Hostname: page-component-586b7cd67f-tf8b9 Total loading time: 0 Render date: 2024-11-27T22:23:49.493Z Has data issue: false hasContentIssue false
  • English
  • Français

The Class Action as an Antitrust Enforcement Device: The Chicago Experience (I)

Published online by Cambridge University Press:  27 December 2018

Benjamin S. DuVal Jr.
Get access

Extract

The common question class action has been a source of division and controversy in the legal community. Hailed by its proponents as both a means for small claimants to obtain redress and a deterrent to corporate wrongdoing, the class action has been attacked by others as tantamount to “legalized blackmail” and as threatening to swamp the already overburdened judicial system with proceedings of extraordinary complexity. Two empirical studies of the class action have also reached diametrically opposed conclusions. A study by the American College of Trial Lawyers found that the common question class action suit “has mandated heavy expenditures of judicial time, effort and expense” and has sacrificed “procedural and substantive fairness to the party opposing the class,” while a study commissioned by the Senate Commerce Committee found that most class actions “proceed with reasonable smoothness in the Federal court.” Proposals to restrict the use of class actions have been advanced. While these proposals have not been adopted, the courts have increasingly limited the circumstances under which class actions may be maintained.

Type
Research Article
Information
American Bar Foundation Research Journal , Volume 1 , Issue 3 , Summer 1976 , pp. 1021 - 1106
Copyright
Copyright © American Bar Foundation, 1976 

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 E.g., Beverly C. Moore, Jr., The Potential Function of the Modern Class Suit, 2 Class Action Rep. 47 (1973); Harry Kalven, Jr., & Maurice Rosenfield, The Contemporary Function of the Class Suit, 8 U. Chi. L. Rev. 684, 684–86, 714–17 (1941); Benjamin Kaplan, A Prefatory Note, 10 B.C. Ind. & Corn. L. Rev. 497 (1969).Google Scholar

2 E.g., Milton Handler, The Shift from Substantive to Procedural Innovations in Antitrust Suits–the Twenty-Third Annual Antitrust Review, 71 Colum. L. Rev. 1, 5–12 (1971); Henry J. Friendly, Federal Jurisdiction: A General View 118–20 (New York: Columbia University Press, 1973); American College of Trial Lawyers, Report and Recommendations of the Special Committee on Rule 23 of the Federal Rules of Civil Procedure, at III, 6–15 (1972) [hereinafter cited as American College of Trial Lawyers Report].Google Scholar

3 American College of Trial Lawyers Report, supra note 2.Google Scholar

4 Note, The Rule 23(6)(3) Class Action: An Empirical Study, 62 Geo. L.J. 1123, 1165–66 (1974).Google Scholar

5 See American College of Trial Lawyers Report, supra note 2; American Bar Association Section of Corporation, Banking and Business Law, Recommendation Regarding Consumer Class Actions for Monetary Relief, Summary of Action and Reports to the House of Delegates, American Bar Association, 1973 Annual Meeting, Washington, D.C., Aug. 6–8, 1973, at 110. The Corporation Section submitted its recommendations to the House of Delegates of the American Bar Association at the 1973 annual meeting. Action on the recommendation was deferred pending study of the issues by an Ad Hoc Committee on Consumer Class Actions. The unanimous recommendation of this committee that (1) “no restrictive changes should be made at this time in the provisions of Rule 23 of the Federal Rules of Civil Procedure or in similar state rules, and any consumer class action legislation adopted by a state in the immediate future should be patterned after Federal Rule 23” and that (2) “no other legislative restrictions should be enacted at this time with respect to consumer class actions for damages” was adopted by the House of Delegates in August 1974. Summary of Action of the House of Delegates, American Bar Association, 1974 Annual Meeting, Honolulu, Hawaii, Aug. 12–16, 1974, at 11, 121.Google Scholar

6 E.g., Eisen v. Carlisle & Jacquelin, 417 U.S. 156 (1974) (individual notice must be sent to all class members who can be identified through reasonable effort; class representative must bear the cost of notice to the class); Zahn v. International Paper Co., 414 U.S. 291 (1973) (each class member must satisfy the jurisdictional amount requirement independently); Eisen v. Carlisle & Jacquelin, 479 F. 2d 1005 (2d Cir. 1973), vacated on other grounds, 417 U.S. 156 (1974) (fluid recovery not a proper remedy; if only remedy is fluid recovery then class action unmanageable); In re Hotel Telephone Charges, 500 F. 2d 86 (9th Cir. 1974) (held common questions did not predominate in part because fluid recovery improper); Abercrombie v. Lum's, Inc., 345 F. Supp. 387 (S.D. Fla. 1972) (where suit alleged illegal tying arrangements, in the absence of a contractual commitment common issues did not predominate).Google Scholar

7 E.g., Perma Life Mufflers, Inc. v. International Parts Corp., 392 U.S. 134, 138–39 (1968); Leh v. General Petroleum Corp., 382 U.S. 54, 59 (1965); Minnesota Mining & Mfg. Co. v. New Jersey Wood Finishing Co., 381 U.S. 311, 317–18 (1965).Google Scholar

8 Professor Posner reports that out of 536 convictions in criminal antitrust cases from 1890 through 1969, a prison sentence was imposed in only 26 cases, most of them involving either violence or union misconduct. He states that the first prison sentence for “pure” price fixing was not imposed until the 1955–59 period. Richard A. Posner, A Statistical Study of Antitrust Enforcement, 13 J. Law & Econ. 365, 388–91 (1970). Although both the number of criminal convictions and the proportion imprisoned have increased, the proportion sentenced to imprisonment remains under 10 percent.Google Scholar

