Bargaining strategies for governance complex games

Extant governance complex models fall under the “outside option” category. This first generation of models seeks to explain why an aggrieved subset of members of an existing international organization would want to exit the organization and establish a new one. These models were initially designed to explain when states renegotiate existing institutions (Lipscy 2017) and when they create new overlapping institutions (Urpelainen and Van de Graaf 2014). These models internalize the first and second steps of the above-mentioned three-step sequence—initial negotiation and reneging—while leaving aside the third step—reconciliation. I discuss these two outside option models and point to their limits.

The logic of an outside option model is straightforward: if a member of an organization entertains the option of improving its payoff by exiting the organization and establishing a new one, it will threaten to exercise that option, forcing the other members to either match that option and retain that member or not match it and lose that member. In either case, the existence of the outside option degrades the agenda setter’s residual value.²

The main difficulty that this line of models faces in predicting regime shifting is a potentially inexplicable lack of foresight: “What prevented the other members of the organization from matching the offer and preempting regime shifting?” Two solutions have been offered: information asymmetry (Urpelainen and Van de Graaf 2014) and renegotiation costs (Lipscy 2017). Information asymmetry means that the other members of the organization do not know the value of the potential quitters’ outside option. Their counteroffer merely rests on beliefs and expected values, with the result that they are bound to guess wrong with some positive probability and let the aggrieved member go.

Another way of getting around the lack of foresight problem, one that does not require asymmetric information, is to combine the outside option with a renegotiation cost. According to Lipscy (2017), some fora are costly to renegotiate because they require supermajorities or consensus from all the concerned parties. In the presence of such high costs, the existence of outside options may lead to exit rather than renegotiation.

Irrespective of whether it is information or renegotiation costs that motivate the exit, outside-option models have one trait in common: they only capture the first two steps of regime shifting—negotiation and reneging—falling short of including the reconciliation stage. It is in that sense that they differ from the broader set of inside-option models, for which the second move is not conceived as a permanent move, but as a temporary tactic setting the stage for the third and potentially final step, that of renegotiation.

Although outside option models were not designed to include the third stage—renegotiation—one may question the pertinence of an outside-option model in an historical context that, in the long run at least, does not offer any real outside options. If regimes exist to solve collective problems that states cannot address alone or in smaller formations (Keohane 1984), then exiting and joining another forum that excludes the status-quo members of the initial forum offers no viable improvement.

At most, outside options may work for powerful states, who do not need to convince the rest to join them (Oatley and Nabors 1998; Gruber 2000; Voeten 2001; Stone 2011; Pratt 2020). But, even then, modeling regime shifting as the exercise of an outside option ignores the gutting or suspension of present law actually inflicted by regime shifting on all players, powerful ones included. It is partial-equilibrium analysis at best, given the fact that bargaining among governments, under one guise or another, never ends. While my adding a third stage of renegotiation and bringing the game to an end still is a simplification of this never-ending historical reality, it is a first step in understanding the dynamics of governance complexes—another key theme of this special issue.

A classic illustration of exercise of an inside option in economics features employees going on strike. More broadly, exercising an inside option involves temporarily degrading the other side’s value for the status quo in order to force a renegotiation of that status quo.

Regime shifting, I argue, involves some form of degrading. For instance, Helfer (2009: 42) describes regime shifting as “decreas(ing) the clarity of international law” and introducing “strategic inconsistencies”. Focusing on intellectual property, he further explains that

As the rules regulating intellectual property become more numerous, nuanced, and contradictory, they provide leeway for states to interpret and implement rules to further their own interests while remaining nominally within the boundaries of what international law requires.

Imprecision and inconsistencies, albeit legal, de facto mean the suspension, whole or in part, of the initial agreement for both sides. An apt illustration of this mechanism is found in Helfer’s study of the intellectual property (IP) regime following the 1995 TRIPs agreement. The debate about intellectual property involved three sets of interests: the North was a net producer of IP and wanted to maximize protection for IP; the South was a net consumer and wanted free access; some countries in the South also were the main custodians of upstream resources that northern corporations often used in the making of IP—the list included plants, other genetic resources, and “the cultural heritages and traditional knowledge of indigenous people”. The TRIPs agreement favored northern interests at the expense of the South by giving blanket protection to intellectual property producers at the expense of both downstream consumers and upstream input providers. The South regrouped and responded by shifting the debate to non-trade fora, where they endeavored to carve exceptions into the TRIPs’ extensive protection for intellectual property. First, the World Health Organization passed resolutions reducing protection for IP in case of wealth emergency. Second, the WIPO, the Food and Agriculture Organization and a few other settings passed resolutions envisaging transfers to the upstream input providers.

