United States Court of Appeals Fifth Circuit F I L E D REVISED FEBRUARY 23, 2005 February 17, 2005 UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT Charles R. Fulbruge III Clerk _______________________

No. 03-30965 _______________________

FRANCES UNGER, ET AL.; Plaintiffs,

WILLIAM PATTERSON, lead plaintiff; GORDON ELLIS, lead plaintiff,

Plaintiffs-Appellees,

versus

AMEDISYS INC; ET AL.,

Defendants-Appellants.

Appeals from the United States District Court for the Middle District of Louisiana,

Before REAVLEY, JONES, and DENNIS, Circuit Judges.

EDITH H. JONES, Circuit Judge:

This case, on review pursuant to FED. RULE CIV. PROC. 23(f),

implicates the standards and procedures used by district courts

when considering certification of securities class actions

dependent on the “fraud on the market” theory. See Basic, Inc. v.

Levinson,

,

,

(1988).

Like our brethren in the Third, Fourth, Seventh and Ninth Circuits,

we hold that a careful certification inquiry is required and

findings must be made based on adequate admissible evidence to justify class certification. Because the district court

erroneously applied too lax a standard of proof to the plaintiffs’

fraud-on-the-market allegations, we must vacate the class certifi-

cation and remand.

BACKGROUND

Amedisys provides home health care, nursing, home infu-

sion therapy, and ambulatory surgery services. The company’s stock

is traded on the NASDAQ Over The Counter Bulletin Board (“OTCBB”).

Approximately ninety percent of Amedisys’s revenue comes from

Medicare. This case stems from the conduct of Amedisys and its

directors in reporting profits based on new Medicare procedures.

Beginning October 1, 2000, Medicare implemented the

Prospective Payment System (“PPS”), which altered the way Medicare

compensated home health care companies. Under PPS, Medicare paid

health care companies like Amedisys a portion of their fees in

advance, based on forward-looking estimates of the cost of

services. After the company provided the service, the remainder of

the fee was paid; alternatively, if the initial payment proved too

high, the company had to reimburse Medicare the difference. To

comply with the new PPS procedures, Amedisys purchased and

implemented new computer software.

Plaintiffs allege that Amedisys willfully manipulated the

PPS program to inflate the estimated costs for certain health

services; that it thereby artificially fueled company earnings;

2 and, ultimately, that Amedisys’s actions wrongfully enhanced its

stock price. On June 13, 2001, Amedisys issued a curative

statement, conceding that it had overstated revenues, but

maintaining that the overstatements were inadvertently caused by

the new software used with the PPS program. The stock price fell.

On August 21, 2001, Frances Unger filed suit against

Amedisys, alleging violations of Sections 10(b) and 20(a) of the

Securities Exchange Act of 1934 and Rule 10b-5 promulgated

thereunder. As is often the case, plaintiffs’ lawyers solicited

potential class members over the Internet and through newspaper

advertisements. Several other suits were consolidated with Unger’s

and five individuals were chosen as lead plaintiffs. Class

certification was requested for “all persons and entities who

purchased the common stock of Amedisys, Inc. between November 15,

2000 through [sic] June 13, 2001.” Discovery occurred to ascertain

the qualifications of the proposed class representatives. At a

hearing, the district court evaluated this evidence and the

plaintiffs’ sketchy evidence in support of the fraud-on-the-market

basis for their presumed reliance on Amedisys’s misrepresentations.

The district court certified the class under Rule 23(b)(3).

The Amedisys defendants timely sought, and this court

granted, an interlocutory appeal raising two issues embodied in the

class certification: the adequacy of the lead plaintiffs’ quali-

fications and the sufficiency of plaintiffs’ evidence to support

the fraud on the market presumption.

3 DISCUSSION

The class certification determination rests within the

sound discretion of the trial court. Gulf Oil Co. v. Bernard,

,

,

(1981). That

discretion, however, must be exercised within the constraints of

Rule 23.

A district court that premises its legal analysis on

an erroneous understanding of the governing law has abused its

discretion. U.S. v. Insaulgarat,

(5th Cir.

2004); U.S. v. Mann,

(5th Cir. 1998).

Rule 23 requires the claims of a proposed class to meet

several requirements before the class can be certified. The party

seeking certification bears the burden of establishing that

all requirements of Rule 23 have been satisfied. Berger v. Compaq

Computer Corp.,

(5th Cir. 2001). First, the

district court must find what has been termed numerosity,

commonality, typicality, and representativeness.1 For class

actions seeking money damages, like this one, the district court

must make additional findings of predominance and superiority.

Rule 23(b)(3). The predominance element requires a finding that

common issues of law or fact “predominate over any questions

1 The specific language of Rule 23(a) is: One or more members of a class may sue or be sued as representative parties on behalf of all only if (1) the class is so numerous that joinder of all members is impracticable, (2) there are questions of law or fact common to the class, (3) the claims or defenses of the representative parties are typical of the claims or defenses of the class, and (4) the representative parties will fairly and adequately protect the interests of the class.

