Discussion: Balancing Justice and Judicial Efficiency

LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM DEFINING THE PROBLEM OF COST IN FEDERAL CIVIL LITIGATION EMERY G. LEE III† T HOMAS E. W ILLGING †† T ABLE OF CONTENTS Introduction .............................................................................................765 I. Is Litigation Too Expensive? ............................................................769 II. Relative to Stakes?............................................................................771 III. Models for Reform? ........................................................................776 IV. Rush to Judgment? ..........................................................................779 Conclusion................................................................................................786 Appendix ..................................................................................................788 I NTRODUCTION We begin with a prediction: At some point in the relatively near future, the 2010 Civil Litigation Review Conference (Duke Conference) 1 will be labeled a failure. 2 We can say this with a high degree of confidence because the conference’s most notable precedent, the National Conference on the Causes of Popular Copyright © 2010 by Emery G. Lee III & Thomas E. Willging.

† Senior researcher at the Federal Judicial Center (FJC). Affiliation is provided for identification purposes only. The views expressed herein represent those of the authors and not the views of the FJC or any other judicial-branch entity; therefore, any use of “we” or “our” in this Article refers solely to the authors. The authors wish to acknowledge the assistance and comments of several FJC colleagues—George Cort, Meghan Dunn, Margaret Williams, and Jill Curry.

†† Retired as a senior researcher at the FJC on July 1, 2010. 1. The Civil Litigation Review Conference, sponsored by the Advisory Committee on Civil Rules of the Judicial Conference, was held at Duke University School of Law on May 10– 11, 2010. An impressive group of judges, attorneys, and researchers convened to discuss the current state of civil litigation in the federal courts and potential reforms of the Federal Rules of Civil Procedure.

2. To be clear, this is decidedly not the authors’ view. LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 766 DUKE LAW JOURNAL [Vol. 60:765 Dissatisfaction with the Administration of Justice (Pound Conference), has been called a failure, most recently in a joint report of the American College of Trial Lawyers (ACTL) and the Institute for the Advancement of the American Legal System (IAALS). 3 Chief Justice Burger called the Pound Conference in 1976 to address a number of issues facing the American legal system, including the problem of discovery abuse. 4 Following the Pound Conference, significant revisions to the Federal Rules of Civil Procedure were proposed, debated, and, in many cases, adopted—decades of what Professor Richard L. Marcus has termed “discovery containment.” 5 But as a joint report by the ACTL and IAALS sharply concludes, “There is substantial opinion that all of those efforts have accomplished little or nothing.” 6 To the list of allegedly failed conferences, one might add the 1997 conference at Boston College, convened by the Judicial Conference Advisory Committee on Civil Rules (Committee) to focus on cost, delay, and abuses in the pretrial discovery process.

Expressing the refrain, “Here we go again,” Judge Paul Niemeyer, then–Committee Chair, reprised the course of changes in discovery rules in the years since the Pound Conference. 7 He asserted that changes to the discovery rules in 1980, 1983, and 1993 aimed “to curtail the expansiveness of discovery, but they have either failed or been so diluted as to have little effect.” 8 3. See A M. COLL . OF TRIAL LAWYERS & INST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., FINAL REPORT ON THE JOINT PROJECT OF THE AMERICAN COLLEGE OF TRIAL LAWYERS TASK FORCE ON DISCOVERY AND THE INSTITUTE FOR THE ADVANCEMENT OF THE AMERICAN LEGAL SYSTEM 9–10 (2009), available at http://www.actl.com/AM/Template.cfm?

Section=Home&template=/CM/ContentDisplay.cfm&ContentID=4008.

4. Am. Bar Ass’n, Report of Pound Conference Follow-Up Task Force, in THE POUND CONFERENCE : PERSPECTIVES ON JUSTICE IN THE FUTURE 295, 318 (A. Leo Levin & Russell R.

Wheeler eds., 1979) (“Substantial criticism has been leveled at the operation of the rules of discovery. It is alleged that abuse is widespread, serving to escalate the cost of litigation, to delay adjudication unduly and to coerce unfair settlements. Ordeal by pretrial procedures, it has been said, awaits the parties to a civil law suit.”).

5. See, e.g., Richard L. Marcus, Discovery Containment Redux, 39 B.C. L. R EV. 747, 747– 48 (1998) (“[S]ince 1976, proposals for amendment to the rules have generally involved retreats from the broadest concept of discovery—in essence to try to contain the genie of broad discovery without killing it.”).

6. A M. COLL . OF TRIAL LAWYERS & INST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., supra note 3, at 10.

7. Paul V. Niemeyer, Here We Go Again: Are the Federal Discovery Rules Really in Need of Amendment?, 39 B.C. L. R EV. 517, 519–21 (1998).

8. Id. at 519. LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 2010] DEFINING THE PROBLEM OF COST 767 A theme begins to emerge. After further rule amendments in 2000 and 2006, the complaints are louder than ever. The ACTL- IAALS joint report stated, for example, The existing rules structure does not always lead to early identification of the contested issues to be litigated, which often leads to a lack of focus in discovery. As a result, discovery can cost far too much and can become an end in itself. As one respondent noted: “The discovery rules in particular are impractical in that they promote full discovery as a value above almost everything else.” Electronic discovery, in particular, needs a serious overhaul. It was described by one respondent as a “morass.” Another respondent stated: “The new rules are a nightmare. The bigger the case the more the abuse and the bigger the nightmare.” 9 Again, rule amendments have only resulted in failure. These complaints did not fall on deaf ears. At its fall 2008 meeting, the Committee voted to proceed with a major conference on the state of civil litigation in the federal courts. In support of that conference, the Committee—following the path it pursued in 1997 10—asked the Federal Judicial Center (FJC) to conduct an empirical study of civil litigation, especially with respect to costs.

The findings of that study have been reported elsewhere, 11 and, 9. A M. COLL . OF TRIAL LAWYERS & INST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., supra note 3, at 2.

10. See Niemeyer, supra note 7, at 521–22 (“The Committee engaged the Federal Judicial Center to study the expense of discovery as well as related questions and to report to the Boston Conference.”); see also Thomas E. Willging, Donna Stienstra, John Shapard & Dean Miletich, An Empirical Study of Discovery and Disclosure Practice Under the 1993 Federal Rule Amendments, 39 B.C. L. R EV. 525, 526 (1998) (“[T]he Advisory Committee on Civil Rules . . .

asked the FJC to conduct research on discovery as part of a Committee decision to undertake a comprehensive examination of that subject.”).