An intended effect of these revisions was to throw just enough uncertainty in the initial TRIPs agreement to make it legally unactionable. Rather than seizing the WTO dispute settlement body, northern countries fell back on signing bilateral agreements known as “TRIPs-plus”. Another intended effect was to force the North to renegotiate the initial agreement. This happened on one account: in 2001, the initial TRIPs agreement was formally renegotiated to make it easier for southern countries to compulsorily license patented medicine in case of national emergency.³

The actual legal mechanism behind the degrading of the TRIPs agreement is unclear. Helfer (2009) attributes it to the accumulation of imprecision and inconsistencies, an argument that touches on the foundation of the rule of law, how something that is not legal becomes so by way of customs.⁴ From a more positivist point of view, however, it is not clear why the Quad countries could not have just ignored all non-WTO resolutions and enforce the pre-existing WTO trade law instead. Rather than taking sides in a field in which, as a political scientist, I have no expertise, I would instead point, along with Steinberg (2002), to the fine line that separates law from power in a self-help system. Ignoring the WHO, WIPO and FAO resolutions in the WTO would presumably come at the cost of further lowering support for divisive issues and increasing footdragging.

While the degrading of the continuation value is typical of the way most inside-option models in the field of economics operate, the present model differs from the economic state of the art by not relying on the steady flow of time as the cause for the degrading of these values. In the case of wage negotiations, for instance, it is each new day spent striking that reduces the aggregate value of an eventual agreement (Muthoo 1999, ch. 6). Rather than money lost through elapsed time, the regime shifting literature focuses on utility lost through unenforced content: how much of the initial deal is made unenforceable by the act of shifting. The time spent negotiating is usually long and of limited consequence to the outcome.

Although regime shifting effectively links issues across international organizations, it is not reducible to a regular issue-linkage. The point of linking two issues is for two governments to reach an agreement (Sebenius 1983; Davis 2004; Mikulaschek 2018). In contrast, the purpose of regime shifting is the sabotage of an existing agreement in order to force a renegotiation. Further, the issues that constitute an issue-linkage are complementary—different sides place the most importance on winning on different issues, enabling logrolling—whereas the issues involved in regime shifting are substitutes—all sides place the same importance on most issues, doing nothing to facilitate a deal.

Regime shifting falls within McKibben (2015) broader category of “confrontational rule-changing” strategy, which she defines as a bargaining move that has the effect of changing the other side’s “no-agreement value”. Modify McKibben’s definition by substituting “continuation value” for “no-agreement value” and it overlaps with the notion of inside option in the present study. More generally, the notion of inside option is not qualitatively different from von Clausewitz’s (1984[1832]) understanding of the role of war, a process in which one side seeks to extract a better bargain in the long run by undermining the other side’s actual (or perceived) continuation value for the political status quo in the short run.⁵

The word “inside” is not coterminous with “internal”. A strike is an inside option that is also an internal move, but a trial is an inside option that is an external move, involving dragging the firm’s executives before a court of law to press them to reach a settlement. Within the field of international organizations, regime shifting is an inside option that involves an external organization whereas footdragging is one that is purely internal.

The key difference between outside and inside option models resides in the strategic intention of the actor who is exercising the option. It does not matter whether the option takes the milder form of footdragging or that of an outright exit; nor does it matter what the outcome is or whether there is full, partial, or no reconciliation. What matters is whether the option is exercised with the intent of being a terminal strategy instead of a temporary move in a multi-period bargaining strategy.

For instance, how to categorize the 1984 U.S.’ exit from UNESCO? Without being privy to the decision, it is hard to tell. If the U.S. left because it had had it with cultural organizations, on account, say, that culture has limited respect for power, then Washington exercised an outside option. In contrast, if the move was meant to trigger a stampede by wealthy countries eager, in the wake of the U.S., to create an alternative to UNESCO or make an example of UNESCO, thereby forcing the remaining members to accept a reform or create a successor organization that would be more willing to accommodate U.S. preferences, then Washington exercised an inside option (Hoffer 1986:167). Merely observing the outcome—the fact that there was no stampede—is insufficient for calling the exit the successful exercise of an outside option as it may very well have been instead a poorly-conceived or badly-executed inside option.

There are three requisites for a government to degrade a deal. First, the implementation of the status quo (the initial deal) must be a source of loss for the party that wishes to degrade the deal. Were the status quo a source of gains instead, suspending it would unadvisedly lower those gains. Therefore, for such a suspension to be desirable, there have to be losses that a suspension will cut. Moreover, these losses cannot have been anticipated but must have instead been inflicted by an unpredictable reversal of fortune.

A second requisite for a government to exercise an inside option is the absence of a better outside option. Were one available, the government facing the reversal of fortune would exercise it in lieu of the inside option.

