Internalizing the International Criminal Court

Transcription

1 Internalizing the International Criminal Court Wayne Sandholtz Department of Political Science University of California, Irvine The International Criminal Court (ICC) possesses authority that traditionally belonged exclusively to domestic courts, namely, the authority to prosecute the nationals of sovereign states. Nearly two-thirds of United Nations member states (121) have ratified the Rome Statute of the ICC, accepting the Court s jurisdiction and committing themselves to cooperate with it. States parties to the Rome Statute are expected to implement their ICC obligations into domestic law via complementarity and cooperation legislation. Complementarity refers to establishing ICC crimes (acts criminalized by the Rome Statute) as crimes under domestic criminal law and conferring on domestic courts the authority to prosecute them. Cooperation refers to putting in place domestic laws that allow states to comply with ICC requests for assistance in conducting its investigations and prosecutions. The question motivating this study is: Why are some states more willing than others to internalize their ICC commitment by enacting complementarity and cooperation legislation? The answers may help us to answer longstanding questions regarding state commitment to the international human rights regime. One puzzle is why ratification of human rights treaties is not broadly associated with respect for human rights in practice. Research has found that democracies do tend to comply with human rights treaties, but that some non-democratic states are insincere ratifiers or false positives (Simmons 2009): they join human rights treaties without intending to comply with them. The Rome Statute of the ICC raises many of the same questions that human rights treaties do about the nature of commitment to the international human rights regime. Some states that ratified the Statute are likely to be more committed to its successful functioning than others. One way to sort out the genuine from the insincere ratifiers is to assess their behavior with respect to implementing treaty purposes and obligations. The more committed states are to the ICC, the quicker they should be to enact complementarity and cooperation legislation, other things being equal. This study seeks to discern the characteristics of the rapid implementers. This study tests the norm congruence thesis: states undertake and implement international commitments that are congruent with their domestic norms and values. More democratic states are more likely to pass the implementing laws. A utilitarian approach would suggest that governments might be reluctant to push for ICC implementing legislation where courts are more independent (able to rule against the government) and more capable of making law. Courts with greater authority to prosecute international crimes and to construct common law norms may be seen as a potential cost or constraint on governments. I test these propositions controlling for characteristics of states that might make legislation in general more difficult and slower. The next section outlines the national implementation measures implied by the Rome Statute. Subsequent sections spell out the theoretical arguments, describe the data, and report on the analyses. The pool of countries in the analyses consists of

2 2 approximately 120 countries that have signed, ratified, or acceded to the Rome Statute, with explanatory variables that could affect the probability, in any given period, that countries will enact ICC implementing legislation. The results broadly confirm the norm congruence thesis. A conclusion pulls together the key findings and implications. What the Rome Statute implies for national legislation The obligations that fall on states that ratify the Rome Statute imply two kinds of domestic implementing legislation, cooperation legislation and complementarity legislation. In principle, the obligations under the Rome Statute become part of domestic law upon ratification. In practice, however, virtually every country that ratifies will have to make some changes in its own laws and institutions. As Broomhall puts it, In most cases... implementation is not so simple, and a State must make specific changes to bring national law into line with its international obligations (Broomhall 1999). With respect to cooperation, states parties are under a general obligation to cooperate : States Parties shall, in accordance with the provisions of this Statute, cooperate fully with the Court in its investigation and prosecution of crimes within the jurisdiction of the Court (Art. 86). Subsequent articles specify the areas in which states are obligated to cooperate and many of the rules governing such cooperation: Surrender of persons to the Court, handling requests from other states for the same person and the same crime, making arrests, providing various forms of judicial assistance (service of documents, taking and preserving evidence, protection of victims and witnesses), and allocating the costs of judicial assistance. Article 88 requires states to make the appropriate provisions in domestic law for all of these types of cooperation: States Parties shall ensure that there are procedures available under their national law for all of the forms of cooperation which are specified under this Part. Though it is possible that the existing laws of some countries might be adequate, without modification, to provide for the kinds of cooperation required by the Rome Statute, it seems far more likely that virtually all states would need to enact new laws or amend existing ones. Some legislation will be necessary simply because many of the Rome Statute obligations are specific to the ICC and would likely not be covered by existing rules. For example, the Rome Statute defines offenses against the administration of justice of the ICC (Art. 70(1)). Article 70(4) obligates states parties to criminalize these offenses in their own laws, to authorize the prosecution of such offenses, and to submit cases of alleged offenses to the appropriate national authorities for prosecution when requested by the ICC. Another example of domestic rules that must be in place and must meet standards specific to the ICC concerns state procedures for arresting, holding, and surrendering persons who are the subjects of ICC warrants. Domestic laws governing these ICC procedures must comply with ICC rules (Art. 55). It seems extremely unlikely that any state could comply with these obligations and others without passing some, albeit modest, cooperation legislation. The second main category of implementing legislation is complementarity legislation. Complementarity refers to the ICC s status as a court of last resort, one that cannot accept jurisdiction over a case if a State with jurisdiction over that case is investigating or prosecuting it (Art. 17). Complementarity is, in part, a recognition that national courts must shoulder the bulk of the burden of prosecuting genocide, war

