The peace vs. deterrence dilemma

A credible exile offer can contribute to domestic peace by reducing an autocrat's incentive to fight to stay in power. Yet such an also offer undermines global deterrence against autocracy. The reason is that autocrats regularly adjust their plans based on information regarding the fortunes of their foreign counterparts, particularly insofar as these fortunes are affected by international actors. ‘Ample evidence’ confirms that autocrats ‘closely follow the fates of their peers’,Footnote ²⁰ especially when assessing their own international retirement prospects. This evidence suggests while facilitating peace, a ‘golden parachute’ offered to any given autocrat is also likely to set back global deterrence against other (potential and actual) autocrats, domestically and globally.

More precisely, my use of the phrase ‘global deterrence against autocracy’ is meant to cover two distinct dimensions of such deterrence. One dimension refers to conduct while in office. Here the thought is that the sight of a given autocrat facing severe legal repercussions for crimes committed while in power leads other countries' autocrats to be much more cautious about engaging in similar crimes.Footnote ²¹ The other dimension is deterrence against the use of criminal methods in the quest for highest office. Here the claim is that individuals who would otherwise be tempted by blatantly illegal tactics in pursuit of political power would be more reluctant to utilize such tactics to begin with – and therefore more committed to the rules of the democratic game – if they see these tactics' legal repercussions for others. The former kind of deterrence has received more robust empirical support than the latter.Footnote ²² But when examining the tension between peace and deterrence, I will assume (at least arguendo) that both dimensions actually obtain.

In turn, my inquiry into this tension draws important but limited inspiration from early classics of the transitional justice literature. Early seminal essays, centered on a ‘peace versus justice’ dilemma, emphasized that a transitioning society can rarely avoid a tragic choice between compromise and confrontation with authoritarian wrongdoers.Footnote ²³ On the one hand, it is precisely a choice of this kind that lies at the heart of my analysis here (hence the inspiration). On the other hand, because the ‘peace versus justice’ debate has been almost purely domestic, it can only take us so far when thinking the international problem of autocratic exile (hence the limit to the inspiration).

For one thing, domestic debates about compromises with authoritarian wrongdoers make no room for salient duties borne by foreign actors. As a result, these debates overlook the fact that some foreign countries might be in a morally distinct position vis-à-vis any possible compromise that a transitioning society might reach with its autocrat. I will argue that this fact is pivotal to a compelling normative analysis of autocratic exile.

Furthermore, purely domestic debates about ‘peace versus justice’ cannot capture the international deterrence costs associated with autocratic exile, nor can such debates capture the distinctive moral justification that these costs require. If autocrats do indeed learn from the fortunes of their foreign counterparts, and if they are therefore emboldened by any given ‘golden parachute’ enjoyed by one of their foreign peers, then it follows that every transitioning society owes justification to foreign countries with their own (aspiring or actual) autocrats, in case it produces headwinds for them by striking an exile compromise with its autocrat. This need for international justification, moreover, stands even if the exile compromise in question does actually lead (eventually) to durable democracy at home. Yet domestic debates about ‘peace versus justice’ do not provide us with any real guidance on how to think about tragic circumstances where the democratic gains to one society come at the expense of enhanced risks to democracy in other societies.Footnote ²⁴

Dismissing the dilemma?

The philosophical guidance that I wish to provide is partly meant to preempt a simple dismissal of the normative dilemma I just identified. More specifically, I wish to preempt three seemingly tempting ways to dismiss the dilemma – all of which have been present in the literature.

First, there is the temptation to assume that policy decisions regarding compromises with autocrats involve few genuinely normative puzzles. On this view, as one political scientist puts it, the ‘critical questions are empirical’.Footnote ²⁶ Accordingly, if we could only have enough confidence in the probabilities we assign to the possible outcomes of different policies over different time-spans, our normative questions would effectively disappear – presumably because the cost–benefit analysis of the relevant policy options would become straightforward. However, a singular focus on cost–benefit analysis simply ignores non-consequentialist approaches to political morality – and can arguably be dismissed on this ground.

The second temptation I wish to preempt is the temptation to associate rigorist treatment of authoritarian wrongdoers with retributivist zeal. Several influential political scientists who have criticized criminal trials for members of authoritarian elites have assumed that only a transitioning society which assigns intrinsic moral value to punishing authoritarian wrongdoers will refrain from pragmatic, forward-looking compromises with such wrongdoers.Footnote ²⁷ This assumption ‘solves’ the policy dilemma by asserting that the dilemma only arises if we adopt (an implausibly rigid form of) retributivist fixation on the past.Footnote ²⁸ But instead of stacking the deck this way, it is more fruitful (and more accurate) to construe both sides to the debate as non-retributivists, who share the claim that criminal punishment can only have instrumental rather than any intrinsic moral value.Footnote ²⁹ The real point of contention between them is not about the past, but rather about whether decisions regarding possible compromises with autocrats should prioritize the global long-run or the local present.

The final temptation that I wish to preempt is the temptation to dismiss this dilemma by dismissing any empirical evidence regarding long-run deterrence as ‘speculative’.Footnote ³⁰ In 2008, for instance, the same prominent political scientist who asserted that the ‘critical questions are empirical’ also asserted the following:

I strongly agree with the claim that we should beware excessive ambitious in social science, and obviously, the further into the future we look, the more ambitious the relevant social science becomes.Footnote ³² Nonetheless, I also believe that we should take seriously the idea of long-term deterrence against autocracy, for several reasons.

For one thing, even where empirical testing is difficult, we immediately recognize the intuitive appeal of similar deterrence hopes in other contexts where unsavory compromises with wrongdoers must be countenanced. Consider, for instance, the familiar claim that that we should not accede to ransom demands – whether coming from cyber-criminals or from ‘classic’ hostage-takers – because doing so would only further incentivize more criminals to try to extract ransoms in the future. In most (if not all) contexts, this claim too may also be hard to assess empirically. Yet despite these empirical obstacles, this claim obviously informs actual policy discourse, and regularly shapes actual policy decisions.Footnote ³³

Furthermore, at least when it comes to atrocity crimes, there is accumulating empirical evidence in support of an actual deterrence effect, resulting from rigorist treatment of wrongdoers. More specifically, in recent years, multiple scholars have offered empirical grounds for believing that trying one autocrat for overseeing atrocities deters other countries' autocrats from resorting to atrocities themselves (if they have not already sunk to such lows).Footnote ³⁴ We therefore have genuine empirical reasons for taking seriously tragic deterrence dilemmas, of the sort emphasized by Daniel Krcmaric:

My final reason for taking the deterrence rationale seriously is perhaps the most important. In seeking to defeat such a rationale, there is value (philosophically and practically) in showing that it will often fail normative scrutiny, even if its empirical basis is solid, and even if international prosecutors' deterrence hopes are therefore cogent. That is what I wish to show here. I aim to establish that there is a crucial set of cases in which powerful democracies ought to facilitate autocratic exile, even if global deterrence against autocracy will indeed be damaged as a result.