The third requisite for a government to exercise an inside option is for it to be legally undeterrable or, at least, de facto left undeterred. For instance, exercising the escape clause in the GATT/WTO is a form of degrading that is easily deterrable because its execution by one member entitles the other members to extract compensation in return, thereby potentially discouraging the exercise of the clause in the first place. In the same way as the trade union from the economic example would be deterred from exercising its inside option if going on strike were illegal or union busting tolerated, a government might be dissuaded from degrading the initial deal if, while so doing, it were to expose itself to a sanction or a reputation loss.

I now present two versions of regime shifting, one that works for the weak (defined as the agenda taker), the other, for the strong (defined as the agenda setter). Both games feature two sides locked into a bargaining relation: while both sides agree to capture gains from cooperation, they disagree on their distribution.

Several empirical illustrations inform the present game. I already presented the evolution of the intellectual property complex in the wake of the TRIPs agreement of 1995. Another illustration involves genetically modified organisms (GMOs). The Uruguay Round generated the Sanitary and Phytosanitary Measures Agreement, prohibiting trade restrictions of GMOs in the absence of a scientific proof of risk to health. Preferring tighter rules, developing countries asked for its amendment. The US refused. Five years after its introduction, developing nations along with the EU adopted the Cartagena Protocol outside the WTO framework, making imports subject to approval by the importing country on the basis of “precaution and socioeconomic factors” (Gehring and Faude 2014: 484). Although there has been no formal renegotiation of the issue within the WTO, the US, Canada and Argentina initiated legal proceedings against the EU before a WTO panel in 2006 with such mixed results that these very same countries were dissuaded from instigating new proceedings (486). Both cases illustrate the three-step sequence highlighted in the introduction—negotiation, regime shifting, and renegotiation—although in both cases, the last step is incomplete.

I offer a model of regime shifting in which the South, who becomes unhappy with the payoff they got in the initial negotiations, uses regime shifting to improve their payoff in the subsequent renegotiation. To get this result, I introduce a random component in the payoffs. In a first period, the players negotiate while having only a vague notion of their future payoffs. Then the payoffs are fully revealed and in the case where one side does not like the result, it may initiate regime shifting.

Note that in the absence of uncertainty, the initial agreement would have anticipated the changes and set payoffs so as to deter regime shifting. Uncertainty is necessary to the occurrence of regime shifting.

The game features two actors bargaining over a single issue—a pie equal to unity—according to a well-structured sequence of moves (the tree is drawn in Fig. 1). The agenda setters (call them North) start by offering x₁ fraction of the pie to the agenda takers (call them South) while keeping 1 − x₁ for themselves, with 0 ≤ x₁ ≤ 1. South accept or reject the offer. In the case where the offer is rejected, the game is over and all players score naught.

This first round of negotiations takes place before Nature modifies the payoffs. After Nature modifies the payoffs, South still get x₁ and North 1 − x₁, but Nature adds (or subtracts) an increment from these payoffs depending on circumstances, which can be “negative” or “positive” for South and vice versa for North. Hence, if Nature draws negative circumstances for South (and thus positive for North) and the outcome is respected, South receive x₁ − θ while North get 1 − x + θ, with 0 ≤ θ ≤ 1. Conversely, if Nature draws positive circumstances for South (and thus negative for North) and the outcome is respected, South receive x₁ + θ while North get 1 − x₁ − θ. Nature draws negative circumstances for South with probability p and positive ones with probability 1 − p.

Although bad circumstances for South are good ones for North and vice versa, henceforward I adopt the Southern perspective and refer to the left branch of the tree as Nature drawing “bad” circumstances and the right side as Nature drawing “good” ones.

Following Nature’s draw, South chooses between “respecting” the initial partition and two additional alternatives: an inside option and an outside option. I develop both successively. The inside option, regime shifting, consists in “partially reneging” on the initial deal. I operationalize regime shifting as a form of partial reneging on one’s promises. This partial suspension of enforceability degrades payoffs for the side that is benefiting from the agreement while cutting losses for the side that is losing from it. More specifically, partial reneging implies that δ percentage of the initial agreement is made unenforceable for both sides. The value of δ is exogenously given and falls somewhere between zero (no reneging) and one (full-scale reneging).

The analysis henceforth focuses on the left branch of the tree. Upon observing reneging, North face two options: a first is to tolerate the fait accompli, yielding final payoffs of \(\left (x_{1}-\theta \right ) \left (1-\delta \right ) \) for South and \(\left (1-x_{1}+\theta \right ) \left (1-\delta \right ) \) for North. A second option is to propose a renegotiation. The renegotiation takes a form identical to the initial negotiation, with North proposing x₂ and 1 − x₂ fractions of the pie respectively going to South and North, with 0 ≤ x₂ ≤ 1.