9 Prior to 1974 the maximum fine for violation of the Sherman Act was $50,000. Act of July 7, 1955, ch. 281, 69 Stat. 282. In 1974 it was increased to $1 million. 88 Stat. 1708, 15 U.S.C. sec. 1 (Supp. IV, 1974). Even a fine of $1 million is dwarfed by the sales and profits of the nation's largest corporations. The mean sales and profits of the 500 largest industrial corpo rations in 1974 were, respectively, $1,668,000,000 and $87,239,000. The Fortune Directory of the 500 Largest Industrial Corporations, Fortune, May 1975, at 209. One million dollars is also small when compared with the damages awarded in some antitrust cases. Charles W. Wolfram, The Antibiotics Class Actions, 1976 A.B.F. Res. J. 251, 254. Moreover, it is the maximum fine which may be imposed; the twenty-fold increase in the maximum permissible fine will not necessarily result in a corresponding increase in actual fines. Most fines prior to the amendment were below the $50,000 maximum. The average fine in cases filed between July 7, 1955, and June 30, 1968, was $13,483 for corporations and $3,068 for individuals. James M. Clabault & John F. Burton, Jr., Sherman Act Indictments 1955–1968: A Legal and Economic Analysis 42 (New York: Federal Legal Publications, Supp. 1968). The tenfold 1955 increase in the maximum permissible fine from $5,000 to $50,000 resulted in less than a fourfold increase in the average fine. Posner, supra note 8, at 390. Fines of 1 to 10 percent of sales and fines of 25 percent of profits have been proposed by commentators. Mark J. Green with Beverly C. Moore, Jr., & Bruce Wasserstein, The Closed Enterprise System: Ralph Nader's Study Group Report on Antitrust Enforcement 175 (New York: Grossman Publishers, 1972), William Breit & Kenneth G. Elzinga, Antitrust Penalties and Attitudes Toward Risk: An Economic Analysis, 86 Harv. L. Rev. 693, 711–13 (1973). In the European Economic Community fines of up to 10 percent of the gross annual sales volume may be imposed. Council of the European Economic Community, Regulation No. 17, art. 15, 1 CCH Comm. Mkt. Rep. para. 2541.Google Scholar

10 15 U.S.C. secs. 13, 14, 18, 19 (1970). If conduct violative of the Clayton Act (or the Federal Trade Commission Act) is also violative of the Sherman Act, it may, of course, be prosecuted.Google Scholar

11 The Antitrust Division expended approximately $13,728,000 in fiscal 1974. Hearings on Departments of State, Justice, and Commerce, the Judiciary, and Related Agencies Appropriations for 1975 Before a Subcomm. of the House Appropriations Comm., 93d Cong., 2d Sess., at 362 (1974) [hereinafter cited as 1975 Appropriations Hearings]. Only a portion of the Federal Trade Commission's appropriations of $38,348,000 in fiscal 1975 was for antitrust enforcement. 30 Cong. Q. Almanac, 93d Cong., 2d Sess., 1974, at 129. Thomas Kauper, head of the Antitrust Division, testified that the division had fewer personnel and a smaller budget in real dollars in 1974 than in 1950, while the gross national product in real dollars had more than doubled. 1975 Appropriations Hearings 1159.Google Scholar

12 21 Cong. Rec. 2456 (1890) (statement of Senator Sherman); 51 Cong. Rec. 9185 (1914) (statement of Congressman Helvering). See cases cited, supra note 7.Google Scholar

13 “The obvious purpose of the section [providing for the recovery of treble damages] was not only to provide redress for private wrongs but also to build into the act the feature of self-enforcement that had been typical in cases of restraint of trade at common law.” Hans B. Thorelli, The Federal Antitrust Policy: Origination of an American tradition 255 (Baltimore: Johns Hopkins Press, 1955).Google Scholar

14 A final judgment or decree heretofore or hereafter rendered in any civil or criminal proceeding brought by or on behalf of the United States under the antitrust laws to the effect that a defendant has violated said laws shall be prima facie evidence against such defendant … as to all matters respecting which said judgment or decree would be an estoppel as between the parties thereto: Provided, That this section shall not apply to consent judgments or decrees entered before any testimony has been taken or to judgments or decrees entered in actions under section 15a of this title.Google Scholar

15 U.S.C. sec. 16(a) (1970).Google Scholar

15 American College of Trial Lawyers Report, supra note 2, at III.Google Scholar

16 Handler, supra note 2, at 9.Google Scholar

17 American College of Trial Lawyers Report, supra note 2, at 20–21; Thomas J. Weithers, Amended Rule 23: A Defendant's Point of View, 10 B.C. Ind. & Coin. L. Rev. 515, 521 (1969).Google Scholar

18 Panel Discussion, Antitrust Class Actions, 41 Antitrust L.J. 321, 344 (1972) (comment of Mr. Haas).Google Scholar

18A. Cook, DeKalb, DuPage, Grundy, Kane, Kendall, Lake, LaSalle, McHenry, and Will counties.Google Scholar

19 Admin. Office U.S. Cts. Annual Rep. 1972, at 291–303; 1973, at A-15, A-17. In fiscal 1974 the Northern District of Illinois ranked fourth in private antitrust filings. Admin. Office U.S. Cts. Annual Rep. 1974, at 397–405.Google Scholar

20 Admin. Office U.S. Cts. Annual Rep. 1974, at 254–58.Google Scholar

21 See, e.g., American College of Trial Lawyers Report, supra note 2, at 4; Friendly, supra note 2, at 118.Google Scholar

22 For cases transferred into the district, the date on which the action was filed in the Northern District of Illinois.Google Scholar

23 One seven-case cluster and one unclustered case.Google Scholar

24 American College of Trial Lawyers Report, supra note 2, at Ill.Google Scholar

25 The 541 private antitrust suits represent less than 3.5 percent of the 15,604 private civil actions filed in the district during the period. See Admin. Office U.S. Cts. Annual Rep. 1967, at 207; 1968, at 199; 1969, at 211; 1970, at 237; 1971, at 269; 1972, at 299; 1973, at A-17. Antitrust actions, however, impose a disproportionate burden on the judicial machinery. A time study by the Federal Judicial Center showed that on the average they consume almost twice as much judge time as the average civil case. Report to the Federal Judicial Center, The 1969–70 Federal District Court Time Study 66E (1971). Administrative Office statistics show that anti-trust cases have also tended to pend longer than most other types of private civil actions. Admin. Office U.S. Cts. Annual Rep. 1971, at 297; 1972, at 337; 1973, at A-28; 1974, at A-30.Google Scholar

26 Fed. R. Civ. P. 42(a).Google Scholar

27 (c) Civil cases are deemed related when a case filed (i) relates to property included in an earlier numbered pending suit, or (ii) involves substantially the same issue of fact or grows out of the same transaction as a pending suit, or (iii) involves the validity or infringement of a patent already in suit in any pending earlier numbered case, or (iv) is brought as a class action and involves substantially the same issues of law and one or more of the same classes as a pending suit, or in the opinion of the Executive Committee is so similar in some other respects to a pending case that time and effort of Court and Counsel would be saved by assigning the cases to one Judge.Google Scholar