Accepting autocratic exile: the transitioning society's prerogative

I wish to show that a transitioning society normally has the moral prerogative of accepting an exile compromise with its autocrat, even when such a compromise is suboptimal from a global perspective, due to its repercussions for global deterrence. The argument in favor of this prerogative can be summarized as follows:

1. (1) If the course of conduct that is impartially best entails grave harm for the agent(s) who might adopt it, it is normally morally optional rather than required.

2. (2) (From 1) The default prerogative: in circumstances where the victims of serious crimes must either suffer further grave harms at the perpetrators' hands or compromise with the perpetrators, the default assumption should be that victims have the moral prerogative of accepting a compromise, even if rejecting a compromise would be impartially best.

3. (3) The manageable risk principle: in certain cases, a compromise between the victims and the perpetrators is impartially suboptimal because it increases the risk of grave harm befalling other parties. But it matters whether these other parties are able, without incurring dramatic costs, to act so as to bring this risk (at least roughly) back to its pre-compromise level. If all relevant other parties have this ability, then this suffices for the victims to be morally permitted to compromise with the perpetrators.Footnote ³⁸

4. (4) Normally, both the default prerogative and the manageable risk principle apply when a transitioning society is considering a compromise with its autocrat.

5. (5) (From 2, 3, and 4): a transitioning society normally has the prerogative of accepting a compromise with its autocrat.

6. (6) Among the compromises that a transitioning society might reach with its autocrat, the exile compromise stands out, given credible-commitment considerations.

7. (7) (From 5 and 6): a transitioning society normally has the moral prerogative of accepting an exile compromise with its autocrat.

We can unpack this argument by considering each of its steps in turn. Premise 1 reflects a bedrock conviction of commonsense morality: because each one of us has a separate life to lead, it is overly demanding to insist that agents must always try to maximize good in the world, regardless of the costs that they might have to shoulder in the process. To insist that ‘the hardships of some’ are always ‘offset by a greater good in the aggregate’ is to ignore what Rawls famously termed the ‘separateness of persons’.Footnote ³⁹ In turn, from this core conviction follows the familiar idea that if doing what is optimal for the world entails heavy costs for them, agents normally have the moral prerogative of prioritizing their own interests instead.Footnote ⁴⁰

Consider next the ‘default prerogative’ in step 2. Victims who have no good choices in the face of perpetrators' threats normally have the same prerogative enjoyed by the rest of us in less tragic circumstances. They too are typically morally permitted to refuse to do what is in the world's best interest, when this means undertaking dramatic sacrifices. Compromising with perpetrators may very well conflict with the interests of humanity writ large. Nonetheless, victims normally have the prerogative of accepting such a compromise, if that is their only way of avoiding extraordinary harms.

Step 3 refers to circumstances where it may initially seem as if this default prerogative disappears. Where victims' compromise with perpetrators increases the risk of grave harm to other parties, it is tempting to think that once the number of such parties crosses a certain threshold, the default prerogative must give way. The manageable risk principle, however, shows that this is not so. According to the principle, a sufficient condition for the continued existence of the prerogative is that the other parties who face heightened risk of grave harm due to the compromise can respond by making adjustments which are not overly costly, and which would return the relevant risk (at least roughly) to its pre-compromise baseline.

As an example, consider a variant of a well-known actual case.Footnote ⁴¹ The scion of an extraordinarily wealthy family is kidnapped by the mafia; the kidnappers tell their victim that they will murder him unless he arranges for a multi-million dollar ransom payment, even going so far as to cut off one of his ears to prove their seriousness. The victim knows that if he accedes to the kidnappers' demands, the public prominence of his case will increase the motivation of prospective kidnappers around the world to target the members of other wealthy families for similar treatment.

However, suppose that, informed that the kidnapped individual has arranged for a ransom to save his own life, every other affluent person affected by this ransom payment can bring the risk of his or her own kidnapping back to its earlier level, through means that each of them can afford (for instance, hiring adequate, reliable personal bodyguards). If this is the case, then the scenario is not one where the (virtually certain) harm suffered by one agent must be weighed against the elevated risk of comparable harm befalling multiple other agents. Rather, the scenario is one where the grave harm to one agent is qualitatively worse than the cost (of rational adjustments) for every other individual agent. Yet in this kind of situation, multiple philosophers have argued that the numbers are beside the point: no matter how much we may increase the number of those who face modest costs, we cannot aggregate these costs so as to outweigh a grave harm that would befall a smaller number of other agents.Footnote ⁴² Accordingly, the kidnapping victim has the prerogative of negotiating with his captors in order to save his own life, no matter how many individuals would have to (each) incur modest costs as a result of him doing so.

With all this in view, consider step 4 of the argument, applying these ideas to a transitioning society's compromises with its autocrat. Just like the hostage, a transitioning society typically cannot avoid grave harms unless it makes at least some concessions to the perpetrators. Yet, again just as in the hostage case, these concessions are impartially suboptimal, insofar as they increase the risk that other parties will suffer grave harms at the hands of prospective perpetrators who will be tempted to imitate the current perpetrators' methods. The question, then, is whether the other parties can reduce this risk through measures that are not too costly. If compromising with a given autocrat incentivizes others to mimic this autocrat, are there non-burdensome policies that could counteract this incentive?