Last to play, South can accept or reject North’s last offer: rejecting keeps the prior reneging payoffs unchanged; accepting yields new payoffs x₂ − θ and 1 − x₂ + θ respectively.

The third choice that is available to unlucky Southern countries besides respecting the initial deal as modified by Nature or exercising their inside option is to exercise their outside option—creating a rival organization. This means that an organized subset of Southern governments offers an alternative organization to all the other members, be they from the South or the North, who each have to make an individual decision about whether to join the new organization or stick to the old one.

The exercise of the outside option opens up a succession that is best modeled as a tipping game. In such a game, two Nash equilibria coexist: everyone joins the rival organization or everyone sticks to the old one. Fueling this all-or-nothing logic are the scale economies that grace collective action. Therefore, for a rival succession to be successful, enough countries must be persuaded to join so as to make joining attractive to everyone else.⁶

To estimate how many southern countries need be persuaded to join, consider Fig. 2. The horizontal axis represents the number k of countries sticking to the old organization, while the vertical axis is an individual Southern member’s payoff. Two curves are drawn. The ascending curve plots a Southern country’s payoff for sticking to the old organization (assuming the number of all countries is n = 100 and the initial deal was x₁ = .3). Reflecting scale economies to organization, this payoff increases with member size k. In contrast, the descending curve plots a Southern member’s payoff for joining the rival organization (assuming y = 1 and β = 1.1). It is drawn in counterpoint to the other curve so that if k members stick to the old organization, n − k join the new one.

Payoffs from the rival succession are equal to \(y\left (\frac {1}{2}-\frac { \beta \left (n-\left (S+1\right ) \right ) }{n}\right ) \) for an individual Southern member and \(\left (1-y\right ) \left (\frac {1}{2}-\frac {\beta \left (n-\left (S+1\right ) \right ) }{n}\right ) \) for each Northern member. There are two components to a Southern member’s payoff. A first is y, the percentage of the new pie that is appropriated by South, henceforth the agenda setter of the rival organization. North’s share is the residual, 1 − y. For the sake of simplification, y is set equal to 1—the South as a whole appropriates all the benefits of the rival successor organization, making North’s payoff equal to zero. Note that payoffs under the rival organization escape Nature’s shock θ.

The second component to a Southern country’s payoff (within parentheses) is the size of the new pie. The key variable is S—the size of the South’ core group. I assume that only a subset S of the Southern block—a core—is able to organize enough to propose a rival organization. Given the scale economies, there is a correlation between the size of the core group, who are committed to join the new organization if that organization is created, and the size of the pie. The larger the core group, the larger the pie, and thus the share of the pie that a non-core member can be expected to receive by joining the rival organization even if no other countries join.

The two curves of Fig. 2 intersect around \(k\simeq 36\). This means that the smallest core group S that would tip the scales in favor of the rival organization is equal to n − 36 + 1 = 63—a tall order. With its lower intercept (\(\frac {1}{2}\) instead of 1) and its steeper slope \(\left (\beta >1\right ) ,\) the payoff function for the rival organization was deliberately designed to be less efficient than the payoff function for the old organization.⁷

Although a bit fastidious to prove (see the Technical Appendix),⁸ the solution of the regime-shifting game is remarkably simple. As shown in Fig. 3, there are two generic equilibria: the regime shifting equilibrium, in which Southern members renege on the initial deal when circumstances are bad but respect the initial deal when circumstances are good; and the outside option equilibrium, in which Southern members exercise their outside option when circumstances are bad but respect the initial deal when circumstances are good. Figure 3 was drawn with the probability of bad circumstances p on the horizontal axis and the intensity of the shock θ on the vertical axis. (The other parameters were fixed at, total population, n = 100, Southern core group size, S = 40, inefficiency index of the rival organization, \(\beta =\frac {11 }{10},\) and reneging rate, \(\delta =\frac {1}{2}\).)

The intuition behind the regime shifting equilibrium goes like this. Regime shifting, which boils down to making the initial deal less enforceable for both players, makes sense on the part of South in poor circumstances because South’s payoff is negative and any degrading of a negative payoff improves that payoff. South’s payoff is negative in the first place because the take-it-or-leave-it bargaining protocol leads North to match South’s reservation value, which, in expectation, is equal to zero and only turns negative if and after Nature chose bad circumstances. Now, because South cut their losses by reneging, they improves their reservation value in anticipation of the incoming renegotiation, which North systematically pursue in order to make up for their degraded payoff. In the renegotiation equilibrium, which also uses a take-it-or-leave-it protocol, North make an offer that matches South’s new reservation value, yielding a final payoff for South that is higher than its “respect” payoff. In contrast, when circumstances are good, South’s payoff from the initial negotiation is positive, making partial reneging suboptimal; South respect the initial deal and the game is over.

