Perhaps the two most salient trends in complex litigation have been the rise of multidistrict litigation (MDL) and the fall of aggregation on plaintiffs’ terms. According to recent statistics, more than one third of federal cases are consolidated within MDLs—meaning that they are being litigated before judges handpicked by the Judicial Panel on Multidistrict Litigation (the JPML or the Panel), which itself was handpicked by the Chief Justice. Meanwhile, decisions on personal jurisdiction, class actions, and other topics have dramatically reduced plaintiffs’ ability to select their preferred forum for complex cases. These trends intersect when jurists and scholars suggest that MDL provides a backstop for aggregate litigation because it is not constrained by rules on personal jurisdiction and class certification. The ultimate choice of the forum in which large-scale cases will be litigated seems to be increasingly in the unconstrained hands of the Panel, and not the plaintiffs. This reliance on MDL as the primary vehicle for aggregation makes it particularly important to know how plaintiffs and defendants fare before the Panel when they argue over where and before whom a new MDL should be heard.
This paper presents the results of our empirical study of the relationship between party preferences and the assignment of MDLs to particular districts and judges. Based on a study of every MDL for a five-year period (2012–2016), we find that party preferences are meaningful for the selection of MDL districts. When plaintiffs and defendants agree on a preferred district, that district is very likely to be the eventual location of the MDL. When they disagree, the Panel sides with plaintiffs and defendants roughly equally. Whether this formal equality implies substantive equality, though, is an issue that merits further attention—and it raises deeper concerns about those forces that are pushing more cases into MDL in the first place. We also examine the characteristics of individual judges to which MDLs are assigned. We find that the Panel has not used its appointment power to engage in partisan behavior, and it has ensured that transferee judges are as diverse as judges overall, although it has not been at the leading edge of diversification.
In sum, we are encouraged by the Panel’s decisions as far as they go, though we believe that these findings call for further scrutiny of federal procedure and judicial administration. The Panel is treating plaintiffs and defendants equally with respect to forum choice, so it matters how frequently MDL is the best (or only) option for aggregation, and it is assigning cases to judges that are representative of federal district judges, so it matters who makes up the federal judiciary. And, of course, the Panel itself is a central actor in this ever-growing segment of federal litigation, so it matters whom the Chief Justice appoints to serve on the Panel in the first place. Understanding these forces is therefore critical to assessing procedural fairness in an era of increasing MDL dominance of the federal judicial system.
The prominence of multidistrict litigation, or MDL, in the federal civil litigation system is no longer a secret. As the federal courts, with assistance from Congress, have increasingly stymied class action litigation, the number of cases in MDL has grown rapidly over the course of the last two decades—a development that has attracted significant attention from both academics and lawmakers. Indeed, in just the last year the House passed a bill proposing major changes to the MDL process, and the Civil Rules Advisory Committee has embarked on a project to consider special Federal Rules of Civil Procedure for MDL cases. Although MDL is now in the spotlight—and the crosshairs—much about how MDL works remains mysterious. Perhaps one of the most frustrating aspects of the movement to “reform” MDL, spearheaded by the corporate-defense bar and political conservatives, is how little empirical support there is for their efforts. As these efforts gain steam and become shrouded in political debate, it has become increasingly important to understand how MDL actually works.
One such area that demands further inquiry is how the Judicial Panel on Multidistrict Litigation (the JPML or the Panel) decides where, and before which judges, to consolidate cases. The MDL statute, 28 U.S.C. § 1407, provides the Panel with almost unlimited discretion to choose the transferee district and judge. Both the Panel and courts have repeatedly held that this discretion is not limited by the increasingly strict restrictions on personal jurisdiction. But the choice is not unguided: before deciding whether and where to consolidate cases, the Panel receives briefing and hears oral argument from parties in cases that would be consolidated. In these settings, the parties express preferences for the potential MDL transferee court and often their preferred transferee judge.
The Panel’s unique procedure gives us a rare look into party preferences, allowing us to study in more detail the relationship between party preferences and JPML decision-making. In this paper, we seek to better understand how party preferences affect the Panel’s decisions about where and to whom MDLs should be assigned. To do so, we studied all 207 instances in which the Panel ordered consolidation and transfer from 2012 through 2016. This included gathering data on 461 motions to consolidate, 2,431 consolidated cases, and 897 party suggestions, as well as gathering demographic information on all transferee and JPML judges during the period.
Our research yields several interesting findings. First, a review of party preferences expressed before the Panel helps explain how the Panel selects transferee districts. We find that the Panel is highly inclined to select a transferee district proposed by a mix of plaintiffs and defendants. For example, the Panel selected a district supported by both plaintiffs and defendants in 80% of cases in which there was such a district.
Perhaps more interesting are cases in which the Panel selects a district supported by only plaintiffs or defendants. These results are relevant to the question whether plaintiffs or defendants “win” in the Panel’s assignment process. When plaintiffs and defendants do not both support one district, we find that each side succeeds roughly half of the time. That is, the Panel chooses a plaintiff-supported district about half of the time, and a defendant-supported district the other half. Other measures of party success in the Panel also converge on fifty-fifty. These findings hold for even the largest MDLs.
What to make of this seeming equipoise of plaintiffs and defendants is an important question for assessing the MDL process overall. Equal performance among plaintiffs and defendants appears fair, and it is consistent with how we would predict parties and unbiased judges to behave. Indeed, having set up a body like the Panel to assign cases, we might hope that its members would be able to remain evenhanded and avoid favoring either plaintiffs or defendants in any identifiable way.
But if we conceptualize MDL consolidation as a mode of forum selection, then putting plaintiffs and defendants on equal footing is a departure from the assumption of plaintiff forum choice that permeates U.S. law. While this departure may be justified for the blockbuster MDLs—for which special treatment may be needed and for which no single “plaintiff choice” exists—we are less convinced that plaintiffs should be deprived of their forum choice in a substantial number of cases that are not so different from mine-run litigation. Indeed, while “mega” MDLs involving widespread personal injury or mass disasters garner most of the attention of scholars and the public, the vast majority of MDLs are much smaller scale and involve a smaller number of litigants and lawyers. We hasten to add that this is not precisely a complaint with the Panel, which treats plaintiffs and defendants equally in these cases. Instead, this is an observation about the forces that are pushing more cases into MDL in the first place—a move that undermines plaintiffs’ traditional control over the initial selection of a forum, sometimes described by the Supreme Court as the “venue privilege.”
Second, although party preferences are expressed in terms of geographic districts, one interesting feature of MDL is that the Panel assigns the case not just to its chosen district for consolidation but also to its preferred judge, assuming the district and the transferee judge agree to accept it. One might wonder, therefore, whether there are trends among transferee judges that are masked when one studies cases in terms of where they are assigned. As Elizabeth Cabraser has memorably put it, the assignment is “not so much a where question, but a who question.” So if one looks only at the physical location of MDL assignments and not the individuals to whom the cases are assigned, one might overlook tendencies by the Panel.
In order to better understand the outcomes of consolidation decisions, we also examined the transferee judges themselves and compared them to the entire population of federal district judges. This not only gives us more information about the plaintiff-defendant contest, but also contributes to ongoing debates about the makeup of the federal bench and about the role of judges as “administrators.”
Here too, the results are somewhat encouraging. We find that the Panel exercised its authority to assign cases in a roughly non-partisan manner. This is true even though the Panel, unlike other similar Judicial Conference committees, skewed markedly toward Democratic appointees during the period studied. The Panel’s choices of transferee judges also roughly mirror the pool of federal district judges with respect to race, gender, political party, and seniority. This matching is encouraging, given the frequent experience of declining diversity in specially selected groups. Moreover, we find that the Panel’s choices of new MDL judges improved the diversity of the transferee-judge pool in terms of gender, though not yet race. This too is moderately encouraging. And while these results do not hold for the very largest MDLs (more than 1,000 cases), there is more diversity in the next largest cohort (between 100 and 999 cases). That said, while the Panel’s choices have mostly matched the racial and gender diversity of the federal bench, that is a low bar that we do not mean to endorse.
In sum, we find that the Panel is highly responsive to plaintiff and defendant agreement on transferee courts and that plaintiffs and defendants are roughly equally successful in persuading the Panel to assign cases to their preferred courts absent such agreement. We also find that MDLs are assigned to judges who are representative of federal district judges overall with respect to race, gender, and political party. We are thus reasonably encouraged by the Panel’s performance given its context—though this environment raises questions about forum selection and judicial demography more generally.
MDL has now been with us for over fifty years. Passed on the consent calendar of both houses of Congress in 1968, MDL has now reached a prominence few envisioned. Indeed, many are astounded to learn how important MDL has become. As of April 2018, there were 227 pending MDLs scattered across more than 50 federal districts and containing more than 120,000 cases. According to recent statistics, the cases within MDLs represent over a third of the pending federal civil cases.
Although those numbers are eye-catching, they may not have been so surprising to the small group that drafted the MDL statute and shepherded it to passage in the 1960s. Designed primarily by Professor Phil C. Neal of the University of Chicago Law School and Judge William Becker of the Western District of Missouri, the MDL statute was created to counter what many judges feared would be a “litigation explosion” in the federal courts. Concerned about the effect on federal litigation of advancements in technology, growth of the population, and new causes of action, the drafters believed large-scale litigation would become the norm rather than the exception and would overwhelm the federal courts. To address their concerns, and in light of recent experiences with mass antitrust claims, they developed a “radical proposal” to centralize pretrial proceedings in separate cases pending in multiple districts that shared a common question of fact. Under their proposed scheme, a new Judicial Panel on Multidistrict Litigation (JPML) would decide whether cases pending in multiple districts should be consolidated and, if so, where the consolidated proceedings would take place. The new Panel would comprise seven judges, all from different circuits, and would be appointed by the Chief Justice “from time to time.” In practice, the judges on the Panel, who come from both circuit and district courts, serve staggered terms of seven years.
