[Update: Kevin Jon Heller replies to Ryan’s post over at Opinio Juris. And Ryan responds to Kevin. Hugo Relva, Legal Adviser at Amnesty International submitted a Letter to the Editor. Ryan Goodman’s response to his Letter is here.]

Would you believe that approximately 76-84% of all states around the world have enacted universal jurisdiction for at least one serious international crime (such as war crimes or torture)? You would if you read Amnesty International’s major survey, Universal Jurisdiction: A Preliminary Survey of Legislation Around the World (2012). But you shouldn’t.

Some background: An annual agenda item of the UN General Assembly’s Sixth Committee is “the scope and application of the principle of universal jurisdiction.” The United States participates actively on this item, as witnessed by the administration’s submission of a helpful report to the Committee describing US practices. Amnesty International prepared its survey for the purpose of informing the Sixth Committee discussions. But Amnesty’s study is also more broadly relevant to any analysis of the legitimacy and spread of universal jurisdiction—as demonstrated by Beth Van Schaack’s post last night. Indeed, many of us who think about these questions turn to the survey.

I do not doubt the general reliability of the survey, but one particular element has always troubled me (in the 2011 and 2012 versions of the survey). The upshot is that some of Amnesty’s totals may be significantly inflated. I wish this were not the case, because Amnesty does such great work, and there are few efforts as worthy as the struggle to end impunity for massive international crimes. But that should not have us shy away from a rigorous and honest analysis of the reported findings.

Here’s what troubles me: Amnesty International appears to count states as having enacted universal jurisdiction if the state is a party to the Rome Statute for the International Criminal Court or, more precisely, if the state has adopted a form of implementing legislation along with ratification of the treaty. Amnesty makes that decision on the stated assumption that the Rome Statute implicitly requires member states to adopt universal jurisdiction corresponding to its core crimes. This raises two problems.

First, the Rome Statute is not based on pure universal jurisdiction. In the treaty negotiations, Germany and South Korea nobly proposed that the ICC should include universal jurisdiction, but that idea was rejected by a vote in the plenary. Instead, the ICC is built on a jurisdictional design limited to violations (1) committed by a national of a member state or (2) on the territory of a member state. So, even if the Rome Statute required states to enact legislation matching its jurisdictional provisions, that framework would not include a member state (e.g., South Africa) adopting legislation for actions of a nonmember state (e.g., the United States) in the latter state’s own territory (e.g., the United States) or on the territory of another nonmember state (e.g., Pakistan).

Second, as Beth’s post deftly shows, the Rome Statute does not include an obligation for member states to enact (extraterritorial) jurisdiction.

Here is the section of the Amnesty report (on page 9) in which the authors describe their methodological choice:

Crimes defined in national law, with reference to treaties.
In some instances, the state has defined a crime under international law, such as genocide, as a crime in national law and provided that its courts have jurisdiction over crimes in treaties it has ratified (some provisions do not specify that the treaty has to have been ratified). In those instances, the state would have jurisdiction not only over crimes in aut dedere aut judicare treaties, but treaties like the Convention for the Prevention and Punishment of the Crime of Genocide (Genocide Convention) and the Rome Statute of the International Criminal Court (Rome Statute) that do not contain an express obligation to exercise universal jurisdiction, although they may contain an implied obligation to do so. Annex I indicates that the state has jurisdiction over the relevant crime (YES).

The strongest argument in favor of Amnesty is that the Rome Statute still involves a form, albeit a modified one, of universal jurisdiction. That is, its framework provides for member states (e.g., South Africa) to exercise jurisdiction – through the Court — for crimes committed in another member state (e.g., the DRC) without any connection to the former. The problem remains, because this “modified” version is probably not what many readers of the survey have in mind. Moreover, this framework would reflect jurisdiction that is “treaty based”—a.k.a. by special agreement and consent of ratifying parties—rather than jurisdiction based on the nature of the offense per se. Regardless, the Rome Statute does not impose an obligation on member states to establish jurisdiction, not even in the way that aut dedere aut judicare treaty provisions do. Hence, the inferences that Amnesty draws are dubious, and so is the tally of universal jurisdiction states that is predicated on those inferences.

[* The original post mistakenly stated that Amnesty’s study reported 86% of states have enacted universal jurisdiction. The Amnesty study states (my emphasis added):

“This updated preliminary survey indicates that 166 (approximately 86%) of the 193 UN member states have defined one or more of four crimes under international law (war crimes, crimes against humanity, genocide and torture) as crimes in their national law. …

… [I]t appears that 147 (approximately 76.2 %) out of 193 states have provided for universal jurisdiction over one or more of these crimes. In addition, at least 16 (approximately 8.29%) out of 193 UN member states can exercise universal jurisdiction over conduct amounting to a crime under international law, but only as an ordinary crime … Thus, a total of 163 states (approximately 84.46%) can exercise universal jurisdiction over one or more crimes under international law, either as such crimes or as ordinary crimes under national law.”

I thank Hugo Relva for bringing the error to my attention in his Letter to the Editor.]