The other equilibrium, the one in which South successfully supplant the existing organization with a rival one in bad circumstances but respect the extant deal and organization in case of good circumstances, only makes sense for high shocks θ and low probability of bad circumstances p (top left corner of Fig. 3). For high shocks θ, North are no longer interested in insuring South against shocks but are better off writing off the left branch of the tree altogether by offering nothing in the initial deal \(\left (x_{1}=0\right ) \) and banking on an expected meagre-yet-positive payoff on the right-hand side.

In addition to a high shock, the outside option strategy also requires a low probability of bad circumstances p. This may sound counter-intuitive at first, because a low p makes it cheaper for North to insure South against negative shocks, but another rationale is also at work: the idea that what is a bad shock for South, is a good one for North—the higher θ, the higher North’s institutional payoff on the left-hand side of the tree. Therefore, as p increases and, with it, the positive incidence of θ on North’s payoff, North are more willing to keep the existing organization alive and, to that effect, bankroll losses incurred by South by passing them on to North through reneging and renegotiation.

Figure 3 also shows expected payoffs for North and South for various values of p and θ. Each capsule shows an upper and a lower value, the upper for North’s payoff, the lower for South’s. These numbers point to the greater efficiency of regime shifting. Indeed, an efficient bargain is one in which the players divide the pie in any way they want without destroying any of it. Such is the case here when the two payoffs add up to one. If the two numbers add up to less than one, the deal generates a deadweight loss, making it inefficient. By this definition, the three regime-shifting equilibria are efficient. In contrast, the outside-option equilibrium is not. This difference reflects a modeling assumption: the creation of an alternative organization suffers from uncaptured scale economies—the core group must convince the next individual to join while assuming the worst case scenario that nobody else does.⁹ This inefficiency is avoided within an existing organization because members, there, move as blocks, not individuals.

The numbers in Fig. 3 also give us some appreciation for the bargaining power released by regime shifting. Keeping in mind that at the beginning of the game, South had a reservation value of zero, an improvement over this value is visible over three-fourth of the solution space. Surely, not all of it is due to regime shifting: high payoffs for South when p is low are an act of Nature, while positive payoffs in the solution area labelled “regime shifting¹” are partly caused by the existence of the outside option (which North must preempt in order to maintain their agenda-setting privileges). But parts of the payoffs in that same area owe their positive value to regime shifting. Moreover, all the positive payoffs around the middle of the area labelled “regime shifting⁰” are the fact of regime shifting.¹⁰

Figure 3 was drawn for the value of a degrading rate δ equal to one half. If this rate were higher, so would South’s payoffs, because the more of the initial deal South can degrade, the less damaging the bad circumstances for South, and thus the higher South’s internal reservation value in anticipation of the incoming renegotiation. It follows that if δ was made to be a choice variable, one that South would choose rather than being set exogenously, its equilibrium value would be δ = 1, allowing South to reset its reservation value to zero, where it was at the beginning of the game. Because I do not believe that extant legal narratives on regime shifting allow for such wholesale reneging, I chose to make δ exogenous and inferior to unity.¹¹

The present game reserves partial reneging to South, leading one to wonder what would happen in the case where both sides enjoyed the option of reneging. The fact is that nothing would change. North would not avail themselves of the opportunity to back out under any circumstances because, being the powerful side, the one that controls the agenda and serves as residual claimant, they can insure themselves a positive payoff, which reneging would merely debase—reneging only improves on a negative payoff.¹²

As mentioned earlier, the consensus in the literature is that regime shifting benefits the powerful as much, if not more, than the weak (Braithwaite and Drahos2000; Drezner 2009; Helfer 2009, Alter and Meunier, 2009; McKibben 2015). But if powerful is defined as the side that sets the agenda, makes a take-it-or-leave-it offer, and is residual claimant to the deal, then it is not clear what extra bargaining leverage the powerful would win by resorting to regime shifting. I try to elucidate this point by formalizing what is considered in the literature as one of the most flagrant case of overbearing organizational maneuvering that the North perpetrated against the South, the so-called “WTO coup”, by which the US and the other Quad countries (EC, Japan and Canada) paired “trade-related” issues, intellectual property being one of them, with a wholesale reneging of all prior trade obligations contracted under the GATT (Steinberg 2002; McKibben 2015 chap. 6). The Quad countries gave the legal sixty-day notice of their intention to terminate all their pre-existing GATT commitments at midnight on December 31st, 1994, and to renew them a second later as part of the new WTO package deal. As a result, the story goes, each country’s reservation value with regard to trade was degraded, for the trade-related component of the payoff no longer was the whole trade “acquis”—a half-a-century worth of trade concessions between all existing GATT members—but a reduced version of this—the trade “acquis” only with the countries that would not denounce the GATT. By putting the trade acquis in the balance, the industrialized countries, it is believed, got the better part of the bargain (Sell 1998; Gruber 2000; Helfer 2004: 24, Jupille et al.2013).

