Loosening the Ties that Bind: A Learning Model of Agreement Flexibility.

The tyranny of an unamendable constitution might lead to revolution; the forced continuance of the legal status quo as between states might make it more difficult to preserve peace. [1]

We all know that future generations will have to live, even on a strictly technological level, in a setting very different from the present one. To imprison them in an iron corset, which could not be adjusted to the changing conditions of history, would in our opinion expose that corset to the danger of bursting. We therefore prefer a steel corset which, being more flexible, could more effectively assure the continuation of the treaty.

--Statement issued by Italian representative Roberto Caracciolo regarding the proposed Nuclear Non-Proliferation Treaty. [2]

How can the corsets of international agreements be made more flexible and therefore more robust? Or, in the words of Jon Elster, how can Ulysses be loosely bound to the mast? [3] This is an important question for the simple reason that international cooperation is important, and international cooperation is often institutionalized in the form of international agreements. Existing international agreements are testament to states' willingness and ability to cooperate despite the international anarchy in which they find themselves. Given the difficulties of cooperation under anarchy documented in the recent international relations literature, understanding how states manage to bring about the formal cooperation embodied in international agreements is of both theoretical and practical interest.

States can make agreements more desirable in prospect and more robust in practice by varying their provisions for duration and renegotiation. These provisions help states account for the uncertain economic, political, and technological contexts in which agreements are made and (ideally) kept.

If we look around, we find that international agreements differ widely in their duration provisions. The Plaza Agreement on foreign exchange intervention, for example, covered a period of just six weeks. At the other extreme, the Universal Postal Union is of indefinite duration and has already lasted over one hundred years. Renegotiation provisions vary widely as well. The International Coffee Agreement was renegotiated about every six years, whereas the Antarctic Treaty had no renegotiation provisions for the first thirty years and renegotiation at the request of any party after that. Importantly, the variation across agreements in these provisions is not random; rather, as I argue here, it results from the considered choices of the parties to each agreement, in light of their preferences and the unique aspects of the context of each agreement.

Nevertheless, the issues of duration and renegotiation have been almost completely ignored in the political science literature on international relations. They have been wholly neglected in theoretical studies of international cooperation, and there has been surprisingly little discussion of these issues from an empirical point of view. In fact, although some discussions of individual agreements cover the issues of duration and renegotiation for the agreement in question, no work exists that attempts to account for or even describe the observed patterns and variation in agreement duration and renegotiation. [4]