N.D. Ill. Gen. R. 108(4).Google Scholar

28 See pp. 1079–80 infra.Google Scholar

29 Table 3 shows 167 clustered antitrust cases filed in fiscal 1967–73 (81 nonclass and 86 class). The discrepancy results from the use of two separate sources to compile these figures. The lower figure was compiled from raw data and is, we believe, the correct one. Since clustered cases have not been used in the statistical analysis, this coding error does not influence our analysis.Google Scholar

30 Illinois v. Harper & Row Publishers, Inc. and related cases, 67 C 1899 (N.D. Ill., filed Nov. 1, 1967). Opinion on certification reported at 301 F. Supp. 484 (N.D. Ill. 1969).Google Scholar

31 Connecticut v. General Motors Corp. and related cases, 71 C 830 (N.D. Ill., filed Apr. 2, 1971).Google Scholar

32 E.g., Para-Tone, Inc. v. Pantone, Inc., 69 C 610 (N.D. Ill., filed Mar. 21, 1969) (dispute between two competing companies over right to publish copyrighted work); Grigsby Barton, Inc. v. C.P. Clare & Co. and related cases, 72 C 2472 (N.D. Ill., filed Oct. 3, 1972) (charging defendant with organizing a group boycott of plaintiff's patent) (cluster involved five cases against different defendants); Research Corp. v. Pfister Associated Growers, Inc. and related case, 63 C 597 (N.D. Ill., filed Apr. 10, 1963) (alleging boycott of plaintiff's patent) (defendant class alleged); Lincoln Laboratories, Inc. v. American Cyanamid Co. and related case, 66 C 1762 (N.D. Ill., filed Sept. 30, 1966) (filed as apparent countermove to patent infringement action) (plaintiff alleged actions designed to drive it from market).Google Scholar

33 Only 49 of which, however, remained in the district and are included in the 164 clustered cases. See note 34 infra.Google Scholar

34 One of these entities consisting of two nonclass cases was transferred back to the originating district almost immediately after being transferred in–a settlement had been worked out prior to the transfer to this district. For purposes of analysis, therefore, Childrens Books consists of nine entities aggregating 49 cases which remained in the district.Google Scholar

35 This number represents actual rather than authorized judges. G. William Foster, Jr. (unpublished research, American Bar Foundation Class Action Study).Google Scholar

36 If these figures are adjusted to omit fiscal 1967 and 1968, during which only three class actions were filed, there were 11.4 judges, 4.3 cases filed per judge, or 0.9 cases per judge per year.Google Scholar

37 See Benjamin S. DuVal, Jr., The Class Action as an Antitrust Enforcement Device: The Chicago Experience (II) in 1976 A.B.F. Res. J. (forthcoming).Google Scholar

38 The Advisory Committee apparently anticipated such a reduction, since it referred to subdivision (b)(3) as encompassing cases “in which a class action would achieve economics of time, effort, and expense ….” Notes of Advisory Comm. on 1966 Amendment to Rules, 28 U.S.C. at 7767, following Fed. R. Civ. P. 23 (1970). On the other hand, it is clear that multiple lawsuits have been more common among cases brought as class actions than among nonclass cases. This may, of course, be the result of differences in the nature of the alleged violations.Google Scholar

39 Information obtained through the attorney interviews tended to confirm the validity of using docket entries to measure burden. Although some cases appeared to be more burdensome and some less burdensome than the number of docket entries suggested, generally the cases that had more docket entries were also the cases that seemed more burdensome on the basis of interview data.Google Scholar

40 Tests of statistical significance are most commonly employed to determine the appropriateness of generalizing from a sample. In the present study, however, sampling in the usual sense was not employed. Data was collected on substantially all of the antitrust treble damage cases in the district during the period of the study. The differentials and correlations reported thus (subject to any errors that might have occurred in data collection and processing) represent actual differences in the behavior of cases in the district. The cases actually filed in the district can be viewed, however, as a sample of a larger hypothetical universe of cases that might have been filed. Significance tests are used to determine the appropriateness of generalizing to this larger hypothetical population. For a more detailed discussion of the use of significance tests in studies dealing with an entire population, see Hubert M. Blalock, Jr., Social Statistics 238–39 (2d ed. New York: McGraw-Hill Book Co., 1972). I wish to stress, however, that statistical concepts do not justify generalizing from the behavior of cases in this district to the behavior of cases in the nation as a whole.Google Scholar

Significance tests were run only for unclustered cases. Because of the small number of entities (24) and the wide variations in burden, it is unlikely that any of the differentials would be statistically significant. The entity information is presented, therefore, to indicate whether the patterns observed in unclustered cases are also present in entities. For similar reasons no attempt was made to test the significance of the differentials presented in table 13.Google Scholar

41 The median number of docket entries for unclustered class and nonclass are, respectively, 108 and 44. Since differences in means are more readily amenable to statistical techniques employed in the analysis, means instead of medians were generally used, but where practical the medians were checked as well.Google Scholar

42 Hereinafter, levels of significance will be indicated in the text as has been done here.Google Scholar

43 The percentage of variation “explained” (variance explained) is a central concept in the analysis. The concept can be most easily understood in terms of prediction. Consider two situations. In the first all that is known of a case is that it is an unclustered antitrust suit. In the second whether or not the case is a class action is also known. In the first situation the most accurate prediction of the number of docket entries in any particular case is the mean number of docket entries in all unclustered antitrust cases. In the second situation the most accurate prediction will vary depending on whether the case is a class action. If the case is a class action the most accurate prediction will be the mean number of docket entries in unclustered antitrust class actions. If the case is not a class action the most accurate prediction will be the mean number of docket entries in unclustered nonclass antitrust cases. If the mean number of docket entries in class and nonclass cases differ, the errors in the second situation will be smaller than the errors in the first situation. The percentage of variation explained is the percentage decrease in these errors. (When we deal with a relationship between two variables (a “bivariate relation”) the percentage of variation explained by a given variable is equal to the square of the correlation coefficient. For a discussion of correlation coefficients see Blalock, supra note 40, at 376–93.)Google Scholar

44 For a discussion of the methods used to evaluate burdensomeness and the difficulties encountered in these evaluations, see Appendix A.Google Scholar

