The Difference Law Makes: Domestic Atrocity Laws and Human Rights Prosecutions

Abstract

This article offers the first systematic analysis of the effects of domestic atrocity laws on human rights prosecutions. Scholars have identified various political and sociological factors to explain the striking rise in human rights prosecutions over the past 30 years, yet the role of domestic criminal law in enabling such prosecutions has largely been unexamined. That is surprising given that international legal prohibitions against human rights atrocities are designed to be enforced by domestic courts applying domestic criminal law. We argue that domestic criminal laws against genocide and crimes against humanity facilitate human rights prosecutions in post-authoritarian states by helping to overcome formal legal roadblocks to prosecution, such as retroactivity, amnesties, immunities, and statutes of limitations. Using original data on domestic atrocity laws and human rights prosecutions in new democracies, we find that atrocity laws increase the speed with which new democracies pursue prosecutions, as well as the overall numbers of trials they initiate and complete.

Over the past 30 years, an increasing number of societies have opted to prosecute former regime officials for human rights violations in domestic criminal courts (Sikkink 2011). This trend is puzzling given long-standing norms that traditionally shielded government officials from criminal accountability. Scholars have sought to explain why prosecutions occur in some post-transition societies but not others, and in doing so have identified a number of factors—including political incentives and regional socialization—that increase their likelihood (e.g., Appel & Loyle 2012; Dancy & Michel 2016; Kim 2012).

Yet despite these findings, one potentially crucial factor has received little attention: domestic criminal law. The international atrocity regime, which criminalizes serious human rights violations like genocide and crimes against humanity, is designed to be enforced primarily through domestic courts. To make such prosecutions possible, treaties obligate states to legislate international crimes into domestic law. These domestic legal provisions, which define and criminalize acts of genocide and crimes against humanity, may be referred to as “atrocity laws.”¹ Many formerly repressive states that since transitioned to democracy adopted such atrocity laws while they were still ruled by abusive governments. Others did not (Berlin 2015). Does having atrocity laws on the books make prosecutions for past abuses following a transition more likely? Or do societies determined to hold former regime officials accountable find ways to prosecute regardless of whether their legal systems have the proper legal provisions? In other words, is the international atrocity regime operating as it was designed to?

In this article, we offer the first systematic analysis of the potential effects of national criminal laws against atrocity crimes. Using a new and comprehensive dataset on the existence of domestic criminal laws against genocide and crimes against humanity adopted since World War II, we find evidence that the existence of atrocity laws, on average, correlates with greater numbers of human rights prosecutions in newly democratic states. States with standing atrocity laws are also quicker to initiate trials and reach more verdicts following periods of regime change than states without such laws. The reason, we argue, is that these laws help relevant legal actors avoid or overcome formal legal roadblocks that often thwart prosecutions.

With these findings, we add to a body of new research arguing that the mobilization of legal institutions has been crucial for the push toward human rights enforcement worldwide (Hilbink 2012; Lake 2014; Michel & Sikkink 2013; Ocantos 2014). As we explain in the next section, the return to legal institutions as an explanatory factor serves as a counterweight to explanations that focus on either rational political elites or transnational socialization. This article demonstrates that legal institutions, once formed, can exert long-term impacts on societies, independent of regional trends or short-term political circumstances. Existing research shows that human rights enforcement is often the product of a confluence of political and social conditions and the work of a variety of different actors, like motivated prosecutors and civil society activists. Yet attention to domestic criminal law highlights how contingent, formal legal resources available to these actors can make the difference between successful and unsuccessful efforts. In this sense, this article provides a more fine-grained view of the legal context and domestic actors that have contributed to the “justice cascade” (Sikkink 2011). Our findings also challenge recent studies finding that formal legal provisions have little effect on the protection of human rights (Chilton & Versteeg 2015; Keith et al. 2009).

Explaining Human Rights Prosecutions

Human rights prosecutions are “any prosecutorial event that reaches a domestic court after an arrest warrant and/or an indictment has been issued for cases related to human rights abuses committed by state agents” (Dancy & Michel 2016: 174). These prosecutions are of two basic types: (1) high-level trials that involve military and state leaders responsible for campaigns of repression or war crimes and (2) low-level trials that prosecute regular police or security service members for administering repressive violence, including torture, illegal imprisonment, disappearances, or sexual violence. Human rights and humanitarian law treaties, along with customary international law, establish positive duties for states to provide remedy to victims of human rights violations or to punish those responsible. Therefore, human rights trials can be understood as a form of compliance with international legal obligations.

Data on human rights prosecutions collected by the Transitional Justice Research Collaborative (TJRC) database—the most complete source on mechanisms to address accountability for human rights violations—show that prosecutions are initiated in various contexts.² As depicted in Figure 1, a large majority (64.6 percent) occur following democratic transitions, like in Argentina,³ while roughly one-third (32.3 percent) occur in post-civil war settings, like in Rwanda.⁴ Around 48 percent of prosecutions occur within two years of a change in a country's individual leadership.⁵ These contexts can overlap. For example, it is possible for a state to undergo simultaneously a democratic and a post-conflict transition—as did Peru in the late 1990s. It is also possible for post-authoritarian or post-conflict contexts to coincide with leadership change, which is often referred to as a “ruptured transition” (Huntington 1991). Only 20 percent of all human rights prosecutions catalogued by the TJRC occur in stable democratic or autocratic contexts where none of these three types of transition occurred.

Note: Percentages do not add up to 100% because these contexts can overlap. Figures are reached by dividing the number of trials in a given context by the total number of trials in the Transitional Justice Research Collaborative dataset.

Figure 1. Proportion of Total Human Rights Prosecutions by Type of Transition, 1970–2010.

Complying with international obligations and initiating human rights prosecutions creates a number of legal and political dilemmas for government leaders in the wake of atrocity, war, or leadership change. State agents traditionally enjoy immunity against prosecution, especially if they are still in office (Fox 2002). Holding current or former leaders accountable for abuses also expends political capital, since leaders often still enjoy the support of large constituencies. Even a maligned former leader like Augusto Pinochet, upon his death in 2006, still commanded the respect of members of the military and thousands of civilians (Franklin 2006). These complications do not end with high-level trials; low-level trials also pose challenges. Governments must simultaneously rely on police and security forces for coercion while still punishing individuals that use excessive violence. This creates a potential risk of backlash against state leaders: if they push too far with punishment, then they will lose the support of their agents. For instance, coups threatened democratic transitions in Argentina and the Philippines in the late 1980s. These coups were reactions to sustained legal campaigns to hold these countries’ armies accountable for past human rights violations (Alfonsín 1993; Mydans). Given these challenges, it is puzzling why high- and low-level trials now regularly occur.

