The Department of Justice’s plan to evaluate immigration judges based on how many cases they close and how fast they hear them might seem like a good approach. The backlog in immigration courts has reached about 700,000 cases, and the average wait time for an immigration case is just more than 700 days. The roughly 375 immigration judges complete 678 cases each per year on average, so the backlog would take roughly 2.6 years to clear even if no new cases were added. The new rules, which were first reported by the Wall Street Journal and are scheduled to kick in on Oct. 1, aim to speed things up by requiring judges to clear 700 cases per year to receive a rating of “satisfactory.” Those who complete fewer than 700 but more than 560 cases would be rated “needs improvement,” while those that complete fewer than 560 cases would be “unsatisfactory.” The DOJ has not yet said what would happen to a judge deemed “unsatisfactory.”
Immigration judges have much larger caseloads than federal district court judges, who operate under very different rules. Still, our research on incentives for federal district court judges suggests that performance-based metrics are inadvisable. When judges are incentivized to work quickly, they are more likely to make errors. This is true even for district court judges with life tenure, whose only penalty for moving slowly is looking bad in front of their peers. If immigration judges’ jobs or promotions are on the line, demanding faster results will likely cause their decisions to suffer even more.
The only incentive we know of that is imposed on federal judges is called the “Six Month List,” a twice-a-year rundown of pending motions. As one retired judge put it, “[N]o judge likes being on this ‘report of shame.’ ” The list, mandated by the Civil Justice Reform Act of 1990, was designed with this kind of shaming in mind—its express purpose was to use peer pressure as an incentive to get judges to make decisions faster.
In recent empirical work, we show that the list has dramatic—and largely negative—effects on judicial behavior. In the last weeks of March and September, when judges and their clerks work overtime to resolve pending motions before the list is compiled, case and motion closures spike by more than 25 percent. We tested the differences between motions filed in August (and that must be decided by March to avoid the “report of shame”) and those filed only a few days later in September (and therefore not “due” for an entire year). We found that judges decided motions that were “due” in March quickly, but waited until the deadline to decide motions due in September. While this is normal for procrastinators, it’s bad for justice, as just a few days’ delay in filing can postpone a decision for months.
Our results reveal that the list does have a modest effect on judicial workflow, reducing the time it takes a judge to decide a motion by between 10 and 40 days. (Our best guess is that the lower estimate is probably the most accurate.) But the list also has significant costs. The most worrisome of these is evidence that it affects case outcomes by increasing the risk of error. Such errors could include wrongly admitting or excluding evidence, ruling incorrectly on substantive motions, misapplying discovery rules, or making legally incorrect decisions. While mistakes are very difficult to identify directly, list-week decisions are 40 percent more likely than non-list-week decisions to be remanded by an appellate court, suggesting a higher error rate (albeit from a very low baseline, because there are only about 300 remands each year).
Our unfavorable assessment of judicial incentives finds theoretical support in the work of economics Nobel laureate Bengt Holmström. Holmström showed that if some aspects of performance are not measurable (here, the quality and fairness of decisions) and so cannot be incentivized, it will often be best to avoid incentives altogether. Incentives focus attention on the aspects of work that are being measured and discourage effort spent on those aspects that aren’t measured or rewarded. (That’s why NBA teams don’t offer players bonuses for scoring more points, which would incentivize excessive shooting.)
Of course we want judges to resolve cases quickly; we also want them to be fair, and we want decisions to be correct and carefully explained. But while it’s easy to keep track of how many matters each judge has disposed of, it’s difficult to quantify whether judges are making mistakes, not explaining their reasoning fully, or giving parties short shrift. Punishing judges for taking “too long” to do their work necessarily directs attention away from these other important goals.
When it comes to immigration judges, the proposed rules, which are scheduled to take effect this October, have a number of obvious drawbacks. In order to reach a just result, judges often need to give the parties or agencies additional time to find and present evidence, procure legal counsel, and do additional research. The added time pressure will make judges more likely to cut out these important features of the process. The new rules also require that judges render the majority of their decisions either at the hearing or within a few days, which will incentivize judges to give oral decisions rather than put their reasons in writing. A 2003 American Bar Association report noted that oral opinions encourage appellate judges to uphold removal orders without examining the merits of the appeal, as those oral rulings often lack the careful reasoning and justification a written opinion demands.
An interesting feature of the new rules—one that might seem to cut against our criticisms—is that there are penalties for judges whose remand rates (the share of opinions that are sent back for a “do-over” by a federal district court judge) are too high. There are reasons to be skeptical, however. As noted, oral opinions will be more difficult to reverse or remand on appeal, and judges can write opinions in ways that make it hard for higher courts to review them. Finally, not all cases present the same remand risk, so judges will have an incentive to be careful only in the kinds of cases in which a remand is likely.
If the immigration backlog is too large, a better solution would be to appoint more (qualified) immigration judges. That’s not cheap—immigration judges earn roughly $150,000 per year—but if we’re serious about reducing the backlog, it’s a price we should be willing to pay. Without more judges, the new incentives are unlikely to accomplish much. As a matter of algebra, making judges work 10 percent harder will only reduce the time to clear the backlog by three months, from 2.7 to 2.4 years. Given all the problems these new incentives will create, it’s hard not to conclude that the best bet is not to introduce them at all.