I analyze the internal consistency of this linkage-cum-succession combination and maps out its conditions of application to conclude that this bargaining strategy was a surprising piece of luck for the North—a fluke.

The linkage-cum-succession strategy indeed presents two shortcomings that limit its use and efficiency. A first is the tipping point Nature of a succession game. Denouncing the GATT was a risky move for the Quad. The southern nations could have refused the terms of the bargain, leaving the Quad countries isolated and outside the GATT at the strike of midnight and forced to renegotiate their reentry into the organization from a position of weakness. Formally speaking, all GATT members, Quad and southern, were players in a tipping point game featuring two equally possible Nash equilibria—everyone sticks to the status quo or everyone shifts to the new WTO—with no prior knowledge of which would prevail. Effectively steering the world toward their preferred outcome forced the Quad countries to make substantial concessions to southern nations.

A second shortcoming of the linkage-cum-succession strategy is that it has to be a surprise in order to work. Had the South anticipated that the North would pursue it while negotiating the Uruguay Round, they would have been more demanding in prior GATT negotiations, de facto cancelling any disadvantage visited by the strategy. The fact that the North took advantage of the South’s bounded rationality means that the strategy worked once but was unlikely to be used anytime thereafter, as Southern governments would most likely learn from their mistakes.

I illustrate these two points by designing and solving a bargaining game between, on one side, industrialized countries led by the Quad, and, on the other side, the developing nations—the South. The North, being the powerful players, control the agenda and the bargaining procedure by making take-it-or-leave-it offers.

Two pies of size one each are at stake. The trade pie was negotiated in a prior GATT game that yielded an absolute payoff of w_zy to an individual developing nation and \(w_{z}\left (1-y\right ) \) to an individual Northern country, with w_z the relative economic weight of individual country z, and y a percentage of the trade pie. To keep the presentation simple, henceforth I normalize all payoffs by dividing them by w_z. Prior GATT payoffs are thus equal to y and 1 − y for North and South respectively.

Negotiations on the other pie, the intellectual property (IP) pie, were deadlocked in the WIPO, leaving each country with a normalized reservation value of 1 for each developing nation and 0 for each Northern country—I assume, for the sake of simplicity, that by not recognizing IP prior to the WTO, each developing nation appropriated the entirety of its share of the pie.¹³

In response to Nature’s shock, North has a choice between doing nothing (“respect”) or initiating a linkage-cum-succession move (offer “x”) in which North link the survival of the initial trade agreement as modified by Nature to South’s reduction of their share of the IP pie to percentage x, with residual 1 − x accruing to North. For the sake of simplicity, I assume that there is no new negotiation with respect to trade, for which past outcomes y and 1 − y obtain.¹⁴

The IP proposal becomes law only for the countries who denounce the GATT and join the WTO. Developing nations are not making a collective decision anymore, but each one is making an individual decision about whether it wants to denounce the GATT and sign on the new WTO package or whether it would rather stick to the GATT and ignore the new agreement on IP.

To stack the deck in favor of the regime-cum-succession narrative, I strengthen the North’s hand by further assuming that economies are of unequal size; more specifically, I model N countries distributed according to the distribution function \(F\left (z\right ) =N-z+1,\) with country z = 1 of size N, country 2 of size N − 1, country 3 of size N − 2 and so on, all the way to country N of size 1. Formally, \(z\in \left [ 1,N\right ] \). I capture the historical correlation between economic size and the production of IP by assuming that the first k countries (1,2,3,...,k) are northern nations favorable to IP regulation while the N − k remainder (k + 1,k + 2,...,N) are southern countries opposing IP regulation. I also assume that the k Northern countries form a core group able to act as a unitary player outside of the GATT. In contrast, the N − k southern countries are disorganized outside the GATT.

Southern countries play last by either accepting or rejecting the new offer. The game tree is shown in Fig. 4.

I explain the payoffs on the left side of the tree. (Right-side payoffs are identical except for the sign of circumstance θ, which is reversed.) The individual payoff for a southern country z that takes the new deal is the weighted sum \(\frac {{\sum }_{z=1}^{k+1}\left (N-z+1\right ) }{ {\sum }_{z=1}^{N}z}\left (y+x_{1}-\theta \right ) \). The fraction is the aggregate market share of countries joining the WTO (the k Northern countries joined by one individual southern country); y is the renewed GATT trade payoff; x₁ is the IP payoff; and θ is Nature’s shock.