Three types of institutionalist approaches—rational choice, sociological, and legal-historical—aim to explain the surprising surge in human rights trials over the last three decades (Dancy & Michel 2016). Rational choice institutionalist explanations assume that actors’ preferences are exogenous, that institutions place constraints on those actors, and within those constraints actors engage others strategically (Acemoglu & Robinson 2006; North 1990; North & Weingast 1989). The first wave of transitional justice research hewed closely to rational-choice explanations centered on the political constraints facing state leaders. This research, which relied primarily on in-depth cases studies, argued that the balance of power between new transitional governments and holdover elements from previous authoritarian or civil war regimes would in large part determine the legal response to human rights violations. In states that experienced ruptured transitions (as opposed to negotiated ones), in which the old regime was forcibly removed from power, it is less likely that veto players associated with the former regime could block litigants’ efforts to bring trials, either by exerting influence over the legal process or by codifying amnesties prior to or following the transition (Huntington 1991; Olsen et al. 2010).

The rational choice institutionalist perspective has two main drawbacks. The first is that its predictions regarding the patterns in human rights prosecutions have not withstood time. Prosecutions increased worldwide over the past few decades, and this increase is not driven solely by cases that experienced a ruptured political transition. Figure 1 shows that at least 52 percent of prosecutions occurred in contexts without leadership change. This suggests that prosecutions are not simply directed by new political coalitions that assume control of government. While shifting political coalitions certainly account for some of these variations, they cannot account for all. The second problem with rationalist accounts is that they attribute little agency to lawyers, prosecutors, or judges. The assumption that human rights prosecutions result from strategic interactions between political elites in the executive and legislative branches implies that courts and litigators are secondary actors. However, recent research shows that lawyers are creative in their methods and judges behave strategically (Hilbink 2007; Moustafa 2007; Ocantos 2014). Theories explaining the rise of human rights prosecutions must account for the behavior of legal actors.

A second approach to explaining variation in human rights prosecutions is sociological. Sociological institutionalism “emphasizes ‘institutional isomorphism’ across countries” attendant to normative trends in world society (Dancy & Michel 2016: 176; Powell & DiMaggio 1991). In one such account, human rights prosecutions are attributable to international convergence on liberal legal norms of justice and accountability for human rights abuses, which are embodied in multilateral legal agreements and actively promoted by a transitional justice “industry” made up of international nongovernmental organizations (INGOs) and wealthy Western donor states (Gready 2010; Nagy 2008). A second sociological account relies on the influence of cultural neighbors. Peer diffusion—that is, the increase in occurrences of trials among a state's cultural and linguistic peers—reinforces the idea that trials are possible and appropriate and helps inspire local actors, like victims groups and other civil society organizations, to push for prosecutions in their own states (Kim 2008, 2012; Sikkink 2011; Simmons & Elkins 2004). In this formulation, elites seek to emulate the socially appropriate behavior of other elites and feel pressure to make choices that resemble those of their peers.

One concern for sociological explanations is that they struggle to account for variation within regions. For instance, the move toward justice in Latin America, a region often characterized as disposed toward of rule of law norms (Hafner-Burton & Ron 2013), was neither uniform nor linear. Global or regional norms may be necessary conditions, but they are certainly not sufficient to explain the pursuit of human rights enforcement.

This article adopts a third, historical institutionalist (HI) approach. In general, HI focuses on the origins and evolution of rules and state institutions (Thelen 1999). For this approach, “dynamic tensions and pressures for change are built into institutions,” and the task for scholars is to root out how and why laws effect change over time (Mahoney & Thelen 2010: 14). An HI analysis of human rights prosecutions would model them not as a direct function of self-interested leaders and their political coalitions, nor as a uniform outgrowth of larger patterns in international society. Instead, such an analysis would examine the ways in which domestic laws cast a shadow, shaping preferences and enabling actions given shifts in political or social circumstances. Recent studies pursue this explanatory strategy. Michel and Sikkink (2013), for example, identify the right of private prosecution—i.e., the right of victims and not merely prosecutors to open and actively participate in criminal investigations—as a crucial mechanism through which civil society pressure can initiate trials and generate further momentum toward more prosecutions (Dancy & Michel 2016).

While recent research advances our understandings of the historical legal conditions that facilitate trials, this article highlights a formal legal factor that has not been subject to testing but which is crucial to how the international regime underpinning many of these trials is designed to operate: domestic criminal law.

How Atrocity Laws Facilitate Human Rights Prosecutions

We argue that domestic criminal laws against genocide and crimes against humanity help facilitate human rights trials by lowering technical legal barriers to prosecution. Consistent with a historical-institutionalist approach, we do not contend that atrocity laws cause prosecutions per se, but that these laws increase the prospects for the types of actions—such as grassroots mobilization and prosecutorial initiative—that do lead to trials. Atrocity laws thus exert a constitutive effect on trials by altering how prosecutors and judges view potential and actual challenges to prosecutions, tipping the balance toward decisions that make trials successful.

To understand how atrocity laws can help lower legal barriers to prosecution, it is necessary to understand the options available to governments who wish to put former regime officials on trial. If a new government wishes to prosecute its own or other states’ former regime officials for human rights violations, it can do so on one of four legal bases: (1) charge individuals under existing, dedicated atrocity laws, (2) adopt new atrocity laws and apply them retroactively to past crimes, (3) directly apply international criminal law domestically in the absence of domestic legislation, or (4) charge individuals under existing provisions for so-called “ordinary crimes.” As we explain below, the first option is most consistent with international criminal law treaties. Pursuing criminal prosecutions under standing atrocity laws also provides an additional advantage: it avoids or overcomes particular legal roadblocks that often hinder prosecutions. As we further explain below, the other three options introduce legal complications that can ultimately block efforts at accountability. In short, prosecuting under standing atrocity laws is the path of least resistance to prosecutions.

Option 1: Dedicated Atrocity Laws

The first option is to prosecute perpetrators under a pre-existing domestic criminal law that explicitly codifies international crimes relating to human rights atrocities, such as genocide, crimes against humanity, forced disappearances, or torture. This is how the international criminal law regime is designed to be enforced (Bassiouni 1983: 29; El Zeidy 2008). International criminal law treaties that criminalize serious human rights violations, such as the Genocide Convention (1948: Article 6) or the Convention Against Torture (1984: Article 7), envision domestic courts as the primary enforcers of their prohibition regimes. Likewise, the Rome Statute of the International Criminal Court (1998: Articles 1, 17) designates national courts as the main enforcers of the treaty, allowing the ICC to prosecute cases only when domestic jurisdictions have shown themselves unable or unwilling to do so.

