Principled and Pragmatic Exit: Understanding Treaty Withdrawal

My book project, based on my NSF-funded dissertation, explores the causes of exit from international agreements as a lens to help understand variation in state responses to international law. To examine the role of legalization in leading to agreement withdrawal, I deploy supervised machine learning on a large dataset of treaty exits. I construct several case studies examining certain treaty exits in greater depth with the assistance of over sixty-five interviews with elites involved in the treaty withdrawal process. Two articles stemming from my dissertation are currently under review and are available upon request.

My central research question is: what leads states to exit international agreements?  Based on my original data collection, states have exited from publicly registered treaties over 4,500 times in the last 100 years.  I argue that the act of exit masks two very different logics of action, which I label principled and pragmatic exit.  Pragmatic exits occur when a state that feels constrained by a treaty exits from it to undermine the agreement’s legitimacy or to avoid the consequences of violation.  An example is Duterte’s decision to withdraw the Philippines from the ICC when he was under investigation for extrajudicial killings that occurred during his regime.  In contrast, principled exits occur when states withdraw from agreements they see as normatively outdated, in which  the fear of punishment associated with violation is less salient.  Recent mass state withdrawals from poorly enforced early 20th century labor treaties that included regressive provisions related to gender, child labor, and colonialism fall into this pattern of exit.  I argue that the interaction between state characteristics, such as political trust, and treaty characteristics, such as legalization, influences the likelihood of each type of exit.

Over the last several years, there have been many high-profile treaty exits, including Brexit; withdrawals from the International Criminal Court (ICC) by Burundi and the Philippines; the United States leaving the Paris Climate Accord, the Intermediate-Range Nuclear Forces (INF) Treaty, and the Iran nuclear deal; and Japanese exit from the International Whaling Commission.  Despite these recent events, scholarship on exit is relatively rare, and often frames exit as form of nationalist backlash.  However, when states withdraw from treaties, they are using the very legal tools granted to them by the international legal order to reject the treaties, suggesting a deep embedding of these principles.  Treaty exit may also be a sign of normative contestation and change: a pathway through which activists can pressure states to reject agreements that no longer fit their values, signaling the end of the norm lifecycle.  We may gain new purchase on the power and limits of international law—and the conditions and beliefs that make it work – through exploring when states announce their decision to no longer be constrained by a treaty.

This project makes several empirical contributions. First, I compile an original dataset on the frequency of exit over time, and illustrate how treaty exit varies by agreement type.  I find that even when controlling for total treaty participation, states with a history of international cooperation, such as Canada, Sweden, and Australia, have some of the highest rates of exit.  Second, to examine the effect of treaty legalization on withdrawal, I coded 2,000 treaties on their degree of precision, obligation, delegation and flexibility, and then, supported by UW’s eScience Institute Incubator, used these human coded labels as the input for a supervised machine learning algorithm I used to label 30,700 publicly registered treaties.  I find that delegation, such as granting power to non-state actors to interpret the treaty, is associated with a higher likelihood of individual state exit.  In contrast, precision, which narrows the possible interpretation of treaties, is correlated with mass exit. To better understand the factors which contributed to recent treaty exits, I spent three weeks interviewing decision-makers in Geneva, Switzerland, and two weeks in the Hague in the Netherlands, supported by two research grants.