11. E MERY G. LEE III & THOMAS E. WILLGING , FED . JUDICIAL CTR., NATIONAL , CASE - B ASED CIVIL RULES SURVEY : PRELIMINARY REPORT TO THE JUDICIAL CONFERENCE ADVISORY COMMITTEE ON CIVIL RULES (2009) [hereinafter L EE & WILLGING , CASE -BASED CIVIL RULES SURVEY ], available at http://www.fjc.gov/public/pdf.nsf/lookup/dissurv1.pdf/$file/ dissurv1.pdf; see also EMERY G. LEE III & THOMAS E. WILLGING , FED . JUDICIAL CTR., LITIGATION COSTS IN CIVIL CASES : MULTIVARIATE ANALYSIS : REPORT TO THE JUDICIAL CONFERENCE ADVISORY COMMITTEE ON CIVIL RULES 2–4 (2010) [hereinafter L EE & WILLGING , MULTIVARIATE ANALYSIS ], available at http://www.fjc.gov/public/pdf.nsf/lookup/ costciv1.pdf/$file/costciv1.pdf (“This report, prepared for the Committee’s March 2010 meeting, presents multivariate analysis of litigation costs in the closed cases.”); T HOMAS E. WILLGING & EMERY G. LEE III, FED . JUDICIAL CTR., IN THEIR WORDS : ATTORNEY VIEWS ABOUT COSTS AND PROCEDURES IN FEDERAL CIVIL LITIGATION 1–2 (2010) [hereinafter W ILLGING & LEE, IN THEIR WORDS ], available at http://www.fjc.gov/public/pdf.nsf/lookup/costciv3.pdf/$file/ costciv3.pdf (“To supplement the multivariate analysis, . . . the Center . . . interview[ed] a LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 768 DUKE LAW JOURNAL [Vol. 60:765 given space constraints, we will not attempt in this Article a summary drained of nuance and precision. Instead, we propose to survey the research, including the FJC reports, prepared for the Duke Conference, to define what, exactly, the problem is with civil litigation. In Part I, we argue that the problem cannot be simply that “litigation is too expensive.” 12 Without a normative standard, it is impossible to say, in any meaningful way, that litigation is too expensive. Moreover, the limited empirical evidence that exists does not support the broad statement that litigation costs, in general, are out of control. In Part II, we discuss our finding that the stakes in the litigation are, empirically, the best predictor of costs. Indeed, in most federal civil cases, the costs appear to be proportionate to the monetary stakes. If so, the problem is not out-of-control costs generally. Nevertheless, there is a desire in some quarters to find general solutions to the as-yet-undefined problem of too-expensive litigation. The usual suspects are the pretrial discovery rules. But, as in Casablanca, 13 the usual suspects are often not the perpetrators. In Part III, we demonstrate that there is little reason to think that the state procedural limits on discovery advanced as models for federal- rules reform accomplish the goals set out for them. Part IV argues, based on empirical research, that there is scant evidence that alternative discovery rules would result in lower costs or shorter processing times in any predictable fashion. This leads to the larger question of whether the pretrial discovery rules are really the cause of the perceived problem of costs. Although we do not answer that question, we end with the suggestion that before any further amendments to the discovery rules are proposed in the name of reducing costs, more effort must be made to define the problem that such rule amendments are supposed to address. 14 Part IV ends, number of the attorneys who responded to the case-based survey. . . . This report documents those interviews, organizing them where possible to track the results of the multivariate analyses . . . .”).

12. See, e.g., I NST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., PRESERVING ACCESS AND IDENTIFYING EXCESS : AREAS OF CONVERGENCE AND CONSENSUS IN THE 2010 CONFERENCE MATERIALS 5 (2010), available at http://civilconference.uscourts.gov/ LotusQuickr/dcc/Main.nsf/$defaultview/7B6B047956592D3A8525771900011F6A/$File/IAALS, Preserving Access and Identifying Excess.pdf (“The collected survey research indicates a very strong consensus among nearly all respondent groups that broadly speaking, the civil justice system is too expensive.”).

13. C ASABLANCA (Warner Bros. Pictures 1942).

14. This is not to say that discovery-rule amendments tied to other goals might not be fruitfully pursued. Specifically, proposed rules for issuing preservation orders or imposing sanctions seem designed primarily to bring clarity and predictability to unsettled procedures, LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 2010] DEFINING THE PROBLEM OF COST 769 fittingly enough, with a call for more empirical research into the costs of litigation.

I. IS LITIGATION TOO EXPENSIVE ?

When asked, large percentages of practitioners agree that “[l]itigation is too expensive.” 15 But it is difficult to know what one is supposed to make of this finding. In one sense, litigation is almost always too expensive. It would often be less expensive to not have the dispute in the first place, or, barring that, less costly to find a way to resolve the dispute without recourse to the courts. 16 But more importantly, our empirical research calls into question the view that litigation is always expensive. 17 We surveyed more than two thousand attorneys of record in federal civil cases terminated in the last quarter of 2008. We excluded large categories of cases from the study because not solely to reduce costs. See, e.g., Memorandum from Judge Mark R. Kravitz, Chair, Advisory Comm. on Fed. Rules of Civil Procedure, to Judge Lee H. Rosenthal, Chair, Standing Comm.

on Rules of Practice & Procedure, Report of the Civil Rules Advisory Committee 12–14 (May 17, 2010), available at http://www.uscourts.gov/uscourts/RulesAndPolicies/rules/Reports/CV05- 2010.pdf (discussing preservation issues on the Committee’s agenda). Moreover, efforts at changing attorney behavior might yield gains that rules changes cannot. See S EVENTH CIRCUIT ELEC . DISCOVERY PILOT PROGRAM , PHASE ONE: OCTOBER 1, 2009–M AY 1, 2010: STATEMENT OF PURPOSE AND PREPARATION OF PRINCIPLES 9–10 (2009), available at http:// www.ilcd.uscourts.gov/Statement - Phase One.pdf (“The goal of the Principles is to incentivize early and informal information exchange on commonly encountered issues relating to evidence preservation and discovery . . . .”); The Sedona Conference, The Sedona Conference Cooperation Proclamation, 10 S EDONA CONF . J. 331, 331 (Supp. 2009) (“This Proclamation challenges the bar to achieve these goals and refocus litigation toward the substantive resolution of legal disputes.”).

15. I NST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., supra note 12, at 5.

16. In another sense, litigation may be too expensive in the aggregate compared to its general societal benefits. Most of the criticisms advanced at the Duke Conference, however, focused on the micro case-level costs of discovery and litigation, and not this macro level of costs. In this Article, therefore, we focus on the claim that the costs in individual cases are too high, not the much more difficult-to-assess claim that the costs of all litigation in the United States substantially outweigh any possible benefits of that litigation. It should be noted, however, that even small cost savings in most cases—as a result of attorney cooperation, for example—would represent substantial savings in the aggregate.

17. For a discussion of discovery costs and the lack of agreement regarding whether discovery limits ameliorate perceived problems, see infra notes 52–66 and accompanying text.

Empirical sources cited in that discussion support the proposition that litigation costs as a whole are not excessive in the typical federal court case. See, e.g., L AWYERS FOR CIVIL JUSTICE , CIVIL JUSTICE REFORM GRP. & U.S. CHAMBER INST . FOR LEGAL REFORM , LITIGATION COST SURVEY OF MAJOR COMPANIES 6–7 (2010), available at http://civilconference.uscourts.gov/ LotusQuickr/dcc/Main.nsf/$defaultview/33A2682A2D4EF700852577190060E4B5/$File/ Litigation Cost Survey of Major Companies.pdf (citing the ACTL-IAALS final report and joining these organizations in the call for “fundamental reforms”). LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 770 DUKE LAW JOURNAL [Vol. 60:765 they generally do not involve discovery, including prisoner civil rights and habeas corpus cases. In cases in which one or more types of discovery was reported, we found median litigation costs, including attorneys’ fees, of $15,000 for plaintiffs and $20,000 for defendants. 18 The median is, of course, the 50th percentile of the distribution of reported costs, so half of plaintiffs’ attorneys reported costs under $15,000, and half of defendants’ attorneys reported costs under $20,000. These medians do not support the claim that the typical case in federal court has escalating costs—indeed, these reported costs are largely consistent with inflation-adjusted figures from the 1997 FJC study. 19 Although it would be inappropriate to infer a trend from two data points, 20 the empirical evidence for out-of-control costs is limited.

