Can the International Criminal Court prevent sexual violence in armed conflict?

Abstract

Human rights activists, international organizations, and certain governments have championed prosecutions as a strategy to prevent conflict-related sexual violence, and the International Criminal Court (ICC) has focused extensively on prosecuting sexual violence crimes during its first two decades of operations. However, even as a growing body of empirical evidence suggests that the ICC contributes to improved human rights practices and the prevention of other atrocity crimes, such as violence against civilians, claims concerning the preventive effects of prosecutions on sexual violence remain largely untested. The purpose of this article is to test these claims. To this end, we analyzed the effects of ICC jurisdiction, interventions, and cross-case actions on sexual violence by government forces in intrastate conflicts from 1989 to 2018, using the Sexual Violence in Armed Conflict (SVAC) dataset. Contrary to the optimistic claims of proponents of prosecutions, we found that ICC jurisdiction and cross-case actions have negligible effects for this category of actors. We also found that ICC interventions are associated with increased sexual violence by government forces in intrastate conflicts. These findings suggest that prevention might require alternative—and, in some cases, potentially costlier—interventions.

Acknowledgments

Previous versions of this research were presented at the Sexual Violence in Armed Conflict Workshop (2014) at Harvard University and the annual meeting of the Midwest Political Science Association (2021).

Notes

1 Broache (2016) and Jo and Simmons (2016) described similar mechanisms applicable to atrocity crimes more generally. Broache (2016) used the terms deterrence and socialization, respectively, to describe these mechanisms, whereas Jo and Simmons (2016) used prosecutorial deterrence and social deterrence (Broache (2016) followed Sikkink (2011) in using the term socialization). Following Broache (2016) and the criminological literature (Apel, 2013; Nagin, 2013), we use deterrence to describe the specific prevention mechanism by which prosecutions are posited to increase the expected costs of crime. This is analogous to prosecutorial deterrence, as described by Jo and Simmons (2016). We use social learning to describe the mechanism analogous to socialization, per Broache (2016), and social deterrence, per Jo and Simmons (2016), to avoid confusion with CRSV as a component of certain armed groups’ socialization processes (Cohen, 2013; Wood, 2018), and deterrence as a distinct prevention mechanism.

2 On the significance of criminalizing wartime atrocities (although not specifically CRSV), see Savelsberg (2010).

3 This logic follows arguments that sexual violence prosecutions in domestic settings can undermine patriarchal social norms. For an overview and critique of these arguments, see Gruber (2009).

4 Article 7 of the Rome Statute defines the enumerated acts as crimes against humanity when occurring as part of a “widespread or systematic attack” against a civilian population, whereas Article 8 specifies these acts as war crimes when “also constituting a serious violation of article 3 common to the four Geneva Conventions” or when “constituting a grave breach of the Geneva Conventions.”

5 The results in Table 2 are robust to restricting our sample to full calendar years after the entry into force of the Rome Statute, 2003–2018 (see Online Appendix, Table A4).

6 For the period covered in our analysis (1989–2018), there are 1,220 conflict-years reported in the dyadic version of the UCDP/PRIO Armed Conflict Dataset, of which 1,118 (97.38%) are intrastate.

7 Among primary belligerents in intrastate conflicts in the UCDP/PRIO ACD from 1989 to 2018, the average value of Combined SV (the sexual violence prevalence indicator used in our analysis) for government actors is .70 versus .25 for rebel groups.

8 These actors are listed as “Type 1” forces in SVAC; this category corresponds to actors listed as “SideA” in the UCDP Armed Conflict Dataset.

9 Due to missing data on several control variables, there are 855 observations in the full models reported in Table 2.

10 For descriptive statistics for all variables, see the Online Appendix, Table A1.

11 It is also plausible that the simple existence of the ICC, independent of jurisdiction or specific ICC actions, may generate deterrence and/or social learning and contribute to preventing CRSV. To account for this, we coded each observation on post-2002, a dichotomous variable coded 1 for all full calendar years in our sample after the Rome Statute’s entry into force. We reestimated Model 1 using this variable and found no significant effect (see Online Appendix, Table A4, Model 8).

12 We additionally generated the variables forced recruitment and troop quality. Forced recruitment is a dichotomous variable coded 1 if the state engaged in “pressganging,” a technique of forcible recruitment associated with CRSV, per Cohen (2016). We did not include this variable in the models reported in Table 2 because Cohen’s (2016) source data covers only 1989–2012. We reported models including forced recruitment for 1989–2012 in the Online Appendix (Table A6). The results are substantively similar for our ICC variables, although ICC global actions and ICC global SV actions are positive and significant in these models. Troop quality was calculated as the quotient (logged) of the state’s annual military expenditure (constant USD 2010), per the Stockholm International Peace Research Institute (2022) Military Expenditure Database, and military personnel. Although indirect, troop quality may also provide insight into the strength of internal command and control, which is associated with CRSV (Hoover Green, 2016). Due to missing data in the SIPRI Military Expenditure Database, we were unable to calculate troop quality for 124 observations in our sample. We reported models including troop quality (omitting military personnel) for the observations for which we were able to construct troop quality in the Online Appendix (Table A7); including this variable does not alter the results for our ICC variables.

13 The military personnel data reported by the World Bank (2022) were originally sourced from the annual military balance reports compiled by the International Institute for Strategic Studies.

14 We additionally estimated models including a lagged dependent variable (see Online Appendix, Table A3). The results are substantively similar for the ICC variables, although ICC interventions is significant only at p < .10 with the reduced sample size (n = 675).

15 This model covers only the years in which the Rome Statute was in force (lagged one year), 2003–2018. To preserve the full range of variation in our ordinal outcome indicator for sexual violence prevalence, we employed a two-stage OLS estimator, rather than a bivariate probit or another estimator that would require dichotomizing this variable. We also noted that all ordered probit results reported in Table 3 are robust to using OLS (see Online Appendix, Table A8).

16 Sitting government officials—most notably, Kenyan President Uhuru Kenyatta—have voluntarily appeared before the Court in response to summonses. However, the Court has yet to secure the arrest of an incumbent government official.

Additional information

Notes on contributors

M. P. Broache

M. P. Broache is an assistant professor of political science at the University of North Carolina Greensboro, where their research focuses on international law and armed conflict.

Juhi Kore

Juhi Kore is an MPhil candidate in the Department of Social Policy & Intervention at the University of Oxford.