In normal circumstances, at least, the answer is quite clearly ‘yes’. This is because our ability to discern certain patterns of authoritarianism which recur across different times and places should also translate into the ability to take pre-emptive action to reduce the risk of future authoritarian regimes. Thus for instance, suppose that we discern a growing international trend of political leaders who acquire authoritarian powers through the same institutional lacuna, such as the absence of (strictly enforced) term limits, or through the same tactics for taking over the mass media. Such recurrent patterns would not necessarily preclude any given transitioning society from compromising with leaders of this sort. Rather, these patterns would reinforce the significance of preventative efforts to fix these recurrent democratic weaknesses in any country where they obtain – ensuring that term limits exist and are rigorously enforced, pursuing reforms and regulations that buttress media pluralism, and so on.

Similar thoughts apply to the central example of military rulers.Footnote ⁴³ According to Sikkink, for example, the fate of military rulers is followed by young military officers, both at home and abroad, who use the trajectories of their seniors' careers to ‘draw conclusions about their [own] future choices’.Footnote ⁴⁴ This claim suggests that the main risk posed by lenient treatment of any given military ruler lies with the increased political ambitions of future military elites. The resulting task is therefore to implement policies that can restrain these ambitions without imposing severe social costs. Such policies can take several forms.Footnote ⁴⁵ In some instances, civilian authorities might be able to check the military's political ambitions by routinely sending officers on international peace-keeping missions.Footnote ⁴⁶ This tactic, pursued with considerable efficacy by post-authoritarian civilian governments from Latin America to sub-Saharan Africa, solidifies officers' professional identity as career soldiers rather than politicians, while also reducing the military's interest and ability to meddle in politics. In other instances, civilian governments might even be able to abolish their military completely (e.g. in the manner of Costa Rica ) without incurring severe social costs. But even where such reform would go too far, there are still available policies which can often contain the risk of military interference in politics without dramatic costs. Abolishing conscription (in the manner of Argentina) is often a natural way to trim the military's wings. Slashing the military's budget might have the same effect in some countries, while in other countries with a history of military rule, regular modest increases in defense spending might actually be a more reliable way of preventing soldiers from forming too much of a taste for politics.

The deeper point here, however, obviously goes beyond particular tactics for dealing with politically risky militaries. Instead, it has to do with the kind of risk that future authoritarian regimes (be they military or civilian) represent, and that decisions regarding compromises with current autocrats must incorporate. The risk of future authoritarianism is not akin to certain processes in the natural world which, once set in motion, are largely impervious to human control. Rather, this is a risk that humans are normally quite capable of preempting. And this possibility of preempting the future danger explains why each transitioning society is normally morally permitted to seek a compromise with its autocrat, if doing so is the only way to avoid grave harm befalling its current members.

In turn, if each transitioning society has the moral prerogative of compromising with its autocrat as a necessary means of avoiding grave harm, then each transitioning society also has, more specifically, the prerogative of accepting an exile compromise, given exile's distinct potential of neutralizing the credible commitment problem which so often inhibits peaceful transitions away from authoritarianism.

Ascertaining popular preferences

Establishing that a transitioning society can permissibly accept an exile compromise with its autocrat is the first step in our argument. The second step concerns the assumptions that foreign governments should make regarding the transitioning society's exile wishes.

I believe that, for practical purposes, foreign governments should approach a transitioning society's exile wishes as follows. In the absence of unequivocal evidence that a majority within the transitioning society rejects an exile arrangement, it is morally appropriate for foreign governments to assume that the transitioning society accepts an exile solution. In turn, for this assumption to obtain in a given case, it need not matter whether the exile solution is initiated by grassroots members of the opposition, by recognized opposition leaders, by the autocrat himself, by outsiders, or by any particular combination of these actors. On any of these variants, foreign countries can permissibly assume that a transitioning society accepts an exile arrangement, barring decisive evidence to the contrary.

Why is that? Why is it sensible for foreign countries to adopt as a default the assumption that transitioning societies ‘opt in’ for autocratic exile, rather than ‘opt out’? The main answer has to do with a central empirical feature of popular protests against autocrats. In the overwhelming majority of cases, such protests explicitly revolve around the demand that the autocrat relinquish power. But this demand is crucially distinct from the (more ambitious) demand that the autocrat submit himself to legal authorities to face trial for any relevant crimes. Moreover, very much related, in those cases where the stand-off between the autocrat and the popular opposition is indeed resolved through exile, mass celebrations typically follow – clearly conveying a sense that the opposition has achieved its primary goal by ending the autocrat's rule, notwithstanding the fact that he is now beyond the effective reach of the country's laws.

I should stress that in highlighting this empirical pattern, I do not mean to ignore the fact that transitioning societies often feature minorities which would prefer accountability for the autocrat over safety. Nor do I mean to reject the thought that an autocrat's specific victims ought to enjoy a prominent role in societal deliberations about his fate – a role which a majority may very well fail to honor.Footnote ⁴⁷ What I do mean to reject is any kind of veto right for specific victims regarding societal compromise with the autocrat.Footnote ⁴⁸

Similarly, I do not mean to suggest that it is entirely unlikely that a transitioning society will oppose an exile solution in a way that ought to matter to foreign actors. Rather, my suggestion is that an exile solution ought to be taken off the table if – and only if – there is clear evidence that a majority within the transitioning society is willing to undertake the risks involved in a dogged pursuit of legal accountability for the beleaguered leader.Footnote ⁴⁹

Extending the scope of our discussion will give us a better sense of how such evidence might look. So far, we have focused on beleaguered autocrats who have so thoroughly distorted democratic institutions that no legal challenge to their rule had any serious chance of success. Some beleaguered incumbents, however, may find that their attempts to hollow out liberal-democratic institutions have only been partially successful, and that conventional legal challenges – including challenges to their personal liberty – are therefore a clear and present danger. These leaders too (whom we might think of as aspiring rather than actual autocrats) might effectively threaten social and political chaos unless granted immunity from the law. Yet when faced with such aspiring autocrats, there might very well be a societal majority which demands not only that the incumbent relinquishes power, but also that he submit to the law. When that is the case, no exile solution protecting the incumbent from the law should be entertained.