In the context of a push for radical change in the rules of procedure to remedy allegedly disproportionate litigation expenses, it seems reasonable that the burden of presenting empirical evidence lies with the proponents of change.

At the Duke Conference, the Lawyers for Civil Justice (LCJ) presented some evidence of escalating litigation costs. Reporting litigation costs from 2000 to 2008 for Fortune 200 companies, the LCJ found that outside legal fees and costs had increased from an average of $66 million in 2000 to nearly $115 million in 2008. 21 For the twenty companies reporting costs for the entire time period, the comparable figures rose from $66 million in 2000 to $140 million in 2008. 22 But given its relatively small sample size of two hundred companies, its response rate of 10 percent, its short nine-year timeframe, and its lack of adjustment for inflation, the value of the study’s findings is limited.

Even taking the LCJ’s findings at face value, it is difficult to know what to make of them. For example, the LCJ also found that, in 2009, thirty-six Fortune 200 companies spent a total of $4.1 billion in U.S., 18. L EE & WILLGING , CASE -BASED CIVIL RULES SURVEY , supra note 11, at 35–36.

19. See id. (discussing the differences between the median inflation-adjusted cost of litigation and discovery in the 1997 and 2008 studies); see also Willging et al., supra note 10, at 531, 548 tbl.3 (offering statistics on reported litigation expenses).

20. As researchers, we are acutely aware that we do not know very much about discovery costs in civil litigation, largely because of the lack of studies, which in turn is the result of the inherent difficulties of studying this topic. See Judith A. McKenna & Elizabeth C. Wiggins, Empirical Research on Civil Discovery, 39 B.C. L. R EV. 785, 796–97 (1998) (discussing the methodological difficulties in studying discovery).

21. L AWYERS FOR CIVIL JUSTICE ET AL ., supra note 17, app. 1 at 2–3, 7 fig.3. Of course, not every Fortune 200 company reported costs to the LCJ. Id. app. 1 at 4.

22. Id. app. 1 at 8–9 & fig.5. LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 2010] DEFINING THE PROBLEM OF COST 771 outside-litigation costs. 23 This finding, however, does not decide the issue; many relevant questions remain. In absolute terms, $4.1 billion is quite a lot. But is it too much? Relative to what? Moreover, how typical are these reported costs for other litigants?

As empiricists, we are not equipped to answer these largely normative questions. That is a matter left to the Committee and other policymakers. But empirical research can shed some light on these questions. Cost is a numerator in search of a denominator. As our study indicates, monetary stakes is a great candidate for that denominator. Running with this, we next report our findings with respect to the proportionality of discovery costs to stakes. Finally, we return to the LCJ report and discuss what it says about the relationship between litigation costs and stakes.

II.

RELATIVE TO STAKES ?

The monetary stakes in a case represent the single best predictor of litigation costs in that case. Graphically depicting the data for private-firm plaintiffs’ and defendants’ attorneys in the FJC study, Figures 1 and 2 24 present scatter plots of the bivariate relationship between monetary stakes 25 and reported litigation costs. The plots demonstrate a strong linear relationship between stakes and costs; variation in the stakes alone explains almost 37 percent of the variation in reported costs for plaintiffs’ attorneys and almost 47 percent of the variation in reported costs for defendants’ attorneys.

This relationship holds in the multivariate analysis as well. For both plaintiffs’ and defendants’ attorneys, and after controlling for other factors including time to disposition, a 1 percent increase in stakes is associated with a 0.25 percent rise in costs. 26 In other words, if the monetary stakes in a case double, all else being equal, the costs 23. Id. app. 1 at 8 fig.4.

24. See infra Appendix. These figures were created using the Lattice package in R. For more information on R, see generally R DEV. CORE TEAM , R: A LANGUAGE AND ENVIRONMENT FOR STATISTICAL COMPUTING : REFERENCE INDEX (2010), available at http://cran.r-project.org/doc/manuals/refman.pdf, and D EEPAYAN SARKAR , LATTICE : MULTIVARIATE DATA VISUALIZATION WITH R (2008).

25. Respondents were asked to estimate, in dollars, the best and worst “likely” outcomes from the point of view of their clients. Stakes were then calculated as the “spread” between these two figures. See L EE & WILLGING , CASE -BASED CIVIL RULES SURVEY , supra note 11, at 41–42 (describing in detail the variables used). For a reprint of the questions used, see id. app. C at 94–95.

26. L EE & WILLGING , MULTIVARIATE ANALYSIS , supra note 11, at 5, 7. LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 772 DUKE LAW JOURNAL [Vol. 60:765 increase by 25 percent. The complete multivariate models explain approximately 62 percent of the variation in plaintiffs’ attorneys’ reported costs and 76 percent of the variation in defendants’ attorneys’ reported costs. 27 Thus, stakes alone account for about 60 percent of the explained variation in the complete models. In a very real sense, there are stakes, and then there is everything else.

In our interviews, we found that many practitioners emphasize the overarching importance of monetary stakes when deciding on discovery and pretrial practices that are likely to increase litigation costs. A number of interview subjects indicated that clients understand the relationship between stakes and costs, too. The following statements are illustrative:

• “Companies are willing to invest more in cases where the stakes are . . . high.” • “Even the client expects an attorney to invest more time in high stakes cases.” • “One has to take into account the possibility of being enjoined from selling a product, which increases the stakes.” • “If there’s a lot of money involved, parties dig in their heels and litigate every little thing.” • “Stakes make a difference in that clients are willing to pay and more likely to dig deeper into discovery.” 28 Our findings indicate that the monetary stakes in the litigation represent the primary cost driver in most civil litigation. This very robust relationship between stakes and costs leads directly to the much-debated question of proportionality. Since it was amended in 1983, Rule 26(b)(2)(C)(iii) has directed courts to limit discovery if they determine that “the burden or expense of the proposed discovery outweighs its likely benefit, considering the needs of the case, the amount in controversy, the parties’ resources, the importance of the issues at stake in the action, and the importance of the discovery in resolving the issues.” 29 This proportionality principle is somewhat broader than monetary stakes but requires, at minimum, that judges and parties consider both the “amount in controversy” 27. Id.

28. W ILLGING & LEE, IN THEIR WORDS , supra note 11, at 5–6 (alteration in original).

29. F ED . R. CIV. P. 26(b)(2)(C)(iii). LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 2010] DEFINING THE PROBLEM OF COST 773 and the value of the “issues at stake in the action” in relation to the burden or cost of proposed discovery.