4 affecting only individual members.”

This requirement,

although reminiscent of the commonality requirement of Rule 23(a),

is “far more demanding” because it “tests whether proposed classes

are sufficiently cohesive to warrant adjudication by

representation.” Amchem Prods., Inc. v. Windsor,

,

,

(1997). Final-

ly, a class action must afford the superior means to achieve “fair

and efficient adjudication of the controversy.” Rule 23(b)(3).

Recognizing the important due process concerns of both

plaintiffs and defendants inherent in the certification decision,

the Supreme Court requires district courts to conduct a rigorous

analysis of Rule 23 prerequisites. Gen’l Tel. Co. v. Falcon,

,

,

(1982).

District courts are required to take a “close look” at the parties’

claims and evidence in making its Rule 23 decision. Amchem,

,

. Class certification hearings

should not be mini-trials on the merits of the class or individual

claims. Eisen v. Carlisle & Jacquelin,

,

, L. Ed. 2d 732 (1974). At the same time,

however, “[g]oing beyond the pleadings is necessary, as a court

must understand the claims, defenses, relevant facts, and

applicable substantive law in order to make a meaningful

determination of the certification issues.” Castano v. Am. Tobacco

Co.,

(5th Cir. 1996). To assist the court in this

process it may sanction controlled discovery at the certification

5 stage. See FED. R. CIV. P. 23 Advisory Committee’s Note to 2003

amendments. The plain text of Rule 23 requires the court to

“find,” not merely assume, the facts favoring class certification.

Rule 23(b)(3).

Appellants first challenge the qualifications of the

class representatives under Rule 23(a)(4). To meet Rule 23

requirements, the court must find that class representatives, their

counsel, and the relationship between the two are adequate to

protect the interests of absent class members. Stirman v. Exxon

Corp.,

(5th Cir. 2002). Class representatives

must satisfy the court that they, and not counsel, are directing

the litigation. To do this, class representatives must show

themselves sufficiently informed about the litigation to manage the

litigation effort. Berger,

.

Nothing in the record indicates that the district court

abused its discretion with regard to the Rule 23(a)(4) requirement.

The district court fully evaluated the evidence, which included

depositions and testimony of the class representatives. The court

was neither clearly erroneous in its factfindings nor in error

legally. To address this argument further would pointlessly

require us to recount the case-specific evidence.

The crux of this appeal lies in the legal basis for and

sufficiency of evidence supporting the district court’s finding of

predominance under Rule 23(b)(3). The district court here ex-

pressed skepticism that Castano, which discussed fraud and other

6 claims raised by a putative nationwide class of tobacco smokers,

should govern securities fraud class actions. Its skepticism was

unfounded. Castano is not logically so limited, and its reasoning

has been approved in the securities fraud context by other circuit

courts as well as by district courts in this circuit. Gariety v.

Grant Thornton LLP,

(4th Cir. 2004); Newton v.

Merrill Lynch, Pierce, Fenner & Smith, Inc.,

, 168 (3d

Cir. 2001); see also Johnston v. HBO Film Management., Inc.,

(3d Cir. 2001); Szabo v. Bridgeport Mach., Inc.,

(7th Cir. 2001); Lehocky v. Tidel Techs.,

Inc.,

(S.D. Tex. 2004); Krogman v. Sterritt,

(N.D. Tex. 2001); Griffin v. G K Intelligent

Sys., Inc.,

(S.D. Tex. 2000).

One of the lessons emphasized by Castano and related

cases is that a district court must perform sufficient analysis to

determine that class members’ fraud claims are not predicated on

proving individual reliance. If the circumstances surrounding each

plaintiff’s alleged reliance on fraudulent representations differ,

then reliance is an issue that will have to be proven by each

plaintiff, and the proposed class fails Rule 23(b)(3)’s

predominance requirement. Castano,

; see also Simon

v. Merrill Lynch, Pierce, Fenner & Smith, Inc.,

(5th Cir. 1973).

Only by invoking the fraud on the market theory can these

plaintiffs establish a classwide rebuttable presumption of reliance

7 on Amedisys’s alleged misrepresentations.2 In Basic, Inc., the

Supreme Court held that reliance may be presumed, enabling 10b-5

class actions to proceed, “when a fraudulent misrepresentation or

omission impairs the value of a security traded in an efficient

market.”3 As the Court explained,

The fraud on the market theory is based on the hypothesis that, in an open and developed securities market, the price of a company’s stock is determined by the available material information regarding the company and its business.... Misleading statements will therefore de- fraud purchasers of stock even if the purchasers do not directly rely on the misstatements.... The causal connection between the defendants’ fraud and the plaintiffs’ purchase of stock in such a case is no less significant than in a case of direct reliance on misrepresentations.