To see why this is not a purely theoretical construct, consider the following case. In 2014, an anti-corruption government taskforce in Papua New Guinea, set up by Prime Minister Peter O'Neill 3 years earlier, issued a warrant for O'Neill's own arrest. The arrest warrant was based on O'Neill's alleged complicity in an attempt by an extremely powerful lawyer to siphon off to Australia millions of dollars that were illicitly extracted from Papua New Guinea's public coffers. Rather than submitting to the warrant, which was endorsed by Papua New Guinea's top judge and police officer, O'Neill ‘took refuge under armed guard in parliament, replaced the police commissioner with a more pliable candidate, and cut off funding to the task force’.Footnote ⁵⁰ It would not have been hard to imagine massive demonstrations by O'Neill's compatriots demanding that he submit to the arrest warrant, even if compelling him to do so would cost lives.Footnote ⁵¹ And such demonstrations, in turn, would have clearly sufficed to make it impermissible for any foreign country (including Australia, keen to court O'Neill for its own reasonsFootnote ⁵²) to offer O'Neill exile as a way of defusing the crisis.

A similar point, in turn, applies to more subtle aspiring autocrats who find themselves in legal peril. Consider, for instance, Israel's longest-serving Prime Minister. Despite – and indeed partly because – of his systematic efforts to remove any constraints on his power, Benjamin Netanyahu was indicted in 2019 on charges of bribery, fraud, and breach of trust that could see him spend years behind bars.Footnote ⁵³ Both before and after the long-expected indictment, Netanyahu and his acolytes violated almost every conceivable constitutional norm, short of explicitly calling for civil war, in an attempt to shield him from the law. Public proclamations that ‘millions will take to the streets’Footnote ⁵⁴ in case of a criminal trial against the Prime Minister were augmented by Netanyahu loyalists physically targeting prosecutors, and by Netanyahu himself, in an incendiary speech outside the courtroom, not only insisting that prosecutors had lunched a ‘coup’ against him but issuing thinly veiled threats at the judges.Footnote ⁵⁵ The attendant social and political upheaval has included an unprecedented series of five national elections in less than 4 years, all forced by Netanyahu in an attempt to secure a parliamentary majority that would provide him with Berlusconi-style retroactive immunity from criminal prosecution.Footnote ⁵⁶ By the summer of 2021, Netanyahu's desperate attempts to cling to power alienated enough of the political class to produce the most heterogenous political coalition in Israeli history, united (temporarily) only by the conviction that Netanyahu must not be allowed to ‘hold the country hostage’Footnote ⁵⁷ in pursuit of personal immunity from the law. Insofar as this has been the central message of numerous anti-Netanyahu protests, it would be wrong for any foreign country to try to defuse the – still ongoing – crisis by offering Netanyahu a golden parachute. This, notwithstanding widely circulated claims that Netanyahu would head abroad if allowed to retire without serving time in prison.Footnote ⁵⁸

Implicated countries: the case of wrongful actions

The first category is one where a foreign government has (a) foreseeably empowered the autocrat through its actions, and (b) these actions have been obviously morally wrong. Consider, for instance, a foreign government which has repeatedly sold crucial surveillance equipment to an autocrat, in full knowledge that the autocrat will rely on this equipment to track and repress peaceful dissidents. In cases of this sort, there is very strong reason to think that the manifestly non-ideal choice now faced by the transitioning society – whether to fight the autocrat or compromise – is forced upon it, at least in part, because of the foreign government's wrongful actions. Nonetheless, according to the deterrence view, these wrongful actions do not change the fact that the foreign government in question is justified in refusing to support any exile compromise, irrespective of whether the transitioning society accepts such a compromise.

In order to assess this view, we should temporarily step back from the specific details of foreign countries' dealings with autocrats, and consider the situation in more general terms. Our interest here is in cases where: (a) victims must either pursue an extremely costly confrontation with a perpetrator or accept a manifestly non-ideal compromise with the perpetrator; and (b) this choice is forced upon the victims, at least in part, because of a third party's wrongful actions. According to the deterrence view, the third party's wrongful actions do not change the fact that it is justified in refusing to facilitate a compromise between the victims and the perpetrators. I want to show that this claim is false.

The path to this conclusion starts with what it means to refuse to support victims' efforts to compromise with perpetrators. At least in the circumstances in which we are interested, such refusal effectively means forcing the victims to confront the perpetrators. This is because no agent can coherently refuse to help victims compromise with perpetrators while expecting another (fourth) party to offer the relevant help instead. Any agent who refuses to help the victims in this way must be willing to defend such a refusal even if it turns out to be universal. But a universal refusal to help the victims negotiate a compromise with the perpetrators would clearly be a way of forcing the victims to confront the perpetrators.Footnote ⁵⁹

The deterrence view, in other words, is committed to the claim that a third party can force the victims to pursue a very costly confrontation due to a predicament that was partly brought about by the third party's own wrongful actions. Moreover, these imposed costs are coming on top of the costs that the victims had already been forced to incur as a result of the third party's wrongful actions. So a third party that effectively compels victims to confront perpetrators violates essential fairness constraints that ought to have moral primacy. Ultimately, it is the deterrence view's inability to account for these fairness constraints which renders it implausible.

To further develop this argument, it would help to elaborate on the relevant fairness constraints. Note, first, that it is widely accepted that a community's pursuit of its collective interests – including its morally significant collective interests – ought to be constrained by considerations of fairness in the distribution of relevant costs among its members. Second, a community's pursuit of its collective interests must be attentive not just to fairness in the distribution of costs among members at any given point in time, but also to fairness over time. This point is familiar in the domestic realm. If, for example, you have performed jury duty multiple times, or if you have already performed a lengthy national service of some form, but are asked to repeat your service yet again, there is plainly a point at which it is reasonable for you to insist that, as a matter of fairness, it is now someone else's turn to bear the relevant burden.

Now, admittedly, the international cases that are our ultimate subject here have a different structure than such familiar domestic cases. In domestic examples, the state as a generally recognized central authority is tasked with allocating the burdens associated with the pursuit of the community's collective interests. This means that the process of allocation is hierarchical in character: the state orders you to serve in a jury, just as it orders you to report to national service. In contrast, international cases feature no effective overarching institution that allocates relevant costs among (the countries that are) the community's members. This means that the allocation is (at least in practice, if not in theory) often left to individual members to negotiate with one another.