To ensure that states’ domestic courts can prosecute atrocity crimes, international criminal law treaties obligate parties to adopt the necessary legislation to prosecute these offenses under domestic criminal law.⁶ Nevertheless, not all states that ratify these treaties follow through with their obligations to adopt domestic implementing legislation. For example, of 147 states that have ratified or acceded to the Genocide Convention, 125 possess national criminal laws allowing their domestic courts to prosecute genocide. When states do adopt such legislation, they often do so years, or even decades, after ratification. Among state parties to the Genocide Convention that have adopted domestic genocide legislation, the median time between ratification and legislation is 16 years, with some states taking as long as 60 years.⁷ The category of crimes against humanity has been incorporated into domestic laws at far lower rates. This is due in part to the fact that crimes against humanity lacks its own dedicated treaty regime. Until the establishment of the Rome Statute in 1998, there was no obligation under international law for states to adopt domestic laws against crimes against humanity. Nevertheless, prior to 1998, as many as 24 states did adopt such laws, with the number increasing to 80 by 2016. Figure 2 shows a map of atrocity law implementation. The light gray countries are those that passed neither genocide laws nor crimes against humanity laws. The dark gray countries are stable regimes that passed at least one. The black countries are transitional states that experienced regime change at some point since 1970 and passed at least one atrocity law.

Figure 2. Atrocity Laws in the World, 2010.

As Berlin (2015) shows, states adopt atrocity laws through one of two legislative pathways. The first pathway involves conventional, targeted legislation that either establishes a new standalone statute or that amends an existing criminal code to introduce the new offenses. The second pathway is for states to adopt atrocity laws as part of wholesale reforms to their existing criminal codes. As with other legal codes, states occasionally modernize their criminal codes by replacing them in their entirety. Though the desire to adopt atrocity laws does not motivate criminal code overhaul, states undertaking sweeping reform sometimes use the opportunity to include atrocity laws as part of the larger package of revisions. This pathway to adoption through comprehensive criminal code reform is especially common for repressive states that are otherwise puzzling cases of adoption. Consistent with a historical-institutionalist explanation, what repressive governments often perceive as the innocuous, technocratic decision to include atrocity laws as part of a broader modernization project can produce unexpected consequences years later. This was the case in Guatemala, where laws introduced in the 1973 criminal code overhaul would later be used to prosecute former government and military officials, including the former head of state Efraín Ríos Montt, for genocide and crimes against humanity.

There are normative reasons why governments, prosecutors, or victims would prefer to prosecute former regime officials under laws explicitly referring to international crimes as opposed to those for ordinary crimes, such as murder or kidnapping. In short, punishing offenders using provisions that explicitly invoke international crimes better serves the retributive, deterrent, and expressive goals that proponents of trials variously ascribe to them (Drumbl 2007). First, from a retributive perspective, “the infliction of punishment rectifies the moral balance insofar as punishment is what the perpetrator deserves” (Drumbl 2007: 61). Crucial differences exist between conduct classified as international crimes and that classified as ordinary crimes in terms of gravity, scale, intent, context, and the involvement of the state organs. By inscribing these offenses into international law, the international community recognizes that certain contextual factors set international crimes apart as particularly heinous. Prosecuting under an international crime provision attaches the appropriate stigma to conduct deserving of the label “atrocity.” The International Criminal Tribunal for Rwanda ruled as much when it denied an offer by Norway to transfer a genocide prosecution to a Norwegian national court because Norway lacked a specific provision criminalizing genocide. Norway had planned to prosecute the defendant, Michel Bagaragaza, under an ordinary homicide law, but the Tribunal ruled that homicide did not appropriately reflect the particular gravity and intent of genocide, which entails “the intent to destroy, in whole or in part, a national, ethnical, racial, or religious group, as such” (The Prosecutor v. Michel Bagaragaza).

Second, from a deterrence perspective, trials help discourage future offenses. International criminal law criminalizes conduct that is often perpetrated by state actors. Prosecutions under provisions for international crimes underscore the role of state actors and institutions in perpetrating offenses. As such, they help undermine long-standing norms of impunity for human rights abuses, and are more likely to put future state actors on notice that there are risks for future abuses (Kim & Sikkink 2010). Finally, from an expressive perspective, the goal of punishment is to cultivate the public's faith in the rule of law. Trials produce “historical narratives, their authentication as truths, and their pedagogical dissemination to the public” (Drumbl 2007: 173). Prosecuting cases under international crime provisions legitimates victims’ claims by publicly and authoritatively affirming that abuses against them were not isolated but systematic and widespread.

Beyond these normative benefits, there is an additional reason why prosecutions under dedicated atrocity laws are preferable to the alternatives: they avoid or overcome formal legal roadblocks that often hinder prosecutions. Absent domestic provisions dedicated to international crimes, governments face one of three alternate options: adopting retroactive laws, directly applying international law, or prosecuting under ordinary criminal provisions. As the next three sub-sections explain, each of these options give rise to particular legal challenges that make prosecutions more difficult compared to those pursued under extant atrocity laws.

Option 2: Retroactive Laws

In the absence of dedicated atrocity laws, a second option for the incoming government is to adopt new laws against international crimes and apply them retroactively to past government crimes. A handful of states have secured convictions through this path. For example, in 1992, following its independence from the USSR, Lithuania adopted a domestic law against genocide, which it used to prosecute former Soviet Lithuanian officials for their roles in atrocities committed against civilians in the 1940s (Quigley 2006: 43–44).

Nevertheless, despite some successful examples, relying on a law adopted post hoc poses fundamental legal and ethical obstacles to prosecution (Elster 2004: 296–299).⁸ Most importantly, criminal justice systems worldwide recognize as a cornerstone principle the idea that one can only be prosecuted for crimes that were defined as offenses at the time they were perpetrated. This is known as the principle of legality, or by the Latin phrase nulla poena sine lege (“no punishment without law”). Courts in some countries have blocked retroactive prosecutions for genocide or crimes against humanity because they violated the principle of legality. For example, in 2011, Uruguay adopted legislation allowing courts to prosecute atrocities committed under the former military dictatorship as crimes against humanity, a category previously absent from Uruguayan criminal law. But in 2013, the Uruguayan Supreme Court declared the law unconstitutional, effectively blocking pending prosecutions, on the grounds that it was adopted long after the alleged crimes were committed, making it an exercise of retroactive punishment (Burt & Lessa 2013). Retroactivity challenges have also raised obstacles to extraterritorial third-party prosecutions based on the principle of universal jurisdiction.⁹ For example, courts in both Norway and Switzerland blocked genocide prosecutions based on universal jurisdiction because their countries’ legal systems lacked domestic genocide provisions (Schabas 2009: 408–409).