45 In discussions of the interview data the terms “class action,”“class action proceeding,” and “case” refer to both single cases and clusters of cases. This has been done for ease of expression and to help maintain confidentiality. The number of “cases” is therefore the number of unclustered cases and of case clusters. The reference to clusters rather than entities reflects the fact that interviews relating to clustered cases were conducted on a cluster rather than an entity basis. While, in theory, interviews on an entity basis would have been preferable, a need to subdivide clusters into entities was not recognized until after the interviews had been completed.Google Scholar

46 For a case that fits this model very well indeed, see Wolfram, supra note 9.Google Scholar

47 “Litigated” as used here encompasses not only the trial but also such pretrial activities as discovery, motions, and briefing. To say that the class issue has been seriously litigated, however, implies that attorney or court activity more substantial than the filing of a motion to certify the class (when the filing of the motion is not accompanied by other activity) has taken place.Google Scholar

Note: The number of cases is not given in tables 17–20 to avoid disclosing the number of clusters. The number of unclustered nonclass cases is 154 and the number of nonclass entities is 13-i.e., the total number of unclustered nonclass cases and nonclass entities. The number of unclustered class actions and of class and mixed entities in each table is quite small.Google Scholar

48 Some cases do not fall squarely into either this category (“Cases in Which Certification of the Class Was Refused”) or the one that follows (“Cases in Which a Class Was Certified or a Settlement Entered in Favor of the Class”). Such cases have been assigned to the category that seemed most accurately to reflect the course of the litigation.Google Scholar

In Pennsylvania v. Ciba Corp., 70 C 2495 (N.D. Ill., filed Oct. 8, 1970) both a consumer class and a class consisting of governmental subdivisions and certain publicly supported institutions were alleged. The court denied certification of the consumer class but accorded class status to the government entity class, and shortly thereafter a settlement terminating the case was entered in favor of the government entity class. Since the consumer class appears to have been by far the more important of the two classes, the case was treated as one in which certification of the class was denied.Google Scholar

In the Concrete Blocks cases (Corona Construction Co. v. Ampress Brick Co. and related cases, 73 C 1082 (N.D. Ill., filed Apr. 26, 1973)) a settlement has been entered terminating four of the five cases. Although no decision on certification has been entered in the remaining case, the settlements disposed of a large portion of the litigation. The Concrete Blocks cases were classified, therefore, as a case in which a settlement had been entered in favor of the class.Google Scholar

In one of the Auto Glass cases (General Glass Co. v. Globe Glass & Trim Co. and related cases, 71 C 921 (N.D. Ill., filed Apr. 16, 1971)) a relatively small settlement was entered by one defendant in favor of the class. None of the cases was terminated as a result of this settlement and, thereafter, certification was denied in each of the class actions in the cluster (in one instance after an initial decision certifying the class). (For an opinion denying class certification in one of the cases, see Hettinger v. Glass Specialty Co., 59 F.R.D. 286 (N.D. Ill. 1973)). In view of the size of the settlement and the subsequent litigation, the cluster was treated as a case in which certification was denied.Google Scholar

In Weit v. Continental Ill. Nat'l Bank & Trust, 70 C 1926 (N.D. Ill., filed Aug. 4, 1970), two classes based on distinct antitrust claims were alleged. Although certification was granted as to only one of these classes and was denied as to the other (60 F.R.D. 5 (N.D. Ill. 1973)), both the size of the certified class and the fact that the case has proceeded as a class action dictated treating the case as one in which certification had been granted.Google Scholar

In the Coal cases (Liebman v. J. W. Petersen Coal & Oil Co. and consolidated cases, 71 C 1340 (N.D. Ill., filed June 2, 1971)) (opinion approving settlement reported 1974 Trade Cas. para. 75,117), the court certified a consumer class but refused to certify a government sub-division class that had been alleged in one of the six cases in the cluster. The cluster was treated as one in which certification had been granted.Google Scholar

49 One defendants' attorney in this case, who stated that less than 25 percent of the time in the case had been consumed by its class aspects, explained that, had the case not been a class action, the discovery would have been about the same but there would not have been the briefing and analysis on the class action issue.Google Scholar

50 The plaintiff's attorney stated that he was unable to estimate the proportion of time devoted to the class aspects because of the extent to which the class aspects were intertwined with the merits. He stressed that he had spent a “tremendous amount of time on discovery” and that this discovery was needed both for the class aspects and for the merits. Consequently it was only the time expended in research and briefing of the class issue that constituted added time. Defendant's attorney, in his estimate of the proportion of time spent on the class aspects, stated that he devoted his time principally to briefing and case analysis. Apparently he did not include the time devoted to factual investigation, which, he said, would have been necessary even if the case had not been a class action.Google Scholar

51 Liebman v. J. W. Petersen Coal & Oil Co. and consolidated cases, 71 C 1340 (N.D. Ill., opinion of May 29, 1973).Google Scholar

52 In a (b)(3) class action the court is required to direct notice to each class member advising that “the court will exclude him from the class if he so requests by a specified date …. Fed. R. Civ. P. 23(c)(2).Google Scholar

53 Included in this category are two cases in which the class allegations were dropped after substantial litigation on the class issue, two pending cases in which the class certification issue was awaiting decision at the time of the interviews, and ten cases in which class status was denied.Google Scholar

54 In one of these relatively little time had been devoted to the class issue, and plaintiff's attorney conceded that the class allegations had turned out to be ill-founded.Google Scholar

55 The nine cases and case clusters in which the class was certified or a settlement was entered in favor of the class and which were more burdensome than the average nonclass case represent less than 5 percent of the number of unclustered treble damage suits and treble damage case clusters filed during the seven-year period.Google Scholar

56 The time required for research of the law relating to class actions may decline as the procedures become more familiar to attorneys and the legal issues become more settled. An attorney in one of the earlier cases filed under amended Rule 23 noted that he and his partner had undertaken an in-depth study of the amended Rule. Since there was no discovery relating to the class aspects of that case, this study presumably accounted for a major part of the 30–35 percent of the time which that attorney estimated had been devoted to the class issues in that case. In contrast, another attorney in a more recently filed suit indicated that it had not been necessary for him to spend much time researching the class issues, since his firm maintained an up-to-date brief of the law on the subject.Google Scholar

57 See, e.g., Thompson v. T.F.I. Companies, Inc., 64 F.R.D. 140 (N.D. Ill. 1974); Abercrombie v. Lum's, Inc., 345 F. Supp. 387 (S.D. Fla. 1972).Google Scholar