Basic, Inc.,

,

(internal

citations omitted) (emphasis added).

To support this rebuttable presumption, a securities

plaintiff must prove, inter alia, that the security at issue is

traded in an “efficient market.”

,

-93.

In many cases, where heavily-traded or well known stocks are the

target of suits, market efficiency will not even be an issue. But

where, as here, the suit involves small-cap stocks traded in less-

organized markets, a demonstration of an efficient market is a

2 To prevail on a 10b-5 claim, a plaintiff must prove (1) a material misrepresentation or omission by the defendant, (2) scienter on the part of the defendant, (3) reliance, and (4) due diligence by the plaintiff to pursue his or her own interest with care and good faith. Stephenson v. Paine Webber Jackson & Curtis, Inc.,

(5th Cir. 1988). 3 Newton, 259 F.3d at 175.

8 prerequisite for certification.4 Without an initial demonstration

of market efficiency, there is no assurance that the available

material information concerning the stock translates into an effect

on the market price and supports a classwide presumption of

reliance. Absent an efficient market, individual reliance by each

plaintiff must be proven, and the proposed class will fail the

predominance requirement. Cf. Castano,

. Because

this inquiry can prove decisive for class certification, and

because, given the realities of litigation costs, certification can

compel settlements without trial, courts have frequently applied

rigorous, though preliminary, standards of proof to the market

efficiency determination. See, e.g., Gariety,

;

Newton, 259 F.3d at 167-69; Szabo,

; Binder v.

Gillespie,

(9th Cir. 1999); In re Seagate

Tech. II Sec. Litig.,

(N.D. Cal. 1994);

Krogman,

; Griffin,

. Courts

have likened the degree of proof required to the standards used in

preliminary injunction hearings, see Gariety,

, or

in FED. RULE CIV. PROC. 12(b)(1) and 12(b)(2) jurisdiction contests,

4 A recent law review article criticizes the efficient market theory adopted in Basic as out of step with current economic analysis and inconsistent with the thrust of recent legislation. See Jeffrey L. Oldham, Taking “Efficient Markets” out of the Fraud on the Market Doctrine after the Private Securities Litigation Reform Act, 97 NW. U. L. REV. 995 (2003). The author contends that “what determines whether investors were justified in relying on the integrity of the market price is not the efficiency of the relevant market but rather whether a misstatement distorted the price of the affected security.” 97 NW. L. REV. at 1035. The article is persuasively argued, but it is the Supreme Court’s job to overrule Basic, in the absence of outright conflict with the Private Securities Litigation Reform Act,

Stat. 737 (1995).

9 Szabo,

. Although the court’s determination for

class certification purposes may be revised (or wholly rejected) by

the ultimate factfinder, the court may not simply presume the facts

in favor of an efficient market.

Courts have relied on several factors to determine

whether a stock traded in an “efficient market”: (1) the average

weekly trading volume expressed as a percentage of total

outstanding shares; (2) the number of securities analysts following

and reporting on the stock; (3) the extent to which market makers

and arbitrageurs trade in the stock; (4) the company’s eligibility

to file SEC registration Form S-3 (as opposed to Form S-1 or S-2);5

(5) the existence of empirical facts “showing a cause and effect

relationship between unexpected corporate events or financial

releases and an immediate response in the stock price”; (6) the

company’s market capitalization; (7) the bid-ask spread for stock

sales; and (8) float, the stock’s trading volume without counting

insider-owned stock. See Cammer v. Bloom,

, 1286-

87 (D.N.J. 1989) (using and discussing the first five factors);

Krogman,

(using the last three factors).

These tools for gauging market efficiency have been used by many

courts throughout the country and within this circuit. See, e.g.,

5 Form S-3 is reserved for companies whose stock is actively traded and widely followed. To file a Form S-3, a company must have filed SEC reports for twelve consecutive months and possess a seventy-five million dollar market capitalization level. See

. By contrast, there is no minimum capitalization requirement to file either Form S-1 or S-2. Further, a company need not even meet the reporting requirements spelled out in § 239.13 to file a Form S-1. See

.

10 Gariety,

; Binder,

; Hayes v.

Gross,

(3d Cir. 1992); Freeman v. Laventhol &

Horwath,

(6th Cir. 1990); Lehocky,

.

Although this does not represent an exhaustive list, and

in some cases one of the above factors may be unnecessary, once a

court endeavors to apply these factors, they must be weighed

analytically, not merely counted, as each of them represents a

distinct facet of market efficiency. Some courts have concluded

that there is not an efficient market as a matter of law for stocks

trading in the over-the-counter market. See In re Data Access Sys.