Conversely, the payoff for a southern country that refuses the WTO deal and sticks instead to the GATT is equal to \(\frac {{\sum }_{z=k+1}^{N}\left (N-z+1\right ) }{{\sum }_{z=1}^{N}z}\left (y-\theta \right ) +1\). The fraction is the aggregate market share of the countries sticking with the GATT. Variable y is the GATT trade payoff. A developing nation’s reservation value for the IP issue—what it gets by not signing to any IP agreement—is equal to 1.

Payoffs for an industrialized country z are the residuals \(\frac {{\sum }_{z=1}^{k+1}\left (N-z+1\right ) }{{\sum }_{z=1}^{N}z}(1-y+1-x+\theta ) \) for choosing the WTO and \(\frac {{\sum }_{z=k+1}^{N} \left (N-z+1\right ) }{{\sum }_{z=1}^{N}z}\) \(\left (1-y+\theta \right ) \) for sticking to the GATT.

The weights in each payoff illustrate each country’s dependence on their peers’ choices. Together, all countries are playing a tipping game, which is a coordination game with two Nash equilibria: either they all agree to the new package, as they did at the end of the Uruguay Round, or they all agree to stay with the status quo, which they could have also done, but were incentivized not to. This is a key part of the argument. Tipping games can go either way, typically leaving the outcome indeterminate. Basically, there is one pivot country whose choice determines everyone else’s, so the goal of the agenda setter is to persuade the pivot to join its side. North must convince one southern country, any would do, to join the WTO, for if that country were to join, all its peers, which would by definition be getting the same deal, would also join. The Quad, of course, targets as pivot the southern country that is the cheapest to persuade.¹⁵

So far, the setup for the present succession does not look much different from that in the prior regime-shifting game (Section 4), except that it is North and not South that are offering the succession. This time, however, I embed the tipping game within a bargaining game: North offer an x share of the IP pie to South, keeping 1 − x for themselves.¹⁶

The game has only one perfect Nash equilibrium: North respect Nature’s shock under any circumstances (see Technical Appendix for the proof). Why do Northern governments prefer respecting circumstances when these circumstances are bad for them to pursuing a linkage-cum-succession strategy? The answer lies in the inefficiencies that are inherent to all succession strategies: winning the tipping contest requires the k Northern members to match the k + 1^th Southern country’s payoff for the status quo without availing themselves of the scale economies afforded by full membership.

Given the present logic, how to account for the actual event that was dubbed the “WTO coup”? The present analysis suggests that its occurrence reflected a lack of foresight: South did not anticipate that North would resort to a linkage-cum-succession strategy in the 1990s while there were negotiating the Nixon Round in the 1980s. In her historical research, McKibben (2015, chap. 6) found that even in 1986 Southern negotiators expected all the issues of the Uruguay Round to be negotiated in isolation from one another. This myopic equilibrium is easily calculated by solving the game in the first period while assuming that there would be no second period following Nature’s intervention. Hence, if one arbitrarily sets the probability of good circumstances to a half \(\left (p=\frac {1}{2}\right ) ,\) the myopic equilibrium has North offering y^∗ = 0 and South taking it. Assuming this latter value as given but then conjuring up a second period in which “emerges” the option for North to pursue a linkage-cum-succession strategy in the case of bad circumstances for North, then the “coup” narrative starts making sense, delivering a much higher utility for North than doing nothing, and this for relatively low values of k, thanks to North’s disproportionate market power.

On the other hand, the conditions for the stratagem to work are not met in a world of rational expectations. Even if we allow for a surprise move, it would only work once, as players tend to learn from past mistakes.

The researcher’s choice between the two equilibria, the myopic and the rational expectations one, eventually is an empirical issue. On the one hand, one cannot expect too much foresight from governments. On the other hand, our tendency to split interstate bargaining into isolated episodes is bound to miss some of the dynamics inherent to global governance complexes—a key theme of this special issue.

How pervasive is regime shifting among international organizations? It depends. Regime shifting for the strong, as I suggested, is bound to be rare. In contrast, regime shifting for the weak is widespread. As already mentioned, Helfer (2004) invented the concept to account for the normative drift of the TRIPs agreement. A similar story could probably be told about the vast array of issues that the US wanted to throw into the GATT/WTO arena: foreign investment with the TRIMs agreement (Margulis 2013), telecommunications and competition with the GATS (Braithwaite and Drahos2000: 566). Gehring and Faude (2014) applied the concept to the Cartagena Protocol on genetically-modified organisms, while Hofmann (2018: 18) applied it to the relations between NATO and the European Security and Defence Policy (ESDP).