However, essentially the same fairness constraint which obtains in the domestic examples also holds in such bilateral international contexts. Here too, the pursuit of what is best for the community as a whole is constrained by fairness in the allocation of the costs of this pursuit, and what fairness requires now will partly be a function of the burdens that each party has borne in the past. In a bilateral setting, more precisely, the fairness constraint is especially sensitive to what each party may have done to the other in the past. So the bilateral fairness constraint is especially pertinent where the current bilateral relationship is being negotiated against the background of wrongful past harms imposed by one party on the other.

To see the practical significance of all this, let us return from abstract reflections to concrete examples. More specifically, let us consider two examples that have important structural parallels to autocratic exile compromises. The first example is an actual, historical case. Throughout the 1980s and early 1990s, Pablo Escobar's infamous Medellin Cartel reaped staggering benefits from cocaine trafficking into the United States, against the background of the US ‘War on Drugs’.Footnote ⁶⁰ With an illicit personal fortune estimated at 30 billion dollars in 1993 (equal to 60 billion dollars in 2019 terms), Escobar was able to buy and/or terrorize a significant portion of Colombia's governing class, compelling successive Colombian administrations to negotiate with him. Thus for instance, in 1985, Escobar funded a notorious raid on the Colombian Supreme Court, which featured the taking of 12 justices as hostages and aimed at preventing drug lords' extradition to the United States. This aim was ultimately achieved in 1991, when a persistent campaign by Escobar and his associates led Colombian leaders to cave and pass a constitutional amendment prohibiting extradition of Colombian nationals. Similarly, rather than insist that Escobar face normal criminal proceedings, President Cesar Gaviria agreed to allow the kingpin to confine himself to a ‘prison’ that was actually an enormous mansion (designed by Escobar himself). In exchange for this safe ‘internal exile’, Escobar was supposed to commit to permanently retire from criminal activities.Footnote ⁶¹

Now, one could very plausibly argue that, from a fully impartial perspective, these compromises were distinctly suboptimal, in light of their deleterious impact on the deterrence of kingpins around the globe. So it is easy to see why the world as a whole would have fared better had all Colombian politicians refused to concede anything to Escobar. According to the deterrence view, then, it would have been entirely appropriate for any country in the world to refuse to aid the Colombian government in compromising with Escobar and his ilk. So the United States, for example, was justified when it pressured the Colombian government not to compromise with Escobar.Footnote ⁶²

Suppose, however, that one accepts the following two claims. First, that the rise of drug lords such as Escobar would have been impossible without the War on Drugs: had the US government not been so hell-bent on its quixotic effort to eradicate all trade in certain substances, commercial exchanges of these substances would not have been nearly as lucrative, and gangsters like Escobar would accordingly lack both the motive and the means to spread violence and chaos. Second, that partly for this reason, the War on Drugs has been a policy disaster, wrongfully harming hundreds of millions of people around the world, among them a very large number of Colombians.

Both of these claims have been endorsed repeatedly by experts on drug policy.Footnote ⁶³ But it is not necessary to defend them here. The point is simply that if they are granted – at least for the sake of discussion – then it follows that the deterrence view is wrong. There is at least one country that was not justified in refusing a Colombian request for help in negotiating with Escobar – and that is, of course, the United States. In line with the fairness view, the US government ought to have facilitated the Colombian government's effort to compromise with Escobar, since the need to pursue such a compromise would not have arisen to begin with had the United States itself not persisted in its harmful anti-drug crusade, with the empowerment of the drug cartels as its foreseeable result.

Now consider a second, hypothetical example. Building on the earlier discussion of hostage crises, and on the aforementioned reference to beleaguered leaders as large-scale hostage-takers, this example directly models the problem of autocratic exile on a small-scale hostage situation:

If these are the only morally relevant facts, then it may be hard to say what Forest Owner ought to do. But suppose that there is another fact that must be accounted for. It is Forest Owner herself who, in full knowledge of Infamous Robber's criminal intentions, sold him the expensive equipment without which Robber could not have broken into the Victims' home. This fact clearly matters morally, for it means that Forest Owner's own wrongful actions have foreseeably empowered Robber at the Victims' expense. In turn, seeing as she herself is implicated in the Victims' predicament in this way, Forest Owner ought to prioritize the Victims' interest in emerging from this predicament alive (assuming there is no clear evidence that they choose to fight Robber), over the community's general interest in deterring future Robbers.

Much the same point, in turn, applies when a foreign government's wrongful actions foreseeably empower an autocrat at his victims' expense. Assuming that the victims prefer the autocrat's flight over a bloody fight, an implicated foreign government owes it to the victims to facilitate this flight. The implicated party – just as in the Escobar and Infamous Robber cases – ought to prioritize its particular moral duties to the victims whose predicament it helped create over the moral value of deterrence.

Implicated countries' wrongful omissions

The previous subsection explored one category of cases where implicated countries ought to facilitate autocratic exile. This subsection examines a second category. As in the first category, here too implicated countries' governments are guilty of serious wrongs. But, unlike the first category, here governments are guilty of wrongful omissions, rather than wrongful actions. More precisely, it is private actors based in the relevant countries whose actions foreseeably empower autocrats. The governments that regulate these private actors simply fail their moral duty to stop this private empowerment of autocratic rule.

In turn, arguably most significant case of private empowerment of autocratic rule is the financial empowerment of kleptocrats by foreign private banks, especially those banks based in the world's most affluent democracies. This particular from of autocratic empowerment is a structural problem, in terms of both supply and demand.

On the demand side, kleptocrats have recurrent incentives to ensure that the largest possible portion of their ill-gotten gains is stored abroad. For one thing, doing so is the most reliable way of complicating the work of any law-enforcement agencies that might seek to track their illicit assets.Footnote ⁶⁴ Furthermore, no kleptocrat will rely solely on his country's financial system to protect his ill-gotten fortunes, since every kleptocrat knows that if he were to lose power, his illicit assets stored at home are bound to be confiscated (whether by a reformed government or by equally grasping rivals). Accordingly, kleptocrats, as corruption scholar J.C. Sharman puts it, ‘have a strange relationship with the rule of law: contemptuous and corrosive of it at home, they are nevertheless keen on locating their wealth in states with strong property rights and effective laws’.Footnote ⁶⁵

The supply side is no more encouraging. There is an abundance of evidence that numerous banks based in affluent democracies have long been – and continue to be – keen to serve as kleptocrats' money-launderers, paying little regard to anti-money-laundering regulations.Footnote ⁶⁶ The other aspect of the supply problem is that in the overwhelming majority of cases, the governments tasked with regulating these banks refrain from meaningful criminal sanctions for the laundering of kleptocrats' ill-gotten gains. This is true even in those instances where governments officially recognize unequivocal evidence of bank violations of criminal prohibitions against such laundering.Footnote ⁶⁷ In most other instances, governments opt for willful ignorance of such violations – ignorance that clearly does not exculpate the government for its blatant failure to act.Footnote ⁶⁸

Now, my suggestion is that the relevant governmental omissions have a similar moral effect on the wrongful government actions discussed in the previous subsection. Here as there, the link between the foreign government's moral failures and autocratic rule generates a fairness-based duty to facilitate autocratic exile – even at the expense of global deterrence.