By design, the Panel operates with wide discretion, both with respect to the decision whether to create an MDL and where to transfer the cases. The Panel drafts and follows its own rules of procedure. Moreover, its decisions are subject to extremely limited appellate review. The statute provides that a decision to deny consolidation is unreviewable; a decision to create an MDL is reviewable only on a petition for a writ of mandamus, and such a writ has never been granted. As a result, the Panel functions in this respect with virtually no oversight.
Moreover, the standard for consolidation is capacious. Given a common question of fact, if the Panel concludes that an MDL will “be for the convenience of parties and witnesses and will promote the just and efficient conduct” of the actions, then the Panel may transfer the cases for pretrial proceedings in “any district.” As a practical matter, the Panel is effectively unrestrained when it comes to its choice of a transferee district. Indeed, the Panel has taken the position since its inception that it is entirely free of restrictions on venue or personal jurisdiction. The few courts that have addressed that question have agreed. Each consolidation order also identifies the Panel’s handpicked transferee judge in the chosen district, subject only to the approval of the district court. All told, the Panel operates with few restraints.
This flexibility was intentional on the part of the judges who designed and supported the passage of the statute. In their view, they required maximum discretion to confront the “litigation explosion” in the federal courts, including when it came to where and to whom the cases should be assigned. This was in part because active judicial case management was both novel and controversial at the time; the supporters of the statute, who mostly populated the early JPML, wanted to be sure they could assign cases to judges who would energetically manage the litigation. (Today, though, managerial judging seems to be the dominant mode.)
The Panel’s transfer orders illustrate just how few restraints exist. For every new MDL, the Panel issues a short opinion giving the reasons for its choice of transferee district. Even cursory review of these opinions demonstrates that the Panel’s reasons vary widely from case to case. Among the factors cited by the Panel are the number of cases to be consolidated already pending in a particular district, party preferences, the experience of the transferee judge, and the workload or resources of the transferee district. In most cases, geography seems to matter little, especially when the controversy is of nationwide scope.
The Panel’s decision is a consequential one. Even though the statute provides for remand of the cases to their original districts at the conclusion of pretrial proceedings, the reality is that nearly all of the action happens in the transferee court. Historically, less than 3% of cases are remanded. The vast majority of transferred cases are resolved, either through settlement or dispositive motion, in the transferee court. Because “pretrial proceedings” have been defined to include literally everything a district court might do prior to trial, the MDL transferee judge has enormous power to manage the litigation, appoint leadership teams, and decide dispositive motions and otherwise important issues such as class certification and motions to exclude expert evidence. And although the Supreme Court has held that the transferee judge may not transfer a case to herself for trial, in reality, MDL judges regularly oversee “bellwether trials,” either because the cases were filed directly into the MDL court or because the parties have consented. Ultimately, then, the selection of the transferee judge is among the most momentous decisions made in the entire litigation.
Paralleling the rise of MDL have been other developments in civil litigation that have seemingly made the Panel’s consolidation choice even more significant. As the Supreme Court has narrowed the forums where plaintiffs can obtain personal jurisdiction and has seemingly made it more difficult to certify a class action, defendants have appeared to be getting an upper hand in forum selection. Aggregation, one of us has written, has become more commonly “on defendants’ terms.”
In response to critics of these developments, some jurists and scholars have suggested that MDL provides an outlet. Because class certification and personal jurisdiction in the transferee court are not prerequisites for an MDL transfer, aggregation remains possible. But this suggestion invites another question: who wins in MDL consolidation? More precisely, when the Panel decides to consolidate cases and transfer them to a district court, which is extremely likely to preside over the final resolution, on whose terms is the case aggregated?
Just as MDL flew under the radar for decades, empirical research on it also was slow to get off the ground. In recent years, scholars have examined the Panel’s decisions to consolidate cases, the Chief Justice’s selection of Panel members, and the Panel’s decisions about where to consolidate cases. This final category, in particular a 2013 study by Margaret Williams and Tracy George, connects directly with the question in this project. Williams and George studied MDL activity up to 2012, with a more detailed examination of cases from 2005 to 2009.
We pick up the timeline in 2012, as MDL was really taking off. Where relevant, we reproduce Williams and George’s analysis to determine whether the MDL process has changed with its increased use and attention. We also seek to answer new questions about party preferences and judge demography, relying on our more recent data.
To be more specific, this paper seeks to better understand the behavior of the Panel and of parties in potential MDLs by reviewing every case that the Panel consolidated from 2012 to 2016. This includes gathering data on 461 motions to consolidate, 207 transfer orders, 2,431 transferor cases, and 897 party suggestions, as well as gathering demographic information on all transferee and JPML judges during the period.
The contributions of this study are roughly divided into two groups. Part III studies the approximately 900 party suggestions to the Panel of potential transferee districts and the Panel’s responses to those suggestions. Part IV takes up transferee judge demography. Rather than reviewing our methods here, we begin each section with a short comment on methodology.
Although our study focuses on decisions by the Panel to assign consolidated cases, along the way we gathered additional information on two other aspects of the process. First, as mentioned above, we collected information on every underlying case consolidated during the period of our study. In the Appendix we provide descriptive statistics about these cases, including their distribution across federal district courts and their relationship to plaintiff preferences in the Panel. Regarding the latter, we find substantial (but not complete) overlap.
Second, we also considered the Panel’s initial decision whether to consolidate at all. During the same period in which the Panel granted 207 motions to consolidate, we also observed that the Panel denied 152 motions, resulting in a grant rate of 57.7%. Though not central to our study of district choice, we note that this is a marked departure from previous periods. Williams and George, for example, found that the grant rate for MDL consolidation hovered around 70–80% for most of the period from the creation of MDL in 1968 until the end of their study. We also asked whether the Panel was more likely to grant defendant or plaintiff motions to consolidate: the grant rates were essentially identical. If there were a plaintiff or defendant advantage in MDL, we did not find it in the Panel’s granting or denying of motions to consolidate. But it may be in the Panel’s decisions on where and to whom to send consolidated cases. It is to those topics we now turn.
When the Panel decides where to consolidate cases, it is making a decision that may have a profound effect on those cases’ ultimate resolution. Transferee judges wield enormous authority—from discovery management, to settlement, to the resolution of dispositive motions. Indeed, the vast majority of cases are resolved in the transferee court by dispositive motion or settlement agreement. It is important to know, therefore, how plaintiffs and defendants compare when suggesting districts for consolidation. In other words: who wins?
In brief, we find that the Panel is highly likely to select districts supported by both plaintiffs and defendants. In cases in which the Panel selects a district supported by only plaintiffs or defendants, we find the Panel is about as likely to select a district supported only by plaintiffs as it is to select a district supported only by defendants. These results hold even if we focus only on the largest MDLs. But, as we discuss below, these results suggest a departure from the default assumption of plaintiff forum choice in federal litigation.
This Part addresses all 207 instances in which the Panel ordered consolidation and transfer from 2012 to 2016. For each decision, we recorded the date, transferee district, transferee judge, and all transferor districts (and the number of transferor cases per district). This information is available on the transfer orders.
We also separately identified every district suggested in every consolidated case. For each suggestion, we recorded which party or parties made the suggestion, and the ordinal rank of preferences for plaintiffs and defendants making the suggestion. This information is also available on the transfer orders. In addition, for each suggested district, we recorded whether the Panel selected that district.
For each MDL, we also identified the total number of cases that comprised the MDL at any time (not just at the time of consolidation). We obtained this information from JPML publications as of April 2018. Many of the MDLs in our study were closed as of April 2018. For ongoing MDLs, there may be additional cases added after this date, though we suspect that those numbers will not substantially affect our results. Because there may be special normative and practical considerations for the largest MDLs, we coded those MDLs comprising more than 1,000 cases, and those with more than 100 but fewer than 1,000 cases. Although the number of cases is not a perfect measure of the complexity of an MDL, it is a reasonable proxy for our first-cut analysis.
Our primary interest in this Part is the relationship between party preferences and the selection of a transferee court. Because this is our primary aim, we are not particularly interested in the frequency that particular districts are selected or nominated. For reference, however, we provide those counts in the Appendix. Were one inclined, these data could be used to identify districts strongly favored by defendants or plaintiffs and districts with high or low “success-rates” in the Panel. We are also not interested, in its own right, in the total number of cases in any particular MDL. But it is notable that the size of MDLs varies greatly, from 2 cases to more than 80,000. Where relevant, though, we explore any differences in the assignment of MDLs with very large numbers of cases (“blockbusters” or “mega MDLs”) as compared to the mine run of MDLs.
Instead, we focus here on the relationship between party preferences and the selection of districts. We approach this question in two ways. First, we study plaintiff versus defendant nominations overall (Subsection 1), and then plaintiff versus defendant ordinal preferences (Subsection 2). The reason for studying overall nominations and ordinal ranks is that both represent plausible ways that parties and judges could interpret nominations: it might matter whether plaintiffs or defendants suggest the district at all, or it might matter whether plaintiffs or defendants ranked the district higher.
We first reviewed the Panel’s orders to identify every time a district was proposed. We coded each proposal as plaintiff, defendant, a mix of plaintiffs and defendants, or a fourth category in which the Panel selected a district not proposed by any party.
This method allows us to report some simple statistics about suggested districts. Defendants suggested 288 districts, slightly more than one per case, including 118 suggestions without plaintiff support. Plaintiffs suggested 609 districts, or almost three per case, including 439 suggestions without defendant support. A mix of defendants and plaintiffs suggested 170 districts.
We can also provide some descriptive statistics about the relationship between party suggestions and the Panel’s decisions on transferee courts. In 165 cases (about 80%), the Panel selected a district supported by a plaintiff. In 149 cases (about 72%), the Panel selected a district supported by a defendant.
As implied by these numbers, the Panel frequently selects districts supported by a mix of plaintiffs and defendants. Indeed, they did so more than half of the time, i.e., in 115 out of 207 cases. (The Panel selected a district recommended by neither party in 8 cases, or about 4%.) Perhaps even more significantly, in cases in which both plaintiffs and defendants support at least one district, the Panel selected one such district in almost 80 % of cases (115 out of 147). In other words, in only about 20% of cases did the Panel reject a district supported by both parties in favor of a district supported by only one party or by neither party.