3 3 crimes, and crimes against humanity. The ICC can only take (and will only have the resources to pursue) a limited number of serious cases that the relevant states are unwilling or unable to investigate or prosecute. Though the Rome Statute does not require states to prosecute ICC crimes, states must be prepared to do so in order for the Court to exercise complementary jurisdiction. The Rome Statute thus implies that states must have in place laws that grant domestic jurisdiction over ICC crimes that occur in their territory and over their nationals who commit ICC crimes abroad. Though it is possible that a country s national laws could, without modification, provide for the prosecution of ICC crimes, it seems unlikely that states could play their role in the ICC s complementary jurisdiction without at least some new legislation. For instance, national laws would need to mesh with the ICC s rules for handling overlapping national and ICC jurisdictions. When a state has asked the ICC Prosecutor to defer an investigation, so that the state can pursue its own investigation and (possibly) prosecution, that state must provide timely information about its investigations and any prosecutions when requested by the Prosecutor (Art. 18(5)). States are likely to need specific laws mandating the reporting required by the ICC. More importantly, virtually all states would require some additions or amendments to their criminal statutes in order for their laws to accord with the ICC s definitions of international crimes. The Rome Statute introduced innovations in international criminal law, specifying offenses that are not defined by customary international law or other treaties. For instance, the Rome Statute developed international criminal law by specifying several sexual offenses and enforced disappearances as crimes against humanity (Art. 7(1)), when the other elements of crimes against humanity are met. As a result, even states whose domestic laws already criminalize genocide, war crimes, and crimes against humanity might need to modify their existing criminal statutes to reflect the ICC s innovations and to enable national authorities to fulfill their role in the ICC s system of complementary jurisdiction. As Amnesty International explains, If the Court is to be an effective complement to states in the international system of justice for such crimes and not overwhelmed by cases, states need to fulfil their responsibilities. They must enact, and then enforce, national legislation which provides that these crimes under international law are also crimes under national law wherever they have been committed, no matter who has committed them or who is the victim (Amnesty International 2010). In short, ICC states undertake a broad range of sometimes highly specific obligations with respect to both cooperation and complementarity (Broomhall 1999). This study proceeds on the assumption that every state that ratifies the Rome Statute will need to enact some cooperation and complementarity legislation, however modest. Implementing the Rome Statute: the global campaign The rule that pacta sunt servanda the customary law foundation of treaty law requires that states adhere to their treaty commitments. In order to fulfill their obligations under the Rome Statute, states must take additional steps to implement it in their domestic laws and institutions. Internalization refers to the processes by which states incorporate international norms and legal obligations into their domestic systems (Koh 1996; Koh 2005). In formally monist states, treaties are directly applicable in domestic courts, as international law and national law together constitute a single

4 4 integrated legal order. As of 2010, the constitutions of 67 states made treaties equal or superior to statute in the domestic legal system (Sandholtz 2012). However, the formal distinction between monist and dualist systems does not always align with judicial practice. Much depends on the doctrines that develop in national courts (Sloss 2009; Sloss forthcoming). Even in states where ratified treaties are directly applicable in domestic courts, full implementation of the Rome Statute may well require additional domestic law-making. For example, the Rome Statute defines offenses against the administration of Justice of the ICC (Art. 70(1)). Article 70(4) obligates states parties to criminalize these offenses in their own laws, to authorize the prosecution of such offenses, and to submit cases of alleged offenses to the appropriate national authorities for prosecution when requested by the ICC. Some of these obligations are likely to necessitate new domestic legal rules, to the criminal statutes, for instance. The working assumption in this study is that full implementation of Rome Statute obligations will require virtually all states, even those in which treaties have direct domestic legal effect, to enact some changes in domestic law. States can internalize international law through a variety of means, including legislative, executive, and judicial action (Koh 1996). It seems unlikely that executive action by itself could accomplish the changes in the domestic legal system required for full implementation of the Rome Statute in any country where at least some checks on executive powers exist. Courts could directly implement some Rome Statute obligations, for instance by allowing domestic prosecution of ICC crimes (genocide, war crimes, and crimes against humanity as defined in the Rome Statute). But even then, legislation will almost certainly be required to establish the institutional mechanism for cooperating with the ICC. This analysis therefore focuses on legislation as a key, and probably indispensable, means of implementing Rome Statute commitments. International efforts to promote implementation of the Rome Statute have, in any case, focused almost exclusively on domestic legislation. Transnational nongovernmental organizations (NGOs) have taken the lead. The Coalition for the International Criminal Court (CICC), the transnational umbrella organization that includes 2,500 member organizations in 150 countries, has made national passage of implementing legislation one of its main foci: For the principle of complementarity to become truly effective, following ratification, States must also implement all of the crimes under the Rome Statute into domestic legislation. As the Court initiates investigations, the existence of solid cooperation and implementing legislation takes on new urgency. The Coalition actively monitors and promotes the development of implementation legislation globally. To assist our membership, and also those governments working on their legislation, the Coalition is committed to providing comprehensive information on the state of legislative drafting and implementation campaigns throughout the world (Coalition for the International Criminal Court 2012). The CICC has also closely monitored national enactment of implementing legislation; this study makes use of the CICC data. Amnesty International, which has supported the creation and development of the ICC since the early 1990s, declares that it actively campaigns for... [a]ll governments to enact effective implementing legislation ensuring that they can prosecute the crimes

