What leads states to exit from treaties? Is treaty exit simply a reversal of the conditions that lead states to commit to a treaty? What factors increase the likelihood a state will withdraw from a treaty and what treaties are they most likely to leave? In this dissertation, I identify broad patterns in state withdrawal from international agreements and theorize that there are two distinct patterns of exit, pragmatic and principled. State proclivity to engage in either pattern of exit is influenced by the interaction of state specific factors, such as diffuse political trust, and treaty legalization. I test this theory through case studies, elite interviews, analysis of speeches and press statements, supervised machine learning with treaty text as data, survival models, and regressions.

I define treaty exit and the conditions under which is it allowed.  I then share data on patterns of treaty exit I compiled from the UN Treaty Archive, which contains all publicly registered treaties between 1919- 2019. I identify over 4,800 treaty withdrawals, and find that treaty exit peaks during times of global transition, such as decolonization and the end of the Cold War. States withdraw from economic, labor, and environmental treaties more frequently than they exit security agreements and agreements which govern physical integrity rights. Somewhat counter intuitively, I discover that the states most prone to exiting treaties, even as a fraction of total treaty participation, are often the states most invested in global cooperation, such as Canada, Sweden, the Netherlands and Japan.Even controlling for total treaty participation and thus the opportunity to exit, European states (and former British settler colonies) leave treaties at higher rates than countries in other regions.

 I theorize that treaty exit is the product of three things: a change related to the treaty, as highlighted by the quote above, the degree to which the treaty is legalized, and  state political trust. I identify two primary distinct patterns of exit, principled and pragmatic. In principled exits, states withdraw from treaties to contest the norms enshrined in the treaty and are not influenced by the likelihood of enforcement under the treaty. In pragmatic exits, the state leaves because they would face a consequence under the treaty for current or future violation. I then hypothesize about how the interaction between political trust and treaty design may lead to patterns of principled or pragmatic exit.

I examine treaty-level factors, specifically legalization. I identify hallmarks of precision, delegation, obligation and flexibility in inter-national treaties. I then use supervised machine learning to label a large corpus of treaties based on the presence of each element of legalization. I find that delegation and precision significantly increase the likelihood of treaty exit, while obligation, especially the presence of dispute resolution mechanisms, may lower it. I briefly examine how legalization may in-fluence exit risk in more detail through examining withdrawal from treaties governing the death penalty, and the denunciation of ILO conventions related to child labor.

I then turn to an examination of state level factors.  I  theorize  that  states  with  high levels of domestic trust are more likely to engage in principled exit but are less likely to engage in pragmatic exit. On the other hand, states with low trust are less likely to engage in principled exit and more likely to carry out pragmatic exit. I find support for the hypothesis that domestic political trust is associated with treaty exit. High political trust decreases the likelihood of pragmatic exit while increasing principled exit

.I then turn from my discussion of treaty and state level factors to an examination of the patterns of principled and pragmatic exit, which integrates these factors.  I explore principled exit through the case of state withdrawal from early 20th century ILO conventions which sought to restrict women's equal opportunity at work under the guise of special protection.  I find that states with strong women's and workers rights are more likely to exit these agreements than their counterparts with weaker rights, suggesting that these denunciations are driven by state conflict with the normative content of these agreements. I then use 42 semi-structured interviews with ILO staff and delegates to better understand the denunciation of these agreements.

Finally,  I study patterns of pragmatic exit by contrasting Burundi's exit from the Rome Statute with South Africa's threatened withdrawal. I find that in both cases, a precipitating event occurred that increased conflict between the state and the International Criminal Court (ICC). However, for Burundi, the precision of the Rome Statute, as well as the delegated authority to the UNSC, created irreconcilable tension that led to exit. In contrast, after deciding to remain in the treaty, South Africa plans to use the dispute resolution provision within the treaty to ask the ICJ to clarify the conflict between obligations under the Rome Statute and customary internationallaw.

Last Modified: 10/24/2022
Modified by: Bree L Bang-Jensen