The rule itself, however, is unclear as to the proper ratio of discovery costs to stakes. The Committee commentary is also not terribly instructive. Professor Arthur Miller, Reporter to the Committee in 1983, described disproportionate discovery with this example: “In a $10,000 damage case, spending $50,000 on discovery is disproportionate.” 30 It seems beyond cavil that a five-to-one ratio of costs to stakes is disproportionate; the much more interesting case is a ratio of one to five. Is that proportionate? As is usually true, it depends. As Miller noted, “Everybody understands you can have a case where the values at stake transcend the economics of the case, so this is not a pure dollar test.” 31 Nor, it would seem, does the rule call for a straightforward ratio.

Lawyers and academic commentators often opine that judges infrequently invoke the proportionality principle. 32 But if the parties limit their requests with an eye toward stakes, or if they negotiate to modify disproportionate requests, judges will have little or no need to invoke the rule. Our research indicates that disproportionate discovery may be less of a problem than critics of the Federal Rules often asseverate. In the survey, we asked respondents to rate the relationship of the discovery costs in a particular closed case to the client’s stakes in the litigation. They were asked to rate proportionality on a seven-point scale, “with [one] being too little, [four] being just the right amount, and [seven] being too much.” 33 A majority of respondents—59 percent of plaintiffs’ attorneys and 57 percent of defendants’ attorneys—answered that the costs of discovery were “just the right amount” compared to the stakes.

34 About one in four respondents—23 percent of plaintiffs’ attorneys and 27 percent of defendants’ attorneys—rated their client’s costs at 30. A RTHUR R. MILLER , FED . JUDICIAL CTR., THE AUGUST 1983 AMENDMENTS TO THE FEDERAL RULES OF CIVIL PROCEDURE : PROMOTING EFFECTIVE CASE MANAGEMENT AND LAWYER RESPONSIBILITY 32 (1984).

31. Id. at 33.

32. See, e.g., K IRSTEN BARRETT , RHODA COHEN & JOHN HALL , MATHEMATICA POLICY RESEARCH , INC., ACTL CIVIL LITIGATION SURVEY , FINAL REPORT 41 (2008) (finding that “76.8% of Fellows agree that judges do not invoke Rule 26(b)(2)(C) on their own initiative”).

33. L EE & WILLGING , CASE -BASED CIVIL RULES SURVEY , supra note 11, at 28 & fig.14.

The unweighted sample sizes were 1,184 for plaintiffs’ attorneys and 1,193 for defendants’ attorneys.

34. Id. at 27–28 & fig.13. LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 774 DUKE LAW JOURNAL [Vol. 60:765 five or higher, but ratings of six or seven were relatively rare. 35 Thus, relatively few respondents believed that discovery costs had been disproportionate in those particular cases. 36 More than three-quarters of the responses, in fact, clustered between three and five on the seven-point scale. 37 In our interviews, attorneys pointed out mechanisms they use to keep costs proportionate to stakes. Some simply kept the stakes in mind: “[O]ne always tailors the amount of discovery to the stakes.

The difference between a $50,000 case and a $500,000 case is always on one’s mind.” 38 Others followed the limits of a scheduling order: “I go as far as the law will allow. There are time constraints . . . .” 39 Attorneys appear to use these and other mechanisms to adjust costs to conform to the stakes of the litigation. 40 These mechanisms, in turn, may help explain why the majority of attorneys surveyed concluded not only that the costs in their case were “just the right amount” in relation to their clients’ stakes but also that they obtained the right amount of information. 41 This case-based finding that costs are generally proportionate to stakes is again at odds with the views expressed by other participants at the Duke Conference. The IAALS-ACTL survey found that ACTL fellows overwhelmingly believe that the costs of litigation are not proportionate to the value of a case. In versions of the survey administered to the members of the American Bar Association (ABA) Section of Litigation and the National Employment Lawyers Association (NELA), the proportionality question was split into one question asking whether litigation costs are proportionate to the value of large cases and one asking whether the costs are proportionate to the value of small cases. Respondents to these 35. Id. at 28.

36. The 1997 FJC study found that 54 percent of all respondents reported that discovery expenses in the closed case were “about right” in relation to stakes, with 15 percent reporting that such expenses were “high” and 20 percent reporting that they were “low.” Eleven percent did not express an opinion. Willging et al., supra note 10, at 531, 550–51 & tbl.8.

37. Id.

38. W ILLGING & LEE, IN THEIR WORDS , supra note 11, at 5.

39. Id. at 22.

40. For a more complete discussion of attorney responses to the question “How much discovery is enough?,” see id. at 21–24.

41. See L EE & WILLGING , CASE -BASED CIVIL RULES SURVEY , supra note 11, at 27 & fig.13 (reporting that 56.6 percent of plaintiffs’ attorneys and 66.8 percent of defendants’ attorneys reported that the discovery in their cases had resulted in “just the right amount” of information). LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 2010] DEFINING THE PROBLEM OF COST 775 questions expressed the general view that litigation costs are disproportionate to the value of small cases but not necessarily disproportionate to the value of large cases. 42 These impressionistic findings are difficult to square with our case-based survey findings. Perhaps the members of the ACTL, ABA Section of Litigation, and NELA have had markedly different experiences than the randomly selected attorneys surveyed by the FJC. Perhaps the responses in the more impressionistic surveys are affected by well-known cognitive biases, such as availability and recall—in other words, the tendency of respondents to call to mind problematic cases when asked a general question. 43 Whatever the explanation for these differences, the case-based surveys provide empirical evidence of greater proportionality in the relationship of discovery costs to stakes than one would predict based on the complaints raised by critics of the Federal Rules. 44 But what of the $4.1 billion in outside legal costs reported by the LCJ? Could costs that high be considered proportionate to anything?

The LCJ report itself does not address the stakes in the underlying cases, but it does present some information about litigation costs for outside counsel as a percentage of companies’ U.S. revenues. Without knowing more about the confidential underlying data, it is difficult to know exactly how to interpret the numbers. But Figure 8 in that report, which excludes outliers, seems informative. For companies in sectors other than health care and insurance, from 2000 to 2008, outside litigation costs as a share of U.S. revenues were relatively constant, hovering around 0.3 percent. 45 In other words, for the reporting Fortune 200 companies in these sectors, outside litigation costs consumed about one in every three hundred dollars of total U.S.

revenues. This percentage varied only between 0.27 percent and 0.4 42. E MERY G. LEE III & THOMAS E. WILLGING , FED . JUDICIAL CTR., ATTORNEY SATISFACTION WITH THE FEDERAL RULES OF CIVIL PROCEDURE : REPORT TO THE JUDICIAL CONFERENCE ADVISORY COMMITTEE ON CIVIL RULES 10–11 & fig.11 (2010), available at http://www.fjc.gov/public/pdf.nsf/lookup/costciv2.pdf/$file/costciv2.pdf.

43. There is a lengthy literature on the role of heuristics in cognition. For an application of this literature in the legal literature, which applies an availability heuristic to risk perception, see, for example, Cass R. Sunstein, Precautions Against What? The Availability Heuristic and Cross-Cultural Risk Perception, 57 A LA. L. REV. 75, 87–89 (2005). For the classic essay on availability heuristics, see Amos Tversky & Daniel Kahneman, Availability: A Heuristic for Judging Frequency and Probability, 5 C OGNITIVE PSYCHOL . 207 (1973).