Option 3: Direct Application of International Law

A third basis for prosecution in the absence of a dedicated law for international crimes is for domestic courts to apply international law directly. Some countries’ legal systems—those in the so-called “monist” tradition—treat international law as an extension of their domestic systems. In theory, international criminal law can be applied in those systems without domestic legislation. In a few rare cases, judges have been willing to do so (e.g., Ferdinandusse 2006: 76–81). However, in practice, even judges in countries whose legal systems seemingly allow for such direct application have been unwilling to directly apply international criminal law or have done so inconsistently (Ferdinandusse 2006 passim). Among the reasons for rejecting direct application is the objection, again, that it violates the principle of legality, given that prosecutions in these cases are based on offenses that were not previously codified in domestic law (Ferdinandusse 2006: 263–267).

Option 4: Laws for Ordinary Crimes

A fourth and final option in the absence of a dedicated law for international crimes is for governments to prosecute former regime officials under provisions for so-called “ordinary” or “common” crimes, such as murder, kidnapping, and assault. Argentina—where the 1985 criminal conviction of Junta leaders helped spark the global “justice cascade”—took this path (Sikkink 2011: 72). This approach is not ideal, both for the normative reasons outlined above and because the definitions of these crimes under international law often contain specific manifestations of the offense for which prosecutors would struggle to find analogies under domestic law (Teitel 2000: 33–34). For example, while a prosecutor could possibly pursue a genocide case involving mass killings under domestic murder laws, an ordinary criminal provision would likely not cover other acts that constitute genocide under the Genocide Convention, such as “imposing measures to intended to prevent births” (Kreicker 2005: 323; United Nations 1948: Article 2).

While this option avoids the problem of retroactivity, there are a number of other technical roadblocks that generally limit the applicability of criminal laws, and these are more likely to hinder prosecutions pursued under provisions for ordinary crimes than for international crimes. In some instances, judges may dismiss cases based on technical issues arising from the application of ordinary criminal provisions, while in others the prospect of these legal challenges may deter prosecutors from pursuing cases from the start. We discuss three such roadblocks in turn: amnesties, immunities, and statutes of limitations.

Amnesties

Formal amnesty laws are common roadblocks to prosecutions of former regime officials. Such laws are widespread. Data collected by Payne et al. (2015) identify a total of 63 amnesty laws in 34 transitional states. Often such laws are adopted as a concession to an outgoing regime to help facilitate a negotiated transition to a new regime. The authors provide evidence that countries with amnesty laws are no less likely on average to complete human rights prosecutions. However, in many cases, amnesty laws prevented, or at least long-delayed, attempted prosecutions (Payne et al. 2015). For example, Brazil's Supreme Federal Court recently upheld the legality of the country's 1979 amnesty law, which was adopted as part of its transition to democracy, blocking accountability for widespread abuses committed under the former military government (Tang 2015).

In cases where prosecutors face an amnesty pursuing prosecutions under dedicated atrocity laws may reduce the likelihood that amnesty laws block or slow prosecutions. One reason is the tenuous position of amnesties for international crimes, like war crimes, genocide, and crimes against humanity, under international law. While, in the words of one expert, jurisprudence on the legality of amnesties under international law remains “patchy,” there is a growing trend among international, regional, and domestic courts to recognize amnesties for international crimes as illegal (Mallinder 2011: 429). For example, in a 2001 landmark case, the Inter-American Court of Human Rights found that Peru's amnesty laws violated the American Convention on Human Rights, which opened the door to new criminal cases in Peru and elsewhere (Barrios Altos v. Peru 2001; Burt 2009; Raimondo 2011). In a 2004 opinion, the Special Court for Sierra Leone declared that amnesties for serious human rights violate customary international law (Prosecutor v. Morris Kallon, Brima Bazzy Kamara 2004: 73). Likewise, domestic courts have also cited international law to invalidate amnesties, as did the Supreme Court of Argentina in the 2005 Julio Simón case (Bakker 2005).

When courts rule that amnesties are illegal, they enable prosecutors to proceed with cases. Yet jurisprudence on amnesties has focused on international crimes, not ordinary crimes. Should a national court rule its country's amnesty law to be invalid for international crimes, it is possible that the court may still uphold the amnesty for prosecutions for ordinary crimes. Under these circumstances, prosecutors are thus better positioned to pursue prosecutions if they can do so under provisions for international crimes. For example, Guatemalan courts rejected challenges to recent prosecutions of former government officials based on the country's 1996 amnesty law. These challenges failed, because the law specifically excluded genocide and crimes against humanity from its amnesty. Pre-existing criminal provisions for genocide and crimes against humanity adopted in Guatemala in 1973 provided the legal basis for subsequent prosecutions to defeat further challenges (Roht-Arriaza 2015: 365). In the absence of dedicated atrocity laws, prosecutors must rely on provisions for ordinary crimes, which courts may rule are still covered by amnesty laws. Alternatively, prosecutors may use retroactive laws, but this invites the types of legal challenges discussed above.

Immunity Doctrines

Similar to amnesties, immunity doctrines are often invoked to shield former government officials from criminal accountability for abuses committed while they were in power. Immunities are a particularly significant issue when it comes to prosecutions based on universal jurisdiction, because international law has long recognized so-called “functional immunity” or “immunity ratione materiae,” that is, the immunity of government officials from being judged in foreign courts for official actions, even after officials are out of office (Akande & Shah 2010: 825–826). However, since the Nuremberg trials, domestic courts have widely recognized that functional immunity does not apply to offenses classified as international crimes, such as war crimes, crimes against humanity, and torture (Akande & Shah 2010: 839–840). A foreign prosecutor attempting to bring a criminal case under universal jurisdiction will more easily overcome the potential challenges posed by claims of immunity if she can pursue that case under domestic laws for international crimes.

Most famously, former Chilean President Augusto Pinochet invoked immunity in his defense to challenge his extradition to Spain on a Spanish arrest warrant alleging torture in Chile under his rule. But the UK House of Lords ruled that Pinochet could not claim functional immunity, because such immunity could not apply to international crimes. Crucially, the Lords ruled that Pinochet's extradition from the UK to Spain could proceed only if the specific international crime for which Pinochet was indicted—in this case, torture—was criminalized in both jurisdictions at the time of the alleged offenses (R v. Bow Street Metropolitan Stipendiary Magistrate 1999). Even though Pinochet was eventually allowed to return to Chile for health reasons, the existence of criminal laws pertaining to torture in the UK and Spain permitted the Lords to find that Pinochet's immunity did not apply and that he could, in principle, be extradited to Spain (Berlin 2013).

