Excerpted From: Andrea Cann Chandrasekher, An Empirical Investigation of Arbitrator Race and Gender in U.S. Arbitration, 35 Yale Journal of Law & Feminism 159 (2024) (124 Footnotes) (Full Document)


AndreaChandrasekherThe need for greater diversity in mandatory arbitration has recently received tremendous public attention. Although, to be fair, mandatory arbitration generally has been subject to public scrutiny for some time now, with critics arguing that it is a rigged system. Consumer protection activists assert that millions of consumers, in their everyday purchases of cell phones, groceries, electronics and other consumer goods, regularly forfeit their right to have any future disputes addressed in court, often without their knowledge. These mandatory arbitration clauses are hidden deep in the fine print of the purchase agreements that consumers implicitly assent to, just by simply making a purchase. At the same time, employee rights groups report that over 60 million American employees are similarly constrained when they, as a term of being hired, must agree to have all future job-related disputes resolved in front of a single arbitrator who is paid by their employer, rather than by a judge or jury of their peers. The mandatory arbitration clauses in employment contracts are typically part of an employee handbook or manual, and as such, rarely come to the attention of a new employee until a dispute arises. According to critics, these arbitration clauses “force” millions of average Americans into an alternative dispute resolution system where adherence to the rules of civil procedure and evidence is not required; where class action is often not allowed; and where the arbitrator's decision is final, binding on all involved parties, and not subject to appeal.

Of late, forced arbitration has come under even greater scrutiny as high-profile cases and national news stories have shined a spotlight on the lack of racial and gender diversity amongst U.S. arbitrators--making an already controversial system seem even more illegitimate. In late 2018, entertainment mogul Shawn Carter, known popularly as Jay-Z, successfully halted an arbitration proceeding before the American Arbitration Association (“AAA”), the nation's largest arbitration provider, on the grounds that the list of AAA arbitrators that he was given to choose from, included only two Black arbitrators out of more than 200 candidates. Mr. Carter argued that “the AAA lacks any meaningful selection of African-American arbitrators who specialize in complex commercial disputes.” As a result, he had no “meaningful opportunity to have [his] claims adjudicated by a neutral decision maker who reflects [his] background and experience,” which denies him “equal protection of the laws” and also violates “New York public policy against discrimination.” Then, in June of 2020, the National Rifle Association (“NRA”) sued the nation's second largest private arbitration firm, Judicial Arbitration and Mediation Services (“JAMS”), to recover fees from an arbitration proceeding that went off the rails after it was discovered that the presiding arbitrator had ties to a white supremacist organization. A year later, in June of 2021, the American Association of Justice (“AAJ”) released a study on diversity in arbitration, later covered by CNBC national news, that concluded that arbitration is a place where “white men rule.”

To be clear, arbitration providers are also strongly in favor of diversifying the field of arbitration. Specifically, both AAA and JAMS have recently taken significant steps towards recruiting more Black, Indigenous, and people of color (BIPOC) and women arbitrators, and are making institutional changes designed to improve the odds that diverse arbitrators will be selected by the parties in arbitration. Still, even though there is a clear consensus from all interested parties that arbitration needs more diversity, as a practical matter, there is very little empirical guidance about where and how to begin in terms of fixing this problem. Indeed, the problem is multidimensional. One part of the issue concerns labor supply (i.e.--the field lacks a sufficient number of diverse neutrals), and so recruitment of larger numbers of diverse neutrals is necessary. At the same time, labor demand issues also exist (i.e.-- even when diverse neutrals are available, they are not hired by the parties). Because of implicit bias and/or a desire to avoid new arbitrators whose reputations have not yet been established, litigants (or more accurately, their lawyers) are less likely to choose diverse neutrals. In response, some scholars have argued that institutional changes to the arbitration selection procedures are needed.

