International Law as Behavior Symposium: Playing the Negotiated Choice Architecture Game – Negative/Positive Listing in Services Trade Agreements
How do negotiators of international treaty regimes engage with the ‘choice architecture’ inherent in the treaties they design? Are they aware of their own susceptibility to cognitive biases and do they take into account the behavioral weaknesses of their counterparts and constituencies? Jean Galbraith’s excellent study on human rights treaty flexibility cast light on this question, very neatly demonstrating on the basis of quantitative research that opt-in/opt-out provisions in treaties have significantly differential effects on subsequent choices. In a paper I am writing with Dr. Shai Moses (a former negotiator and affiliated with the Université de Genève) for a forthcoming handbook on trade in services (edited by Martin Roy and Pierre Sauvé), we explore the behavioral dynamics of negotiated choice architecture in the context of international services trade liberalization, and in particular in the ongoing negotiations towards a plurilateral Trade in Services Agreement (TiSA). Here are some of our initial observations on what seems to be going on in Geneva, from a behavioral perspective.
Negative/Positive Listing in Services Liberalization: The Rational Choice Puzzle
Since the advent of the international liberalization of trade in services three decades ago, an issue of continued systemic interest has been the question of positive listing versus negative listing of liberalization commitments. Generally, in a positive list approach, no commitments to liberalize are undertaken by a state unless explicitly inscribed in a schedule of commitments, and to the extent of the inscription. This was the overarching method adopted in the GATS with respect to specific commitments and in subsequent negotiations during the Doha Round. In contrast, in a full-blown negative list approach, all services sectors and disciplines are absolutely committed unless expressly excluded, to the extent of the exclusion. This is often associated with the model pursued in the North American Free Trade Agreement (NAFTA). To be sure, as services experts have correctly noted, the line between negative and positive listing is not as bright as just described: in GATS, sectors and modes of service distribution are indeed ‘positively’ listed in GATS schedules, but the content of schedules is ‘negative’ – as derogations from full liberalization of market access and/or national treatment in those listed sectors. Nevertheless, the starting point for commitment is different in each approach.
Why has this issue garnered so much attention, when a standard rational choice perspective would tell us that the positive/negative listing distinction is immaterial (subject to transaction cost issues, that I discuss below)? From a technical-legal perspective, the approaches are quite the same. After all, in principle, the same level of commitment to liberalization can be formulated in either positive or negative terms. If we begin our theoretical analysis within a standard rational choice, contract theory framework, in the context of trade in services, states will make credible commitments to liberalize international trade in a given service (operationalized primarily through the disciplines of market access, national treatment and most favored nation treatment (MFN)) only if the expected benefits exceed the expected costs. The costs of a concession can be understood as a combination of (a) the economic costs of the concession (e.g., loss of market share by domestic service providers); and (b) constraints on regulatory autonomy. The benefits of a concession derive from reciprocal – but not necessarily symmetrical – concessions made by trade partners, benefiting services exporters. In Coasean economics, in the absence of transaction costs, the initial assignment of resources or entitlements is not determinative of their ultimate allocation. Cost-benefit bargaining between rational actors will assign the entitlements efficiently to the actor to whom they hold the most value, irrespective of the starting point. Thus, from a Coasean bargaining perspective, the outcome of negotiations should reflect an aggregation of the preferences of the participating states as represented by negotiators, regardless of the point of departure of discussions.
In services trade terms, it should not make a difference if the starting point is zero-commitment to liberalize (or an existing level of commitment), gradually negotiated towards an enhanced list of commitments (positive list), or, conversely, a full commitment to liberalize, whittled away by the derogation of sectors, modes of supply and/or disciplines.
A Behavioral Approach to Services Liberalization and the TiSA
Anyone who has been following the TiSA negotiations (or negotiations on services liberalization in the CETA, TPP or TTIP) – not an easy task given their overall transparency deficit – knows that discussions about the design of commitments within the negative versus positive listing framework (‘modalities’) are extremely intense, often taking up lots of negotiation time, even more than sectoral negotiations. In a Coasean reality this wouldn’t make much sense, unless the transaction costs of adopting one approach over another were significant (and there are transaction costs – e.g., the US is comfortable with a negative list approach because of the NAFTA, whereas the production of a negative list would be costly for the EU, which has advocated the need to utilize the GATS model in TiSA). Turning to behavioral theory might provide a better explanation for the dominance of the listing debate.
Behavioral economics (whose potential implications for international law were discussed in a previous series of posts) challenges the assumptions underlying neoclassical economics and rational choice theory, arguing that human rationality is in fact bounded and susceptible to systematic divergences from perfect rationality. In particular, Prospect Theory questions the validity of the Coase Theorem, Experimental observations have shown repeatedly, in varying circumstances, that initial assignments of entitlements do matter, significantly, in that they influence actor’s decisions, in particular, their willingness to part with their entitlement. In Coasean terms, 10$ (or any other thing of value – such as unfettered rights to restrict international trade in services) have equal worth, whether gained or lost. Yet in real life, people do not regard losses and gains of equal size indifferently. For example, they will often invest more in the prevention of loss than in the generation of gains of the same amount.