Returning to our initial question, we are interested in the rate at which plaintiffs and defendants succeed in their suggestions. (We take up the meaning of this “win rate” below.) In 84 cases, the Panel selected a district recommended by only plaintiffs or only defendants. Among those 84 cases, the Panel selected a plaintiff-supported district in 50 cases, or 59.5%. The Panel selected a defendant-supported district in 34 cases, or 40.5%. Regression results track the descriptive statistics.
Table 1: Transferee District and Party Suggestions, 2012–2016
|Transferee district suggested by plaintiffs||Transferee district not suggested by plaintiffs|
|Transferee district suggested by defendants||56% (both)||16% (defendant only)|
|Transferee district not suggested by defendants||24% (plaintiff only)||4% (neither)|
As suggested above, we also inquired whether these results varied when looking at much bigger MDLs—cases that tend to attract attention because they involve issues of broad public concern, such as mass disasters (like the Deepwater Horizon oil spill) or a defective drug or medical device that allegedly caused injuries to thousands (like defective hip implants). Among these cases, we found that the Panel was still very likely to select a district supported by plaintiffs and defendants. In MDLs with more 1,000 cases, the Panel selected a case supported by both parties 55% of the time. In cases where there was at least one district supported by both plaintiffs and defendants, the Panel selected such a district 80% of the time. We also found that among bigger MDLs, the Panel was more likely to select a plaintiff-only district than a defendant-only district. In MDLs with more than 1,000 cases, when the Panel selected a district supported by only one side, it selected plaintiff-only districts 66 % of the time, and defendant-only districts 34% of the time. We should note, though, that these large MDLs represent a small sample size (22 MDLs during our period).
The previous analysis treated all plaintiff and defendant suggestions equally. We also collected data on plaintiff and defendant ordinal preferences. Relying on the Panel orders, we ranked every plaintiff- and defendant-suggested district. This allows us to see whether plaintiffs or defendants “win” even in cases in which both plaintiffs and defendants suggested the same district. This is also helpful to account for the fact that plaintiffs suggest more districts than defendants.
To analyze these data, we categorized each transferee district as recommended higher by plaintiffs, higher by defendants, equally by plaintiffs or defendants, or by neither plaintiffs nor defendants. We find that the transferee district was ranked higher by plaintiffs in 56 cases, higher by defendants in 60 cases, equally in 83 cases, or by neither in 8 cases. So in cases where we identified a “winner,” plaintiffs won in 48%, and defendants won in 52%.
Table 2: Transferee District and Ordinal Rank, 2012–2016
|Transferee district suggested higher by plaintiffs||Transferee district not suggested by higher plaintiffs|
|Transferee district suggested higher by defendants||40% (tie)||29% (defendant higher)|
|Transferee district not suggested higher by defendants||27% (plaintiff higher)||4% (neither)|
We also separately identified those cases in which the transferee district was the first choice of plaintiffs or defendants, because the Panel often emphasizes the top choice on each side. In 42 cases, the Panel selected the plaintiffs’ first choice and not the defendants’ first choice. In 53 cases, the Panel selected the defendants’ first choice and not the plaintiffs’ first choice. By this measure, plaintiffs “won” in 44% of cases, and defendants “won” in 56%. Here, too, regression results are consistent with these descriptive statistics.
We again considered larger MDLs separately. For MDLs with more than 1,000 cases, we find that the district was rated higher by plaintiffs in 36%, higher by defendants in 23%, and equal in 41%, for a plaintiff “win rate” of 62%. Among these MDLs, when the Panel chose one party’s first choice over the other’s, we find the plaintiffs’ win rate to be 55%. For MDLs with more than 100 cases but fewer than 1,000, we find that the district was rated higher by plaintiffs in 24%, higher by defendants in 20%, and equal in 56%, for a plaintiff “win rate” of 55%.
Taking all of these findings together, two issues jump out to us as important and relevant to the debate that has accompanied the growing prominence of MDL in the federal litigation scheme.
First, the Panel frequently selects districts suggested by both plaintiffs and defendants. Our descriptive results reveal that this was true in 115 of 207 cases, and in 80% of cases in which there was such a district. Regression analysis tracks these results. And these results persist when we consider larger MDLs separately. Though perhaps not earthshattering, we think this finding is highly significant to understanding the Panel. (Though it is possible that these findings support concerns that plaintiff and defense counsel are collaborating too much, we cannot from our data separate beneficial from malevolent convergence of preferences.)
Second, and returning to where this part began, we are interested in the zero-sum competition between plaintiffs and defendants when they disagree, or at least do not provide the Panel with a consensus option. This Part provides various ways to approach this question. Using the descriptive statistics, we offered various measures that suggested different “win rates,” but virtually all of the results are within sixty-forty. In other words, we think it is fair to say that plaintiffs and defendants perform relatively equally in the Panel. And, again, the results hold in the larger MDLs. So while we can find some hints of defendant or plaintiff success, we are not convinced that the results differ dramatically from even.
These findings are consistent with empirical studies in two other areas that share some similar features to MDL. First, Clermont and Eisenberg analyzed federal venue transfer motions under Section 1404(a), and they find that courts grant these motions in about 50% of cases. Second, various scholars have looked at motions to dismiss for forum non conveniens, and they too have found a 50% grant rate.
Our findings also line up with how we would expect parties to behave. Plaintiffs and defendants express their preferences in a strategic environment. It would be shocking if they did not consider how the Panel would react to their suggestions. In other words, plaintiffs and defendants are likely to suggest districts that are both favorable to their side but also reasonable in the eyes of the Panel. The effect would be results that converge on fifty-fifty—which they do.
Equal treatment of plaintiffs and defendants is also consistent with how we would expect the Panel to behave. As just mentioned, Panel behavior interacts with party behavior, feeding back into a 50-50 split. Moreover, equal treatment for plaintiffs and defendants is intuitively appealing. Nothing in the history of MDL suggests that consolidation should favor one side or the other. Even if we were to assume that the judges on the Panel were inclined to act strategically, they might select an even-handed approach that would insulate them from criticism. And while we have not seen any evidence that the Panel is consciously striving for equal treatment, we suspect that the Panel judges might be pleased to learn that their decisions have not had a pro-plaintiff or pro-defendant bias. For this, the Panel should be praised.
However, from the perspective of system design, these results depart from the background condition in which plaintiffs usually select the forum. According to the Supreme Court in the famed decision Piper Aircraft v. Reyno, “there is ordinarily a strong presumption in favor of the plaintiff’s choice of forum.” Indeed, in many ways, the systems of forum and venue in U.S. courts are built on an assumption that the plaintiff chooses the forum. Although defendants may use Section 1404(a) or forum non conveniens to get cases out of particular federal courts, estimates suggest that those motions arise in at most 5% of cases. In other words, in the vast majority of unconsolidated cases, plaintiff forum choice prevails. Indeed, as the old saw goes, “The plaintiff is the master of the complaint.”
MDL is a different story. Once the Panel has agreed to consolidate cases, plaintiff forum choice seems to be at parity with defendant forum choice. That result differs substantially from unconsolidated litigation. From the perspective of any given case within an MDL, the numbers look even more extreme. For example, in our period, only about one-third of transferor cases were consolidated in the district in which they were first filed—meaning that plaintiffs’ original forum choice was disrupted in two-thirds of cases. It is quite a switch from plaintiffs picking almost every forum to plaintiffs picking only half or a third of the time. This development would be consistent with the perception that recent trends favor defendants in aggregate litigation and civil litigation more generally. And, importantly, it means that when decisions about class actions or personal jurisdiction push more cases into MDL, those same forces are concomitantly reducing plaintiff forum choice overall.
The increasing reliance on MDL thus reduces the plaintiffs’ traditional venue privilege in individual and aggregate litigation. That is, in cases that would otherwise be one-on-one, plaintiffs may find themselves involuntarily swept into an MDL in a distant forum. In cases that would otherwise be class actions, plaintiffs not only lose the benefits of Rule 23 but also lose their venue privilege when their cases are pushed into MDLs.
This departure from the presumption of plaintiff forum choice might make sense in exceptional cases. In cases like these, where the overall stakes are extremely high and the parties are distributed throughout the country, one could be comfortable with the Panel departing from the status quo ante of plaintiff forum choice. In such cases, the national interest in consolidated pretrial proceedings arguably outweighs the plaintiffs’ interest in choosing the forum in the first instance. One of us has argued that this national interest in efficient resolution of nationwide controversies, a concern that was central to the statute’s drafters, will typically mean that the effectively nationwide personal jurisdiction created by the MDL statute passes constitutional muster under the Fifth Amendment, even at the cost of the plaintiff’s venue privilege.
In such cases, plaintiffs may even prefer quick resolution in a single nationwide forum over choosing the forum. Since the Supreme Court’s decision in Goodyear, and supported by language in the Court’s recent decision in Bristol-Myers Squibb, aggregation by plaintiffs is likely only available in a state where the defendant is “essentially at home” and therefore subject to general jurisdiction. At best, plaintiffs must choose between extremely costly individual litigation or aggregation in forums pre-selected by the defendant as “home”: the state of incorporation and the principal place of business. In such circumstances, a forum chosen by a neutral Panel may not only be constitutionally justifiable but preferable as to the alternatives. In other words, fifty-fifty may be just right.
But for many MDLs, departing from the assumption of plaintiff forum choice seems much less justified. Consider MDLs with fewer than ten cases, which describes 46 of our 207 cases. When the Panel selected a district supported by only plaintiffs or only defendants for these MDLs, the Panel selected a district supported by plaintiffs 43% of the time. It is hard to explain why we would depart so dramatically from a presumption of plaintiff choice in these cases, which could have been small class actions or separate suits in forums of plaintiffs’ choosing. Again, this is not the fault of the Panel, but a result of parties needing to use MDL rather than some other mode of aggregation.
