Cases Commenced, Terminated, and Pending

We begin by examining the total filings, terminations, and pending cases per court per year, divided by the number of authorized judgeships on a given court. Figure 1 presents averages of each of these series across all courts annually.Footnote ⁵

Figure 1.

Means of measures of commenced, terminated, and pending cases per authorized judgeship across all district courts, 1964–2012.

As we see from the figure, there was reason to be concerned about the mounting caseloads across the district courts from the mid-1960s until the mid-1980s, as all three series point to a steady increase until the mid-1980s. Following that period, caseloads began to fall and have since remained relatively stable, with a spike in pending cases from 2008 to 2012. Comparing each individual series, we observe that across much of the period, commenced cases take on the highest value. In recent years, however, pending cases have exceeded commenced and terminated ones. As one can see, the series are collinear; moreover, they are highly correlated with one another.Footnote ⁶

Of course, the circumstances facing judges on a particular court may have little relationship to the national averages. We illustrate the variance in caseload through a distributional plot, showing the means and 1 standard deviation above and below the mean annually from 1964 through 2012 (see fig. 2). Here we plot civil and criminal terminations. From this figure, one can see that the trend for the mean of civil cases terminated is similar to what we observed in figure 1 for total terminations, which is not surprising given that civil cases account for a greater proportion of cases terminated than criminal ones. The mean for criminal terminations was actually highest in the period from 1964 to 1979, then declined, and then rose steadily from 1994 to 2011. Looking at the standard deviations, we can see from civil terminations that the variance has increased considerably from the late 1990s to the present, with a pronounced spike in 1999. The relatively stable mean in the 1990s and 2000s, shown here and in figure 1, obscures the fact that some courts are burdened with high caseloads.

Figure 2.

Means and standard deviations of civil cases terminated and criminal cases terminated per authorized judgeship across all district courts, 1964–2012. Bars represent the value of 1 standard deviation plus or minus the mean. Note the two figures do not share the same y-axis, as there are more civil cases than criminal ones.

Although the cases’ commenced, terminated, and pending data can reveal interesting patterns about caseload in the district courts, there are several reasons for scholars of the courts to be cautious when using these data to make inferences about the work environment of judges. First, certain types of cases can distort the workload of a given court. This is particularly true for pending cases, where, for example, class action suits can remain unresolved for a number of years. To illustrate such an effect, we present pending case data per authorized judgeship for two district courts in figure 3: the District of South Dakota and the Middle District of Louisiana.Footnote ⁷

Figure 3.

Comparison of pending cases per authorized judgeship for two district courts, 1964–2012.

Appearances would suggest that judges in South Dakota have had fairly low caseloads compared with judges in the Middle District of Louisiana, particularly from the mid 1990s to 2007. Although it may be true that judges face less docket pressures in South Dakota than in Louisiana Middle (LAM), the spike in pending cases in LAM in the 1990s can be accounted for by a class action suit where more than 9,900 plaintiffs ultimately filed. Although litigated jointly, the cases were filed individually, a common practice in the federal courts in class action and multidistrict litigation. Though the case was settled in 1997 and some plaintiffs were terminated at that time, nearly half of those plaintiffs were not terminated until 2007 (the district was authorized for an additional judgeship that year as well). Thus using pending cases per authorized judge does not best capture the caseload of judges in LAM. More generally, in districts where class action suits have been filed, or where the Judicial Panel on Multidistrict Litigation (MDL) has assigned an MDL case, the number of filings, pendings, and terminations will all appear to spike in response to several hundred (or thousand) cases, cases that share both a common defendant and a legal question. Of course, such a spike will occur only in certain years for filings and terminations, whereas the spike can persist for pending cases until the cases are resolved. Nonetheless, the existence of such spikes creates problems in making inferences using nominal filings, pendings, or terminations data. Such data may not accurately capture the docket pressures of a judge on a given court.