5 5 before national courts and cooperate fully with the Court (Amnesty International 2012). Amnesty International also published a Checklist for Effective Implementation of the ICC Statute (Amnesty International 2012). Human Rights Watch has similarly made ratification and implementation of the Rome Statute one of its campaigns: As part of our ratification campaign, we regularly travel to different parts of the world to meet with and offer technical assistance to parliamentarians, government members, senior officials and local groups to raise awareness of the ICC and to assist with the process of ratification and domestic implementation of the Rome Statute (Human Rights Watch 2012). Human Rights Watch has tracked implementing legislation and published a Handbook for Implementing the Rome Statute (Human Rights Watch 2001). The International Centre for Criminal Law Reform and Criminal Justice Policy also published a Manual for the Ratification and Implementation of the Rome Statute (International Centre for Criminal Law Reform and Criminal Justice Policy 2008). So far, the rate of enacting implementation legislation has been slow but steady, as shown in Figure 1. As of 2009, 109 countries had ratified the Rome Statute. Of those, 58 had passed complementarity legislation and 42 had enacted cooperation legislation. The next section identifies factors that might affect the probability that states will enact implementing legislation. Theories, hypotheses, and data Human rights treaty commitments are puzzling because the incentives for participating in such treaties must differ sharply from the motivations states have for joining, for example, trade agreements. Whereas the gains of free trade can only come through

6 6 interaction with other states, gaining human rights does not require international exchange or reciprocity states can protect human rights on their own. And whereas states benefit materially from regimes that nudge other states toward compliance with free trade rules, states do not benefit so clearly and directly from rules and institutions that encourage other states to respect human rights at home. The puzzle of state commitment to human rights treaties is linked to the problem of state compliance. One argument has been that, because states will ratify treaties with which they expect to comply, the decision to commit is also the decision to comply (Downs, Rocke et al. 1996; von Stein 2005). Responding to that challenge, research began to test the notion that states only ratify treaties that they expect to comply with anyway. Scholarship on the effects of treaties on human rights performance has consistently shown that some states that ratify continue to violate human rights at home. Some analyses indicated that ratification of human rights treaties was not associated with better human rights performance (and was sometimes associated with worse performance) (Hathaway 2002; Hafner-Burton and Tsutsui 2005; Hafner-Burton and Tsutsui 2007). Other analyses have found that human rights treaties are associated with better respect for human rights in practice among at least some categories of states or under certain conditions (Landman 2005; Neumayer 2005; Simmons 2009). Analysis of the effects of human rights treaties is, in any case, necessarily linked to analysis of human rights treaty commitment (Hathaway 2003; Landman 2005; Goodliffe and Hawkins 2006; Hathaway 2007; Simmons 2009; Sandholtz 2012). One fairly consistent finding has been that democracies are more likely to ratify than non-democracies, a relationship that this analysis will test with regard to implementing the Rome Statute. Ratification of the Rome Statute raises questions similar to those surrounding ratification of human rights treaties. A broad and active international campaign pushed for ratification of the Rome Statute. The umbrella NGO, the Coalition for the International Criminal Court, vigorously pushed for ratification of the treaty, lobbying specific governments. Other prominent NGOs, including Amnesty International and Human Rights Watch, also mobilized in the effort to secure ratifications. The Organization of American States, in both its General Assembly and its meetings of heads of state, urged its members to ratify. The European Union went even farther, making ICC ratification a talking point in its discussions and negotiations with states and other regional organizations, and issuing 131 diplomatic demarches to governments specifically urging ratification and implementation of the Rome Statute (Goodliffe, Hawkins et al. 2012). Some states, like Canada, actively promoted ratification through meetings of experts. Given this broad international campaign, it is possible that some states ratified the agreement insincerely, for reputational or diplomatic credit. The norm-congruence thesis Enacting the implementing legislation that membership in the ICC, however, requires a greater level of commitment. Passing complementarity and cooperation legislation is not cheap talk, as it requires changes in domestic laws and institutions. The changes in themselves are not costless to make legislating always entails the investment of time, energy, and political capital. Once in place, the implementing laws and institutions could produce additional consequences that governments would regard as costly, for example, in domestic prosecutions of government officials, or in the arrest and surrender to the ICC of suspects found in the national territory. In other words,