In short, it makes sense that the Panel treats plaintiffs and defendants equally. Indeed, that is what we might hope it would do, especially when the statute does not direct any new presumption in favor of one side or the other. But it is important to acknowledge that once cases are in the Panel, plaintiff forum choice may be sacrificed. So, when other forces drive more cases into MDL, we should attribute to those forces a decline in plaintiff forum choice. One’s views on this trend will correspond to what one believes is the greater evil: forum shopping by plaintiffs or tilting the scales in favor of defendants. Regardless, both the case law and our data demonstrate that the balance has shifted.
When the Panel consolidates cases into an MDL, the consolidation order includes the name of the district judge to whom the cases will be assigned (subject to approval of the district court). In so doing, the Panel often refers to the experience or interest of the transferee judge as a justification for the assignment decision.
This judge-focused aspect of Panel decision-making means that there might be important trends among assigned judges that are not apparent from districts alone. Geography may tell part of the story, but the particular judge may make even more of a difference, given her discretion in pretrial proceedings. It is important, therefore, to study the transferee judges themselves.
This examination contributes to important ongoing conversations in multiple areas of study. For example, there has been substantial recent attention to administrative or bureaucratic decisions made by judges, especially the Chief Justice of the United States. The Chief Justice picks the members of the Panel, but it is the Panel judges who select the transferee judge. How the Panel performs in this capacity is worthy of study. There also has been interest in the demography of judges, “elite judges,” the Supreme Court bar, and others. Our work adds an additional data point by examining the judges selected to handle multidistrict litigation, an increasingly important component of federal litigation generally.
This Part reviews our results on transferee judge demography. We find that transferee judges look a lot like the general population of federal district judges—for better or for worse.
We then narrow our focus to two subsets of judges: judges in the largest MDLs and first-time MDL judges. In the largest MDLs, transferee judges are more likely to be male and white than federal district judges overall. Meanwhile, new MDL judges are slightly more likely to be female and liberal than the pool overall, but not dramatically so. New MDL judges are not more racially diverse than the pool overall.
Finally, when we compare transferee judges to the Panel itself, the most striking results relate to partisanship. While the Panel is more liberal than the overall pool, the transferee judges are more conservative than the Panel (and similar to the overall population of federal district judges). In other words, the Panel does not seem to have wielded its nearly limitless appointment authority to empower transferee judges who match its ideological preferences. We also find that the Panel in our period is much more diverse on gender than race—a result that parallels our findings for the new MDL judges that the Panel selected.
In sum, we raise two cheers to the Panel. The Panel has not used its appointment power to engage in partisan behavior, and it has ensured that transferee judges are as diverse as judges overall. But the Panel has not been at the leading edge of diversifying the legal profession either.
This Part addresses all 207 cases in which the Panel ordered consolidation and transfer from 2012 to 2016. Relevant here is the identity of the judge to which the case was transferred. Although cases may be reassigned to another judge after the transfer, these reassignments lie outside of the Panel’s initial choice; our study cares about choices within the Panel’s control.
To determine the demographic characteristics of judges, we rely on the data provided by the Federal Judicial Center (FJC), the research arm of the federal judiciary. The FJC’s biographical information comes from a range of primary sources identified in the “Biographical Directory of Federal Judges.” We use these data to code the transferee judges and the population of federal district judges. Note that for judicial ideology, we separately computed results using multiple measures of ideology in addition to the judge’s appointing president, but none dramatically changed our substantive results.
To identify judges assigned an MDL for the first time, we rely on Panel documents that list terminated and pending MDLs. This list reveals whether an assigned judge ever supervised a terminated or pending MDL as of July 2017. To identify judges handling large MDLs, we code each MDL for the total number of cases as of April 2018.
Finally, as noted above, we compare MDL judges to Panel judges and to the total population of federal district court judges. To identify Panel judges, we rely on the Panel’s listing of members and dates. To identify the population of district court judges, we relied on the aforementioned FJC sources to identify all active (including senior) district court judges on July 1, 2014, the midpoint of our sample. We also calculated results for judges serving on the first day, last day, or any time during our study period, though we do not reflect those results in our tables. We then use our usual data sources to determine relevant information for these judges.
This section reviews our results regarding transferee judge demography. We begin in Subsection 1 with all transferees. Subsection 2 looks only at the largest MDLs, and Subsection 3 looks only at judges handling their first MDL.
First, we considered all 207 transferee appointments from 2012–2016. The basic demographic information of the transferee judges is provided in the following table. As shown, transferee judges are predominantly white (82%) and male (71%). The median transferee judge was born in 1951 and began federal judicial service in 2001. About 17% of transferee judges had senior status at the time of appointment, and 22% had been a Chief Judge prior to the appointment. With respect to ideology, about 54% were appointed by Democratic presidents.
Reporting these data in a vacuum tells us only so much. A more useful question is how transferee judges compare to all district judges. As noted, we have selected the midpoint of our time period as the comparison set. The results are also reported in the following table.
Table 3: Comparing Transferee Judges, 2012–2016
|Category||Transferee judges||All district judges|
|Birth year (median)||1951||1949|
|Commission date (median)||2001||1999|
|Senior status||17%||41% (estimated 15% of docket)|
|Party of president||54% Democrat
The most striking finding, to us, is how closely the transferee judges match the overall population of federal district judges. Transferee judges and district judges are about equally likely to be white and male, and they are similar in age and commission date. (Although a much larger proportion of all district judges have senior status, the proportions are closer to equal when adjusted for caseload.) Transferees are slightly more likely to be appointed by a Democrat, but this result is statistically significant only as compared to all district judges at the start of our sample—i.e., before Democratic President Barack Obama appointed a substantial number of judges who were eligible to be, and were selected as, transferee judges.
As suggested above, we find these data mostly encouraging. Unlike the appointments to other elite judicial institutions and positions, the appointments of MDL judges do not reflect less diversity than the general population from which members get selected. Indeed, as discussed later, in this way transferee judges differ from the Panel itself. But as suggested, these results should temper any nascent enthusiasm about the Panel breaking glass ceilings in the federal judiciary.
Finally, we should say a word about plaintiff and defendant preferences. Though we cannot code every party suggestion for judge identity, we can look at those transferee judges from districts suggested only by plaintiffs or defendants. On race and gender, the differences are minimal: plaintiff-only transferees are 72% male and 88% white, and defendant-only transferees are 76% male and 79% white. More noticeably, and contrary to what some might expect, 48% of plaintiff-only transferees were Democratic-appointees while 68% of defendant-only transferees were Democratic-appointees. These differences, though, are not statistically significant.
For various reasons, it is possible that the Panel treats larger MDLs differently. These MDLs, simply by virtue of involving more parties and lawyers, may be more complicated and attract disproportionate attention. These MDLs also may eventually involve plaintiffs scattered throughout the country, meaning that geography and the location of already-pending cases may be less relevant to the Panel. These cases may therefore present different considerations.
In light of these observations, we identify those MDLs that comprised more than 1,000 cases, and those that comprise more than 100 but fewer than 1,000 cases. We then compare the transferee judges in these cases to all transferees, as reflected in the following table.
Table 4: Comparing Transferee Judges from Large MDLs, 2012–2016
|Category||Transferee judges||100-999 cases (25 MDLs)||1000+ cases (22 MDLs)|
|Birth year (median)||1951||1952||1953|
|Commission date (median)||2001||2002||1998|
|Party of President||54% Democrat
These results reveal interesting patterns. Among the very largest MDLs (more than 1,000 cases), transferee judges are 100% white, and decidedly more male (86%) than all transferee judges (71% male). Though not reported in this table, transferee judges in these very largest cases are less likely to be new MDL judges than transferee judges overall (23% new versus 32% new), though there are still some new judges among these cases.
Meanwhile, if we shift to the second largest category (more than 100 but fewer than 1,000 cases), these effects disappear. This group has somewhat more women than all MDLs, is roughly equal with respect to nonwhite judges, and has many more Republican-appointed judges. Twenty-eight percent of these judges are new to MDL.
The Panel has been criticized for sending too many cases to the same judges while not democratizing (or diversifying) the pool of transferee judges.
Using the above-described methodology, we identified sixty-six cases in our sample that were assigned to “new” MDL judges. This comprises approximately 32% of MDLs in our period. The following table reports the demography of new MDL judges as compared with all transferees and with all district judges.
Table 5: New MDL Judges versus Transferees and All District Judges
|Category||New transferee judges (66 MDLs)||Transferee judges||All district judges|
|Birth year (median)||1954||1951||1949|
|Commission date (median)||2008||2001||1999|
|Senior status||3%||17%||41% (or 15%)|
|Party of President||64% Democrat
New MDL judges are not so different from federal district judges overall. The racial breakdown is about the same. New judges are younger and less senior, but this is not surprising given that they are new to MDL. New judges are more likely to have been appointed by a Democratic president, but it should be unsurprising that Obama appointees are over-represented in this group. Indeed, thirty-one of the sixty-six new MDL judges were Obama appointees.
But new judges markedly differ in terms of gender. New judges are more likely to be women than all transferees or district judges. About 39% of new MDL judges were women. Interestingly, although President Obama has been praised for appointing more women to the bench, the new-judge effect is driven more by non-Obama appointees than by Obama appointees.
Finally, as mentioned, the selection rate of new judges varies by MDL size. New judges were less common in MDLs with more than 1,000 cases (23% of transferees were new) than MDLs with 100–999 cases (28% new) or all MDLs (32% new). The median MDL assigned to a new judge was about 30% smaller than the median MDL assigned to a judge who had previously received an MDL.
Bringing together the findings so far, the following figures show how each of the groups—all district judges, all transferee judges, large-case judges, and new MDL judges—compare on gender, race, and party of appointing president.
Figure 1: Gender
Figure 2: Race
Figure 3: Party of Appointing President
Transferee judges overall appear quite similar to all district judges. For gender, new MDL judges are more likely to be women, while judges in the largest MDLs are more likely to be men. On race, judges in the largest MDLs are more likely to be white than in the other case groupings. Judges in MDLs of 100 to 999 cases are more likely to be Republican, but this finding does not hold for smaller or larger MDLs.