Weighted Case Filings

There is a second concern related to the validity and reliability of the measures across courts and across time. When examining civil and criminal cases commenced, terminated, or pending across courts, one finds evidence that courts can differ substantially in the balance of criminal and civil cases undertaken. And followers of the courts recognize that the resources invested in these cases differs (Lombard and Krafka Reference Lombard, Patricia and Krafka2005). Concerning comparability over time, one finds that in the period from 1964 to 2012, many courts shifted in terms of the kinds of cases they were reviewing, with attention to civil or criminal cases fluctuating even within a given court. For these reasons, one should be cautious in making claims about the relative docket pressures in even one court over a significant time period.

The Administrative Office of the US Courts publishes a measure of caseload that attempts to account for the comparison of cases across courts and over time. These data are weighted case filings.Footnote ⁸ The Federal Judicial Center (FJC) has calculated weights for each nature of suit and offense in civil and criminal cases filed, respectively. According to the FJC, case weights “are a measure of the judicial work required by cases of different types. They indicate how much more or less time-consuming one type of case is compared to other cases” (Lombard and Krafka Reference Lombard, Patricia and Krafka2005, 1). In order to calculate case weights, historically the FJC relied on time diaries compiled by judges, which were adjusted approximately every 10 years. The most recent iteration of the study, completed in 2004 and reported to the Judicial Conference in 2005, used “event-based case weighting,” where researchers relied on monthly reports of judge time on trial-related activity and “the consensus assessments of experienced district judges, providing estimates of time spent in nontrial proceedings and chambers activities” (Lombard and Krafka 2005, 3). As with any periodic data collection of an ongoing process, weighted case filings in years that are more temporally proximate to the data collection studies by FJC will be more accurate indicators than measures of weighted case filings long after a revision of the case weights (see also Gillespie Reference Gillespie1974).

All nominal measures can be criticized as lacking comparability across districts or over time, whereas the weighted filings are an attempt to deal directly with comparability issues by accounting for the case mix and by regularly adjusting the case weights. Thus the weighted case filings offer an improvement, albeit an imperfect one, over the nominal filings, pendings, and terminations data. And, importantly, the measures are a component of what the Judicial Conference uses when evaluating requests by district courts for additional judgeships to deal with mounting caseloads (see Government Accountability Office 2003, 8).Footnote ⁹ Concerning the scholarly literature, weighted filings data have been used in some analyses (see, e.g., Binder and Maltzmann Reference Binder, Sarah and Maltzman2009), although such measures appear to be less commonly referred to than nominal filings, terminations, or pending data.

To illustrate the value of the weighted case filings measures, we return to the districts of South Dakota and Louisiana Middle. In figure 4, we plot case filings and weighted case filings over time. Both have the number of authorized judgeships as the denominator. The dashed lines correspond to raw filings data, and the solid lines, to weighted filings. South Dakota’s data are presented in black, and Louisiana Middle’s, in dark gray.

Figure 4.

Comparison of case filings and weighted case filings per authorized judgeship for two district courts, 1964–2012.

The figure draws attention to the difference in weighted and unweighted case filings. As one can see, for Louisiana Middle, the raw filings data overestimate the workload of a judge in that district. That is, the dotted gray line is higher than the solid gray line for almost every year. We do observe a similar spike in the 1990s that we saw in figure 3, but we can see that the spike is lower for weighted filings than for nominal filings. Turning to South Dakota, from the 1990s to the present, we see that the raw filings data underestimate the workload of a judge in that district.

