The Market for Private Dispute Resolution Services--An Empirical Re-Assessment of ICANN-UDRP Performance

The impressive growth of the Internet in the 1990s and the boom of the e-economy generated a competition for domain names in the most coveted of the top level domain names, i.e., the .com space. The other original generic top-level domain names (gTLDs) open to commercial use, .org, and .net, were also in demand from businesses. Other types of top-level domain names, especially the country code TLDs (ccTLDs), were of little commercial value, and their registrations were not as important as the gTLDs. In 1997, partly because of the expansion of the Internet to the international sphere, the U.S. government delegated the management of numbers and names of the Internet to a non-profit corporation based in California, the Internet Corporation for Assigned Names and Numbers (ICANN). Even though ICANN is not the only organization managing Internet names and addresses, it is enormously influential in fashioning the actual structure of the Internet. The relevance and power of ICANN in enacting new policies for the Internet are based on two main characteristics—its monopoly of the main Domain Name System on the Internet and the lack of technological compatibility between competing Domain Name Systems which has prevented other private firms from competing with ICANN. One of the main problems in the medium term on the Net was the creation of a system to handle the growing number of problems among users because of the, sometimes indiscriminate, registration of domain names that collided with already established trademarks in real life markets. These disputes grew at the same pace as Internet commerce boomed in the late nineties. The usual mechanism to solve these kinds of disputes, i.e., courts, were handicapped to handle cases in which parties came from different jurisdictions and laws, enforcement of court judgments was weak, and court procedures were slow and expensive. One of the main tasks of ICANN was to provide a fast and inexpensive system to resolve domain name disputes. In 1999, after a series of consultations with many interest groups, ICANN created the Uniform Dispute Resolution Policy (UDRP), a decentralized regime for dispute resolution in which ICANN created the general rules and a series of competing private providers were authorized to manage and resolve disputes. ICANN, because of its role as the only manager of the domain name system, could exert almost perfect enforcement of the UDRP providers’ decisions. Nonetheless, after a few years, harsh criticism has been leveled at ICANN/UDRP from scholars and commentators. Most of the empirical studies of the UDRP have been based on the analysis of cases handled by providers and the results of the panels’ decisions. The most common critiques are that the UDRP providers are biased, that they have an incentive to favor complainants and trademark owners, and that the rules have been designed to favor proprietary interests on the Internet. Some of these facts are a direct consequence of the political structure of ICANN, which we (and others) have analyzed elsewhere. In this paper, we present a thorough empirical study of the performance of UDRP providers. We analyze the decisions of the complainants in deciding to send their claims to particular UDRP providers. Using a multinomial logit regression model to determine whether complainants select the provider based on bias or duration of the dispute resolution procedure, we show that duration is at least as important as bias in the selection of providers. This is a key finding since our results show that the emphasis of theoretical and empirical work, which has been exclusively concentrated around the effects of bias, is misplaced. As we demonstrate, more attention should be paid to other performance and efficiency indicators, such as those proposed in this paper. We identify the duration of the domain name disputes, i.e., the entire period of time taken by panels to decide these cases is one of the main variables that determines the efficiency of the dispute resolution system and influences the choice of UDRP provider by complainants. From our empirical results, we use the duration of cases as the variable to measure the general efficiency of each UDRP provider. Hence, our study goes beyond the empirical questions regarding the final results of the cases by looking at the actual performance of providers. Among our main findings, we claim that the UDRP providers have different duration functions, implying a different process or technology in treating cases, which sets up the existence of forum shopping. The existence of forum shopping based on the performance of providers is different from the forum shopping mentioned in the UDRP literature which is based on the bias of the providers towards complainants. Second, the providers have an unambiguous bias for specific countries. This finding is very important because most of the literature discusses the bias between individuals. Nonetheless, the bias towards countries of origin of the UDRP providers could be an important element to take into account in the design of a general dispute resolution system such as the UDRP. Furthermore, the evidence of such a bias delivers a hard blow to ICANN’s claim that the system is intended to handle the most diverse claims on the Internet regardless of the countries of origin of the parties. Third, we also find that some panelists have completely different duration times for deciding cases, as compared to the rest of the cases under any private provider. That said, the structural differences among providers can have an influence on the performance of the judges. This evidence regarding some panelists, however, calls into question the actual system by which providers assign cases to panelists in the sense that the selection of the panelist is not an innocuous decision in terms of efficiency. Fourth, the performance of the providers is affected by the proofs presented by complainants and respondents. This is an indication that decisions are based on the proofs presented according to the rules of the UDRP. Finally, we find that three member panels are just as efficient as single member panels. Accordingly, a change to a general three member panel system could be beneficial in terms of fairness, without having a negative impact on efficiency