7 domestic implementing legislation is likely to be more costly than ratifying the Rome Statute, and therefore less likely to be merely cheap talk. Analyzing the factors that affect the probability of enacting implementing legislation is, therefore, one way of sorting sincere from insincere ratifiers. Kelley (2007) has distinguished levels of commitment to the ICC by analyzing post-ratification behavior. Kelley assesses the likelihood that states enter into bilateral non-surrender agreements with the United States. The agreements undermine the purposes and functioning of the ICC and Kelley hypothesized that countries that were more committed to the goals and values embodied in the ICC would be less willing to sign non-surrender agreements. She finds that countries with the greatest affinity for the ICC ( congruence in the terms of this study), as captured by democracy, human rights, and membership in the group of like-minded states, are less likely to sign bilateral non-surrender agreements with the United States, controlling for other factors that might make them vulnerable to U.S. pressure. This study applies a similar argument to the analysis of ICC implementing legislation. The proposition to be tested here is the norm congruence thesis, which holds that states will accept and implement international commitments that fit with their domestic values and norms (Sandholtz 2012). This argument connects with the broader family of approaches that emphasize a logic of appropriateness or fit in a social (which is to say, normative) context (March and Olsen 1989). The norm congruence thesis implies that democracies, which tend to be more committed than non-democracies to respecting human rights domestically, will be more likely than other states to implement the Rome Statute through domestic legislation. This argument is in line with research that finds that the more democratic a state, and the more it upholds human rights, the more likely it is to ratify human rights treaties (Landman 2005; Neumayer 2005; Simmons 2009; Sandholtz 2012). With regard to the ICC, Kelley finds that states with an affinity for the norms and purposes of the ICC were more likely to ratify the Rome Statute. Democracy and respect for human rights both raised the probability of ratification (Kelley 2007). Goodliffe et al. find that the effect of democracy on the probability of signing or ratifying the Rome Statute is weaker once network dependence is included in the models. With network dependence included, democracy is significant in three of six models of ICC ratification (Goodliffe, Hawkins et al. 2012). Democracy might have a greater effect on treaty implementation than it has on ratification, as democracies should be more aligned with the norms embodied in the ICC than non-democracies. States whose domestic norms and values are congruent with those of the ICC will be more willing to bear the potential downstream costs of implementing legislation. Legislation implementing the Rome Statute opens the possibility that government officials could be prosecuted domestically under the implementing (complementarity) statutes, or that a state might cooperate with the ICC in ways that the government did not favor. Autocratic governments would be more uncomfortable with these kinds of outcomes than would democratic ones. Democracies are more likely to be supportive (in principle if not in specific instances) of domestic prosecutions for ICC crimes and abhor immunity for such crimes, given their domestic norms and values. In other words, democracies are more likely than non-democracies to accept the possibility of domestic prosecution for ICC crimes and cooperation with the ICC. 7