To shed light on these issues, this paper uses an originally-collected data set of arbitrator race, ethnicity and gender from JAMS and AAA. The data were collected using public data sources and cutting-edge machine learning techniques. This is the first-ever scholarly effort to empirically estimate the race and ethnicity of arbitrators for both the JAMS and AAA populations. Equipped with these data, this paper investigates two commonly discussed issues in the recent conversations about diversity in arbitration: labor supply issues and labor demand issues.

First, on the labor supply side, this study seeks to comprehensively estimate the statistical profile of the U.S. arbitrator population. While almost all academic and news-related articles on the topic mention the low diversity in arbitration, no paper has sought to quantitatively measure what the population of arbitrators looks like, both along the race/ethnicity and gender dimensions, as well as along other important dimensions (age, education, judicial experience, etc.). Relatedly, this paper seeks to compare the demographic profile of arbitrators with the demographic profile of the U.S. population, as well as the population of other jurists such as lawyers, state judges and federal judges. Much of the discussion concerning the underrepresentation of diverse arbitrators has proceeded without the guidance of a benchmark for what we would expect the level of diversity to be in the absence of discrimination. This comparative analysis seeks to fill that void.

Moving on to labor demand issues, the paper's second objective is to quantitatively assess whether and to what extent diverse neutrals are systematically under-selected by parties for their arbitration matters. Unlike in court cases, parties to arbitration have the unique ability to choose the neutral that will preside over their dispute. If the parties have a preference for white male arbitrators, either because of discrimination or because they are hesitant to put their matter in the hands of a relatively unknown neutral, then supply-side increases in diversity will not be impactful. To investigate this issue, this study not only collected detailed arbitrator demographic records, but then took the next step of matching these arbitrator records with publicly available arbitration case records by arbitrator name. This matching allows for a demographic analysis of which categories of arbitrators are selected and at what rates.

The analysis in this paper yielded the following findings. First, on supply issues, I find that the population of JAMS and AAA arbitrators is overwhelmingly white and male, and the extent of white-male overrepresentation is greater than what the JAMS and AAA aggregate demographic surveys report. Moreover, my comparative analysis suggests that the population of arbitrators is less diverse than the population of other American jurists (lawyers, federal judges and state judges) and substantially less diverse than the general U.S. population. Women and people of color are both underrepresented relative to these benchmarks, but people of color are underrepresented to a greater extent.

Second, on demand issues, I study selection along two different dimensions. Specifically, I study whether or not a diverse arbitrator is ever selected to arbitrate from the arbitration roster within my five-year sample period (2015-2019) (the ever-selected dimension) and, conditional on ever being selected, I study the frequency with which a diverse arbitrator is selected to arbitrate (the frequency-of-play dimension). Furthermore, I assume that, in the absence of discrimination, all sub-groups of diverse arbitrators would be selected in proportion to their roster share (for the ever-selected dimension) or their expected share conditional on ever being selected (for the frequency-of-play dimension). For both JAMS and AAA, on the ever-selected dimension, I find no evidence of under-selection. All groups of diverse arbitrators studied (Asian, Black, Hispanic, and female arbitrators) are selected at a rate that is proportional to their (very low) representation on the arbitrator roster. Second, along the frequency of play dimension, I find different results for JAMS and AAA. For JAMS, I find that, conditional on being selected to arbitrate at least once in the sample period, Asian and Black arbitrators receive fewer cases than their proportional share, and female arbitrators receive slightly more cases than their proportional share. Moreover, arbitrators that were formerly judges receive more cases than their proportional share. For AAA, the selection analysis is hampered by limited data availability. However, the data that I do have suggest that diverse neutrals are selected for cases at a rate that is at or above their proportional share.

Given the first two results, my data suggest that diversity issues exist both along the labor supply dimension and the labor demand dimension within U.S. arbitration.

The paper is organized as follows: Section II presents a primer on U.S. domestic arbitration--what arbitration is, how pervasive arbitration is within the lives of average Americans, how it works and the dominance of AAA and JAMS within the market of arbitral institutions. Section III discusses the importance of arbitrator diversity, while Section IV summarizes the related literature. Section V introduces the originally-collected data sources. Section VI presents the empirical analysis. Section VII concludes. The tables and figures are provided in Section VIII.

