I want to start by thanking each of the commentators for their kind, forgiving and thought-provoking comments on my article, and by recognizing that the work of Dapha, Michael and Tom set the foundations and served as the inspiration for this work. I will organize my responses into three different clusters: methodology, extensions and other, more general comments.
I will first address Tom’s comments on the record of appointment as a proxy for displayed preferences. Can we ignore the fact that an appointment does not necessarily indicate that the arbitrator was the appointer’s first choice? I like to think of this question as a job offer to a candidate who decides not to accept an offer. The truth may be that certain candidates are in high demand and that many companies are offering jobs to the same well-qualified candidates. Hence it is possible that parties who nominate arbitrators and may be trying to enlarge the diversity of the pool of arbitrators in ICSID may be somewhat restricted by the broader market of arbitration professionals. This is true especially where, as Michael points out (and as confirmed by arbitrators during my interviews), the compensation provided by ICSID is lower than that of other arbitration opportunities and venues. This is a limitation of the assumptions that can be made from the ICSID appointments. We are left with only the record of appointments as a second-best proxy.
A second point raised by Tom as to why the mid-2000s witnessed the first appointments of today’s power-brokers is also relevant. We must take into consideration that these years saw a boom of investor-state cases and, hence, more appointments. It is also true that this is when we started seeing more awards discussed publically. Although I did not report these statistics in this paper, measures that reflect the connectivity of the network start stabilizing and changing in less intense ways during this period. Thus I’m not surprised that the network acquired self-organizing qualities during this time. One plausible hypothesis for future exploration is how the development of transparency rules may have helped cement public knowledge of decisions and with that, some of the insights that come with understanding the decision-making philosophy of arbitrators.
On the methodology of determining the average compensation per arbitration of US$200,000, I admit that the number is not incredibly precise. However, it is the best approximation I could divine. I followed a complex, yet not foolproof strategy to arrive at what I consider a very rough average. First, I computed the amount in 2010 dollars from roughly 70 awards that provided information on compensation in my sample of ICSID and ICSID Additional Facility awards. I divided this amount by the months of duration of each case and averaged the result. Using the dollars per month estimate, I calculated the possible amount paid to an arbitrator on settled or dismissed cases based on an average duration of the case. I confirmed that this rough estimate was in the right ballpark with one scholar who has done extensive empirical work with ICSID data as well as through my interviews with arbitrators. It is not an exact appraisal, but what is relevant for the purpose of the article is whether ICSID’s rate completely changes the incentives to accept an appointment and hence whether the distribution of ICSID appointments is completely different from that of other institutions. We cannot know the answer to this question for certain, but with a grain of salt, my educated guess (taking into account what I gleaned from working at ICSID) is that it does not. Most appointments are accepted and most rejections occur when a conflict exists. In my experience it was only in very few cases that, for other reasons (such as being too busy) an arbitrator decided not to accept an appointment.
The extensions of this work may be limited and Michael’s suggestion of taking the challenge of testing Anthea’s theory using network analysis is right-on. Another possible area for future exploration is to better understand how the social structure of the arbitration bar influences outcomes. Daphna has written on this and her post provides a good explanation of the mechanism in which the characteristics of the social structure (core-periphery) may affect outcomes. I believe her explanation is persuasive, but in this article I did not want to get too much into the complicated mechanism of how exactly the interactions may structure outcomes. My article goes as far as to acknowledge that, based on the appointment records, the members at the core are unlikely to escape the observation of other members of the core, but may remain insulated from outside influence. These two insights suggest that piercing the core is very difficult and requires strategic behavior, including voting, acceptance and appointing strategies, and that dissenting can be very costly.
This leads to another point raised by Tom. Is it a bad thing that most cases end-up being decided by a core of arbitrators? My answer to this is that, on a routine basis, no. In fact, I agree with Tom’ argument that this is a characteristic of well-functioning legal systems. Most cases are argued, tried and decided by a limited number of actors, and we should probably not expect or desire the opposite. In fact, a remarkable consequence of the system is that the self-organizing tendencies that result from the highly interconnected core provide much needed stability and predictability, as the same arbitrators are appointed again and again. However, this also may create some problems when arbitrators also act as counsel, including conflicts of interest and intense strategic voting. In other words, it is not necessarily bad if some of the mechanisms for transparent third-party decision-making of public law disputes are present. Certainly some of these mechanisms may still be lacking in ICSID procedures.
This leads to another point made by Tom: if “the evil that we know is best”, does this imply that when choosing an arbitrator it does not matter whether they are good or bad lawyers, how well they know investment law, how much experience they have in arbitration, etc.? My answer to this is also no. Litigants and the appointing authority value all these qualities, I’m sure. However, in exploring their choices, appointers are restricted by two important variables: 1) the limited actual information about many potential arbitrators; and 2) the presence of biases that nudge the appointment in a particular direction. For instance, imagine two potential arbitrators with similar qualifications, backgrounds, expertise and positive reputations: would the appointers choose the candidate with 2 or the candidate with 10 appointments? Everything else being equal, I argue based on the evidence that the number of appointments serves as a signaling device in a system where very few final decisions are public. The work of scholars like Susan Franck, Anne Van Aaken and Tomer Broder who are exploring how biases work in international law may provide the foundations to study these phenomena and serve as the basis for future research.
Finally, I will address some additional general comments. First, Michael raises an important question: How important are the initial conditions and timing of the emergence of particular individuals as super arbitrators? I agree with his observation that it is likely that there is considerable unpredictability in these appointments. However, timing may be a relevant factor, and currently it is more difficult for new entrants to join the super-arbitrator echelon than it was in the past. How much more difficult is an empirical question that is difficult to assess, but the probability of the appointment of a new entrant is now 1 in 30 appointments!
Michael also raises another important question, one that I’m currently working on: would the same individuals be at the core of the arbitration network if we went back in time and started from scratch? The answer is impossible to determine. What we can do, however, is create a mathematical model for the formation and emergence of the network as it exists now. Using such a model, we can see how the network could look in 10, 20 or 50 years or how the network would look if, for instance, gender notions had been introduced earlier in time. My bet is that if we were to run the model with, let’s say, four more women in the 70’s the gender distribution today would be much closer to even!
Finally, I would agree with Tom that arbitrators at the core are powerful and influential. They have the power to decide specific legal issues and interpret treaty rules. Arbitrators are often unsupervised by any superior entity, and their decisions bind states and affect their citizens. They have the task of deciding important matters in a range of potential topics. Hence, I do think the core group of arbitrators has a special responsibility, and I believe many, if not most, members of ICSID’s arbitration bar are conscious of this power. Many have endorsed transparency initiatives such as the UNICTRAL Convention, the new arbitration intelligence resource and/or the adoption of best practices. I do not believe arbitrators are the villains of this movie. To the contrary, a few have been the superheroes that have saved ICSID from recurrent legitimacy crises (as I’ve argued and documented here).