58 Similarly, in a case in which the judge had ordered that discovery be limited to the existence of an effect on interstate commerce, one defendants' attorney stated that the order had little impact on the extent of discovery since it could not really prevent discovery which went to the merits or the class issue.Google Scholar

59 See discussion at p. 1071 infra.Google Scholar

60 To separate out the effect of discovery from the effect of whether the case is a class action we employed the statistical technique known as multiple regression analysis, discussed in note 63 infra.Google Scholar

61 See, e.g., Fortner Enterprises, Inc., v. United States Steel Corp., 394 U.S. 495 (1969); Standard Oil Co. v. United States, 337 U.S. 293 (1949).Google Scholar

62 See discussion pp. 1068–69 supra.Google Scholar

63 To determine the extent to which each of these factors-the numbers of plaintiffs, plaintiffs' attorneys, defendants, and defendants' attorneys-has an effect on burden independent of the other three, we employed the statistical technique known as multiple regression analysis. This technique builds up hypothetical cases that differ in one of the factors, e.g., number of plaintiffs, while remaining constant in the other factors. Since the two cases are the same in other respects, any difference between the two cases in docket entries must be attributed to the effect of the particular factor which is varied. The first column in the tabulation below measures the difference in mean docket entries associated with an increase of one unit in the number of plaintiffs, plaintiffs' attorneys, defendants, and defendants' attorneys when these variables are considered separately–i.e., without controlling for the remaining variables. For instance, an increase of one defendants' attorney is associated with an increase of over 25 docket entries. This measure may understate or overstate the impact of a variable on variation, however. While an increase of one plaintiff's attorney is shown to be associated with a relatively large increase in the number of docket entries, there is little variation in the number of plaintiffs' attorneys, most cases being brought by a single attorney. To determine the relative influence of each variable on burden it is necessary to standardize them. The figures in the second column of the table describe the relative influence of each variable in units of measurement that have been standardized on the basis of variability (i.e., change in z scores). When the units of measurement have been standardized it becomes clear that differences in the number of defendants' attorneys are a more important factor in explaining variations in docket entries than variations in the number of plaintiffs' attorneys.Google Scholar

The two foregoing measures both indicate only the effect of each factor on docket entries when they are considered separately. The partial regression coefficient in column 3 of the table measures the mean differences in docket entries associated with a one-unit difference in each of the four variables when the effects of the other three variables are controlled. For instance, an increase of one defendants' attorney is shown to be associated with an increase of 16.8 docket entries when the effects of the other variables are controlled as contrasted with an increase of 25.3 docket entries when they are not controlled. The final column measures in standardized units the effect of each variable when the remaining variables are controlled. The relative magnitude of the standardized partial regression coefficients indicates the relative importance of the direct effect of each variable. As the table shows, while an increase of one plaintiffs' attorney results in a greater increase in mean docket entries than an increase of one unit in any other variable, even after controlling for the remaining three variables, variations in the number of plaintiffs' attorneys is the least important of the four factors in explaining variations in the number of docket entries.Google Scholar

64 Occasionally a single defendant will be represented by more than one firm–usually, in such instances, one Chicago firm and one out-of-state firm. Multiple firms in these cases may present fewer problems of coordination because one firm is likely to be in control. They will certainly not present the same problems of duplicated filings.Google Scholar

65 See note 63 supra. Although the number of plaintiffs has a greater impact on the number of discovery entries than does the number of plaintiffs' attorneys, the direct effect of the number of plaintiffs' attorneys is greater than the direct effect of the number of plaintiffs. This is because the number of plaintiffs affects the number of discovery entries in large part through its association with other variables.Google Scholar

66 In one of these cases a plaintiffs' attorney stated that the multiplicity of plaintiffs' attorneys had not significantly added to the burden because only a few had ever actively participated in the proceedings. In the other case, there were only two defendants' law firms involved. The defendants' attorney interviewed stated that the multiplicity did not “unduly” complicate matters, noting that the two defendants filed joint briefs on all common questions.Google Scholar

67 In two other cases attorneys stressed that only two firms had been involved to any substantial extent. In one case, which appears to have consumed no greater time (and perhaps less time) than the average nonclass treble damage suit, one of the principal defendants' attorneys stated that while there had been duplication, the multiplicity of defendants had not created any great problem. In the other, which was more burdensome than the average nonclass suit, one of the defendants' attorneys, while not directly commenting on the impact of multiplicity in the case, stressed that only two law firms had been involved and that there had been good cooperation among the defendants. Both attorneys emphasized that in other cases where more firms had been involved, the added burden had been much greater.Google Scholar

68 One prominent defense attorney declared: “I have a personal feeling that most defendants are overrepresented to their disadvantage. I think most defendants would do much better if they settled on one counsel … but clients seem to feel that they want to have their own lawyer ….”Google Scholar

69 One of the prerequisites of certification of the class is that the court find that “the representative parties will fairly and adequately protect the interests of the class.” Fed. R. Civ. P. 23(a)(4). Although the number of named plaintiffs is generally not considered decisive, some courts have relied in part on this factor. See 7 Charles Alan Wright & Arthur R. Miller, Federal Practice and Procedure sec. 1766 (St. Paul: West Publishing Co., 1972); 3B James Wm. Moore & Jo Desha Lucas, Moore's Federal Practice para. 23.07[4] (New York: Matthew Bender, 1975). One plaintiff's attorney, while not subscribing personally to the need for multiple named plaintiffs, stated that including them might reflect a belief that multiplicity would help convince the court that the named plaintiffs were representative. He added that if you represent one plaintiff and he is “knocked out … you're out of court” while if you have two or three plaintiffs it is “less likely that they'll [all] be knocked out.”Google Scholar

70 In one franchisee case a large number of the original plaintiffs dropped out of the action, allegedly because of pressure from the franchisor.Google Scholar

71 The difference is not statistically significant at the .05 level and may be the result of chance. On the other hand, the likelihood of such a distribution resulting from chance is less than 20 percent.Google Scholar

72 Compare tables 22 and 27. Without controlling for discovery these variables collectively explain 61 percent of the variation in docket entries. If variations in discovery are first taken into account they explain an additional 29 percent of the variation in docket entries.Google Scholar