We find the overall results roughly consistent with how a neutral Panel might be expected to behave. For the most part, the Panel appoints MDL judges that reflect the pool of federal district judges. This is, relatively speaking, a good thing. It would not have surprised us to learn that judges selected to join the “elite club” of MDL judges were less diverse than judges overall. Although we find that transferee judges in the very largest MDLs were less diverse (and although we do not endorse that result), it is not so surprising that for these special cases, the Panel turned to experienced players who, themselves, are less diverse than judges overall.
That said, these results should not overshadow the fact that the pool of federal district judges is decidedly not diverse with respect to race and gender. The lack of diversity is a systematic problem that deserves systematic attention, but again, it is not limited to the Panel. Indeed, as noted, the Panel has increased female representation by appointing more women as new MDL judges than their share of federal district judges would predict. The Panel is, in a sense, more responsible for the lack of diversity among transferee judges in the mega-MDLs, though this may be a function of which judges have experience with large MDLs (which is a function of the transferee-judge selections of previous iterations of the Panel).
As a policy matter, we think efforts by the Panel to increase MDL-judge diversity are all to the good, especially when it comes to giving experience to first-time MDL judges who might be tapped again for future cases. And as the pool of experienced judges expands, we hope that the Panel continues to diversify the judges handling the largest MDLs. At the same time, whether and how seven Panel judges handpicked by the Chief Justice should make these choices remains open to question. The MDL statute provides only two criteria for consolidating and transferring cases: the convenience of the parties and the just and efficient conduct of the litigation. Moreover, it would be ahistorical to assume that the all-white and male judges who drafted the statute in the 1960s had any implicit goal of enhancing diversity when establishing MDLs. The historical record indicates that it was not on their radar.
In our view, the MDL statute is capacious enough to embrace a goal of diversifying the pool, at the very least in proportion to federal district judges overall, as long as the selection does not conflict with the statutory mandate. But it will be interesting to see if explicit discussions of judge diversity make their way into arguments before the Panel.
Finally, before leaving our analysis of judicial demography, we consider one important set of judges that has, so far, been left out of the analysis. This section considers the characteristics of Panel judges themselves as they compare to transferees.
As mentioned above, the Chief Justice of the United States selects the members of the Panel. Because Panel membership changed dramatically during the study period, we report the weighted averages for the Panel. We calculated weighted averages by assigning members a weight proportionate to the share of days they served on the Panel.
Table 6: JPML versus Transferees
|Category||JPML (weighted)||Transferee judges||New transferees||100-999 cases (25 MDLs)||1000+ cases (22 MDLs)|
|Race||100% White||82% White
|Birth year (median)||1948||1951||1954||1952||1953|
|Comm. year (median)||1995||2001||2008||2002||1998|
|Party of president||73% D
Figure 4: Gender, Race, and Appointing President for JPML and Transferees
The first set of observations relates to the Panel itself. The Panel was 100% white during this period. It was close to evenly divided among men and women. And it was almost three-quarters Democratic appointees.
Next we ask if the Panel’s composition differs from that of the transferee judges it selects. On race, the Panel appointed judges more diverse than the Panel—but as noted above, no more diverse than the district court judges generally. The Panel’s selections for the largest MDLs (more than 1,000 cases) are 100% white, but as racially diverse as all district judges for the second largest group of MDLs (100 to 999 cases). As suggested above, it is possible that the Panel views the very largest MDLs as a special category, and thus it falls back on judges with whom it is familiar, and those judges tend to be less diverse. On gender, the Panel appointed more female first-time MDL judges than would be predicted by the population of federal district judges. The Panel also appointed more women for MDLs comprising 100 to 999 cases (but not for MDLs with more than 1000 cases).
One potential (though speculative) explanation for these patterns is that a Panel that is almost 50% female (but 0% nonwhite) might be more attentive to correcting gender imbalances than racial ones. This view suggests that the Chief’s appointments to the Panel may have a trickle-down effect on the diversity of transferee judges, at least when it comes to race and gender.
On age and seniority, the Panel is—unsurprisingly—older, longer serving, more likely to have served as a Chief Judge, and more likely to have senior status. This latter point merits a further comment: senior judges frequently serve on the Panel even though one might think that they would be a bad match for the additional work of Panel service.
Finally, we observe clear ideological differences. Panel members are more likely to have been appointed by a Democratic president than the population of federal judges. One potential concern is that the JMPL could use its unfettered assignment powers to select judges ideologically similar to itself. Although it is certainly possible that the Panel is engaged in some as-of-yet undetected skewing, these data show that the Panel has not produced a cadre of MDL judges that mirror the Panel’s ideological preferences. Instead, transferee judges roughly match the pool of federal district judges, making transferee judges more conservative than the JMPL, which consists of 70% Democratic appointees.
So, while the identities of the Chief’s Panel appointments may have some effect on the race and gender of MDL judges, JMPL appointments have not appeared to affect the ideologies of transferee judges.
MDL has become the centerpiece of the federal civil litigation system. The selection of the transferee district and judge is one of the most important moments of any MDL because the MDL judge wields enormous power and discretion over pretrial proceedings. This power includes the ability to rule on dispositive motions, preside over bellwether trials, and generally manage the litigation, often toward a global settlement. And in practice, MDL cases are exceedingly unlikely to ever leave the transferee court. Because the Panel has almost unlimited discretion over where and to whom an MDL is transferred, empirical research into how the Panel makes such momentous decisions is timely and warranted.
In this paper, we find that the parties’ preferences for transferee district are especially important to the Panel. When parties agree on a preferred district, that district is likely to be the eventual location of the MDL. When the parties disagree on the district, the Panel sides with plaintiffs and defendants roughly equally. And finally, we find that the Panel assigns cases on a roughly non-partisan basis and selects judges fairly representative of the pool of federal district judges. While none of these findings are cause to canonize the Panel, on the whole we are encouraged by the ways that this nearly unchecked body has managed complex litigation. Absolute power, it seems, has not corrupted absolutely. That said, although our results indicate that the Panel is doing its job evenhandedly, these results raise challenging questions for lawmakers as they consider whether the current system demands change.
Appendix Table A: Total Transferor Cases by District (Alphabetical)
|District||Total Cases||District||Total Cases||District||Total Cases|
|M.D. Ala.||8||E.D. Ky.||10||S.D. Ohio||36|
|N.D. Ala.||27||W.D. Ky.||20||E.D. Okla.||1|
|S.D. Ala.||5||E.D. La.||43||N.D. Okla.||4|
|D. Alaska||1||M.D. La.||10||W.D. Okla.||13|
|D. Ariz.||15||W.D. La.||30||D. Or.||8|
|E.D. Ark.||9||D. Me.||2||E.D. Pa.||207|
|W.D. Ark.||7||D. Md.||17||M.D. Pa.||19|
|C.D. Cal.||147||D. Mass.||43||W.D. Pa.||9|
|E.D. Cal.||19||E.D. Mich.||18||D.P.R.||1|
|N.D. Cal.||174||W.D. Mich.||2||D.R.I.||7|
|S.D. Cal.||79||D. Minn.||50||D.S.C.||35|
|D. Colo.||15||N.D. Miss.||10||D.S.D.||0|
|D. Conn.||18||S.D. Miss.||14||E.D. Tenn.||20|
|D. Del.||36||E.D. Mo.||32||M.D. Tenn.||33|
|D.D.C.||16||W.D. Mo.||23||W.D. Tenn.||4|
|M.D. Fla.||75||D. Mont.||3||E.D. Tex.||42|
|N.D. Fla.||18||D. Neb.||5||N.D. Tex.||16|
|S.D. Fla.||62||D. Nev.||18||S.D. Tex.||9|
|M.D. Ga.||11||D.N.H.||12||W.D. Tex.||9|
|N.D. Ga.||46||D.N.J.||108||D. Utah||11|
|S.D. Ga.||8||D.N.M.||6||D. Vt.||7|
|D. Haw.||2||N.D.N.Y.||3||E.D. Va.||15|
|D. Idaho||3||S.D.N.Y.||156||W.D. Va.||6|
|C.D. Ill.||5||W.D.N.Y.||2||E.D. Wash.||10|
|N.D. Ill.||148||E.D.N.C.||11||W.D. Wash.||38|
|S.D. Ill.||59||M.D.N.C.||3||N.D. W.Va.||6|
|N.D. Ind.||7||W.D.N.C.||18||S.D. W.Va.||16|
|S.D. Ind.||35||D.N.D.||1||E.D. Wis.||10|
|N.D. Iowa||3||D.N. Mar. I.||0||W.D. Wis.||6|
|S.D. Iowa||6||N.D. Ohio||30||D. Wyo.||1|
Appendix Table B: Total Transferor Cases by District (Ranked)
|District||Total Cases||District||Total Cases||District||Total Cases|
|E.D. Pa.||207||D. Conn.||18||N.D. Ind.||7|
|N.D. Cal.||174||N.D. Fla.||18||D.R.I.||7|
|S.D.N.Y.||156||E.D. Mich.||18||D. Vt.||7|
|N.D. Ill.||148||D. Nev.||18||S.D. Iowa||6|
|D.N.J.||108||D. Md.||17||W.D. Va.||6|
|S.D. Cal.||79||D.D.C.||16||N.D. W.Va.||6|
|M.D. Fla.||75||N.D. Tex.||16||W.D. Wis.||6|
|S.D. Fla.||62||S.D. W.Va.||16||S.D. Ala.||5|
|S.D. Ill.||59||D. Ariz.||15||C.D. Ill.||5|
|D. Minn.||50||D. Colo.||15||D. Neb.||5|
|N.D. Ga.||46||E.D. Va.||15||N.D. Okla.||4|
|E.D.N.Y.||45||S.D. Miss.||14||W.D. Tenn.||4|
|E.D. La.||43||W.D. Okla.||13||D. Idaho||3|
|D. Mass.||43||D.N.H.||12||N.D. Iowa||3|
|E.D. Tex.||42||M.D. Ga.||11||D. Mont.||3|
|D. Del.||36||D. Utah||11||M.D.N.C.||3|
|S.D. Ohio||36||E.D. Ky.||10||D. Haw.||2|
|S.D. Ind.||35||M.D. La.||10||D. Me.||2|
|D.S.C.||35||N.D. Miss.||10||W.D. Mich.||2|
|M.D. Tenn.||33||E.D. Wash.||10||W.D.N.Y.||2|
|E.D. Mo.||32||E.D. Wis.||10||D. Alaska||1|
|W.D. La.||30||E.D. Ark.||9||D.N.D.||1|
|N.D. Ohio||30||W.D. Pa.||9||E.D. Okla.||1|
|N.D. Ala.||27||S.D. Tex.||9||D.P.R.||1|
|W.D. Mo.||23||W.D. Tex.||9||D.V.I.||1|
|D. Kan.||22||M.D. Ala.||8||D. Wyo.||1|
|W.D. Ky.||20||S.D. Ga.||8||D. Guam||0|
|E.D. Tenn.||20||D. Or.||8||D.N. Mar. I.||0|
|E.D. Cal.||19||W.D. Ark.||7||D.S.D.||0|
Appendix Table C: Percentage of Transferor Cases in the Plaintiff’s First-Choice
This table identifies the percentage of transferor cases in the plaintiffs’ first-choice district using 10% bands. If cases involved a percentage that was a multiple of ten (e.g., 20%), we included that with the higher band (e.g., 20–30%).