8 The proposition that democracies are more normatively inclined to accept ICC implementing legislation does not reduce to an argument that governments in democracies simply face a lower likelihood of such prosecutions ever taking place. The probability of a domestic prosecution under ICC complementarity laws is a function of two component probabilities: the probability of an ICC crime occurring and the probability of domestic courts proceeding to prosecute such a crime. In a democracy, the probability of an ICC crime occurring is presumably lower (human rights are broadly respected) and the probability of domestic courts prosecuting such a crime is high, given that democracy tends to be associated with independent courts and the rule of law. In an autocracy, the probability of an ICC crime occurring is presumably higher, but the probability is low that such a crime would be prosecuted domestically, given that governments are more able to manipulate or block prosecutors and courts. For both democracies and autocracies, the product of the two component probabilities is in some middle range (high * low is roughly equal to low * high); they do not differ markedly on the probability of a domestic prosecution. The thesis is, therefore, that democracies (as measured by the Polity2 score) will be more open to ICC implementing legislation because of norm congruence. Some kinds of democracies may be more rights-supporting than others. Arend Lijphart has suggested that consensus democracies generally will be associated with kinder, gentler, and more generous policies than will majoritarian democracies (Lijphart 1999). Institutional features of consensus democracy produce sharing of power and compromise, which lead in turn to policy outcomes. Consensus democracies imprison less, abolish the death penalty more, provide more foreign aid, and spend more on social welfare. In the end, though, consensus democracy is not just a set of institutional and policy characteristics. It is a culture, a constellation of values and orientations. Consensus democracies tend to have a strong community orientation and social consciousness (Lijphart 1999). As a result, they may also be more in tune with the principles and norms of the global human rights regime. Research has found that consensus democracies (as indicated by proportional representation) have been more receptive to the international human rights movement to ban the death penalty (McGann and Sandholtz 2012). This study explores the proposition that consensus democracies will also be more supportive of the norms and purposes of the ICC. Because proportional representation implies the kind of multi-party, negotiated government evoked by Lijphart s consensus democracy, the analysis includes a binary indicator for proportional representation (PR). An additional measure of norm congruence with the ICC is membership in the group of like-minded states. The like-minded states worked closely with the Coalition for an International Criminal Court (CICC), starting in the preparatory phase ( ) and continuing through the Rome Conference and after. The Like-Minded Group consistently pushed for completion of the treaty and for broader conceptions of the Court s mission and authority (Struett 2008). Kelley included membership in the Like- Minded Group as one of her measures of affinity for the ICC, and found that membership in the Like-Minded Group was associated with a higher probability of ratifying the Rome Statute (Kelley 2007). The like-minded state variable serves an additional important function. Countries that are more enthusiastic about the ICC to begin with are probably also more likely to pass implementation legislation more quickly. 8

9 9 The like-minded state variable helps to control for the level of commitment to the ICC itself, as it does in Kelley s analysis (Kelley 2007). To summarize, implementation is almost certain to be more costly than ratification, thus weeding out insincere ratifiers. The norm congruence thesis leads to the following hypotheses. Hypothesis 1: Democracy will be associated with a higher probability of enacting complementarity and cooperation legislation. Hypothesis 2: Democracies with proportional representation will be more likely to enact complementarity and cooperation legislation. Hypothesis 3: Membership in the Like-Minded Group will be associated with a higher probability of enacting complementarity and cooperation legislation. World Society World society approaches suggest that world cultural norms shape the institutional forms and purposes of states. Such norms define what it means to be a modern state, and states therefore enact world cultural scripts (Meyer, Boli et al. 1987; Meyer, Boli et al. 1997). Human rights have become a core part of world culture, and human rights laws and institutions have accordingly diffused globally (Boli 1987; Elliott 2007; Beck, Drori et al. 2009; Koo and Ramirez 2009). World cultural influences may therefore also push states toward passage of implementing legislation. Three variables are included as measures of world culture. The first (GLOBAL HUMAN RIGHTS REGIME) captures the extent of the global human rights regime. It represents the cumulative total ratifications of 18 human rights treaties. Over time, as the number of treaties expands and the number of states parties to those treaties grows, the global human rights regime is expanding in both scope (rights covered) and breadth (countries participating). The more developed is the global human rights regime, the more likely it is that countries will conform in their own rules and institutions with its norms, in this case by enacting ICC legislation. The second measure of world society is a count of the number of memberships in international non-governmental organizations (INGOs). Finally, new regimes are especially likely to embody world society norms and institutional forms. World society provides the model for modern, legitimate statehood. New regimes will be likely to enact the world society script for statehood, in this case by implementing the ICC. The NEW REGIMES variable marks states that have undergone a regime change in the previous ten years. The domestic costs thesis Though I have argued that norm congruence will be the primary driver of states willingness to enact implantation legislation, governments can also be sensitive to the potential costs or constraints that passing such legislation could entail. States with common law legal traditions may be more reluctant to pass implementing legislation, given the more complex interactions between statute, precedent, and judicial decisionmaking. Courts in common law countries make law through precedent, which governments do not control. In principle (though practices vary widely), courts in civil