Statutes of Limitation

Finally, statutes of limitations frequently hinder prosecutions (Roht-Arriaza 2015). Prosecutions for human rights violations often take place years or even decades after abuses were committed, as the political conditions that make prosecutions possible can take years to emerge. Many countries in the world recognize statutes of limitations for even the most serious ordinary crimes, such as murder and rape. Yet under international law, statutes of limitations do not apply to the most serious international crimes. Specifically, the UN Convention on the Non-Applicability of Statutory Limitations to War Crimes and Crimes Against Humanity (1968) and its European equivalent (Council of Europe 1974) both bar the application of statutes of limitations to these crimes.

Nevertheless, in some states that lacked dedicated laws for international crimes, attempts to prosecute atrocities under ordinary criminal provisions have been thwarted by statutes of limitations. For example, following the end of the Cold War, the Supreme Court of Bulgaria ruled that prosecuting Communist-era atrocities based on ordinary murder laws would violate statutes of limitations, which had expired (Hristov & Kashumov 2012). One cannot say for sure, but presumably if Bulgaria had legislation against atrocity crimes at the time the offenses were committed, this legal challenge would have failed. In contrast, in France, the prosecution of the Nazi war criminal Klaus Barbie was made possible by the Court of Cassation's ruling that customary international law prohibited the application of statutes of limitations to crimes against humanity. France was then able to prosecute Barbie under France's legislation for crimes against humanity (Wexler 1994).

For a prosecutor, the option to charge individuals under a dedicated atrocity law helps deter or defeat challenges invoking statutes of limitations by establishing a clear link to international law and its prohibitions against such limitations. But in the absence of such dedicated laws, prosecutors must rely on provision for ordinary crimes, and statutes of limitations may then obstruct such prosecutions.

Hypotheses

Atrocity laws are a constitutive condition that facilitates human rights prosecutions. They provide relevant legal actors with formal statutory resources that increase the possibilities for initiating and completing prosecutions. For this reason, atrocity laws should be correlated with more attempted and completed prosecutions. This expectation rests on an assumption that two types of relevant legal actors—judges and prosecutors—are influenced by the availability or existence of atrocity laws. While we do not claim to comprehensively theorize these actors’ behaviors, which would be outside the scope of this article, research does lend support to these assumptions.

An enormous body of research establishes that multiple and sometimes competing factors—such as judges’ political attitudes and the prospects for government compliance—influence judicial decision making. Among these factors is doctrine, or legal principle (Kapiszewski 2011). Given that our argument focuses on the ways that formal laws can undermine legal challenges, we rely on the assumption that legal principle is one factor that influences how judges rule. There is reason to believe that legal principle will be especially influential in contexts where atrocity prosecutions are in greater demand, like newly established democracies. Judges in these regimes face the need to signal a break with their societies’ repressive pasts and demonstrate their new commitment to the rule of law going forward (Kalb 2013). Research also suggests that public prosecutors seek to balance potentially competing priorities, such as their commitment to serve the state of which they are agents and their principled commitment to the rule of law (Brinks 2007). One common consideration for either public or private prosecutors when weighing a potential case is the prospect of success; prosecutors prefer to avoid highly complex and politically fraught cases that will be dismissed easily once challenged (Michel-Luviano 2012). Given the potential for legal roadblocks, the existence of standing atrocity laws is one factor that should influence prosecutors’ calculations.

To test our argument, we derive two hypotheses. The first hypothesis concerns the initiation of human rights prosecutions. When atrocity laws are available to a prosecutor, she will be less deterred from initiating cases in the first place, since she will anticipate a lower likelihood of legal challenges based on retroactivity, amnesties, immunities, or statutes of limitations. Thus, we expect the existence of atrocity laws to be associated with an increase in the likelihood that states initiate prosecutions in the first place, as well as an increase in the numbers of trials and the speed with which they are initiated. The second hypothesis concerns the ability for human rights prosecutions, once initiated, to reach the verdict stage. Where atrocity laws exist, a prosecutor will more often reach the verdict stage of trials without cases being dismissed based on retroactivity, amnesties, immunities, or statutes of limitations. Thus, we expect the existence of atrocity laws also to be correlated with more verdicts.

Hypothesis 1: States with domestic atrocity laws will be more likely to initiate human rights prosecutions and will do so more quickly and more frequently than states without such laws.

Hypothesis 2: States with domestic atrocity laws will reach more verdicts than states without such laws.

Research Design

Sample

We test the above hypotheses using an original dataset on atrocity laws and human rights prosecutions over a forty-year period. One way to test our hypotheses would be on a global sample of all countries, including stable democracies, transitional democracies, and authoritarian states. We provide such a test in the Supporting Information Online Appendix, but for a number of reasons, it makes more sense to test the argument on a sample of only transitional democracies, which are states that recently moved from authoritarianism to democracy. First, restricting our analysis to transitional democracies enables us to compare our results to existing studies of the determinants of human rights trials, which have mostly centered on these cases. Second and more importantly, focusing on transitional democracies allows for a more controlled test of our theoretical expectations. Legal roadblocks to prosecutions should matter most in places where courts adhere to legal principles amid difficult circumstances, which is more likely in democratic regimes attempting to establish rule of law after periods of weakness under autocratic rule. Courts that are subservient to authoritarian governments, like China, have no problem overriding amnesties and statutes of limitations if it suits political interests, while stable democracies, like Sweden, are less likely to face demands for criminal prosecution of state agents in the first place.¹⁰ Finally, because post-authoritarian states exhibit a high demand for justice, roughly 65 percent of all observed human rights prosecutions occur in countries that have transitioned to democracy since 1970 (See Figure 1). By analyzing these cases, which represent the majority of human rights prosecutions, we can capture variation in prosecutions while still maximizing inferential leverage.

A country enters our sample the year it experiences any minor or major democratic transition or gains independence from an autocratic state between 1970 and 2010.¹¹ Once a transitional democracy enters, it remains in the dataset for every year through 2010, even if it experiences autocratic reversion. This is done to avoid selecting only cases predisposed to democratic institutions, thus biasing the results in favor of our findings. In the event that the country experiences a subsequent transition following autocratic reversion, the original panel is right-censored in the last year of autocracy, and the country re-enters the dataset as a distinct panel. The resulting data set includes 125 post-transition democracy panels in 100 countries, amounting to 2,103 total country-year observations.

Dependent Variables

The dependent variables for the analyses refer to a specific type of human rights trial. We choose only those cases that are characterized as transitional domestic human rights prosecutions and verdicts. These are human rights prosecutions that aim to hold previous leaders and state agents accountable for crimes committed prior to democratization (Kim & Sikkink 2010). Transitional human rights trials (TRTs) are more difficult to initiate than common proceedings against low-level state agents because they often target elite planners of authoritarian repression who still retain some power. The dataset includes a total of 548 new TRTs recorded in 267 country-years; the number of yearly prosecutions ranges from a minimum of 0 to a maximum of 10. Verdicts are registered in the year that the court renders a decision on guilt or innocence, which, in some cases, can be many years after the prosecution begins.¹² The total number of verdicts in the dataset is 358.