[. . .]

Relying on publicly available arbitrator profiles, court documents, artificial intelligence machine learning algorithms, and publicly available records from forced arbitrations, this article presents an original data set of AAA and JAMS arbitration cases and individual-level arbitrator demographic characteristics, including gender, judge status, and--for the first time ever-- race and ethnicity. Equipped with these novel data, this study has four main findings.

First, with respect to the diversity of the AAA and JAMS rosters, this study suggests that women and people of color are both underrepresented relative to all important benchmarks (the U.S. population, and the populations of American lawyers and judges), but people of color are underrepresented to a greater extent.

Second, the analysis suggests that selection issues exist but only along the frequency of play dimension. For both JAMS and AAA, I found no evidence of statistically significant under-selection along the ever-selected dimension for any group of diverse arbitrators. Of course, this is a low bar, requiring only that the proportion of diverse arbitrators that arbitrate at least one time in the five-year period is equal to that group's roster share. When one considers that the roster share for diverse neutrals is already low, parity along the ever-selected dimension is even less satisfying. Conditional on being ever-selected, the frequency of play analysis asks how often a group of arbitrators gets to play given that they were selected to play at least one time in the sample period. Here, I find some differences between JAMS and AAA.

For JAMS, I find that Asian and Black arbitrators receive fewer cases than what one would expect given their share of the selected arbitrator pool. Over the five-year sample period, the analysis suggests that, if cases were randomly assigned to JAMS arbitrators that were selected at least once, each Black arbitrator would have received three more cases and each Asian arbitrator would have received seven more cases. Of course, it is possible that nondiscriminatory reasons could explain these differences, but the existence of the differences does merit further explanation. One interesting finding is that Hispanic, white male and white female former judges are all over-selected relative to their expected rates, but Black former judges are not, despite having the most judicial experience. A possible explanation for this finding is that Black former judges have less JAMS tenure than other groups of former judges. To the extent that this relative under-selection is due to networking or connections, campaigns to promote and advertise these highly qualified arbitrators would seem to help not just these arbitrators but also the parties that could benefit from their services. More generally, as the diversity of the JAMS arbitrator roster (hopefully) increases over time, it seems that the challenge will be to increase the frequency of play for new diverse arbitrators.

Unlike JAMS, AAA does not publish public web profiles for its arbitrators, so what we can learn about selection dynamics is necessarily limited. For the ever-selected dimension, only Black arbitrators can be studied. I find slight over-selection of Black arbitrators (2.96% selection rate vs 2.76% roster rate--a 7% difference). As previously mentioned, though, over-selection relative to the very low roster rate of 2.76% is a low bar to meet, especially when considering that the only requirement is that a Black arbitrator be chosen at least once in the five-year sample period. For the frequency of play dimension, I am able to study dynamics for all diverse groups of neutrals, not just Black arbitrators. I find that some categories of diverse arbitrators (Asians, Blacks, and women) are over-selected along the frequency of play dimension. Future work should investigate parity between diverse arbitrators and white male arbitrators along other important dimensions besides selection, such as compensation. For example, I find that AAA Black and female arbitrators claim a large share of AAA employment cases (41%). White males arbitrators, on the other hand, while still claiming the majority of employment cases, are slightly under-selected relative to their expected share. What is unclear, however, is whether Black and female arbitrators are compensated for their arbitration services at the same rate as their white male counterparts. Certainly, true parity requires not just proportional selection but also parity in compensation.

Third, given the first two results, my data suggest that diversity issues exist both along the labor supply dimension and the labor demand dimension within U.S. arbitration.

Fourth, finally, and perhaps most importantly, this study finds that future empirical work on diversity in arbitration will be severely hindered unless more and better data are available to researchers

Professor of Law and Martin Luther King Research Scholar, UC Davis School of Law. PhD (Economics), UC Berkeley. MA (Statistics), UC Berkeley. JD, Stanford Law School.