|Plaintiffs’ first-choice district had 0% of transferor cases||9 cases|
|Plaintiffs’ first-choice district had 1–10% of transferor cases||13 cases|
|Plaintiffs’ first-choice district had 10–20% of transferor cases||26 cases|
|Plaintiffs’ first-choice district had 20–30% of transferor cases||36 cases|
|Plaintiffs’ first-choice district had 30–40% of transferor cases||30 cases|
|Plaintiffs’ first-choice district had 40–50% of transferor cases||12 cases|
|Plaintiffs’ first-choice district had 50–60% of transferor cases||29 cases|
|Plaintiffs’ first-choice district had 60–70% of transferor cases||19 cases|
|Plaintiffs’ first-choice district had 70–80% of transferor cases||13 cases|
|Plaintiffs’ first-choice district had 80–90% of transferor cases||11 cases|
|Plaintiffs’ first-choice district had 90–100% of transferor cases||4 cases|
|Plaintiffs opposed consolidation and offered no preferred transferee district||5 cases|
Appendix Table D: Nominations by District
This table identifies the frequency with which only plaintiffs, only defendants, and a mix of plaintiffs and defendants s nominated particular districts. It also identifies the total number of nominations, the percentage of those nominations from plaintiffs, and the frequency that district was selected as the transferee district (among cases in which it was nominated).
|District||P-only||D-only||P&D||Total||P %||Transferee %|
|D. Me.||0||0||1||2||50%||10 0%|
|D.N. Mar. I.||0||0||0||0||0%||0%|
|N.D. W. Va.||2||0||0||2||100%||50%|
|S.D. W. Va.||1||1||3||8||50%||80%|
Appendix Table E: Comparing 2005–09 and 2012–16
This Table compares our results to Williams and George Table 3. We do not include the second column from Williams and George because it does not account for the number of suggestions per case, which at least in our period, dramatically affects the results such that they are not substantively meaningful.
|Variable||Percentage of all potential districts with characteristic (2005–09)||Percentage of all potential districts with characteristic (2012–16)|
|District supported by plaintiffs only||68.39%||59.64% (426)|
|Districts supported by defendants only||37.76%||16.0% (117)|
|Districts supported by both plaintiffs and defendants||11.65%||23.6% (170)|
|District represented on Panel||11.76%||20.25% (148)|
|District represented on Panel in the past||41.64%||50.88% (374)|
Appendix Figure A: Race Detail
Zachary D. Clopton & Andrew D. Bradt: Professor of Law, Northwestern Pritzker School of Law; Professor of Law, University of California, Berkeley School of Law (Boalt Hall). Thank you to Steve Burbank, Dawn Chutkow, Kevin Clermont, Scott Dodson, Maggie Gardner, William Hubbard, Sam Issacharoff, Linda Mullenix, Jeff Rachlinski, and participants in the Federal Courts Workshop and the NYU Center on Civil Justice Symposium. Thanks also to Melissa Decker, Rohini Tashima, and Doug Wagner for superb research assistance, and to the California Law Review for all of their excellent work on this Article.
. See generally Elizabeth Chamblee Burch, Judging Multidistrict Litigation, 90 N.Y.U. L. Rev. 71 (2015) (noting the importance of MDL) [hereinafter Burch, Judging]; Abbe R. Gluck, Unorthodox Civil Procedure: Modern Multidistrict Litigation’s Place in the Textbook Understandings of Procedure, 165 U. Pa. L. Rev. 1669 (2017) (same).
. Andrew D. Bradt, Something Less and Something More: MDL’s Roots as a Class Action Alternative, 165 U. Pa. L. Rev. 1711, 1740 (2017) (“with MDL’s ascendance has also come increased scrutiny”); see, e.g., Arthur R. Miller, The Preservation and Rejuvenation of Aggregate Litigation: A Systemic Imperative, 64 Emory L.J. 293, 312 (2014) (questioning certain procedural innovations in MDL); Martin H. Redish & Julie M. Karaba, One Size Doesn’t Fit All: Multidistrict Litigation, Due Process, and the Dangers of Procedural Collectivism, 95 B.U. L. Rev. 109, 111 (2015) (criticizing MDL’s protection of substantive rights).
. H.R. 985, 115th Cong. § 104 (2017); Press Release, Goodlatte House Office, Goodlatte Introduces Major Litigation Reform Bill to Improve Access to Justice for American Consumers (Feb. 10, 2017), http://goodlatte.house.gov/news/documentsingle.aspx?DocumentID=809 (promoting introduction of bill in House Judiciary Committee). The bill ultimately languished in the Senate and the now Democratically controlled House has (predictably) not revisited the statute. For more on the fate of the bill and the proposals within it, see Howard M. Erichson, Searching for Salvageable Ideas in FICALA, 87 Fordham L. Rev. 19 (2018).
. Advisory Committee on Civil Rules, Agenda Book (Nov. 7, 2017), at 477 (“[T]he question is whether the time has come to undertake an effort to generate rules specially adapted to MDL proceedings.”). For criticism of these efforts, see Zachary D. Clopton, MDL as Category, 105 Cornell L. Rev. (forthcoming 2020) [hereinafter Clopton, MDL as Category], https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3354742 [https://perma.cc/DW86-MDZT].
. See In re FMC Corp. Patent Litig., 422 F. Supp. 1163, 1165 (J.P.M.L. 1976) (per curiam) (“Transfers under Section 1407 are simply not encumbered by considerations of in personam jurisdiction and venue.”).
. See, e.g., In re “Agent Orange” Prod. Liab. Litig., 996 F.2d 1425, 1435 (2d Cir. 1993) (noting that transfers under Section 1407 are not encumbered by personal jurisdiction concerns); In re FMC Corp. Patent Litig., 422 F. Supp. at 1165 (same).
. See Margaret S. Williams & Tracey E. George, Who Will Manage Complex Civil Litigation? The Decision to Transfer and Consolidate Multidistrict Litigation, 10 J. Empirical Legal Stud. 424, 436 (2013) (“Parties often agree that transfer and consolidation is appropriate, but they just as often sharply disagree about where and in front of whom.”). In our study, the Panel selected a district suggested by one or more parties in more than 96% of cases.
. See Heyburn, supra note 9, at 2230 (“Only 37 % out of about 300 active MDLs comprise more than 100 constituent actions and only ten have more than 1000. By contrast, about one-half of all open MDLs are comprised of ten or fewer actions.”). See generally Statistical Information, U.S. Jud. Panel on Multidistrict Litig., http://www.jpml.uscourts.gov/statistics-info [https://perma.cc/DEQ6-JHAD] (collecting reports describing MDL cases).
. Audio Recording: Elizabeth Cabraser, MDL Problems, Proceedings of the Section on Litigation, Annual Meeting of the American Association of Law Schools (Jan. 6, 2017) (on file with the American Association of Law Schools).
. See Stephen B. Burbank & Sean Farhang, Rights and Retrenchment: The Counterrevolution Against Federal Litigation 244 (2017) (noting that “a series of Chief Justices, all of whom were appointed by Republican presidents . . . have ensured that a greatly disproportionate share of [committee] appointments went to judges who were themselves appointed by Republican presidents”); infra Part IV.D (providing data on JPML demographics).
. Notably, during the same period, the Panel itself was almost half women but 100 % white, suggesting that the Chief Justice’s appointments to the Panel may have a trickle-down effect for gender and racial diversity among transferee judges.
. Andrew D. Bradt, “A Radical Proposal”: The Multidistrict Litigation Act of 1968, 165 U. Pa. L. Rev. 831, 906 (2017) (describing circumstances of passage of the statute) [hereinafter Bradt, Radical Proposal]; Linda S. Mullenix, Aggregate Litigation and the Death of Democratic Dispute Resolution, 107 Nw. U. L. Rev. 511, 552 (2013) (describing MDL as once a “judicial backwater”); Judith Resnik, From “Cases” to “Litigation”, 54 Law & Contemp. Probs. Summer 1991, at 5, 47 (describing MDL as a “sleeper”).
. See MDL Statistics Report – Distribution of Pending MDL Dockets by Actions Pending, U.S. Jud. Panel on Multidistrict Litig. (Apr. 16, 2018), https://www.jpml.uscourts.gov/sites/jpml/files/Pending_MDL_Dockets_By_Actions_Pending-April-16-2018.pdf; Statistical Tables for the Federal Judiciary, U.S. District Cts., at tbl. C (Dec. 31, 2017), http://www.uscourts.gov/statistics/table/c/statistical-tables-federal-judiciary/2017/12/31 [https://perma.cc/Q239-53PQ]. For important context relevant to these statistics, see Clopton, MDL as Category, supra note 4.