44. See supra text accompanying note 9.

45. See L AWYERS FOR CIVIL JUSTICE ET AL ., supra note 17, app. 1 at 11, 12 fig.8 (“[A]ll other industries report an average of 0.33 percent of revenue.”). LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 776 DUKE LAW JOURNAL [Vol. 60:765 percent over the time period and was at almost exactly the same level in 2000 as in 2008. 46 Health care and insurance companies, on the other hand, reported increasing percentages of revenues consumed by litigation, from 0.45 percent in 2000 to 1.15 percent in 2008, with a particularly sharp increase between 2004 and 2005. 47 Even so, the 2008 figure for health care and insurance enterprises represents about one in every one hundred dollars of U.S. revenues consumed by outside litigation costs.

It is not our place to say whether one in every one hundred—or every three hundred—dollars of revenues is a disproportionate amount for the largest U.S. companies to spend on litigation. Once again, the problem is the lack of a normative standard for evaluating the cost data. But the question should be put to the authors of the LCJ report to specify what the appropriate outlay for litigation would be, if, as the LCJ argues, the reported costs are too high. The answer would be informative.

III. M ODELS FOR REFORM ?

Given the lack of empirical evidence for a cost problem, it should not be surprising that the empirical support for some of the solutions put forward by rule critics is also pretty weak. The IAALS, for example, puts forward the limits on discovery in the Arizona and Oregon state-court procedures as models for the Committee to consider. 48 But the IAALS studies themselves suggest that neither of these systems accomplishes the goals that the IAALS sets out for them. In both systems, for example, attorney respondents indicate that litigation in the state courts is still too expensive, despite the limits on discovery. In Arizona, 84 percent of respondents agreed 46. Id. app. 1 at 12 fig.8.

47. Id.

48. See I NST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., SURVEY OF THE ARIZONA BENCH & BAR ON THE ARIZONA RULES OF CIVIL PROCEDURE 26–27, 29–42 (2010) [hereinafter I NST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., THE ARIZONA RULES OF CIVIL PROCEDURE ] (discussing the Arizona Rules of Civil Procedure’s presumptive limits on discovery); I NST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., SURVEY OF THE OREGON BENCH & BAR ON THE OREGON RULES OF CIVIL PROCEDURE 31–43 (2010) [hereinafter I NST .

FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., THE OREGON RULES OF CIVIL PROCEDURE ] (discussing restrictions on discovery under the Oregon Rules of Civil Procedure).

Oregon, in addition, adheres to a form of fact pleading, which the IAALS also supports. See Rebecca Love Kourlis, Jordan M. Singer & Natalie Knowlton, Reinvigorating Pleadings, 87 D ENV . U. L. REV. 245, 266–67 (2010) (“The Oregon Supreme Court has repeatedly reaffirmed the state’s commitment to fact pleading.”). LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 2010] DEFINING THE PROBLEM OF COST 777 with the statement that civil justice in Arizona superior court is “too expensive”; 49 in Oregon, 79 percent of respondents agreed that civil justice in Oregon circuit court is “too expensive.” 50 Somewhat oddly, the IAALS cites these state figures in a document calling for reform of the Federal Rules. 51 That experienced practitioners in these model systems think civil justice is too expensive despite the discovery limits, however, undercuts the argument that adopting similar rules would reduce the perceived unnecessary expenses in federal court litigation.

More tellingly, however, Arizona and Oregon practitioners seem rather cool to the purported advantages of limited discovery. This is not to say that limits on discovery do not have any effect. For example, when asked about the potential benefits of Arizona’s presumptive limits on discovery as a whole, 64 percent of respondents agreed that the limits focused discovery, and 58 percent agreed that the limits actually reduced the volume of discovery. 52 With respect to whether the presumptive limits make costs more predictable or reduce costs, time to disposition, or the threat of forced settlement, however, opinions were mixed, or even negative. Forty-seven percent of respondents agreed that the presumptive limits reduced costs, but 44 percent disagreed—within the reported margin of error of 3.5 percent. 53 In other words, not even a statistically significant plurality of surveyed Arizona practitioners agreed that the Arizona discovery limits reduce costs. Furthermore, 53 percent disagreed that the limits reduce time to disposition, and 55 percent disagreed both that the limits make costs more predictable and that they reduce the threat of forced settlement. 54 This does not make a very strong case for reform based on the Arizona model.

49. I NST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., THE ARIZONA RULES OF CIVIL PROCEDURE , supra note 48, at 44 & fig.44.

50. I NST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., THE OREGON RULES OF CIVIL PROCEDURE , supra note 48, at 54 & fig.51.

51. See I NST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., supra note 12, at 4–5 (grouping the Oregon and Arizona surveys with the ACTL, ABA Section of Litigation, and NELA surveys).

52. I NST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., THE ARIZONA RULES OF CIVIL PROCEDURE , supra note 48, at 37 & fig.35. This latter advantage is somewhat tautological. It would be strange if respondents did not agree that limits on discovery actually limit discovery. Interestingly, 35 percent of respondents expressed just this view. Id.

53. See id. at 7, 37 & fig.35 (providing a 95 percent confidence level at plus or minus 3.54 percent for the entire sample of 767 valid responses).

54. Id. at 37 & fig.35. LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 778 DUKE LAW JOURNAL [Vol. 60:765 Perhaps one could argue in rebuttal that Arizona practitioners with federal-court experience prefer state court to federal court, thus indicating the superiority of state courts and their limited discovery.

Almost half of the attorney respondents with federal-court experience—49 percent—expressed a preference for Arizona state court over federal court, whereas only 25 percent expressed a preference for federal court. 55 A “two-to-one ratio” 56 is a large margin in favor of the Arizona state courts. But the report’s assertion that “nearly three-quarters of respondents either prefer the state forum or have no preference” is a bit much. 57 Although about half of attorney respondents with federal-court experience prefer state courts, about half either prefer federal court or have no preference. One can spin these results, but this is not a knockdown case for reform. 58 The results of the Oregon survey are very similar. The Oregon discovery limits as a whole fare about as well as the Arizona limits.

Again, a majority of respondents—64 percent—agreed that the limits on discovery reduce the volume of discovery. 59 Similarly, 51 percent agreed that the limits focus discovery, whereas only 41 percent disagreed. 60 But there is simply no agreement among respondents that the limits make costs more predictable or that they reduce costs, time to disposition, or the threat of forced settlement. Forty-three percent agreed that the limits make costs more predictable, but 48 percent disagreed.

61 Forty-six percent of respondents agreed that the limits 55. Id. at 12 fig.6.

56. Id. at 12.

57. Id.

58. Moreover, there are many reasons why attorneys may prefer one forum to another; procedural rules are just one—and probably not the most important one. In a study of choice of forum in class action litigation, for example, 78 percent of plaintiffs’ attorneys identified the source of the claims (state law) as a reason for filing in state court. Only 28 percent identified the favorableness of discovery rules, and only 31 percent identified lower costs of litigation.