10 10 law countries apply the laws enacted by the government. The analyses include a binary variable that marks countries with a common law legal tradition. Judicial independence could imply that courts might be more willing and able to hear cases in which government officials are prosecuted for ICC crimes, and more willing to convict. Following a similar logic, Powell and Staton found some evidence that effective courts would make governments more reluctant to ratify the Convention Against Torture (Powell and Staton 2009). Governments in countries with independent domestic courts may be slower to enact complementarity and cooperation legislation. A measure of judicial independence tests this argument. Domestic politics and institutions Legislating takes time and political effort, and the passage of ICC implementation legislation might take longer in some countries for purely institutional reasons, regardless of how committed the government is to the ICC. The analysis therefore includes variables that capture the general difficulty of enacting legislation, including a measure of the number of obstacles to passing laws in a given political system (CHECKS). Presidential systems can take longer to legislate (compared to parliamentary systems) because the executive is not automatically of the same party that controls the legislature. Federal states may take longer to pass new laws because sub-national units possess the institutional means to slow or delay proposed laws. The analyses include indicators for both presidential and federal systems. Conflict Countries whose nationals could be vulnerable to prosecutions for ICC crimes might be more reluctant to enact ICC implementation legislation. Wars, both international and civil, provide the context for many if not most ICC crimes. Countries that are involved in conflicts have more nationals who could be involved in ICC crimes and who therefore could be subject to investigation or prosecution for ICC crimes, whether in domestic courts or in the ICC itself. The analyses include indicators of involvement in both kinds of war in the immediate past. International diffusion Analyses of treaty commitment routinely include variables measuring the expansion of ratifications of the treaty in question. Following Simmons, the argument is that countries tend to follow their neighbors, whether for reasons of socialization or of competition (Simmons 2000; Simmons 2000; Simmons 2009). The hypothesis is that the larger the share of a country s neighbors have enacted ICC legislation, the more likely that country is to do so as well. A similar logic applies to the worldwide rate of enacting ICC legislation. For each country-year, two variables (REGIONAL RATE, WORLD RATE) capture these rates, as a percentage, excluding the country itself. Economic variables The analyses also evaluate a number of economic variables: economic size (total GDP, logged); wealth (GDP per capita, logged); trade intensity (trade as a share of GDP); dependence on foreign economic assistance (Development Assistance Committee (DAC) assistance as a share of GDP); and dependence on foreign direct investment (FDI as a share of GDP).

11 11 Analysis and results A look at univariate survival functions serves as an initial probe of the main hypotheses. Survival functions represent the probability that a subject (in this case, a country) will have experienced the event being studied (enactment of complementarity or cooperation legislation) by a given period. Figure 2 offers the results. The graphs clearly support the main hypotheses regarding norm congruence and implementation legislation. When the three main variables of interest (Democracy, Proportional Representation, and Like-Minded State), are present, countries are substantially more likely to have enacted the domestic ICC legislation (that is, less likely to have survived without passing the laws), whether that is complementarity or cooperation legislation. For states with the variables present, the probability of having enacted the legislation at the end of the analysis time is close to or greater than 50 percent. A far smaller share of states without these characteristics have enacted the legislation by the end of the analysis time. Multivariate models will be helpful in seeing whether the relationships depicted in Figure 1 continue when other variables are added. Cox proportional hazards models are well suited for testing the hypotheses proposed above. Cox models permit the analysis of variables that affect the probability that a state will, in any given time period, enact complementarity and cooperation legislation, given that it has not already done so. The pool of countries in the analysis includes those that have ratified the Rome Statute and are therefore expected to enact implementation legislation. Countries that have not ratified the Statute would have no reason to pass ICC implementing legislation. Countries enter the risk pool when they ratify or accede to the Rome Statute; they exit the risk pool when they pass the relevant legislation. The analysis period ends in 2009; countries that enacted implementation legislation after 2009 are right censored (treated as not having passed the legislation during the study period). Data are annual.

12 12 Figure 2: Kaplan-Meier Survival Estimates, ICC Implementing Legislation Probability of not enacting Democracy: complementarity legislation Analysis time Probability of not enacting Proportional representation: complementarity legislation Analysis time Probability of not enacting Likeminded states: complementarity legislation Analysis time Non-democracy Democracy No PR PR Not likeminded Likeminded state Probability of not enacting Democracy: cooperation legislation analysis time Probability of not enacting Proportional representation: cooperation legislation Analysis time Probability of not enacting Likeminded states: cooperation legislation analysis time Non-democracy Democracy No PR PR Not likeminded Likeminded state

13 Table 1 reports the analyses for complementarity legislation. The entries in the table are hazard ratios, which have a multiplicative effect on the probability of enacting the legislation. The analyses provide strong support for the norm congruence thesis. Two of the norm congruence variables Democracy and Proportional Representation are significant in all six models; both increase the probability of passing complementarity legislation. Democracy raises that probability by about 25 percent and Proportional Representation by almost 300 percent. The third norm congruence variable, Like-Minded State, is significant only when the World Society variables are included. That is, for a given level of the Global Human Rights Regime, and taking into account whether a regime is new, being a like-minded state increases the probability of legislating by more than 250 percent. The World Society variables are also significant. New regimes are about two and a half times as likely to pass complementarity legislation as older regimes. Because the unit for the Global Human Rights Regime variable is one treaty ratification by one country, the effect looks small. But one hundred additional human rights treaty ratifications increases the probability of enacting ICC complementarity legislation by about 30 percent. In contrast, the domestic costs thesis finds no support. Judicial Independence is insignificant, and Common Law has the sign opposite the hypothesized one: common law tradition increases the likelihood of legislating. None of the domestic institutional variables designed to capture the difficulty of the legislative process itself is significant. Wars do not have a significant effect, nor do regional and international diffusion. None of the economic variables is significant. Countries that are more dependent on foreign aid and investment are not more likely to pass the legislation. 13