Tabulating Judges Serving on a Court

Thus we argue that the choice of how one measures the number of judges on a court, the denominator, is of equal importance to assessing the numerator, the number of cases considered. Accordingly, we offer three indicators scholars may use for the denominator. First, we simply recorded the number of authorized judgeships per court per year, the denominator used by the Administrative Office of the US Courts in the weighted case filings measure and the denominator used in the earlier figures we displayed.Footnote ¹³ We next tabulated a denominator that takes into account vacancies. Although this strategy may sound straightforward, accurately assessing how many judges are on the court for any particular period of time requires considerable data collection. Here we relied on detailed biographical descriptions for each judge’s service available from the FJC. The FJC records the date of commission for each district court judge, the date of their transition to senior status.Footnote ¹⁴ Thus we monitored the history of each judge’s service over time and recorded precisely the length of tenure from commission date to date of senior status by month.Footnote ¹⁵ The virtue of this method is that we are also able to account for judges who began midterm or ended their careers midterm. For district courts in the period from 1964 to 2012, this effort amounted to tracking the duration of the careers of more than 2,000 judges.Footnote ¹⁶

We considered a final denominator, capturing both vacancies and the work of senior status judges. We assume here that senior judges work a caseload equal to 25% of that of active judges. In this we follow work by Vining (Reference Vining2009a, Reference Vining2009b) for the courts of appeals and Boylan (Reference Boylan2004) for district courts—both of which impute a 0.25 caseload for senior judges, although we recognize that many senior judges oversee substantially more cases, and some oversee less, and thus this denominator is also subject to a degree of measurement error. As we noted earlier, according to the US code, judges can retain eligibility for any salary increases granted to active status by performing work equal to 25% of the work of an active judge. We know that many judges in senior status take on more than the 25% caseload. In fact, Burbank et al. (Reference Burbank, Jay Plager and Ablavsky2012) find that “on average, district court judges serving in senior status were carrying a 43.5% case workload in the 1990s and a 50.3% case workload in the 2000s” (29). Alternatively, the US Code also allows judges to remain in senior status without hearing cases under some circumstances. That is, senior judges may also perform administrative work for the courts or elsewhere in the government on a full-time basis (or in combination with caseload or noncourtroom duties) or be certified as disabled and retain eligibility for raises that are granted to active status and senior judges.Footnote ¹⁷ Lacking systematic data over our entire period of analysis on precisely how many cases senior status judges oversee at the district court level, we impute a 0.25 caseload.Footnote ¹⁸

To tabulate the denominator, we returned to the detailed biographical descriptions of judges’ careers available through the FJC. Here, to the data already collected, we added information on the month when the judge transitioned from active to senior status and information on the final exit from the court via retirement, resignation, or death. We allotted a workload of 0.25 for each full year of service in senior status.Footnote ¹⁹ The virtue of our method for calculating this third denominator is that, similar to before, we are able to account for those whose service was for less than the entire year and for vacancies, but now we are also able to account for the work of senior status judges.Footnote ²⁰

It is useful to consider an example to clarify how we tabulated this last denominator. Although any of the judges serving between 1964 and 2012 would be appropriate, we offer Judge Aubrey Eugene Robinson Jr. from the District of Columbia. Judge Robinson Jr. was commissioned on November 3, 1966, and remained in active service until March 1, 1992, when he transitioned to senior status. His service was ultimately terminated on February 27, 2000, owing to his death. To tabulate this one judge’s career, we did the following: 0.667 service for the year July 1, 1966, to June 30, 1967 (i.e., 8 months active service); 1.0 service for each full year through June 30, 1991; 0.833 service for the year July 1, 1991, to June 30, 1992 (i.e., 9 months active service plus 3 months one-quarter time service); 0.25 service for each full year in the period July 1, 1992, to June 30, 1999; .167 service for the period July 1, 1999, to February 27, 2000 (i.e., 8 months senior status service). To calculate the number of judges on the US District Court for the District of Columbia, we repeated this exercise for each judge on that court. We then repeated this process for each judge on every district court between 1964 and 2012, more than 2,100 judges total.

We now illustrate the effect of how one measures judges’ serving by displaying weighted case filings with our three unique denominators in figure 5. We present national averages—weighted case filings across all courts—for each year in our analysis. Again, our three denominators are the number of authorized judgeships; the number of active judges serving (capturing vacancies); and finally, the number of active judges serving (capturing vacancies) plus the number of senior status judges.

Figure 5.

Means across all district courts of weighted case filings data with three alternative denominators, 1964–2012.