Respondents could, however, identify more than one reason for filing in a particular forum. See T HOMAS E. WILLGING & SHANNON R. WHEATMAN , FED . JUDICIAL CTR., AN EMPIRICAL EXAMINATION OF ATTORNEYS ’ CHOICE OF FORUM IN CLASS ACTION LITIGATION 18 tbl.2 (2005). Without knowing more about why attorneys prefer the state courts to federal courts, it is not clear what one should conclude from this finding with respect to procedural rules. The IAALS report includes a long list of reasons why respondents preferred the state forum, but these are not given particular weights. See I NSTITUTE FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., THE ARIZONA RULES OF CIVIL PROCEDURE , supra note 48, at 13.

59. INST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., THE OREGON RULES OF CIVIL PROCEDURE , supra note 48, at 41 fig.33.

60. Id.

61. Id. LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 2010] DEFINING THE PROBLEM OF COST 779 reduce costs, but 45 percent disagreed that they did so. 62 Forty-five percent agreed that the limits reduce time to disposition, but 46 percent disagreed. 63 And 42 percent agreed that the limits reduce the threat of forced settlement, but 44 percent disagreed. 64 Moreover, the Oregon attorneys with federal-court experience do not show much of a preference for state court. Forty-three percent of respondents indicated a preference for state court, and 37 percent expressed a preference for federal court 65—a difference of only six percentage points in a study with a reported margin of error of 4.5 percent for the entire sample, of which these percentages represent a subset. We doubt that this result is statistically significant, and the report does not say. And although the report asserts that “[a]lmost two-thirds of respondents either prefer the state forum or have no preference,” 66 it is also true that almost 60 percent prefer federal court or had no preference.

In short, Arizona and Oregon attorneys, in evaluations based on their experience, revealed the shortcomings of limits on discovery as a solution to the perceived problems of cost and delay. The responses provided by these attorneys simply do not provide much support for adopting such rules at the federal level.

IV. RUSH TO JUDGMENT ?

A careful review of the empirical research prepared for the Duke Conference begs the question with which we began—what, exactly, is the problem to be solved? The usual suspect, at least since the 1976 Pound Conference, has been pretrial discovery costs. But empirical research does not support the charge.

Empirical research has not provided support for the prevailing view that discovery costs are necessarily the major cost driver in litigation. The 2009 FJC case-based study found that the median percentage of total litigation costs accounted for by discovery was 20 62. Id.

63. Id.

64. Id. The Oregon report also notes a number of commonly held complaints about the Oregon discovery limits that are worth considering. Almost 40 percent of respondents complained that the complete absence of interrogatories in Oregon “diminishe[d] counsel’s ability to prepare for trial,” and “a majority indicated that the absence of expert discovery decreases counsel’s ability to prepare for trial.” Id. at 2.

65. Id. at 12 fig.6.

66. Id. at 12. LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 780 DUKE LAW JOURNAL [Vol. 60:765 percent for plaintiffs’ attorneys and 27 percent for defendants’ attorneys. 67 Although that figure has struck some as low—a claim to which we will turn shortly—the point to remember is that the FJC estimate is generally consistent with previous studies. The Columbia Project study from the 1960s estimated that discovery accounted for between 19 and 36 percent of litigation costs, depending on whether a party was a requesting or requesting-and-producing party. 68 The Civil Litigation Research Project in the 1970s found that, in the ordinary case, 16.7 percent of attorney time, a proxy for cost, was spent on discovery. 69 And in the 1990s, the RAND Corporation found that “lawyer work hours on discovery are zero for 38 percent of general civil cases, and low for the majority of cases,” making discovery “not a pervasive litigation cost problem for the majority of cases.” 70 Focusing on cases lasting longer than 270 days, RAND found that postfiling discovery consumed a little over one-third (36 percent) of attorney work hours. 71 The highest estimate of the percentage of total cost associated with discovery was that reported in the 1997 FJC study—a median estimate of 50 percent for both plaintiffs’ and defendants’ attorneys.

72 The LCJ report presented at the Duke Conference may also shed some light on the percentage of total litigation costs allocated to discovery in “major cases,” which that study defines as cases with attorneys’ fees exceeding $250,000. 73 Figure 11 in that report provides the average discovery costs, for at least some corporations, for major cases closed from 2004 to 2008. 74 Figure 10 provides the average outside legal fees for major cases closed during the same years.

Unfortunately, the number of corporations included in the two figures is not the same, making it impossible to draw any meaningful 67. L EE & WILLGING , CASE -BASED CIVIL RULES SURVEY , supra note 11, at 38–39 tbls.6 & 7.

68. W ILLIAM A. GLASER , PRETRIAL DISCOVERY AND THE ADVERSARY SYSTEM 180 tbl.43 (1968).

69. David M. Trubek, Austin Sarat, William L.F. Felstiner, Herbert M. Kritzer & Joel B.

Grossman, The Costs of Ordinary Litigation, 31 UCLA L. REV. 72, 91 tbl.3 (1983).

70. James S. Kakalik, Deborah R. Hensler, Daniel F. McCaffrey, Marian Oshiro, Nicholas M. Pace & Mary E. Vaiana, RAND INST . FOR CIVIL JUSTICE , DISCOVERY MANAGEMENT : FURTHER ANALYSIS OF THE CIVIL JUSTICE REFORM ACT EVALUATION DATA , at xx (1998).

71. Id. at xxi tbl.S.2.

72. Willging et al., supra note 10, at 548 tbl.4.

73. L AWYERS FOR CIVIL JUSTICE ET AL ., supra note 17, app. 1 at 13–14.

74. Id. app. 1 at 15 fig.11. Note that for 2006 and 2007, the number of responding corporations is only four. LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 2010] DEFINING THE PROBLEM OF COST 781 comparisons. It is impossible, for example, to assess how the average discovery costs for major cases closed in 2008 for twenty reporting corporations—$621,880—relate to the average outside legal fees for major cases closed in 2008 for thirty reporting corporations— $2,019,248. 75 But with that caveat, simply dividing the average discovery costs per major case closed in 2008 by the average outside legal fees per major case closed that year yields an estimate of discovery costs as a share of outside legal fees of 30.8 percent. That is surprisingly close to the 2009 FJC report’s figure for defendants’ attorneys (a median of 27 percent) and is well within the bounds of previous empirical research. Moreover, because this estimate does not appear to include in-house legal costs or other litigation costs beyond legal fees in the denominator, it may actually overstate the percentage of costs associated with discovery. 76 The empirical studies of discovery costs, in short, indicate that in the typical case—and perhaps even in the typical major case, although that data is very limited—one should expect discovery costs to account for more than 20 percent, on the lower end, and maybe, on the higher end, about half of the total litigation costs. There will be some more discovery-heavy cases, of course, but 20 to 50 percent is what we would expect in a typical case.

The impressionistic studies provide a very different picture, however. In the ACTL, ABA Section of Litigation, and NELA surveys, the median estimate of the percentage of litigation costs attributable to discovery in cases not going to trial was 70 percent. 77 Again, this figure is rather hard to square with the other studies. The likely reason for the disparity is cognitive bias—respondents are providing answers based on problematic cases or on what can be described as the conventional wisdom. 78 Although it is plausible that discovery costs account for 70 percent of total litigation costs in some cases, the weight of empirical evidence indicates that it is unlikely to reach 70 percent in many cases. In the October 2009 FJC report, the 95th percentile for the reported percentage of litigation costs incurred in discovery was 80 percent for both plaintiffs’ and defendants’ 75. Id. app. 1 at 14–15 figs.10 & 11.