Independent Variables

The main independent variable for our analyses is the existence of a dedicated atrocity law prior to democratic transition (Pre-Transition Atrocity Law). We construct this as a binary summary variable that indicates whether a country has a domestic criminal law against either genocide or crimes against humanity on the books prior to the moment of regime change. On a conceptual level, it makes more sense to use a single binary summary variable to indicate the existence of either law rather than either using an additive index or analyzing the effects of each law separately. That is because we are seeking to assess whether the existence of any nexus between international atrocity law and domestic criminal law helps facilitate prosecutions. The record of domestic human rights prosecutions suggests that litigants often find ways to “fit” whichever atrocity statutes exist to past acts of abuse, so there is little reason to assume that more atrocity laws will have an additive effect.

Our decision to focus on atrocity laws adopted prior to a democratic transition is consistent with a historical-institutionalist approach that emphasizes the long-run effects of contingent legal institutions. As explained earlier, many autocratic states adopted atrocity laws as part of technical modernizations of their criminal codes, so variation in adoption across autocratic states can be seen as exogenous to future attempts at prosecution. In effect, operationalizing the variable this way omits laws that may have been adopted following democratic transitions for the express purpose of targeting past regimes. This operationalization makes our tests more conservative by limiting the variable to those laws whose adoption is independent of the choice to prosecute.

Focusing on pre-transition atrocity laws has an additional advantage; it allows for a unique test of the roadblock mechanism when the effects of these laws are compared to atrocity laws passed after a democratic transition (Post-Transition Atrocity Law). Post-transition laws are endogenous to the general move to accountability after democratization. Therefore, including Post-Transition Atrocity Law in the models allows us to test one specific component of our roadblock mechanism—the overcoming of legal challenges based on retroactivity. If Pre-Transition Atrocity Law is correlated with prosecutions while Post-Transition Atrocity Law is not, then it will lend support to this component of our theory by suggesting that new democracies with previously existing atrocity laws have an easier time prosecuting past abuses than those that adopted with the intent of retroactive application.

Data on atrocity laws are taken from Berlin (2015), who documents the existence and timing of national criminal laws against genocide and crimes against humanity in every country in the world that has adopted them since World War II (See Section 3 of the Supporting Information Online Appendix for more on how these data were collected). A criminal law against genocide is a provision that either defines a crime that it refers to as “genocide” or one that otherwise defines a crime in a way that closely resembles the definition of genocide from Article 2 of the Genocide Convention (1948).¹³ At minimum, the definition must refer to the intent to destroy, in whole or in part, either a national, ethnic, racial, or religious group. The second condition is important, because some criminal laws include more or less the definition of genocide from the Genocide Convention, yet they omit the specific label “genocide,” and instead either do not include any label or use another label, such as “crimes against humanity.”¹⁴

In contrast to genocide, which refers to one specific type of offense, the concept of crimes against humanity applies to a variety of acts committed against a civilian population perpetrated by state or state-like actors as part of a widespread or systematic attack. A crimes against humanity law is defined as one that meets both the following conditions: (1) The law explicitly refers to “crimes against humanity,” “offenses against humanity,” or “crimes against the peace and security of mankind,” either in the text of its section heading, title, or body; and (2) The law includes at least one of the enumerated acts contained in Article 7 of the Rome Statute of the ICC (1998) committed against a civilian population. The second part of the rule is necessary, because in some instances national criminal codes contain a section titled “crimes against humanity” that only includes specific offenses against genocide or war crimes while omitting any further reference to distinct acts considered crimes against humanity under international law. These cases do not count as instances of crimes against humanity in domestic criminal law, because they merely refer to other crimes that do not embody the particular meaning of “crimes against humanity.”¹⁵

Model Choices and Controls

Hypotheses 1 and 2 speak to the likelihood that transitional countries initiate a TRT in the first place, the speed with which they do so, and the numbers of trails pursued and verdicts reached. We thus analyze the data in two ways. To model the numbers of prosecutions and verdicts, we specify negative binomial models to estimate variations in counts across transitional country-years.¹⁶ These models are superior to regular regressions, because they account for the distribution of count data, which follows a Poisson curve. To model the likelihood that and speed with which newly democratic countries initiate cases, we use a binary cross-section time-series Logit model, which analyzes the number of years between a transition to democracy and the occurrence of the first human rights prosecution. Analogous to a discrete time survival model (Box-Steffensmeier & Jones 2004), this test operates like a normal Logit, except that it controls for time dependence by including a measure of time (Time) between the onset of a democratic transition and the occurrence of the event in question, along with squared and cubed measures of Time (Carter & Signorino 2010).¹⁷

We are interested in the effects of atrocity laws on prosecutions and verdicts. However, because atrocity laws and legal prosecutions are endogenous factors, the potential for selection bias is high. That is, those countries more prone to adopting atrocity laws may also be more prone to holding human rights trials. To address this selection issue, which may undermine causal inference, we use a technique called coarsened exact matching (CEM) (Iacus et al. 2012). Because our data are drawn from the observed world, it is not possible to randomly assign atrocity laws to some countries and not others; therefore, we are not able to study “treated” groups in comparison to a control group, as is done in carefully administered experiments (Lupu 2013). However, CEM allows one to manipulate the dataset in a way that creates more balance in the covariate means between control and treatment groups.

CEM starts by binning observations into strata based on shared characteristics that may be correlated with the presence of an atrocity law. For this analysis, the likelihood of having an atrocity law is modeled as a function of four predictors shown in research to be correlated with the likelihood a country will adopt these laws. These factors include level of democracy (measured by the Polity2 scale), a high number of INGOs (see operationalization below), ratification of the ICC Rome Statute, and the number of a country's regional peers that have adopted new criminal codes in the past five years (Berlin 2015). After creating strata based on these factors, CEM assigns weights to each observation based on the number of treated observations in each stratum. By including these weights in the models, we are able to control for the fact that some countries are more likely than others to pass atrocity laws in the first place, and thus arrive at better estimates of the average effect of having such laws (Blackwell et al. 2009).