14 Table 1: Complementarity legislation Norm congruence Democracy 1.251*** 1.309*** 1.300*** 1.220*** 1.219*** 1.197** (0.006) (0.001) (0.001) (0.003) (0.004) (0.021) Proportional representation 2.938** 2.617* 2.463* 2.869** 2.801** 2.949** (0.038) (0.060) (0.067) (0.031) (0.040) (0.032) Like-minded state ** 2.693** (0.239) (0.038) (0.041) (0.170) (0.164) (0.102) Potential domestic costs Common law 4.016** 3.855** 4.052** 3.319** 3.290** 3.664** (0.020) (0.017) (0.012) (0.033) (0.032) (0.021) Judicial independence (0.428) Domestic politics and institutions Presidential system (0.664) Left executive (0.209) Federal system (0.991) Checks (0.282) World society New regime 2.410* 2.571** (0.061) (0.035) Global human rights regime 1.003*** 1.003*** (0.005) (0.005) INGOs (ln) (0.648) Conflict International war (previous 3 years) (0.823) Civil war (previous 3 years) (0.238) International diffusion Regional rate (0.924) World rate (0.936) Economics GDP/capita (ln) (0.402) GDP (ln) (0.380) Trade/GDP (0.174) DAC assistance/gdp (0.742) FDI/GDP (0.429) Observations Number of countries Number of events Pseudo-R Log-likelihood Chi Note: Cox proportional hazard models, complementary log-log function. Regional dummies included but not reported. Hazard ratios with probabilities in parentheses. *** p<0.01, ** p<0.05, * p<0.1 14

15 For cooperation legislation, the results are quite similar, as Table 2 reports. Again, the norm congruence thesis finds support: Democracy increases the probability of passing the legislation. Whereas being a like-minded state was significant in two of the models of complementarity legislation, it is significant across the board with cooperation legislation. Proportional representation is not significant. The difference, as compared with complementarity legislation, may be that cooperation legislation deals with cooperating with the ICC, that is, with external relationships. Complementarity legislation is about bringing ICC crimes into domestic criminal law. If consensus democracies are more supportive of human rights, they might be more interested in criminalizing ICC crimes in their own laws than with external cooperation. One world society variable, New Regime, is significant for cooperation legislation but the Global Human Rights Regime is not (though it was for complementarity legislation). In neither set of models were INGOs significant. Just as in the complementarity models, the regional and world diffusion variables are not significant. But in contrast to the first set of models, GDP/capita is consistently significant and increases the probability of enacting cooperation legislation. The other economic variables are, as before, insignificant. 15

16 16 Table 2: Cooperation legislation Norm congruence Democracy 1.311** 1.210* 1.188* 1.219** 1.204** 1.224** (0.028) (0.069) (0.052) (0.031) (0.038) (0.030) Proportional representation (0.512) Like-minded state 3.820** 3.533** 3.031* 3.475** 3.562** 3.730** (0.020) (0.034) (0.051) (0.029) (0.025) (0.019) Potential domestic costs Common law (0.620) Judicial independence (0.448) Domestic politics & institutions Presidential system (0.354) Left executive (0.315) Federal system (0.633) Checks (0.375) World society New regime 3.702** 2.914* 3.228* (0.036) (0.087) (0.057) Global human rights regime (0.546) INGOs (ln) (0.175) Conflict International war (previous 3 years) (0.890) Civil war (previous 3 years) (0.313) International diffusion Regional rate (0.243) World rate (0.582) Economics GDP/capita (ln) 2.653** 2.095** 3.271*** 2.809** 3.188*** (0.038) (0.030) (0.004) (0.031) (0.005) GDP (ln) (0.896) Trade/GDP (0.535) DAC assistance/gdp (0.929) FDI/GDP (0.863) Observations Number of countries Number of events Pseudo-R Log-likelihood Chi Note: Cox proportional hazard models, complementary log-log function. Regional dummies included but not reported. Hazard ratios with probabilities in parentheses. *** p<0.01, ** p<0.05, * p<0.1