. 28 U.S.C. § 1407(d) (2012) (providing that “[t]he judicial panel on multidistrict litigation shall consist of seven circuit and district judges designated from time to time by the Chief Justice of the United States, no two of whom shall be from the same circuit”).
. Brooke D. Coleman, One Percent Procedure, 91 Wash. L. Rev. 1005, 1032 (2016) (noting practice that Panel members serve seven-year terms); Heyburn, supra note 9, at 2227 (noting that “then-Chief Justice William H. Rehnquist imposed some regularity and predictability on the appointment process by establishing staggered seven-year terms for each member”).
. 28 U.S.C. § 1407(e) (“No proceedings for review of any order of the panel may be permitted except by extraordinary writ . . . . There shall be no appeal or review of an order of the panel denying a motion to transfer for consolidation or coordinated proceedings.”).
. See Andrew D. Bradt, The Long Arm of Multidistrict Litigation, 59 Wm. & Mary L. Rev. 1165, 1169–70 (2018) [hereinafter Bradt, Long Arm] (noting that the Panel can locate an MDL “anywhere in the United States, essentially without limitation”).
. See, e.g., In re FMC Corp. Patent Litig., 422 F. Supp. 1163, 1165 (J.P.M.L. 1976) (“Transfers under Section 1407 are simply not encumbered by considerations of in personam jurisdiction and venue.”).
. See Bradt, Radical Proposal, supra note 20, at 839 (noting that the supporters of the statute recognized that the control of MDL cases should be centralized in the hands of judges “committed to strong pretrial case management”).
. See generally Judith Resnik, Managerial Judges, 96 Harv. L. Rev. 374, 376 (1982) (noting the attitude transition of federal judges from a “relatively disinterested pose” to a “more active, managerial stance”).
. See generally Bradt, Long Arm, supra note 33, at 1215–21 (arguing that although geography is a factor, finding a “willing and able” judge is more dispositive for the choice of transferee jurisdiction).
. See Bradt, Long Arm, supra note 33, at 1169 (“While the cases are in the MDL court, the MDL judge has all of the powers that the transferor court would have, including the power to decide dispositive motions, and typically, the litigation is resolved by a mass-settlement agreement reached within the MDL.”).
. See 15 Charles Alan Wright & Arthur R. Miller, Federal Practice and Procedure § 3866.2 (4th ed. 2019) (noting that Section 1407 does not afford a transferee court the authority to try cases without parties’ consent); Eldon E. Fallon et al., Bellwether Trials in Multidistrict Litigation, 82 Tul. L. Rev, 2323, 2354–55 (2008) (observing that while a transferee court cannot try transferred cases without the consent of the parties, the transferee court may obtain trial authority by allowing out-of-district plaintiffs to file cases directly in the transferee court).
. See Andrew D. Bradt & D. Theodore Rave, Aggregation on Defendants’ Terms: Bristol-Myers Squibb and the Federalization of Mass-Tort Litigation, 59 B.C. L. Rev. 1251, 1268–74 (2018) (noting that the Supreme Court has restricted general and specific personal jurisdiction in recent years).
. John C. Coffee, Jr., Entrepreneurial Litigation: Its Rise, Fall, and Future 116 (2015) (describing MDL as “group litigation that is the functional equivalent to a class action [that] has come to supplant the class action in the mass tort field”). During oral argument in the personal-jurisdiction case Bristol-Myers Squibb v. Superior Court, the United States supported a tight standard for personal jurisdiction in part because MDL provided a backstop for mass torts. See Transcript of Oral Argument at 28, Bristol-Meyers Squibb Co. v. Superior Court, 137 S. Ct. 1773 (2017) (No. 16-466) (“Justice Breyer: What is your solution to mass torts? [Counsel for United States]: So we think there are a couple solutions. Claims like this, which are, I think, mass tort claims can be brought in a jurisdiction of general jurisdiction. They can also be brought in Federal courts and consolidated for schemes like the MDL scheme. These are both solutions that provide efficiency.”).
. See Bradt & Rave, supra note 45, at 1318–19 (concluding that, in light of the difficulties in obtaining class certification and personal jurisdiction, MDL has become “the best available playing field for mass-tort litigation”).
. See generally Dawn M. Chutkow, The Chief Justice as Executive: Judicial Conference Committee Appointments, 2 J. L. & Cts. 301–25 (2014); Tracey E. George & Margaret S. Williams, Designing Judicial Institutions: Special Federal Courts and the U.S. Judicial Panel on Multidistrict Litigation 11–21 (2013) (unpublished manuscript), https://www.vanderbilt.edu/csdi/events/George.pdf [https://perma.cc/GWZ7-DFXL] (summarizing the characteristics of the MDL Panel members).
. More specifically, retracing some analyses for recent years is important for at least two reasons. First, the importance and attention to MDL has only grown, so it is not inconceivable that recent MDL is fundamentally different from the prior period. Second, it would be surprising if the story of MDL were static. Certainly parties (and their lawyers) have an incentive to learn how the Panel operates and respond strategically. Moreover, there is reason to believe that the Panel itself is conscious of its prior performance. Indeed, we suspect that the Panel is aware of Williams and George’s study in particular. For these reasons, further study is warranted.
. See Appendix. These data reveal that in about three-quarters of cases, the plaintiffs’ first-choice district is also the district with the most transferor cases (and vice versa). This is consistent with our intuition that plaintiffs are expressing preferences for certain districts by filing cases there and by suggesting these districts to the Panel. However, this intuition alone does not explain why in almost a quarter of cases we do not see this identity. And even when plaintiffs’ first-choice district is also home to the most transferor cases, one should not assume that the district is home to all or even nearly all transferor cases. For example, in less than a sixth of cases do we observe that 70% or more transferor cases are from the plaintiffs’ first-choice district. See id.
. For data for these figures, see Calendar Year Statistics: January through December 2016, U.S. Jud. Panel on Multidistrict Litig., http://www.jpml.uscourts.gov/sites/jpml/files/JPML_Calendar_Year_Statistics-2016.pdf [https://perma.cc/AZ3K-NC73]. Note also that the denominator—the number of motions filed—is also declining in recent years.
. Looking at MDL transfer motions filed between 2012 and 2016 (a slightly larger population than looking at MDLs consolidated during the period), we find that the Panel transferred (rather than denied) 63.2% of plaintiff motions (122 out of 193) and 63.6% of defendant motions (63 out of 99). These results do not include those motions on which the Panel did not rule, for example because the case settled after the motion but before resolution. For completeness, we note that the Panel did not rule on about 32% of motions to consolidate. Plaintiffs and defendants filed about equal numbers of those not-ruled-upon motions. Again, these motions are excluded from our percentages above. See Statistical Information, supra note 13.
. Somewhat relatedly, we observed little party preference in cases in which the Panel granted motions to consolidate over the objections of one or parties. In particular, in cases that were ultimately consolidated, we found forty-two MDLs in which one or more plaintiffs (and no defendants) opposed consolidation, and twelve MDLs in which a mix of plaintiffs and defendants opposed consolidation.
. For purposes of our regression analysis, we also collected a few control variables that have been used or identified in prior research. See generally Williams & George, supra note 10. In particular, for each suggested district, we recorded whether it was represented on the Panel at the time of the decision or any time prior to the decision, its share of transferor cases, its weighted average of cases per judgeship, and various features of the ultimate transferee district. Most of this information is available on the transfer order. Current and historical JPML membership was determined by relying on the Panel’s public information about membership. The weighted average of filings per judgeship is a statistic published by the Administrative Office of U.S. Courts. We used the averages from the twelve-month period ending September 30, 2014, which we believe is a reasonable proxy for our study period.
. And were one further inclined, one might call these “MD-Hellholes.” Cf., Judicial Hellholes, Am. Tort Reform Found., http://www.judicialhellholes.org [https://perma.cc/89JL-HE82] (characterizing pro-plaintiff jurisdictions as “hellholes”).
. Of course, these data do not account for the number of suggestions per case—e.g., for purposes of “success rate,” a case in which ten districts were nominated is treated the same as a case in which two districts were nominated.
. Note also that because the decision to consolidate often occurs relatively early on, the number of cases that wind up in the MDL can be much larger than the number at the moment of JPML action, or the number might stay the same.
. Williams and George asked similar questions for the period 2005 to 2009. We compare our results to theirs in the Appendix. Of note, the percentage of districts supported by both plaintiffs and defendants is twice as high in our period, reducing the percentage of defendant-only suggestions dramatically and the percentage of plaintiff-only suggestions somewhat. See Appendix E.
. We do not report above the line any statistics about the “effectiveness” of nominations. We find those statistics not particularly helpful because parties (especially plaintiffs) may suggest multiple jurisdictions in a single case. So, for example, if a plaintiff suggested four districts, the best she can hope for is a 25 % success rate. In any event, for the interest of readers, we calculate those statistics here. For plaintiffs, 165 of 609 suggestions were accepted, or 27 %. For defendants, 147 of 288 suggestions were accepted, or 51 %.
. Common explanations for selecting districts in these cases include the presence of related cases or MDLs in the district (that would not be formally consolidated with the transferee cases), geographic centrality, or proximity to the defendants’ headquarters.
. Of those 32 cases, the Panel selected a district supported only by plaintiffs in 23 cases, a district supported only by defendants in 4 cases, and a district supported by no party in 5 cases. Note, though, that plaintiffs nominate dramatically more districts than defendants.
. Using a logit model that roughly tracks prior research, we find that districts supported by a mix of plaintiffs and defendants are very likely to be selected. See Williams & George, supra note 10. We also find weak evidence that defendant-only suggestions outperform plaintiff-only suggestions, but these results are far from certain. Because these results add little to the description statistics, we do not reprint them here. Instead, results are on file with authors.
. See, e.g., In re: Biomet M2A Magnum Hip Implant Prods. Liab. Litig., 896 F. Supp. 2d 1339, 1340 (J.P.M.L. 2012); In re: DePuy Orthopaedics, Inc., Pinnacle Hip Implant Prods. Liab. Litig., 787 F. Supp. 2d 1358, 1360 (J.P.M.L. 2011).