One way to see this duty is to recall an earlier observation, regarding the meaning of a foreign government's refusal to facilitate autocratic exile. Such refusal, we previously noted, is akin to forcing the transitioning society into a very costly confrontation with the autocrat. The deterrence view, in turn, holds that such forcing is warranted because it serves the global interest in deterring autocracy: the world's moral labor, as it were, happens to be divided in such a way so as to require that the transitioning society bear the dramatic costs of confrontation. But, in the circumstances in which we are interested, the transitioning society can reasonably complain that it would not be facing this extraordinary burden to begin with, had the foreign government in question been taking care of its own – far more modest – share of the world's moral labor, by preventing private actors under its jurisdiction from empowering the autocrat.Footnote ⁶⁹ The availability of this complaint, in turn, explains why the culpable foreign government ought not impose the burden of confronting the autocrat on the transitioning society. Instead, so long as it is sensible to assume that the transitioning society accepts an exile compromise, the culpable foreign government is under a duty to facilitate such a compromise.

Unfair imposition

In the previous subsection, our focus was on a foreign country's wrongful omissions. But the fairness view suggests that an implicated country can have a duty to facilitate autocratic exile even in certain cases where its policies empowering the autocrat's rule are assumed to be justifiable rather than wrongful, all things considered.

The primary example here once again concerns foreign countries' lenience toward private banks, and the way in which such lenience supports foreign crime. More specifically, this example concerns lenience extended to ‘systemically important’ banks that routinely launder the proceeds of crime. Senior law-enforcement officials have publicly sought to justify such lenience, by pointing to the economic repercussions of a collapse of major banks in the aftermath of severe criminal sanctions targeting them.Footnote ⁷⁰ However, our earlier conclusions remain intact even if we assume, arguendo, that these officials are in fact correct.

The main reason has to do with a simple observation. The fact that implicated countries' governments have repeatedly refused to prosecute major banks for mass-scale money laundering (even in the face of overwhelming evidence of such laundering) has caused profound damage to the deterrence of financial crime.Footnote ⁷¹ But this damage, in turn, undermines any attempt by the same governments to invoke deterrence of financial crimes as a reason to block a transitioning society's exile compromise with its kleptocrat.

One way to see this point is to consider yet another angle on unfair distribution of costs in the pursuit of collective ends. Normally, a necessary condition for the justifiability of imposing serious costs on others in pursuit of collective ends is a willingness to incur comparable costs oneself in pursuit of the same ends. If one is unwilling to incur comparable costs for the sake of the relevant ends, then one cannot appeal to these ends to justify the imposition of such costs on others.

This point arguably obtains even in a hierarchical setting. Suppose, for example, that a senior manager at a large non-profit announces a junior-employee pay-cut, in the face of a deep financial crisis. By way of moral justification, the manager proclaims that all junior employees ought to be willing to prioritize the organization's socially valuable mission over their personal finances. It would clearly be morally untenable for the same manager to then dismiss calls to cut his own pay. Were he to be dismissive in this way, the natural reaction would be to accuse the manager of unfairly ‘prioritizing’ the organization's social mission only so long as someone else is paying the cost of this mission.

The same point, in turn, applies even more clearly in non-hierarchical settings, such as our international case, featuring sovereign countries that are supposed to relate to one another as equals. As was emphasized earlier, foreign countries that refuse to facilitate an exile compromise between a transitioning society and its autocrat must be willing to defend such a refusal even if it is universal, thus effectively imposing on the transitioning society the dramatic costs involved in a full-blown conflict with the autocrat. But foreign countries clearly cannot appeal to the deterrence of financial crime to justify imposing such costs on a transitioning society, if their own law-enforcement decisions regarding the largest banks show that they are patently unwilling to incur even lesser (economic) costs for the sake of the same end.

The future of international law: the kleptocrats, the murderous, and the ICC

Let us take stock of our findings so far. The preceding arguments yield an important prerogative for transitioning societies that are considering an exile compromise with their autocrat, as well as important duties incumbent upon certain foreign countries. Each transitioning society normally has the moral option of seeking an exile compromise with its autocrat, notwithstanding the fact that such a compromise sets back global deterrence. In turn, the fact that a given foreign country has been entangled in an autocrat's rule is normally sufficient to generate a duty for the relevant country to facilitate autocratic exile in the interest of domestic peace and democracy – once again, even at the expense of global deterrence.

In this final section, I take up at length an important objection regarding the scope of my argument. This objection has to do with the distinction, noted at the beginning, between autocrats who have been implicated in mass atrocities (call these ‘murderous’ autocrats) and ‘non-murderous’ kleptocrats, who have not been involved in such atrocities. As we have seen, since the turn of the millennium, the exile prospects of these two kinds of autocrats have differed markedly. But this difference has been conspicuously absent from my analysis. This absence (according to the objection) is problematic. The reason is that if exile should be entertained for the murderous just as it should be entertained for non-murderous kleptocrats, then my account seems to pull the rug from under the feet of the evolving regime of international criminal accountability for atrocity crimes. Am I really willing to countenance such damage to international bodies with obvious progressive potential, such as the ICC?

Ultimately, the answer is ‘yes’. However, since I recognize that this is a radical conclusion, I want to approach it gradually. My opening step is to preempt an intuitive but erroneous thought: that even if my reasoning has been cogent regarding kleptocratic exile, this reasoning does not carry over to the case of murderous autocrats, because here the practical stakes are much higher.