73 A path diagram tracing the relationship between whether a case is a class action, the number of plaintiffs, defendants, plaintiffs' attorneys, defendants' attorneys, discovery entries, and docket entries is set forth below. The arrows in this diagram indicate actual or possible causal relationships. Thus, it is assumed that whether a case is a class action may affect the number of plaintiffs and plaintiffs' attorneys but that the number of plaintiffs and plaintiffs' attorneys will not affect whether the case is brought as a class action. (For an explanation of path analysis see Jae-On Kim & Frank J. Kohout, Special Topics in General Linear Models, in Norman H. Nie et al., SPSS: Statistical Package for the Social Sciences 368, 383–97 (2d ed. New York: McGraw-Hill Book Co., 1975).) Some possible paths that were found to be of de minimis effect and that were of no theoretic interest have been eliminated from the diagram in the interest of simplicity. The effect of any of the independent variables on burden through any of the paths can be determined by multiplying the coefficients on that path. For example, the extent to which an increase in the number of defendants results in an increase in the number of docket entries through its impact on the number of defendants' attorneys is determined by multiplying .5197 by .2755.Google Scholar

The impact of whether a case is a class action on the number of docket entries can be decomposed into the following paths: CA→NP→DE; CA→ND→DE; CA→AP→DE; CA→AD→DE; CA→SE→DE; CA→DE. The last of these paths represents the extent to which whether a case is a class action has a direct effect on docket entries–an impact independent of the association of the class action variable with any of the other variables considered. In terms of substantive theory this impact can be equated with the impact of the class aspects of the case. In this decomposition the effect of certain indirect paths has been assigned to less indirect paths. For instance, class actions have more defendants' attorneys in part because they have more defendants. Thus one logical path is CA→ND→AD→DE. This path is subsumed in the tabulation below under “indirect effect through No. of defendants' attorneys.” The result of this procedure is to assign the maximum effect to that mediating variable proximately linked with docket entries.Google Scholar

The suggestion that there may be a direct causal link between whether a case is a class action and the number of defendants' attorneys requires explanation. In one sense class actions have more attorneys only because they have more defendants. Defendants hire attorneys, not attorneys defendants. The number of defendants' law firms involved in a case, however, should, in theory, reflect more accurately the number of independent defendants than does the total number of defendants. It was, therefore, to be expected that if whether a case was a class action had an effect on the number of defendants' attorneys, it would be more highly correlated with the number of defendants' attorneys than with the number of defendants.Google Scholar

74 The direct impact of the class action variable on docket entries just misses being significant at the .05 level. There is, consequently, a slightly better than 5 percent probability that the variation between class and nonclass cases which exists after controlling for the number of parties, the number of attorneys, and the amount of discovery would occur on a random basis. The interviews suggest, however, that the variations represent modest increments to burden resulting from the class aspects of the case.Google Scholar

75 United States v. Trenton Potteries Co., 273 U.S. 392 (1927); Dr. Miles Medical Co. v. John D. Park & Sons Co., 220 U.S. 373 (1911).Google Scholar

76 United States v. Socony-Vacuum Oil Co., 310 U.S. 150, 223–24 & n.59 (1940) (dictum). See United States v. McKesson & Robbins, Inc., 351 U.S. 305 (1956).Google Scholar

77 United States v. Socony-Vacuum Oil Co., 310 U.S. 150, 224 n.59 (1940) (dictum). See Plymouth Dealers' Ass'n v. United States, 279 F. 2d 128, 132 (9th Cir. 1960).Google Scholar

78 United States v. Trenton Potteries Co., 273 U.S. 392 (1927). Cf. Goldfarb v. Virginia State Bar, 421 U.S. 773 (1975) (not necessary to show that fee schedule raised prices).Google Scholar

79 United States v. Socony-Vacuum Oil Co., 310 U.S. 150, 221–22 (1940).Google Scholar

It has been held too often to require elaboration now that price fixing is contrary to the policy of competition underlying the Sherman Act and that its illegality does not depend on a showing of its unreasonableness, since it is conclusively presumed to be unreasonable. It makes no difference whether the motives of the participants are good or evil; whether the price fixing is accomplished by express contract or by some more subtle means; whether the participants possess market control; whether the amount of interstate commerce affected is large or small; or whether the effect of the agreement is to raise or to decrease prices.Google Scholar

United States v. McKesson & Robbins, Inc., 351 U.S. 305, 309–10 (1956).Google Scholar

80 The concerted refusal of competitors to buy from, sell to, or otherwise deal with a particular firm or class of firms either absolutely or except on discriminatory terms. See Radiant Burners, Inc. v. Peoples Gas Light & Coke Co., 364 U.S. 656 (1961); Klor's, Inc. v. Broadway-Hale Stores, Inc., 359 U.S. 207 (1959). Although there is some authority that not all concerted refusals to deal fall within the per se rule (see, e.g., United States v. United States Trotting Ass'n, 1960 Trade Cas. para. 69,761 (S.D. Ohio 1960); Section of Antitrust Law of the American Bar Association, Antitrust Law Developments 18 (Chicago: American Bar Association, 1975) for purposes of the present analysis all concerted refusals falling within the definition given above were classed as per se restraints.Google Scholar

81 United States v. Topco Associates, Inc., 405 U.S. 596 (1972). Although Topco was the first Supreme Court decision squarely to decide the issue, there was considerable prior authority for treating market division as a per se violation. See, e.g., White Motor Co. v. United States, 372 U.S. 253, 263 (1963) (dictum); Northern Pac. Ry. Co. v. United States, 356 U.S. 1, 5 (1958) (dictum); Las Vegas Merchant Plumbers Ass'n v. United States, 210 F. 2d 732, 748–49 (9th Cir.), cert. denied, 348 U.S. 817 (1954); Kingman Brewster, Jr., Antitrust and American Business Abroad 79 (New York: McGraw-Hill Book Co., 1958); Report of the Attorney General's National Committee to Study the Antitrust Laws, 26 (March 31, 1955).Google Scholar

82 In United States v. Arnold, Schwinn & Co., 388 U.S. 365 (1967), the Court held that “[w]here the manufacturer retains title, dominion, and risk with respect to the product and the position and function of the dealer in question are, in fact, indistinguishable from those of an agent or salesman of the manufacturer, it is only if the impact of the confinement is 'unreasonably' restrictive of competition that a violation of sec. 1 results …” but that “lol nce the manufacturer has parted with title and risk, he has parted with dominion over the product, and his effort thereafter to restrict territory or persons to whom the product may be transferred … is a per se violation ….” 388 U.S. at 380, 382. For present purposes, however, no attempt was made to distinguish between distribution by consignment and distribution by sale and resale. Thus, the category “territorial and customer restrictions” includes not only arrangements which constitute a per se offense but also some which do not.Google Scholar

