensitivity to fluctuations in G7 unemployment, benefiting the most when G7 conditions are strong, but also facing the steepest increase in the probability of protest when those opportunities wane. By our estimates, this is a fairly drastic swing from a predicted probability of about 0.19 when G7 unemployment is at 6%, to just over 0.54 when G7 unemployment reaches 8.8%.
Figure 4 illustrates an additional dimension of risk faced by authoritarian regimes that allow greater freedom of foreign movement. It shows the distributions for the expected count of protests based on 20,000 simulations, generated using the Model 5 estimates.
21
Panels A and B show the expected protest count for states that fully restrict foreign movement versus those that impose no restrictions at 6% and 8.8% unemployment, respectively. As Panel A shows, when economic conditions in the G7 states are good, authoritarian states allowing full freedom of foreign movement have a lower expected occurrence of protest than states imposing full restrictions. Alternatively, Panel B shows that when economic conditions in the G7 states deteriorate, authoritarian states allowing full freedom of foreign movement now have a higher expected protest count than those states where foreign movement is tightly restricted. What is striking, however, is the tremendous variance in the expected protest count for these states. While the mean predicted count is roughly 1.8, it ranges from approximately 1 to over 3. Alternatively, states restricting Simulated protest counts based on Model 5 (results based on 20,000 simulations)
Finally, it should be noted that across all three samples of ‘non-democracies’, the control variables generally perform as before. One notable difference is that the presence of war is now estimated to have a significant and positive association with protest. This may simply reflect the difference between democracies and non-democracies in both the ability of people to challenge the government and the types of conditions that are necessary before such antigovernment action becomes plausible. Protest may be treated as an absolute alternative to civil war in democratic societies, whereas it may more often coincide with war in non-democratic societies where the threshold for challenging the government, regardless of method, is typically higher to begin with. Also, the economic globalization index and its interaction term fail to attain statistical significance in Models 4 and 5. Although these variables are highly significant and in the expected direction for the sample of institutionally non-democratic regimes, we find less evidence that these dynamics impact antigovernment protests in either of the more restrictive subsets of non-democracies. Still, the presence of these variables in our model gives us greater confidence that our findings reflect the hypothesized relationships, and not a spurious correlation.
Conclusions
The results presented above are instructive. They show that greater freedom of foreign movement can, at times, act as a safety valve to release domestic pressures that might otherwise explode on the streets, and thereby challenge authoritarian regimes. When states lower the barriers to exit, it can provide alternatives for those determined to secure a better life. Rather than bear the high costs of challenging the regime from within, they may opt instead for the lower-cost option of finding greater sociopolitical and economic satisfaction abroad. Those that do leave may also funnel remittances back home, thereby ‘buying’ the loyalty and silence of friends and family (Frank, 2009; Ahmed, 2012). These are positive outcomes from the vantage point of an authoritarian elite bent on political survival. However, it requires that individuals actually have opportunities abroad to pursue. When such opportunities are limited, the benefit of allowing foreign movement may quickly disappear, and it can even become a liability, as would-be emigrants are faced with the added frustration of dashed expectations, and the inflow of resources from friends and family overseas dries up. This poses several difficult questions for authoritarian leaders, and points to some fascinating trade-offs inherent in the liberalization of foreign movement. While a leader’s inclination may be to exert more control by restricting cross-border movements, permitting greater freedom of foreign movement can actually alleviate antiregime pressures and thereby strengthen her hold on power. 22 But there is a catch. The autocrat who disincentivizes voice by reducing the de jure costs of exit necessarily makes herself dependent on factors beyond her control. This strategy can backfire if the de facto costs of exit make a liberal migration policy a moot point.
A broader implication of these findings is that global integration through liberalization entails myriad and dynamic consequences, both beneficial and costly. It can at times serve to reinforce entrenched regimes and existing power structures, and at others broadside them when problems in foreign countries reverberate through the channels that connect them. These nuances are often neglected in studies of integration and ‘globalization’, and future research should explore them further.
Finally, our theoretical argument draws heavily on the literature concerning the interrelationship between exit and voice (e.g. Hirschman, 1970; Gehlback, 2006), applying it to antigovernment activity in a cross-national context. As emigration is an implicit factor in this, the research presented here should also be relevant to those more specifically interested in migration and its political ramifications. Indeed, while there is extensive research on the costs and benefits of immigration at the national and community levels, and the problems of cultural and political assimilation in host states, 23 far less is known about the political effects of emigration on emigrants’ home states. 24 This is unfortunate, as the consequences of emigration likely extend beyond the economic costs often associated with the so-called ‘brain drain’. Indeed, as our results suggest, the relative costs of exit and voice have important implications for social stability and antigovernment activity more broadly.
Footnotes
Replication data
Replication data, do files, and the online appendix can be found at http://m-flynn.com/Research.html and
.
Acknowledgements
The authors would like to thank Olga Shvetsova and the participants of the Binghamton University World Politics Workshop, the participants of Binghamton University's 2011 Dr Han-Jyun Hou Conference on The Politics of Immigration, and the editors and anonymous reviewers for their helpful comments and suggestions.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