This thought does not align with the facts. The practical stakes involved in exile for ‘mere’ kleptocrats are at least as high as the stakes involved in exile for the murderous. This is not only because even ‘mere’ kleptocrats, who have never before engaged in large-scale violent repression, will be tempted to fight the opposition to the bitter end, if they know that fleeing the country is simply not an option.Footnote ⁷³ Less obviously, the practical stakes in terms of lives saved are at least as high when it comes to ending kleptocratic rule, because kleptocracy too is a mass killer. There are multiple possible examples, ranging from corruption-stricken healthcare systems that have collapsed in the face of the COVID-19 pandemic,Footnote ⁷⁴ to corruption-induced construction shortcuts that routinely lead to tens of thousands of avoidable deaths from moderate earthquakes.Footnote ⁷⁵ So if there is a fundamental normative difference between exile for the murderous and exile for ‘mere’ kleptocrats, this difference cannot be due to any ‘lower stakes’ in ending kleptocratic rule. Independently of any differences in how we judge their respective characters, or indeed in how we judge the severity of their respective crimes, we can confidently say that in the case of ‘mere’ kleptocrats just as with the murderous, numerous lives depend on decisions regarding autocratic exile. It follows that if my arguments have been compelling with regard to exile for ‘mere’ kleptocrats, they should also generate (at the very least) a strong presumption in favor of analogous conclusions regarding exile for the murderous.

With this presumption in mind, let us now focus more squarely on the normative significance of the international accountability regime for mass atrocities. At the heart of this regime is the idea of external actors serving as ‘backup enforcers’ of the rights of individual victims, when domestic law-enforcement is either unable or unwilling to hold perpetrators accountable.Footnote ⁷⁶ Such backup enforcement often means that international judicial bodies effectively override the decision of domestic officials not to prosecute (or to pardon) offenders. Am I committed to the idea that it is always problematic for international actors to act with such determination, when a transitioning society accepts a compromise with serious wrongdoers?

No. There certainly could be circumstances in which international actors are justified in defending individual rights that the relevant domestic government effectively refuses to protect. In particular, if it becomes clear that a given government is unwilling to use the law to uphold the basic equality of the members of systematically threatened groups, then it is entirely appropriate for international actors to take a whole slew of actions, going (at the limit) all the way to armed intervention. But this is not an apt description of standard cases where a transitioning society agrees to let an autocrat (and his cronies) off the hook in the name of domestic peace. Officials who refrain from prosecuting the elites of the old regime in order to maintain a fragile ‘pacted’ transition are not akin to bigoted officials who systematically refuse to prosecute certain crimes because their victims have (say) the ‘wrong’ racial, religious, or gender identity.

Still, some critics might worry that the preceding paragraphs elide the most crucial concerns about my position. For one thing, when it comes to exile for murderous autocrats, I seem to be calling on the world's most powerful democracies to actively defy international criminal law and its institutions, which were, after all, established by the international community. But this conclusion smacks of dubious Western exceptionalism. Moreover, such ‘international civil disobedience’ on the part of powerful democracies would surely be fatal to the entire project of international law. ‘International law’, as Ronald Dworkin for example puts it, ‘is fragile, still nascent, and in critical condition. The proposition that a sense of moral duty can justify violations of international law threatens to strangle the child’.Footnote ⁷⁷

I do not want to pretend that these concerns are entirely immaterial: I cannot claim that my position is an uncomplicated friend to international law. Nor, however, is it an unabashed enemy. To see why, note first that my claims do not depend on any particular view of ‘international civil disobedience’. Perhaps such ‘disobedience’ can be justified in certain instances, such as the NATO intervention in Kosovo, famously carried out in the absence of Security Council authorization.Footnote ⁷⁸ But my argument does not hinge on this possibility. Much more modestly, the argument points to mechanisms which are already built into existing international law, but which have been (massively) under-utilized.

More specifically, I have in mind Article 16 of the Rome Statute underlying the ICC. This Article authorizes the UN Security Council to postpone ICC prosecutions for 1-year renewable periods in ‘the interests of peace and security’.Footnote ⁷⁹ Yet the Security Council has never utilized Article 16, despite (among other things) the formal efforts of the African Union to set this Article in motion on two separate occasions.Footnote ⁸⁰ I am suggesting, with the African Union and against the Security Council, that Article 16 should receive much more serious consideration. This is true, at least, in those cases where foreign countries have been entangled in a murderous autocrat's rule, and where they consequently have a duty to his victims to facilitate his exile.

Is this appeal to Article 16 merely a form of legalistic dishonesty, effectively dismantling the ICC while preserving its façade? No, because the very existence of this Article is deeply pertinent to the objection at hand. This objection trades on the intuitive appeal of policies that sustain and bolster international legal institutions such as the ICC. But this intuitive appeal, in turn, is inseparable from the hope for a global future in which these institutions are genuinely analogous to the domestic legal institutions of consolidated liberal democracies. The very existence of Article 16, however, reminds us that such a future is (at best) extraordinarily distant. No consolidated democracy, after all, would ever even entertain a version of Article 16 as part of its own domestic law. No consolidated democratic government, even one that accords its prosecutors extensive discretion as to which criminal cases to bring, would ever announce formal provisions allowing legal officials to spare violent criminals simply due to fears that these criminals would unleash more violence if prosecuted. Such fears are not credible in a consolidated democracy – in fact, that is precisely what makes a democracy consolidated. Any consolidated democracy, by definition, enjoys the coercive capacities that are necessary to ensure the effective subjection of perpetrators to the law – no matter how willing they are to resort to violence.

Article 16, however, foregrounds the fact that international legal institutions manifestly lack these crucial capacities. The Article thus orients our normative comparisons in the right direction. It is implausible to compare the domestic benefits that compromises with murderous autocrats might yield to a global ‘fantasy upon fantasy’,Footnote ⁸¹ of the sort explicitly urged by Dworkin, featuring a fully empowered ICC, perfectly capable of enforcing its edicts through some form of reliable global police force. Rather, the morally appropriate comparison is between the domestic benefits of compromise with the murderous and the actual international criminal system – shot through with (at best) amoral power politics, routinely accused of anti-African and other biases, and fundamentally lacking independent coercive capacities. Yet once this is the comparison on the table, I see no obvious reason to think that the domestic benefits highlighted by the fairness view are bound to lose out to any relevant international gains.