An earlier example of regime shifting, one that spans all three steps, is the interaction between the GATT and the UNCTAD in the 1960s. LDCs established this competitor to the GATT in 1964 within the UN system, where their numerical superiority gave them control of the agenda, to address issues related to trade from a developmental perspective. Although UNCTAD resolutions were not binding, the sideways move was effective in 1966 in helping LDCs renegotiate GATT’s rules on preferential treatment for developing countries (Jupille et al. 2013: 140-159).

Commodity agreements offer a broad field of application for the notion of reneging. International agreements on the production of tin, aluminum, rubber, cocoa, sugar, coffee, jute, wool, bananas, and olive oil, among others, are renegotiated on a regular basis, after cheating undermined the extant deal (Gilbert 1995). Cheating in a cartel is a prototypical form of degrading because, being hard to observe, it is legally unactionable and thus difficult to deter.

Many cases of IO succession are the successful outcome of a regime-shifting strategy. A good example is the wine regime, which until the 1990s was regulated by the International Office of Vine and Wine, referred to in le milieu du vin as OIV, after its predecessor, the Office International du Vin.¹⁷ Dominated by the so-called “old wine countries”, Spain, Italy, Germany, and France in particular, all wedded to the French notion of terroir¹⁸ and its afferent production quotas for upscale wine, OIV failed to anticipate the free-market aspirations of the “new wine countries”, Chile, Argentina, Australia, New Zealand, and, most of all, the United States. Under the latter’s leadership, the new wine countries exploited the antiregulatory spirit of the Uruguay Round to renege on some of their obligations under the OIV (Hannin et al. 2006). They also created the World Wine Trade Group (WWTG) to lobby the WTO and the OIV and also to use as a new forum for its members to sign trade agreements among each other. Faced with the potential degradation of the wine regime and the risk of being displaced by the WTO and the WWTG, in 2001 the old wine countries agreed to renegotiate the OIV into the more market-oriented International Organization of Vine and Wine (still referred to as OIV!) to keep the new countries into the fold. However, France and its allies made only enough concessions to attract the new wine countries that, like Chile, are highly dependent on European markets; the United States, which consumes most of what it produces, has so far stayed out.

Global governance complexes offer their member governments options to improve their bargaining power. One of them is regime shifting. Regime shifting can be analyzed as a bargaining strategy in which a player temporarily undermines another player’s value for an existing deal in order to extract a renegotiation and a more favorable deal. I offered two versions of regime shifting that members of a governance complex can resort to in order to strengthen their bargaining position: one that works for the powerful, the other that works for the weak, with powerful and weak defined as agenda setter and taker respectively. Both strategies involve degrading the other side’s payoff for some initial agreement, thereby giving that side an incentive to take a worse deal. In the strategy that works for the weak, the weak sabotages some of the powerful’s payoff by rendering part of the existing deal unenforceable. In the strategy combining issue linkage with succession, the powerful lowers the weak’s reservation value by denouncing pre-existing deals.

Regime shifting occurs in a layered governance complex, characterized by duplication of substantive domain and membership. Although duplication usually connotes less efficiency, regime shifting levers duplication into flexibility, without which few formal agreements can hope to survive for long in the face of unpredictable future payoffs. Although a more complete assessment of the efficiency-enhancing or -reducing effect of regime shifting remains to be done, the present analysis suggests that regime shifting may strengthen and extend the life of international organizations.

Regime shifting also has a distributive component: it favors the weak more than the powerful. Because powerful countries typically control the agenda and, consequently, are residual claimants, there is little that is left for them to do to improve outcomes. The strategy consisting of combining issue linkage with succession is a possible option, but one with strict boundaries. The powerful must be organized and either numerous or graced with disproportionate market power. These conditions were apparently met in 1994/5, but that was before China joined the WTO in 2001. The strategy must also take the weak by surprise, making it hard to repeat.

In contrast, institutional complexity can benefit the weak. In the days before governance complexity, when each regime was self-contained, the weak’s only leverage came from their capacity to say “no” (Schneider 2011; Mikulaschek 2016) or exercise an outside option, one that was typically not good enough to elicit a good inside offer. With governance complexity, they can now punch above their weight by invoking an inside option and leveraging a payoff that is disproportionate to their outside option. The only condition that must be met to justify regime shifting is undeterrability.

All degrading strategies involve exercising an inside option in the sense that they are temporary disagreements designed to realign players’ values for the status quo with an eye to improving one’s bargaining power in an ongoing negotiation. They are to be contrasted with outside option strategies, which consist in one party improving its payoff by threatening to permanently stop bargaining. Inside option strategies, I believe, are more adapted to the diffuse reality of governance complexes than outside option strategies because, if one were to compare countries to stars and fora to constellations, governance complexes would be black holes, from which nothing escapes.