Returning to services liberalization, negotiators intent on making as few liberalization commitments as possible (while gaining as much access to foreign markets as possible (Bagwell and Staiger, 2002: 59-62)), are apt to consider specific commitments made on a clean slate (or vis-à-vis a pre-existing level of commitment, as the case may be in ongoing liberalization talks) (ie., positive listing) as a loss, much as the very term ‘concession’ implies. In contrast, a reservation made to a general commitment to liberalize (i.e., negative listing) would be considered a gain. Thus, even if the liberalization outcome were identical, negotiators might tend to evaluate the positive listing of a specific commitment as a loss greater in size and value than the gain attained through the equivalent negative listing. This in itself suggests that positive listing is less conducive than negative listing to the collective negotiation goal of increased liberalization – a position held by some of the TiSA negotiators. The claim is hard to substantiate empirically, given the relatively small pool of services agreements outside the GATS, and several complicating factors. But the perception in this case may be more influential than facts. In any case, we consider that certain cognitive biases – and more importantly, as I will explain below, perceptions of cognitive biases may be in play.
First, framing effects – the way treaty options are presented is clearly a significant negotiating issue. At one level of analysis, as an extension of the prospect theory hypothesis above, negotiators may be less willing to make a commitment that is framed as a loss (positive list) than as a gain (negative list). Indeed, framing is part and parcel of the willingness to ratify, although the question of how it empirically affect willingness to liberalize at the outset and later on is still open. At another level of analysis, however, negotiators may have strong intuitions about the effects of status quo bias – the experimentally proven phenomenon whereby, all things considered, people actually have a tendency to prefer an existing state of affairs over alternatives that might leave them better off. Status quo bias in contracts translates into the finding that default rules matter. This means that a positive list cannot be expected to grow over time. Indeed, the main concern of TiSA negotiators has been that the architecture question will pre-determine the level of ambition and the final outcome – one way or another. There is a lot more to be said here – but that will be left for the full paper.
The Dual Role of Services Trade Negotiators: Choice Architects, Choice Inhabitants
It bears mentioning that ‘choice architecture’ and the associated concept of ‘nudging’ as applied to individuals in society, has come under fire on ethical and political grounds, certainly with respect to its legal and social implications, as an expression of libertarian paternalism. However, the choice architecture inherent in services trade agreements regarding the positive versus negative listing question (and indeed, in other international regimes), is different in one important respect. Regular choice architecture is imposed on individuals operating within its environment by an external actor, such as government ministries or marketing agencies – pushing people to make certain choices regarding their health, lifestyle, purchases, or civic participation (such as payment of taxes). In contrast, the choice architecture of a trade in services agreement is negotiated on behalf of the same states that will subsequently be its subjects; and by negotiators who are, institutionally if not individually (as they often are), the same negotiators who will be tasked with making the decisions within the choice architecture conditions that they agree upon. In other words, negotiators are not only choice architects. They will also have to inhabit the choice architecture they negotiate.
This not only neutralizes, or at least mitigates, the problem of paternalism (while possibly raising questions of democratic legitimacy) but suggests an almost dialectic behavioral dimension of negotiations. Trade negotiators are engaged in a process whose chief purposes are forward looking – hand tying regarding future domestic policies and the regulation of ongoing liberalization. When negotiators deliberate the choice design of a services liberalization agreement, behavioral phenomena come into play in two distinct ways. First, the cognitive biases of negotiators may subliminally influence their own initial preferences. Second, negotiators may speculate about the ways in which a particular choice architecture will impact upon future decisions – their own, and those of their successors and counterparts. Put differently, if a negotiator knows, thinks, or thinks that she knows, that a certain choice architecture (such as negative or positive listing) will either promote or hinder certain subsequent choices and outcomes, she will push for the adoption of that architecture.
The observation that international treaty negotiators are not only themselves subject to cognitive biases, but are also acutely aware of the role that behavioral choice architecture can play in the success of the regime (whether on an intuitive or informed basis) is hardly clear cut and parsimonious. There are many complicating factors in trade negotiations – transaction costs, adjustment costs, negotiating asymmetries, multilateral/regional interactions, the services/investment nexus and more – that make this difficult to distill much further on a field study basis (lab-work with elite decision-makers, or semi-structured interviews with negotiators (for an excellent example of such qualitative analysis of other issues in the WTO, see Sivan Shlomo Agon, Clearing the Smoke: The Legitimation of Judicial Power at the WTO, 49(4) JOURNAL OF WORLD TRADE (forthcoming, August 2015), may be potentially more rewarding in this regard). The negotiation dynamics of the TiSA are, however, indicative of the role behavioral considerations play even in a very messy setting. The results of the TiSA ‘choice architecture’ (to the extent that it has been finalized) reflects this complexity. Negotiators have agreed on a hybrid approach including (i) a horizontal commitment for national treatment, subject to appropriate exceptions and reservations (a negative list approach pushing parties to avoid discriminatory measures in services regulation in all sectors); and (ii) a GATS approach to listing commitments for market access (a positive list approach more geared towards letting parties avoid making new commitments in unliberalized sectors). Whether the same outcome would have been achieved in a perfected Coasean framework, we cannot tell; what is certain, however, that it is a compromise between negotiators who were playing the behavioral choice architecture game.
*On another note, it is a real pleasure contributing to this online symposium, soon after OJ’s 10th anniversary, not least because I played a small role in what was one of the first OJ online symposia, in 2006.