A further, related point is also worth stressing. It is a sad but simple truth that a fully empowered ICC (in the sense just identified) could never materialize, if none of the world's most powerful countries were ever willing to place their military might at the service of international criminal justice. This observation is not meant as a license for messianic military interventions, nor as a reversal of the position that I have been defending throughout this essay. Rather, this observation is relevant here because it points to a necessary condition for the moral permissibility of any powerful foreign country actively sabotaging an exile compromise with a murderous autocrat. If what I said earlier is correct, and any foreign country that opposes an exile compromise with an autocrat is effectively seeking to compel the autocrat's victims to fight him, then fairness arguably requires that any such powerful country also be willing to – literally – join the fight. Certainly, a government's willingness to sacrifice the lives of some of its own citizens for a global cause may not suffice to make such sacrifice legitimate, nor suffice to legitimate the government's demand that other societies sacrifice some of their members too. But we are already in a position to see why such willingness is morally necessary. In the absence of such collective willingness to sacrifice, it is implausible to demand that other societies make sacrifices for a global goal – in general, and particularly when other societies will be forced to pay an extraordinarily heavy price in pursuit of this goal.Footnote ⁸²

In turn, taken together, the last few paragraphs also allow us to contest the charge of worrisome Western exceptionalism. The best way to do so is to consider a specific scenario, combining elements from several actual African cases. Suppose that, because of amoral geopolitical calculations, specific Western powers spend years arming, financing, and otherwise supporting a remote dictator, in full knowledge of the mass-scale atrocities that the dictator has repeatedly committed against his own people (as the United States and France did with Chad's Hissene Habre, for instanceFootnote ⁸³). Years later, facing a real risk of defeat in a civil war, the dictator must choose whether to fight on, or to try to flee abroad. His former Western backers, no longer willing to support him, now proclaim themselves firm adherents to international criminal law, and consequently insist that the dictator must face trial for the atrocities that he has overseen, effectively blocking an exile solution, and thus forcing an actual war between the dictator and his domestic opposition. The African Union, protesting this newfound Western rigorism as a real threat to the dictator's victims, initiates a formal process to activate the ICC's Article 16, in order to salvage peace negotiations. The same Western powers, however, scuttle this process, and the dictator's victims consequently endure more violence – all while the relevant Western governments would not even entertain the idea of committing their own troops to any international military force that might join the fight against the dictator. After several years of civil war, the dictator, facing imminent violent defeat, actually flees abroad only to be sent to The Hague following intense pressure from the same Western powers. The dictator's eventual trial and conviction are officially celebrated by the same powers, who herald the resulting global deterrence while only offering passing rhetorical acknowledgement of their own past entanglement in his crimes, without any substantive policy implications.

Now compare this – eminently realistic – scenario to one where the relevant Western governments genuinely own up to their entanglement in the dictator's rule, accordingly support the use of Article 16, and facilitate exile for this dictator to spare his victims further harm. Is it really clear, even from a moral perspective heavily concerned with Western exceptionalism, that the former, deterrence scenario, is preferable to the latter, exile route? I confess that I fail to see why.

I am strongly aware that my reasoning here points to decidedly grim conclusions, and that these conclusions are likely to induce a lingering sense of unease. After all, I seem to be calling on the world's most powerful democracies – long perceived as the essential (if deeply flawed) ‘stewards’Footnote ⁸⁴ of liberal values in international politics – to turn their backs on the morally inspiring international project of ending impunity for politically mighty criminals. The resulting unease, in turn, is likely to be felt most acutely by those who cling on the hope that the ‘nascent’ international legal system can generate its own momentum, even in the enduring absence of anything like a world state with a global police force. Some philosophers of international law, for example, believe that the very existence of international prosecutions for crimes against humanity can contribute to the transition toward a more cosmopolitan future, in which ‘humanity’ itself becomes a more salient locus of identity at the expense of separate national identities.Footnote ⁸⁵ Am I really willing to scuttle this cosmopolitan hope? To derive, from affluent democracies' sordid past, policy prescriptions that take us further away from an uplifting vision for the global future?

My main response to this question is to note commonsensical limitations on the practical reach of the entanglement duties that I have emphasized here. Entangled parties' duties toward those whom they have harmed normally take precedence over their general moral reasons to make the world a better place. But ‘normally’ is not the same as ‘always’. There could arise drastic circumstances in which the (future) good of the world does have moral priority over past sins.

Suppose, for example, that during World War II, the Axis were on the verge of invading India.Footnote ⁸⁶ Suppose further (at least arguendo) that India's inability to protect itself from this threat on its own, and consequent desire to compromise with the Axis, stemmed in large measure from the centuries of colonial pillaging to which India has been subjected by the British. Even against this sordid historical background, it would have still been morally appropriate, all things considered, for the British to compel India to avoid compromising with the Axis – even by using force if necessary. Although the British would have clearly been entangled in India's predicament, the impartial good of the world writ large would have weighed so heavily, that the duties borne by the British toward India would have had to take a backseat to the fight for a world not dominated by the Nazis and their allies.

The general lesson of this example is that in global ‘supreme emergencies’,Footnote ⁸⁷ the past does not morally govern the shape of the present and the future. Therefore, if someone could muster the evidence necessary to show that the linear advancement of international criminal justice, entirely unencumbered by past wrongs perpetrated by powerful democracies, also qualifies as a global ‘supreme emergency’, then I would be willing to revise the normative guidance I have offered here. I am simply skeptical that the requisite evidence exists.

To be sure, even the last few paragraphs may not bring much solace to those who want to hope that a better world may come about simply through the progressive realization of a normative vision, rather than through ‘extra-theoretical’ calamities. It would surely be better, for example, to achieve more robust supranational institutions (such as a solidifying European Union) because of the peaceful development of genuine cosmopolitan sentiment, without relying on the exigencies of an appalling war. Similarly, it would be much better to progress toward the ideal of a world state simply because a majority of humans come to see themselves first and foremost as citizens of the world, rather than because global environmental calamity has made the idea of national governments obsolete. But the mere fact that we would prefer non-tragic pathways to a better global future does not mean that such pathways will be available. In the world that we actually inhabit, tragedy is all around us. Even if participants in the debate on how to deal with the world's worst rulers can agree on little else, this much has been – and should remain – common ground.