Domestic law and human rights treaty commitments: The Convention against Torture

ABSTRACT

How human rights treaties will be incorporated and applied domestically must affect how eager states will be to ratify those treaties. This article focuses on two characteristics of domestic legal systems that shape the relationship between international law and domestic law: whether treaties are directly incorporated into domestic law and whether treaties can override ordinary statute. The analysis probes two arguments as to why domestic legal institutions influence ratification decisions, one emphasizing the potential costs associated with ratification and the other emphasizing congruence between domestic values and treaty norms. Survival analysis of ratification of the Convention against Torture reveals that both judicial independence and making treaties equal or superior to statute increase the likelihood of ratification, which is consistent with the norm-congruence thesis. The results suggest new avenues for investigating the relationships between human rights treaties and domestic legal institutions.

Acknowledgments

The author is grateful to Mila Versteeg, Zach Elkins, Tom Ginsburg, and James Melton for sharing data and to Benjamin Graham, Darren Hawkins, Yonatan Lupu, Brian Rathbun, Mila Versteeg, the Irvine Comparative Sociology Workshop, the Faculty Workshop at the University of Southern California Gould School of Law, and two anonymous reviewers for the Journal of Human Rights for exceptionally helpful comments on earlier drafts.

Notes on contributor

Wayne Sandholtz is the John A. McCone Chair in International Relations at the University of Southern California.

Appendix A

Table A. Countries where treaties are equal or superior to statute.

Appendix B

Table B. Descriptive statistics.

Appendix C

Table C. CAT ratification, test of proportional hazards assumption, and Cox Proportional Hazards Model with time-varying covariates.

Appendix D

Figure D1. Schoenfeld residuals.

Notes

1. In international law, this is the distinction between “monist” and “dualist” systems. In the former (in the ideal type), ratified treaties are automatically part of domestic law, directly applicable in domestic courts. In dualist systems (again, an ideal type), treaties must be incorporated into domestic law via legislation.

2. The non-common law (specifically, non-British) legal traditions include the French, German, and Scandinavian traditions (La Porta et al. 2008).

3. “Compliance costs” refer to the possibility “that a new treaty or any new commitment will be used against [a state] in unintended and unwanted ways,” including the risk that domestic judges will apply international treaty rules against government actors or policies (Goodliffe and Hawkins 2006: 363–364).

4. Data on formal constitutions are available, whereas broadly cross-national and cross-temporal data on domestic judicial practices regarding conflicts between treaty law and domestic statute do not yet exist.

5. I am grateful to Mila Versteeg for sharing her data.

6. I am grateful to Zach Elkins, Tom Ginsburg, and James Melton for sharing their data.

7. For a similar argument, see Simmons (2009: 68–71). With respect to the data, see C. Beck et al. (2009) and Keefer (2010).

8. To verify that the probabilities (p values) reported in Table 2 were not the product of testing numerous hypotheses simultaneously, I controlled for the false discovery rate. The false discovery rate is the fraction of tests found significant in a study but for which the null hypothesis is actually true. I calculated adjusted p values using the Benjamini-Hochberg (B-H) method, which generates p values such that, for the set of tests taken together, the proportion of significant p values when the null hypothesis is actually true remains below a specified level (here, p < .05). See Benjamini and Hochberg (1995). For each variable, the B-H adjusted p value confirmed the original p values (results not reported but available from the author). The findings are therefore not produced by simultaneous hypothesis testing.

9. This figure and the next two are based on Model 2 in Table 2.

10. The data used for these comparisons are averages of the independent variables over the analysis period.