17 17 Additional hypotheses: categories of democratic states The results so far powerfully support the hypothesis that democracies are more likely to implement their ICC commitments with complementarity and cooperation legislation. This section tests more fine-grained propositions about regime types. Moravcsik predicts that new democracies will participate in human rights regimes more readily than either autocracies or mature democracies. New democracies are motivated by a desire to lock in democratic institutions and human rights norms that may still be fragile (Moravcsik 2000). Simmons has suggested that regimes in the middle of the regime spectrum are more likely to ratify human rights treaties than either fully autocratic regimes or strong democracies (Simmons 2009). Hathaway finds that democracies without human rights abuses are likely to ratify human rights treaties, but that democracies in which human rights violations do occur are less likely to do so. The logic is that democracies presumably have functioning legal systems and relatively independent courts. The courts could therefore hold governments accountable for the abuses. In order to avoid the potential for courts to enforce human rights treaties against them, governments are reluctant to commit to the treaties (Hathaway 2007). A similar logic could apply with respect to ICC implementation. Finally, Simmons and Danner evaluate combinations of democracy and recent experience of civil war in their analysis of ratification of the Rome Statute. Recent civil wars have different implications for democracies as opposed to non-democracies. Democracies tend to have effective mechanisms (well-functioning legal institutions) for holding government officials accountable for ICC crimes; they therefore have less to gain from delegating prosecution of such crimes to the ICC. Non-democracies lack effective domestic legal accountability mechanisms and therefore see more to gain from the ICC. The data analysis supports these propositions (Simmons and Danner 2010). A similar logic could apply to enactment of ICC implementing legislation. Tables 3 and 4 display results from analyses testing the various arguments concerning democracies. The lock-in thesis does not find much support. I employ two different ways of marking new democracies. The first, reported in column 1, includes countries that have been democratic for 25 years or fewer (about one generation). Mature democracies have been democratic for more than 25 years. On this measure, for complementarity legislation, mature democracies are more likely to ratify than new democracies, though the reverse holds for cooperation legislation. However, the complementarity legislation would be more vital for new democracies hoping to lock-in respect for human rights, as it embeds ICC norms in domestic law. The second measure contrasts countries that have been democratic continuously since World War II (mature democracies) with countries that were not always democratic during the analysis period ( ) but were democratic as of These are countries that have gone through some kind of democratic transition. On this measure, mature democracies are more likely to ratify than new democracies, for both types of implementation legislation. The balance of evidence, then, contradicts the lock-in hypothesis.

18 Table 3: Complementarity legislation and democracies Mature democracy 4.017** (0.043) New democracy 2.921* (0.092) Democracy post *** (0.004) Democratic transition 4.565** (0.017) Democracy *** 1.389*** (0.135) (0.000) (0.000) Democracy (0.648) Anocracy (0.191) Strong democracy with civil liberties abuses 0.178** (0.024) Strong democracy with physical integrity rights abuses 0.311*** (0.008) Democracies with civil war in previous 3 years (0.204) Democracies with no civil war in previous 3 years 3.441** (0.040) Democracies with international war in previous 3 years 3.347* (0.079) Democracies with no international war in previous 3 years 3.092* (0.050) Observations Number of countries Number of events Pseudo-R Log-likelihood Chi Note: Cox proportional hazard models, complementary log-log function. Probabilities in parentheses. *** p<0.01, ** p<0.05, * p<0.1. All models include regional controls and the other explanatory variables included in Table 1, column 3. The only exception is that the model in column 1 in this table does not include New Regime, which would be collinear with New Democracy. 18

19 19 Table 4: Cooperation legislation and democracies Mature democracy (0.226) New democracy 4.739* (0.088) Democracy post ** (0.010) Democratic transition ** (0.026) Democracy ** 1.302** (0.167) (0.030) (0.022) Democracy (0.423) Anocracy (0.126) Strong democracy with civil liberties abuses (0.138) Strong democracy with physical integrity rights abuses 0.381* (0.052) Democracies with civil war in previous 3 years (0.730) Democracies with no civil war in previous 3 years 5.125* (0.075) Democracies with international war in previous 3 years (0.136) Democracies with no international war in previous 3 years (0.110) Observations Number of countries Number of events Pseudo-R Log-likelihood Chi Note: Cox proportional hazard models, complementary log-log function. Probabilities in parentheses. *** p<0.01, ** p<0.05, * p<0.1. All models include the explanatory variables included in Table 2, column 6; those coefficients are not reported here. I use two measures to assess Simmons s hypothesis regarding middle regimes. The first is the square of the democracy score (Polity 2). If middle regimes are more likely than either strong democracies or full-blown autocracies to accept human rights commitments (a version of the lock-in thesis), the hazard ratio for the squared term should be less than one (higher values should be associated with lower probabilities). But the reverse is true. The second measure (Anocracy) is a binary indicator marking regimes in the middle of the Polity 2 range, from -5 to 5. It is not significant. To test the argument that democracies with human rights abuses will be less likely to accept international human rights commitments, I try two different measures of human rights. Both are combined with an indicator of democracy that identifies strong democracies, that is, those in which legal institutions are most likely to be able to