76. The report itself states, however, that its numbers “understate . . . discovery costs, because many discovery costs may go unmeasured and unreported.” Id. app. 1 at 15.

77. I NST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., supra note 12, at 13.

78. For a discussion of the role of heuristics in cognition, see supra note 43 and accompanying text. LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 782 DUKE LAW JOURNAL [Vol. 60:765 attorneys. 79 Additional analysis of that data shows that, in cases with any reported discovery event, only 10 percent of respondents reported that discovery costs were as much as 70 percent of total costs. This is the result of only one study, of course, but 10 percent of cases with discovery as a share of total costs as high as 70 percent is not really typical of federal cases in general.

If one frames the problem as discovery costs accounting for 70 percent of litigation costs in the typical case, then there is little empirical evidence that any such problem exists on a wide-scale basis.

Only the impressionistic survey responses support the view that this is a problem in the typical federal case. There is nothing else, unless one wants to credit survey responses stating that discovery is “too expensive.” 80 Has the case for wide-ranging reform of the discovery rules been made? The argument is being made insistently, for sure. But, as District Judge Lee Rosenthal, chair of the Judicial Conference Committee on Rules of Practice and Procedure and former chair of the Advisory Committee on Civil Rules, recently observed, “Since their inception in 1938, the rules of discovery have been revised with what some view as distressing frequency. And yet the rulemakers continue to hear that the rules are inadequate to control discovery costs and burdens.” 81 The critics of the Federal Rules believe that the long-term project, arguably initiated with the Pound Conference, to devise cost-controlling discovery rules has failed. 82 There are, however, two ways in which the project may have failed. On the one hand, the project may not have gone far enough, largely because the potential for gamesmanship inherent in the system makes it difficult to achieve perfection or even “excellence.” 83 Attorneys will exploit the rules and the rulemaking process to seek advantages over their 79. L EE & WILLGING , CASE -BASED CIVIL RULES SURVEY , supra note 11, at 38–39 tbls.6 & 7.

80. See, e.g., I NST . FOR THE ADVANCEMENT OF THE AM. LEGAL SYS., supra note 12, at 14 (“At least 70% of respondents . . . agreed that discovery in general is too expensive.”).

81. Lee H. Rosenthal, From Rules of Procedure to How Lawyers Litigate: ’Twixt the Cup and the Lip, 87 D ENV . U. L. REV. 227, 228 (2010).

82. See supra notes 6, 8 and accompanying text.

83. See Paul D. Carrington, Making Rules to Dispose of Manifestly Unfounded Assertions:

An Exorcism of the Bogy of Non-Trans-Substantive Rules of Civil Procedure, 137 U. PA. L. REV.

2067, 2070 (1989) (“Not only is perfection [in procedural rulemaking] impossible, but even excellence is unstable, especially so in a system dependent on the adversary tradition, because of changing circumstances and the corrosive effect of perpetual exploration and exploitation of systemic weaknesses by adversaries.”). LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 2010] DEFINING THE PROBLEM OF COST 783 adversaries, to the detriment of the system and society as a whole. 84 Thus, only radical changes can make the strategic exploitation of the rules impossible. Yet radical change appears unlikely to survive the rulemaking process.

On the other hand, the project may have failed to reduce costs because it does not address the actual drivers of cost. Perhaps the procedural reforms have not reduced the purportedly high costs of litigation because those costs have a source other than the Federal Rules themselves. Professor Charles Silver argued just that in an article published in 2002: “[P]rocedural reforms have not reduced [litigation] costs because adjudicatory procedures are not generating these costs.” 85 “Costs,” he continues, “instead reflect the need to figure out how much claims are worth and the difficulty of bargaining, which in turn reflect, respectively, properties of claims and of relationships between claimants and respondents. Empirical studies support this idea.” 86 Our study also supports Silver’s argument. The monetary stakes in litigation best predict the costs. 87 Claims that are worth more also cost more to litigate, primarily because the parties will spend more when more is at stake. It seems difficult to believe that any but radical rule changes could affect this basic pattern.

But other properties of claims impact costs in ways that rule changes are unlikely to affect. 88 Factual complexity, for example, is associated with higher costs, probably because it makes it harder for the parties to uncover the evidence necessary to price claims and bargain toward settlement. Respondents in the 2009 FJC case-based survey rated the factual complexity of the closed case on a seven- point scale, with one being “[n]ot complex at all,” four being “[a]verage complexity,” and seven being “[e]xtremely complex.” 89 Even controlling for other factors, each one-unit increase in this subjective scale of factual complexity was associated with an 11 84. Id.

85. Charles Silver, Does Civil Justice Cost Too Much?, 80 T EX. L. REV. 2073, 2074 (2002).

86. Id.

87. See supra Part II.

88. Interestingly, we find that, in general, nature-of-suit categories of cases do not affect costs, though there are a few notable exceptions. We find that, once other factors are controlled for, tort cases are less costly for plaintiffs and that intellectual property cases are much more expensive for defendants than other kinds of cases. But the other nature-of-suit categories are no more or less costly, on a consistent basis, than the baseline. L EE & WILLGING , MULTIVARIATE ANALYSIS , supra note 11, at 6, 8.

89. L EE & WILLGING , CASE -BASED CIVIL RULES SURVEY , supra note 11, app. C at 96. LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 784 DUKE LAW JOURNAL [Vol. 60:765 percent increase in costs for plaintiffs and a 13 percent increase in costs for defendants. 90 The most complex cases, in short, are going to have higher costs than the least complex cases, even after one controls for other factors related to the Federal Rules.

Contention among the parties also increases costs, at least for defendants. Respondents were asked to rate the contentiousness between the parties on a seven-point scale, with one being “[n]ot contentious at all,” four being “[a]verage contentiousness,” and seven being “[e]xtremely contentious.” 91 All else being equal, defendants reported an 8 percent increase in costs for each one-unit increase in contentiousness. 92 This suggests that difficulties in litigating and bargaining that result from the relationship between the parties can generate much higher costs than other factors can explain.

The economics of the legal practice also affect costs in a way not directly related to specific procedural reforms. In general, the larger the law firm handling the case—as measured by the number of attorneys—the higher the costs, even after controlling for factors such as stakes, complexity, and levels of discovery. Using a solo practitioner as the baseline and holding everything else constant, costs for a firm of more than five hundred attorneys would be more than double for both defendants (156 percent higher) and plaintiffs (109 percent higher). 93 Hourly billing was also associated with higher costs for plaintiffs.94 Our interview subjects said a number of interesting things on this front. One put it quite graphically: “You have to feed the tiger first before defendants will settle a case.” 95 This is not to say that pretrial discovery does not explain some of the variation in costs—it does. But once one accounts for nonrules factors, the effects of discovery on costs are much more mixed than one might expect based on the criticisms of the Federal Rules. The bête noire of the critics—electronic discovery—shows a very 90. L EE & WILLGING , MULTIVARIATE ANALYSIS , supra note 11, at 6–7.