83 Although it is not one of those practices generally thought of as a per se offense, sec. 8 of the Clayton Act makes it unlawful for any person to be simultaneously a director of any two corporations either of which has capital, surplus, and undivided profits aggregating more than $1,000,000 if “such corporations are or shall have been theretofore, by virtue of their business and location of operation, competitors, so that the elimination of competition by agreement between them would constitute a violation of any of the provisions of any of the antitrust laws.” 15 U.S.C. sec. 19 (1970). No effect or probable effect on competition need be shown. United States v. Sears, Roebuck & Co., 111 F. Supp. 614 (S.D.N.Y. 1953).Google Scholar

84 Standard Oil Co. v. United States, 337 U.S. 293, 305–6 (1949) (Standard Stations case) (dictum), quoted in, e.g., Fortner Enterprises, Inc. v. United States Steel Corp., 394 U.S. 495, 498 (1969). Although tying arrangements have been characterized as a per se violation (see, e.g., Northern Pac. Ry. Co. v. United States, 356 U.S. 1, 5 (1958)), neither in the Northern Pacific case nor in any other has proof of the practice been deemed sufficient to establish a violation.Google Scholar

85 Fortner Enterprises, Inc. v. United States Steel Corp., 394 U.S. 495, 499 (1969). More-over, where either of these standards is met the tying arrangement violates sec. 3 of the Clayton Act. See Times-Picayune Publishing Co. v. United States, 345 U.S. 594, 608–9 (1953); Advance Business Systems & Supply Co. v. SCM Corp., 415 F. 2d 55, 61 (4th Cir. 1969), cert. denied, 397 U.S. 920 (1970). In United States v. General Dynamics Corp., 258 F. Supp. 36, 65 (S.D.N.Y. 1966), the court analogized reciprocal dealing agreements to tying agreements but ruled that the reciprocal dealing agreements proven by the government affected an insubstantial volume of commerce. Cf. FTC v. Consolidated Foods Corp., 380 U.S. 592 (1965) (reciprocity is “one of the congeries of anticompetitive practices at which the antitrust laws are aimed”).Google Scholar

86 See United States v. Jerrold Electronics Corp., 187 F. Supp. 545, 554–58 (E.D. Pa. 1960), aff'd per curiam, 365 U.S. 567 (1961); Dehydrating Process Co. v. A. O. Smith Corp., 292 F. 2d 653 (1st Cir.), cert. denied, 368 U.S. 931 (1961).Google Scholar

In Fortner the Court declared that when the case went to trial “[i]t may turn out that the arrangement involved here serves legitimate business purposes and the U.S. Steel's subsidiary does not have a competitive advantage in the credit market.” 394 U.S. at 506.Google Scholar

87 Brown Shoe Co. v. United States, 370 U.S. 294 (1962); Standard Oil Co. v. United States, 337 U.S. 293, 306–7 (1949).Google Scholar

88 See, e.g., United States v. Philadelphia Nat'l Bank, 374 U.S. 321 (1963); United States v. E. I. duPont de Nemours & Co., 353 U.S. 586 (1957); Standard Oil Co. v. United States, 337 U.S. 293 (1949). Although some lower courts have construed Tampa Elec. Co. v. Nashville Coal Co., 365 U.S. 320 (1961) as mandating a rule of reason approach to exclusive dealing contracts, see, e.g., Curly's Dairy, Inc., v. Dairy Cooperative Ass'n, 202 F. Supp. 481, 484–85 (D. Ore. 1962), Tampa Electric has also been read to hold only that where the market share foreclosed is not sufficient to warrant the application of a rule of presumptive illegality, this does not end inquiry into whether the contract is illegal. Mytinger & Casselberry, Inc. v. FTC, 301 F. 2d 534 (D.C. Cir. 1962); cf. Fortner Enterprises, Inc. v. United States Steel Corp., 394 U.S. 495 (1969) (failure to allege sufficient economic power over tying product or substantial dollar volume of tied product would not warrant dismissal since tying arrangement might still be unlawful under general rule of reason standards). In Philadelphia Nat'l Bank the Court in constructing a rule of presumptive illegality for horizontal mergers relied in part on Standard Stations, in which, according to the Court, a presumptive illegality test was applied. 374 U.S. at 366–67.Google Scholar

89 15 U.S.C. sec. 13 (1970).Google Scholar

90 Neither the prohibition against payment of brokerage in sec. 2(c) nor the prohibition against discrimination in promotional allowances or in services or facilities requires any showing of anticompetitive effect. To establish a claim of illegal price discrimination, however, plaintiff must show that the effect of the discrimination “may be substantially to lessen competition or tend to create a monopoly in any line of commerce, or to injure, destroy, or prevent competition with any person who either grants or knowingly receives the benefit of such discrimination, or with customers of either of them ….” 15 U.S.C. sec. 13(a) (1970). The effect of price discrimination on competitors of the recipient of the lower price (secondary line injury) may be inferred from the existence of a substantial persisting price difference, at least where profit margins are low. See, e.g., FTC v. Morton Salt Co., 334 U.S. 37, 45–47 (1948); United Biscuit Co. of America v. FTC, 350 F. 2d 615 (7th Cir. 1965), cert. denied, 383 U.S. 926 (1966); Foremost Dairies, Inc. v. FTC, 348 F. 2d 674 (5th Cir.), cert. denied, 382 U.S. 959 (1965); Section of Antitrust Law of the American Bar Association, supra note 80. Compare American Oil Co. v. FTC, 325 F. 2d 101 (7th Cir. 1963), cert. denied, 377 U.S. 954 (1964) (price discrimination temporary).Google Scholar