. These results are roughly consistent with the results for MDLs with more than 100 cases. Among MDLs with more than one hundred cases but fewer than one thousand cases, the Panel selected a district supported by plaintiffs 72% of the time, or 86% of the time when there was such a district. We also separately test products liability cases, which tend to be larger. Similar results obtained (and are on file with authors).
. Here, too, the results are roughly consistent with those for MDLs with more than 100 cases. Among MDLs with more than one hundred cases but fewer than one thousand cases, when the Panel selected a district supported by only one side, it selected plaintiff-only districts 71% of the time, and defendant-only districts 29% of the time. We also separately test products liability cases, which tend to be larger. Similar results obtained (and are on file with authors).
. Here, we used a logistic regression with an index variable to determine whether the suggestion was more favored by plaintiffs or defendants: (i) district ranked higher by plaintiffs than defendants; (ii) district ranked higher by defendants than plaintiffs; (iii) district ranked equally by plaintiffs and defendants. Ties were mostly likely to be selected, and there is some (weak) evidence of defendant advantage. Results are on file with authors.
. Clermont and Eisenberg found that the success rate of § 1404(a) transfer motions is about 45%, with reported cases skewing toward 50%. Kevin M. Clermont & Theodore Eisenberg, Exorcising the Evil of Forum-Shopping, 80 Cornell L. Rev. 1507, 1529 (1995).
. Donald Earl Childress III, When Erie Goes International, 105 Nw. U. L. Rev. 1531, 1563 (2011) (finding a forum non conveniens grant rate of 52%); Michael T. Lii, An Empirical Examination of the Adequate Alternative Forum in the Doctrine of Forum Non Conveniens, 8 Rich. J. Global L. & Bus. 513, 526 (2009) (finding a forum non conveniens grant rate of 52%); Christopher A. Whytock, The Evolving Forum Shopping System, 96 Cornell L. Rev. 481, 502 (2011) (“U.S. district courts dismiss transnational claims on forum non conveniens grounds at an estimated rate of 47.1% in their published decisions.”).
. This is not technically an application of the Priest-Klein hypothesis, but it shares some of its intuitions. See generally George L. Priest & Benjamin Klein, The Selection of Disputes for Litigation, 13 J. Legal Stud. 1 (1984); George L. Priest, Reexamining the Selection Hypothesis: Learning from Wittman’s Mistakes, 14 J. Legal Stud. 215 (1985).
. See Andrew D. Bradt & Zachary D. Clopton, MDL v. Trump: The Puzzle of Public Law in Multidistrict Litigation, 112 Nw. U. L. Rev. 905, 926 (2018) (discussing the benefits of a lack of politicization of the Panel).
. Of course, plaintiff forum choice might be disrupted vertically through removal. See 28 U.S.C. §§ 1441–54 (2012). But removal can precede any venue transfer option (or none)—it is not a unique problem for MDL plaintiffs.
. Looking at the Panel’s historical data, we also observe that about 40% of cases were filed in the transferee district, with 60% having been filed elsewhere. See Statistical Information, supra note 13.
. See, e.g., Zachary D. Clopton, Procedural Retrenchment and the States, 106 Calif. L. Rev. 411 (2018); Bradt & Rave, supra note 45. Importantly, we cannot say that plaintiffs in MDL perform worse on the merits than in normal federal litigation. Cf. Clermont & Eisenberg, supra note 76, at 1511–12 (“In recent federal civil cases, the plaintiff wins in 58% of the nontransferred cases that go to judgment for one side or the other, but wins in only 29% of such cases in which a transfer occurred.”). Nor do we express an opinion on whether federal-court merits decisions should be more or less plaintiff-friendly.
. We, therefore, might want to pay special attention to MDL cases resolved without resorting to Rule 23. See, e,g., Elizabeth Chamblee Burch, Procedural Justice in Nonclass Aggregation, 44 Wake Forest L. Rev. 1 (2009); Linda S. Mullenix, Policing MDL Non-Class Settlements: Empowering Judges Through the All Writs Act, 37 Rev. Lit. 129 (2018).
. And because these very large MDLs are drawing cases originally filed around the country, it is likely impossible to respect all plaintiffs’ forum preference while also allowing plaintiffs to aggregate.
. See, e.g., Stephen B. Burbank & Sean Farhang, Federal Court Rulemaking and Litigation Reform: An Institutional Approach, 15 Nev. L.J. 1559, 1571–76 (2015); Chutkow, supra note 51; Maxwell Palmer, Does the Chief Justice Make Partisan Appointments to Special Courts and Panels?, 13 J. Empirical Leg. Stud. 153 (2016); James E. Pfander, The Chief Justice, The Appointment of Inferior Officers, and the “Court of Law” Requirement, 107 Nw U. L. Rev. 1125 (2013); Judith Resnik & Lane Dilg, Responding to a Democratic Deficit: Limiting the Powers and the Term of the Chief Justice of the United States, 154 U. Pa. L. Rev. 1575 (2006); Theodore W. Ruger, The Judicial Appointment Power of the Chief Justice, 7 U. Pa. J. Const. L. 341 (2004); Elizabeth Thornburg, Cognitive Bias, the “Band of Experts,” and the Anti-Litigation Narrative, 65 DePaul L. Rev. 755, 765–66 (2016); Charlie Savage, Roberts’s Picks Reshaping Secret Surveillance Court, N.Y. Times (July 25, 2013), https://www.nytimes.com/2013/07/26/us/politics/robertss-picks-reshaping-secret-surveillance-court.html [https://perma.cc/SK5R-ZGJ2].
. See, e.g., Chutkow, supra note 51; Zachary D. Clopton, Making State Civil Procedure, 104 Cornell L. Rev. 1 (2018) [hereinafter Clopton, Making State Civil Procedure]; Brooke D. Coleman, A Legal Fempire?: Women in Complex Civil Litigation, 93 Ind. L.J. 617 (2018); Brooke D. Coleman, #SoWhiteMale: Federal Civil Rulemaking, 113 Nw. U. L. Rev. 407 (2018); Katherine Shaw, Friends of the Court: Evaluating the Supreme Court’s Amicus Invitations, 101 Cornell L. Rev. 1533 (2016).
. The comparison between the overall population of judges and MDL judges may be affected by the statutory requirement that both the Chief Judge of the transferee district and the transferee judge accept the MDL assignment. Because the Panel does not report rejections of MDL cases, it is difficult to know how much this statutory requirement matters. But we should acknowledge that the Panel’s choices can be an interactive process, rather than purely unilateral decisions by the Panel.
. For more on the sources used to compile the directory, see Biographical Directory of Article III Federal Judges: About the Directory, Fed. Jud. Ctr., https://www.fjc.gov/history/judges/biographical-directory-article-iii-federal-judges-about-directory [https://perma.cc/86MG-Q7R7].
. Based on the way the Panel reports its data, we would not see judges that had an MDL but then had it taken away before termination. However, to avoid these errors, we also searched our data for any repeat judges, and we reviewed all JPML explanations for picking judges to identify any language that contradicted our coding. The first method revealed one judge that we changed from “new” to “old.” The second method did not reveal any potential errors.
. Of course, not all judges were potential transferees every day during the study. We selected the midpoint as a simple approximation of the potential judges available for MDL assignments across our study period. Though, we discuss alternative approximations where relevant.
. One additional datum not reported in the table: of the 207 transferees, four were made to judges currently on the Panel, three were made to judges who previously served on the Panel, and seven were made to judges later serving on the Panel.
. Because not all senior judges accept a full caseload, the pure number of senior judges is misleading. The Administrative Office of U.S. Courts estimates that senior judges handle approximately 15 % of the federal docket. See FAQs: Federal Judges, U.S. Cts., http://www.uscourts.gov/faqs-federal-judges [https://perma.cc/ZA4T-QCK4]. But see generally Stephen B. Burbank et al., Leaving the Bench, 1970–2009: The Choices Federal Judges Make, What Influences Those Choices, and Their Consequences, 161 U. Pa. L. Rev. 1 (2012).
. Using a Fisher’s exact test, the party of the appointing president is significantly different from the appointing president of all district judges on the first day of our sample (p = 0.0073) but not on the middle day (p = 0.2876) or last day (p = 0.8190).
. There is a difference in the composition of nonwhite judges. For new MDL judges, nonwhite judges were either African American (18%) or Hispanic (3%). Among all transferees, nonwhite judges were more evenly distributed as African American (8%), Hispanic (5%), Asian American (4%), and other/mixed (less than 1%).
. The means are even more extreme—the mean MDL assigned to an old judge had four times as many cases as the mean MDL assigned to a new MDL judge—but these numbers are skewed based on a few very large MDLs assigned to older judges.
. For example, looking at MDLs consolidated prior to our period, we find 23 terminated MDLs that had more than 1,000 cases. All 23 were closed by white judges, and 19 of 23 (83%) were closed by male judges. Expanding the pool to MDLs with 500 cases or more, we find only one nonwhite judge (out of 34) and the percentage of men goes up slightly to 85%. We compiled these results using the Panel’s list of terminated MDLs from July 2018 and the usual sources for judge demography. See supra Part III.
. Although new MDL judges are more likely to be appointed by a Democratic president than all district judges, they are closer to the overall population than the Panel. And because President Obama appointed many judges who were MDL candidates, it is unsurprising that new MDL judges were often Obama appointees.
. We do not have a clear explanation for the partisan patterns. It may be that the experienced judges interested in unpaid service on the Panel happen to be Clinton appointees. Or that the Panel judges are conscious of the risk of partisan bias. Or that the Chief Justice and the Panel judges he selects are committed to the efficient and effective resolution of complex cases without regard for partisan valence. Indeed, it is possible that the Chief consciously selects judges for the Panel that he has confidence will not use their power ideologically. Whatever the causal explanation, ideology has not been a major force in transferee judge selection—a trend that we hope will continue. But see Bradt & Clopton, supra note 79 (identifying risks for bipartisan MDL).