We see that although the series are collinear, there are notable differences. First, one finds that the difference in caseloads observed by comparing our first and second denominators—the series with authorized judgeships as the denominator versus active judges—grows considerably over time. At times when these series are close to one another, the number of vacancies across the federal courts is small. But one can see that beginning in the 1980s, vacancies on the district courts have shifted even the mean workload of judges across courts. The growing difference in recent years reflects the fact that nominations to the district courts have become more contentious and politicized over time (Binder and Maltzman Reference Binder, Sarah and Maltzman2009). In fact, the gap between the two series from 2008 to 2012 is striking. Second, when comparing our first and third denominators, one can see that the work of senior status judges has helped to alleviate docket pressures, even in the face of vacancies. When looking from the late 1990s to the present, the weighted per active and senior status judges series is lower than that of the authorized judgeship series. The work of so many senior status judges may have helped mitigate a court crisis.

We offer two final figures to provide context for the effect of the differing denominators. First, we provide a similar figure to figure 5 but here focusing only on the weighted case filings for the Southern District of New York. In figure 6, one can see that caseloads in this district have grown considerably over time. And one notes that, at first, the series track fairly closely with one another. But when one compares the authorized judgeship denominator series against the active judges series from 1980 to 2000, as we discussed earlier, one sees the pronounced effect of vacancies, particularly in the early 1990s. However, it is also evident that the work of so many senior status judges has helped alleviate caseload pressures, as one sees in the weighted per active and senior status judges series.

Figure 6.

Weighted case filings data for the southern district of New York with three alternative denominators, 1964–2012.

Our final figure compares the disparity between first and third denominator for a single year, 2012. We subtract the number of weighted filings per active and senior status judges from the number of weighted filings per authorized judgeship, which allows us to evaluate the difference between the published weighted filings data and our measure that accounts for both vacancies and senior status judges. We display the outcome in quintiles in figure 7.

Figure 7.

Comparison of the difference in weighted case filings per authorized judge against active judges and senior status judges serving, 2012. Negative values (shown in darker shades) indicate districts where vacancies outweigh assistance by senior judges. Positive values (shown in lighter shades) indicate districts where assistance by senior judges outweighs vacancies.

Even for this one year, one can observe the considerable effect of calculating workload based on judges serving rather than authorized judgeships. Negative values, indicated by darker shading, show those districts where the number of vacancies is great enough that the assistance by senior judges is insufficient to replace the contribution that would be made were the vacant judgeships filled. Positive values, indicated by lighter shading, reflect those districts where the number of senior judges serving is greater than the number of vacant judgeships. In this map, one should not assume that districts near zero reflect the optimal district. Rather, these are districts where the weighted filings per judgeship closely match the weighted filings per judge serving.

Summary of Measures Available

We conclude our discussion of the measurement of caseload by providing a summary of the measures available to scholars. We have argued in favor of using weighted case filings as the numerator, and we have pointed to problems associated with authorized judgeships as the denominator. That said, it may be that scholars have theoretical reasons for preferring, for example, the use of terminated civil cases as the numerator and authorized judgeships as the denominator. Our article makes publicly available the following measures as numerators:

• Civil cases commenced

• Criminal cases commenced

• Total cases commenced

• Civil cases terminated

• Criminal cases terminated

• Total cases terminated

• Civil cases pending

• Criminal cases pending

• Total cases pending

• Weighted case filings

For each of these numerators, our data collection includes the following measures as denominators:

• Authorized judgeships

• Active judges serving

• Active judges and senior status judges serving

Thus we offer scholars the possibility of 30 measures of caseload for each district court from 1964 to 2012. We argue, however, that the numerator of weighted case filings offers an improvement over nominal filings, terminations, and pending cases because the measure best allows for comparison across courts and over time, and that the denominator that accounts for both vacancies and the contributions of senior status judges better represents the judicial resources available on a given court. That said, the appropriate caseload measure depends on the question asked, and scholars should recognize that any measure is imperfect.