91. L EE & WILLGING , CASE -BASED CIVIL RULES SURVEY , supra note 11, app. C at 96.

92. L EE & WILLGING , MULTIVARIATE ANALYSIS , supra note 11, at 7.

93. Id. at 6, 8.

94. Id. at 6. Attorneys representing plaintiffs in the closed cases who reported using hourly billing reported costs 25 percent higher than other attorneys, all else being equal. There were so few defendants’ attorneys using alternative fee arrangements that we cannot say what the effects of hourly billing are on defendants’ costs. Fewer than 5 percent of defendants’-attorney respondents reported using a billing method other than hourly billing. Id. at 8.

95. W ILLGING & LEE, IN THEIR WORDS , supra note 11, at 10 (internal quotation marks omitted). LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 2010] DEFINING THE PROBLEM OF COST 785 interesting pattern. All else being equal, a request by plaintiffs for production of electronically stored information (ESI) appears to be associated with higher costs for those plaintiffs, regardless of whether the plaintiff is also a producing party. The costs for plaintiffs who were requesting-only parties with respect to ESI were approximately 37 percent higher than for plaintiffs in cases without electronic discovery, and costs for plaintiffs both requesting and producing ESI were 48 percent higher than for plaintiffs in cases without electronic discovery. 96 For defendants, however, producing-only and requesting- only parties did not have higher costs, after controlling for other factors, than parties in cases without electronic discovery. 97 But in cases in which the defendant was both a producing and requesting party with respect to ESI, costs were approximately 17 percent higher than in cases without electronic discovery. 98 The results for plaintiffs make a great deal of sense. The requesting plaintiff incurs higher costs as a result of electronic discovery. But how could it be that defendants producing ESI did not have consistently higher costs, once other factors were accounted for?

The answer, it appears, is that the costs of producing ESI are highly variable from respondent to respondent. 99 In other words, defendants producing ESI do not consistently face higher costs than similarly situated defendants in cases without electronic discovery. Factors internal to the company and its information systems, not the Federal Rules, are responsible for some of these costs. 100 With that said, however, it is important to note that electronic- discovery disputes are costly when they occur.

101 All else being equal, for plaintiffs and defendants alike, each reported type of dispute over 96. L EE & WILLGING , MULTIVARIATE ANALYSIS , supra note 11, at 5. Not surprisingly, very few plaintiffs’ attorneys reported that their clients were producing-only parties. Id.

97. Id. at 7.

98. Id.

99. This finding is consistent with those of the RAND study, which was also presented at the Duke Conference. Email from Nicholas M. Pace, Behavioral/Soc. Scientist, RAND Corp., to Thomas E. Willging, Senior Researcher, Fed. Judicial Ctr. (June 28, 2010, 16:12 EDT) (on file with the Duke Law Journal). For a video of Mr. Pace’s presentation at the Duke Conference, see Civil Litigation Conference, D UKE UNIV . SCH . OF LAW (May 10, 2010), http://www.law.

duke.edu/webcast (follow “Civil Litigation Conference – 2” hyperlink).

100. See supra note 99.

101. This is not as often as some might believe. L EE & WILLGING , CASE -BASED CIVIL RULES SURVEY , supra note 11, at 24 (reporting that the overwhelming majority of plaintiffs’ and defendants’ attorneys—72.4 percent and 78.3 percent, respectively—reported that no disputes had occurred in the closed case). LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 786 DUKE LAW JOURNAL [Vol. 60:765 ESI was associated with a 10 percent increase in costs. 102 To the extent that rule changes make such disputes less common or less costly to resolve, the Committee would potentially provide a great deal of cost savings for parties.

Finally, a ruling on summary judgment adds considerably to the cost of litigation for both plaintiffs and defendants. All else being equal, including duration, summary judgment added 24 percent to plaintiffs’ costs and 22 percent to defendants’ costs.

103 Though the procedures for ruling on summary judgment are rule based, the costs are associated with a ruling on a motion. Thus, the costs seem to be, in substantial part, a product of a party’s decision to move for summary judgment and not of the procedural mechanics of Rule 56. 104 Moreover, the participants in the Duke Conference did not entertain any serious suggestions for reforming, let alone eliminating, summary judgment procedures.

Despite the findings of the FJC study, however, there is still much to learn about the costs of litigation. The Duke Conference’s focus on empirical research may inspire additional work in this area.

No one knows better than us the inherent difficulties of studying the costs of litigation. But given that this will likely continue to be an area of interest to the judiciary for many years to come, there is a need for more information to inform the policy debate.

C ONCLUSION The FJC study found that discovery and overall litigation costs were largely proportionate to stakes, and that the stakes in a case were the single best predictor of overall costs. In general, litigation costs were lower than one might have expected, given what Professor Linda S. Mullenix has called “[t]he [p]ervasive [m]yth of [p]ervasive [d]iscovery [a]buse.” 105 Participants in the Duke Conference—and many others—will disagree with the myth characterization. But this raises its own questions. First, why is this belief so enduring, when it 102. L EE & WILLGING , MULTIVARIATE ANALYSIS , supra note 11, at 5, 7.

103. Id. at 6, 8.

104. Many plaintiffs’ attorneys argue that summary judgment practice is not tethered by the constraints of Rule 56(c)(2) that there be “no genuine issue as to any material fact.” See W ILLGING & LEE, IN THEIR WORDS , supra note 11, at 29–31 (quoting plaintiffs’ attorneys who argue that summary judgment is overused).

105. Linda S. Mullenix, The Pervasive Myth of Pervasive Discovery Abuse: The Sequel, 39 B.C. L. REV. 683 (1998). LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 2010] DEFINING THE PROBLEM OF COST 787 has never been supported by a single empirical study of costs, as opposed to beliefs about costs? Second, have the critics of the Federal Rules carried their burden of persuasion? Is there really a need for sweeping, radical procedural reforms, as opposed to more-focused reforms of particular federal rules?

Although we do not endorse Professor Silver’s conclusion, we think that it deserves the consideration of the rulemakers. His argument is an increasingly plausible alternative to the widespread belief that the procedural reforms enacted since the Pound Conference have failed to reduce costs. Perhaps the procedures that have been reformed were not causing the problem in the first place.

Instead of pursuing sweeping, radical reforms of the pretrial discovery rules, perhaps it would be more appropriate to pursue more-focused reforms of particularly knotty issues (such as preservation duties with respect to ESI) and additional, credible research on the relationship between pretrial discovery and litigation costs.

Otherwise, we may simply find ourselves considering an endless litany of complaints about a problem that cannot be pinned down empirically and that never seems to improve regardless of what steps are taken. In other words, we might find ourselves, again, right back where the Pound Conference set off almost four decades ago. Déjà vu, indeed. LEE & WILLGING IN FINAL. DOC 11/29/2010 7:06:19 PM 788 DUKE LAW JOURNAL [Vol. 60:765 A PPENDIX Figure 1. Relationship Between Monetary Stakes and Total Litigation Costs, Private-Firm Plaintiffs’ Attorneys (N=828) Figure 2. Relationship Between Monetary Stakes and Total Litigation Costs, Private-Firm Defendants’ Attorneys (N=715) Stakes Stakes Costs Costs