91 25 Stan. L. Rev. 764 (1973); Note, The Cost-Internalization Case for Class Actions, 21 Stan. L. Rev. 383, 418 (1969). See Albertson's, Inc. v. Amalgamated Sugar Co., 62 F.R.D. 43 D. Utah 1973); Shaw v. Mobil Oil Corp., 60 F.R.D. 566 (D.N.H. 1973); Bel Air Markets v. Foremost Dairies, Inc., 55 F.R.D. 538 (N.D. Cal. 1972). But see Gold Strike Stamp Co. V. Christensen, 436 F. 2d 791 (10th Cir. 1970). For the plaintiff to establish liability he must prove that each class member paid a higher price than some competing purchaser and that competition with the favored customer was injured. Moreover, a showing that the price differentials reflected differences in cost or that the lower price was a good faith effort to meet the price of a competitor is a defense. To the extent that each purchaser receives a different price, a purchaser who is disfavored in relation to one competitor may be favored when compared with another, creating conflicting interests among the members of the class. Plaintiff must establish that each member of the alleged class is in competition with a favored purchaser. Although there may be instances in which this issue can be disposed of on a unitary basis, it more often requires individualized treatment as does, generally, cost justification if it is asserted as a defense.Google Scholar

92 The following types of claims were assigned to this category: customer and use restrictions in patent licenses, agreements by firms participating in a joint enterprise to deny competitors access to some service or facility under the control of the joint enterprise, unilateral refusals to deal pursuant to a vertical agreement, predatory pricing, price squeeze, other predatory practices, single firm monopolization, fraudulent procurement of patents, other franchise restrictions, other patent licensing restrictions, attempted monopolization, general allegation of conspiracy in restraint of trade, other federal antitrust.Google Scholar

93 Warren Chem. Co. v. Warren Cook Chem., Inc., 68 C 1056 (N.D. Ill., filed June 10, 1968) (the defendants included a competitor of the plaintiff and 19 of its employees, all of whom were represented by a single attorney); Kramer Prods. Co. v. Elgin Scientific Glass Co., 69 C 1453 (N.D. Ill., filed July 11, 1969) (predatory pricing and unilateral refusal to deal).Google Scholar

94 See text p. 1081 supra.Google Scholar

95 See Donald F. Turner, The Definition of Agreement Under the Sherman Act: Conscious Parallelism and Refusals to Deal, 75 Harv. L. Rev. 655 (1962).Google Scholar

96 What constitutes an agreement to fix prices within the meaning of the per se rule can, at times, present considerable difficulty. Compare United States v. Container Corp. of America, 393 U.S. 333 (1969), with Belliston v. Texaco, Inc., 455 F. 2d 175 (10th Cir.), cert. denied, 408 U.S. 928 (1972). Compare Virginia Excelsior Mills, Inc. v. FTC, 256 F. 2d 538 (4th Cir. 1958), with United States v. Columbia Pictures Corp., 189 F. Supp. 153 (S.D.N.Y. 1960).Google Scholar

97 Section of Antitrust Law of the American Bar Association, supra note 80, at 258–60.Google Scholar

98 One of the least burdensome class actions included in the study charged defendant with a combination to monopolize. No per se, quasi per se, or Robinson-Patman violations were alleged. Plantiff's attorney was inexperienced and lacked the resources necessary for a serious effort. Defendant's attorney declared that “we really didn't treat it [this case] in the same way we treat the major antitrust cases that we have.”Google Scholar

99 As was indicated in the preceding section, whether a case is a class action may affect burden both directly, as a result of time expended on the class aspects of the case, and indirectly, as a result of the tendency of class actions to have greater numbers of parties and attorneys. That differences in the legal theories invoked in class and nonclass cases have no substantial impact on the comparative burdensomeness of class and nonclass cases is most clearly demonstrated by decomposing the effect of class actions into its constituent elements. As the tabulation below shows, controlling for the nature of the legal theory does not substantially increase or decrease either the direct or indirect effect on burden of whether a case is a class action.Google Scholar

100 Although the variance in docket entries, discovery entries, and days pending explained is statistically significant at the .001 level, the increase in variance explained by taking into account the legal theory alleged is not statistically significant.Google Scholar

101 In deriving these percentages each entity which contained a class action was deemed a class entity. The docket entries and discovery entries in the case with the largest number of entries in the entity and the total number of entries for all cases in the entity were used as minima and maxima.Google Scholar

102 See note 25 supra.Google Scholar

103 There were, however, a number of instances in which experienced antitrust attorneys disagreed as to the burdensomeness of the case.Google Scholar

104 Although some cases were viewed by attorneys as less burdensome than the statistical measures of burden would indicate, and much less often as more burdensome, cases viewed by attorneys as particularly burdensome generally ranked high in docket entries and days pending while cases low in docket entries and days pending were generally thought to be not especially burdensome.Google Scholar

105 For instance, in one case one attorney estimated that 25 percent of the time was devoted to the class aspects while another said that 40 percent of his time had been so devoted, but stressed that this was a very rough guess. In another case one attorney estimated that one-third of the time had been devoted to the class aspects while the other attorney stated that relatively little of the time was devoted to them.Google Scholar

106 Some additional burden is imposed on the appellate courts by appeals which relate to the class aspects of the case. The Eisen case, where there were two rounds of appeals, represents an extreme instance. Eisen v. Carlisle & Jacquelin, 41 F.R.D. 147 (S.D.N.Y. 1966) (class status denied), motion to dismiss appeal denied, 370 F. 2d 119 (2d Cir. 1966), cert. denied, 386 U.S. 1035 (1967), rev'd and remanded, 391 F. 2d 555 (2d Cir. 1968), 52 F.R.D. 253 (S.D.N.Y. 1971) (class status granted), rev'd, 479 F. 2d 1005 (2d Cir. 1973), vacated, 417 U.S. 156 (1974). Such appeals have been infrequent, at least in this district. In only 2 of the 31 unclustered class actions was an appeal taken on the class aspects. The low number of appeals in this district may result in part from the rulings by the Court of Appeals for the Seventh Circuit that a decision granting or denying certification of the class is not appealable as of right, pending termination of the case. King v. Kansas City Southern Industries, Inc., 479 F. 2d 1259 (7th Cir. 1973); Thill Sec. Corp. v. New York Stock Exch., 469 F. 2d 14 (7th Cir. 1972); Lupia v. Stella D'Oro Biscuit Co., No. 73–1026 (7th Cir., order entered May 18, 1973), cert. denied, 417 U.S. 930 (1974). For a discussion of the appealability of class action orders see 7A Wright & Miller, supra note 69, at sec. 1802. Probably of equal importance, however, is the fact that there have been relatively few contested decisions on the class issue in this district.Google Scholar

107 In actuality, through inadvertence, the full number of docket entries was utilized in several instances, although the error introduced is minimal.Google Scholar