Matching on these variables significantly decreases the imbalance between control and treatment groups. But these groups are not perfectly balanced; therefore, we also include additional confounders in our main analysis to account for remaining variation. To account for rationalist explanations, we control for a number of endogenous political and structural factors that might contribute to or hinder the push for prosecutions. The first is whether the state experienced a change in individual leadership in the previous two years (Leader Change), coded from the Archigos Data Set (Goemans et al. 2009). We include this variable to capture the potential effects of a clean break from the previous regime (Huntington 1991). Second, because regimes that generally respected human rights prior to democratic transition are unlikely to see public demand for human rights trials, we include a normalized measure of pre-transition human rights protections in the country (Pre-Transition Human Rights), averaged over the last ten years of autocratic rule (Fariss 2014).¹⁸ A third set of political and structural controls measure the degree of democratic institutionalization and wealth. This includes Linzer and Staton's (2015) level of Judicial Independence (measured on a scale of 0–100); the 21-point Polity2 scale of Democratic Institutionalization (Marshall et al. 2013); and a binary measure (Low Income) that indicates whether the country's income level is listed by the World Bank as either “low income” or “lower middle income.”¹⁹

To account for sociological explanations, we first include a measure of regional diffusion, Regional Trials. Modeled after Hun Joon Kim's formulation, this variable is operationalized as the percentage of a state's cultural neighbors that have undertaken prosecutions, lagged one year (Kim 2008, 2012).²⁰ We also include two measures of pressure from domestic and international civil society: a logged count of the number of INGOs operating in the country (Hathaway 2007) taken from the Yearbook of International Organizations (Union of International Associations 1967–2015); and a binary measure of whether a UN peacekeeping mission was present (Kim 2012).²¹

Finally, we include controls that capture aspects of legal history and precedent in the country. These include a binary measure of whether a Legal Amnesty was in place to block or delay prosecutions, since as explained in the theory section, outgoing regimes often pass amnesties to prevent their own future prosecution.²² Also, following other studies (e.g., Dancy & Sikkink 2011), we include a measure of British Legal Tradition, taken from Mitchell et al. (2013), to account for the possibility that differences in the criminal process between common law systems and those of other legal families (civil law, Islamic law, and mixed systems) may influence the prospects for human rights prosecutions.

Results

Model 1

The results of the statistical analyses, detailed in Table 1, lend strong support to the argument that standing atrocity laws help facilitate human rights prosecutions. Model 1 pertains to the speed with which transitional democracies initiate their first prosecution. The results show that atrocity laws passed prior to democratic transitions are significantly correlated with an accelerated “failure rate” (Hypothesis 1), meaning that countries with such laws are more likely to initiate a prosecution following a transition and to do so more quickly than other countries. However, it is important to note that the p value is only 0.08. When treated as a two-tailed test, this is short of the 0.05 level. A one-tailed test, which is arguably justified given the directional nature of Hypothesis 1, would suggest a p value of approximately 0.04. Though this finding should be interpreted in light of the significance level, it is a fair indication that prosecutions occur more quickly in countries with pre-transition atrocity laws. The positive coefficient supports the argument that the existence of atrocity laws is associated with a lower likelihood that prosecutors will be deterred from pursuing cases early in the transitional regime. When exponentiated, the coefficient is 2.14, indicating that regimes with pre-transition atrocity laws are 114 percent more likely, or over twice as likely, to initiate prosecutions than states without such laws.

Table 1. Determinants of Transitional Human Rights Trials, 1970–2010

Note: Models 1 is a binary cross-section time series Logit test of speed to prosecutions. Models 2 and 3 are negative binomial tests of counts. Robust standard errors clustered by country in parentheses. ^***p < 0.01; ^**p < 0.05; ^*p < 0.10.

Figure 3 illustrates the magnitude of this effect using a Kaplan-Meier function of the amount of time that transitional states go without initiating at least one transitional human rights prosecution. In countries with pre-transition atrocity laws, survival rates drop off steeply around year three; only half of states with such laws make it three years without holding some kind of human rights prosecution. By year six, that figure drops to 25 percent. Overall, the presence of an atrocity law increases the baseline hazard of prosecution by a little over 100 percent.

Figure 3. Kaplan-Maier Plot of Probability of No Human Rights Prosecution.

Other explanations emphasized in existing studies receive mixed support in Model 1. INGOs is positively and significantly associated with a higher likelihood and speed of prosecution, lending support to sociological explanations that emphasize social pressure (e.g., Kim 2012; Sikkink 2011). However, Leader Change is not significant, offering little support to the rationalist explanation that justice comes quickly in transitional countries that have made a clean break from the former regime (e.g., Huntington 1991).

Models 2 and 3

Models 2 and 3 analyze the correlates of trial and verdict counts. Consistent with Hypotheses 1 and 2, atrocity laws are associated with greater numbers of prosecutions initiated (Model 2) and verdicts reached (Model 3). Newly democratic regimes with pre-transition atrocity laws hold and complete more trials than states without such laws. These findings support our arguments that atrocity laws contribute to higher counts of prosecutions and verdicts by helping to defeat legal challenges. When atrocity laws are available, prosecutors will pursue more cases overall—perhaps encouraged by successes in overcoming challenges to initial cases—and more cases will reach the verdict stage without being dismissed by judges on legal challenges.

As before, other explanatory variables receive mixed support in the count models. Regarding sociological explanations, INGOs is positive and significant in Models 2 and 3, whereas UN Peacekeeping is only significant in Model 3, which analyzes verdicts. Regional Trials is not substantively significant in any of the models. The effects of INGOs suggest a diffusion of norms, but in light of the uneven effects of UN Peacekeeping and Regional Trials, they may merely reflect domestic activism or a greater capacity for legal mobilization. Rationalist factors receive more support in Models 2 and 3, though inconsistently. As is evidenced by the negative coefficients on Pre-Transition Human Rights, greater human rights protections under the previous regime are associated with fewer prosecutions and fewer verdicts. These countries simply experience less demand for prosecutions. Leader Change reaches significance in Model 2, though it fails to do so in Model 3. Additionally, more institutionalized democracies (as measured by Polity2 scores) hold more prosecutions, as do wealthier countries (Model 2), though only the latter variable reaches significance in Models 3. Counterintuitively, countries with higher levels of judicial independence have fewer prosecutions. This finding merits further investigation, but it likely reflects a selection effect; countries with high scores on judicial independence probably experienced fewer human rights violations in the first place and face lower demand for prosecutions.

Unlike rationalist and sociological factors, Pre-Transition Atrocity Law is consistently significant across models. Figure 4 depicts the average expected effects of Pre-Transition Atrocity Law on counts of prosecutions and verdicts. The average yearly count of prosecutions for countries without existing atrocity laws is 0.182, but increases to 0.514 for transitional countries with such laws.²³ This amounts to 2.82 times as many prosecutions, or a 182 percent increase, meaning that countries with atrocity laws will pursue almost three times as many prosecutions in any year than those without. Likewise, the average count of verdicts for countries without existing atrocity laws is 0.134 but increases to 0.301 for countries with such laws, amounting to 2.25 times as many verdicts, or a 125 percent increase, for any given year.

Figure 4. Changes in Expected Counts of Prosecutions and Verdicts.

