“When Matching Markets Unravel? Theory and Evidence from Federal Judicial Clerkships”, 2019-07-01 (; similar):
We study the judge-clerk match, a market plagued by ‘unraveling’. Evidence from an unique dataset on match and production shows that (1) agents on either side have similar preferences over those on the other side, (2) the matching game for judges is close to zero-sum, (3) this fierce competition among judges explains the unraveling in this market.
We develop a theoretical model investigating how homogeneity of preferences (and competition) affects unraveling in matching markets. We show that a static mechanism, as proposed in many previous reforms, is impossible to solve the problem of unraveling in a market with a high degree of homogeneity. By contrast, a dynamic mechanism that takes advantage of judges’ repeated participation in the market over time is proven promising.
Based on our findings, we propose a new market design for the judge-clerk match.
…A law clerk assists judges on a range of tasks including researching issues, drafting opinions, and making legal determinations. Most law clerks are recent graduates who performed near the top of their class in law school. The positions are highly sought after as they can lead to professional opportunities. Some federal judges receive thousands of applications for a single position and even the least sought-after clerkship will receive over 150 applications. Each judge presently hires 4 clerks for a year, which leads us to a many-to-one matching problem: There are roughly 167 judges (similar to firms), each of whom is hiring 4 law students on a one-year contract from a much larger pool of candidates. The matching can be considered as a non-transferable utility problem because each clerk receives fixed salary.
While the National Federal Judges Law Clerk Hiring Plan recommends when judges may receive applications and when they may contact, interview, and hire clerks, generally many do not follow this schedule and hire law students quite early, in some time periods, as early as right after the first year of law school. Due to extreme competition, by judges to get the best candidates and by candidates to get the best judges, sometimes judges can require a candidate provide an answer to the question, “Will you accept an offer?” prior to scheduling an interview. It goes without saying that job offers are expected to be accepted on the spot. To defer would be a sign of disrespect that can stigmatize the year-long relationship.
Several failed reforms have been attempted to regulate the earliest date at which law students could be hired. The market promptly unraveled in each of these prior reforms, in 1983, 1986, 1990, and 2005. While the reforms varied in their specific implementation, they generally had a deadline like “no job offers, tentative or final, shall be made to law clerk applicants before May 1st of the applicant’s second year” or “judges should not consider applications before September 15 of the students’ third year of law school.” These failures have sparked an active theoretical and experimental literature (for example, Avery et al 200123ya, Avery et al 200717ya; Fréchette et al 200717ya). This literature observes that some Circuits (Fifth, Seventh, and Eleventh) were noted to “cheat” in the reform years.