Testing the Mechanism

In all, the effects of Pre-Transition Atrocity Law in our models lend strong support to the argument that these laws help facilitate prosecutions. But the roadblock thesis posits that atrocity laws exert influence in a particular way, that is, by helping lower formal legal barriers to prosecution. What evidence is there for this specific mechanism? In the theory section, we provided some support for our underlying theory by offering illustrations from specific cases where atrocity laws (or their absence) made a difference. Given the nature of our quantitative data and the lack of availability of crucial indicators, we are limited in our ability to systematically verify all the possible pathways through which the roadblock mechanism should lower barriers to prosecution. That said, the statistical results do speak to the verifiability of two of those roadblocks: retroactivity and amnesties.

We can assess atrocity laws’ impact on retroactivity challenges by examining the effects of Post-Transition Atrocity Law. If the coefficient on this variable were significant, it would indicate that the impulse to pass atrocity laws is endogenous to the motivation to prosecute. However, Post-Transition Atrocity Law fails to reach statistical significance in any of the three models. Combined with the consistent significance of Pre-Transition Atrocity Law, these findings suggest that states that adopt atrocity laws prior to transitions can more easily prosecute former regime officials than states that adopt and apply atrocity laws retroactively. These findings lend additional support to the argument that atrocity laws facilitate prosecutions by helping to overcome retroactivity challenges. Additionally, the lack of significance for Legal Amnesty in all three models suggests that atrocity laws are associated with prosecutions and verdicts regardless of whether amnesty laws exist, again adding additional support to the argument that atrocity laws help overcome the problems that stem from the existence of amnesty laws. Unfortunately, we are not able to similarly verify the statute of limitations and immunity roadblocks given the lack of systematic data on these laws. Nevertheless, the evidence for the retroactivity and amnesty roadblocks, along with the above case illustrations, increases our confidence in the roadblock thesis.

What about evidence for an alternative thesis that would link atrocity laws to prosecutions in ways that do not involve lowering legal roadblocks? One possible alternative thesis is that atrocity laws lead to prosecutions by helping to raise legal consciousness among civil society organizations of the possibility and appropriateness of prosecutions. The codification of categories like “crimes against humanity” and “genocide” in domestic law may help victims view their situations as not merely isolated incidents of government abuse, but as part of larger systems of repression for which legal systems provide labels and attach a special form of condemnation. As McCann (2006: 25–26) writes, legal norms “can become important elements in the process of explaining how existing relationships are unjust, in defining collective group goals, and in constructing a common identity among diversely situated citizens.” It is plausible that the existence of these specialized legal categories in criminal law helps foster understandings among victims and civil society organizations that accountability is possible. We attempted to test this alternative thesis by interacting Pre-Transition Atrocity Law with INGOs. If atrocity laws work by raising legal consciousness, then an observable implication would be that atrocity laws exhibit their greatest effects in countries that have greater capacity for interested actors to make demands of their governments, that is, where there is a more robust civil society. For the sake of space, we present the full results in the Supporting Information Online Appendix. In short, the interaction does not reach statistical significance in any of the models. In other words, the association between atrocity laws and prosecution is not mediated by civil society. By discounting this alternative mechanism, the interaction models thus lend additional support to the roadblock thesis.

Conclusion

The international atrocity regime is designed to work primarily through domestic courts enforcing national legislation that codifies international crimes. That is why international human rights organizations in recent years have directed advocacy campaigns promoting the adoption of national atrocity laws (Schroeder & Tiemessen 2014). But are these moves by states to adopt atrocity laws helping the international atrocity regime to function as intended? In other words, do domestic atrocity laws increase the likelihood of criminal prosecutions? This study represents the first systematic analysis of this question, and the findings provide strong evidence that the answer is “yes.” While controlling for a range of alternative explanations and possible selection effects, survival and count models indicate that the existence of a national criminal law against either genocide or crimes against humanity roughly doubles the likelihood of initiating a prosecution, and produces more than twice as many average prosecutions and verdicts in any given transitional country-year. We have argued that these laws provide legal resources to overcome or avoid formal legal obstacles that often hinder prosecutions. At the same time, the findings offer measurable but inconsistent support for the influence of other factors—such as civil society pressure and a ruptured transition—that receive attention in existing research (e.g., Huntington 1991; Kim 2012; Sikkink 2011).

This study's main findings underscore the importance of formal legal factors in the pursuit of accountability for past human rights violations. Several studies in recent years have cast doubt on whether mere words on a page can exert any influence on state compliance with international human rights law (Chilton & Versteeg 2015; Keith et al. 2009). While the findings here do not speak to the question of whether atrocity laws deter abuses, they do suggest that formal laws make a difference for the likelihood that abuses will be remedied after the fact, which constitutes one form of compliance with international law. Our findings suggest that, contrary to pessimistic assumptions, efforts by advocacy organizations to promote the adoption of national atrocity legislation are worthwhile, particularly in authoritarian states, where future transitional regimes could take advantage of these provisions for pursuing accountability for past abuses. These findings contribute to a new body of research on how legal institutions can condition human rights outcomes in ways that are independent of rationalist or sociological factors (Collins 2010; Hilbink 2012; Lake 2014; Michel & Sikkink 2013; Ocantos 2014). Contrary to rationalist approaches, our theory brings attention to the agency of prosecutors and judges in deciding how to anticipate or rule on legal challenges. And contrary to sociological approaches, our theory can help explain variation in prosecutions among cultural and regional peers. The findings also complement sociological approaches by highlighting formal legal conditions that can tip the balance from unsuccessful to successful efforts at accountability.

While our analysis focuses on prosecutions in newly democratic states, our findings likely shed light on variation in atrocity prosecutions in other types of states, such as those that have experienced armed conflict. In other words, the findings here suggest that states in the midst of a conflict may be more likely to prosecute military personnel for war crimes and crimes against humanity if their legal systems have dedicated provisions for these offenses. For example, Lake (2014) notes that what, in part, has allowed military courts in the Democratic Republic of Congo in recent years to prosecute a surprisingly large number of cases of sexual violence was the adoption of new laws recognizing sexual violence as a crime against humanity. Future research should explore whether states emerging from armed conflict and which have relevant atrocity statutes are more likely to prosecute military and government officials for war crimes and crimes against humanity than states that lack such laws.

The findings in this article also highlight the importance of domestic legislation for the enforcement of international law. In recent decades, international legal regimes have increasingly sought to regulate domestic policy areas, such as human rights and environmental protection, so the need to institutionalize international legal norms into domestic law has become ever more important for international legal regimes to function properly (Betts & Orchard 2014; Yoo & Boyle 2015). Yet these moves receive relatively little attention. More research should scrutinize whether domestic legislation of international law actually makes enforcement more likely. This article provides initial evidence that domestic institutionalization is indeed consequential for the enforcement international law.