Sec. Litig.,

(D.N.J. 1984), rev’d on other

grounds by

(3d Cir. 1988); Epstein v. Am. Reserve

Corp., No. 79 C 4767,

(N.D. Ill. Apr. 21, 1988). We

need not go so far here, but such holdings are indicative of the

wide gulf between the type of market for stocks that trade millions

of shares daily, e.g., Basic,

,

, and the much less active market for stocks like Amedisys.6

6 There is no requirement for expert testimony on the issue of market efficiency, but many courts have considered it when addressing this determi- nation, which may often benefit from statistical, economic, and mathematical analysis. See, e.g., Bell v. Ascendant Solutions, Inc., No. Civ. A. 301-CV-0166- N,

(N.D. Tex. July 1, 2004); Lehocky,

; Krogman,

. Although courts are not to insist upon a “battle of the experts” at the certification stage, see Manual for Complex Litigation (4th ed. 2004) § 21.21, one court explained that In many cases, it makes sense to consider the admissibility of the testimony of an expert proffered to establish one of the Rule 23 elements in the context of a motion to strike prior to considering class certification. In order to consider Plaintiffs’ motion for class certification with the appropriate amount of scrutiny, the Court must first determine whether Plaintiffs’ expert testimony

11 Unfortunately, the district court in this case devoted

insufficient attention to evaluating the market efficiency factors.

The court’s determination that, during the time in question,

Amedisys stock traded in an efficient market, was predicated on its

finding of three factors: high stock trading volume, market makers

trading the stock, and a cause-and-effect relationship between

corporate events and price movement.

A high weekly stock trading volume suggests the presence

of active, informed investors. In evaluating the stock trading

volume, however, the district court never ascertained — and the

plaintiffs never proved — the actual number of Amedisys shares

being regularly traded. Accepting the plaintiffs’ naked claim as

to this analytical starting point cannot yield a reliable result.

The court first “found” that the average weekly trading volume was

3.9% of the outstanding shares, but then conceded that the figure

could be cut in half. Because the court appears to have based its

determination only on two printouts from the Internet, the court

did not determine the mathematically correct average weekly trading

volume. As commentators observe, however, trade volume can be

grossly exaggerated on some exchanges through double-counting,

sometimes by over fifty percent. M. Barclay & F. Torchio,

A Comparison of Trading Models Used for Calculating Aggregate

Damages in Securities Litigation, 64 LAW & CONTEMP. PROBS. 105, 106

supporting class certification is reliable. Bell,

, at *3-*4 (citations omitted).

12 (Summer 2001). At the certification stage, reliance on

unverifiable evidence is hardly better than relying on bare

allegations.

The district court also found that the presence of

twenty-two “market makers” for Amedisys stock weighed in favor of

a finding of market efficiency. To support this conclusion, the

court relied on a single Internet printout, coupled with affidavits

by plaintiffs’ witnesses that were admitted without opportunity for

cross-examination. Moreover, the court failed to acknowledge

growing concern that the mere number of market makers, without

further analysis, has little to do with market efficiency. See,

e.g., Krogman,

(noting that the “number of market

makers” factor has in practice proven an unreliable measure of

market efficiency unless tied to trade volume and price); Griffin,

; Serfaty v. Int’l Automated Sys., Inc.,

(D. Utah 1998); O’Neil v. Appel,

(W.D. Mich. 1996) (“The economic literature has criticized

reliance upon the number of market makers as an indicator of

efficiency.”); see also Brad M. Barber, et al., The Fraud-on-the-

Market Theory and Indicators of Common Stock’s Efficiency, 19 J.

CORP. L. 285, 307 (1994). The district court erred when it did not

consider the questionable relevance of this finding.

The district court also found a causal connection between

Amedisys corporate events and the movement of the stock price, but

did not take into account the many other factors that could affect

13 the price of Amedisys stock. The court correctly identified the

causal connection as one of the most important market-efficiency

factors. It goes to the heart of the “fraud on the market” theory:

In an efficient market, where information is nearly perfect,

material misstatements alter a stock’s price almost immediately.

In such circumstances, “it is easy to see how injury can befall a

person who is unaware of the deceit.” See Eckstein v. Balcor Film

Investors,

(7th Cir. 1993). Demonstrating

that market reactions are caused by company press releases should

not, however, be an exercise in post hoc, propter hoc logic. Many

variables have the potential to and do affect a stock price — the

daily market average; national, local and industry-specific

economic news; competitors’ activities; and on and on. The overall

volatility of the stock price and the speed of its reaction to

company news may also be significant. See, e.g., Krogman,

. To this end, expert testimony may be helpful

because of the utility of statistical event analysis for this

inquiry. See supra, n.6.

Instead of recognizing the complexity of this cause-and-

effect factor, the court relied on a showing that on March 1 and

May 1, 2001, the stock price rose following positive announcements

issued by Amedisys on those days, and the price dropped the day the

company announced that its earnings would be restated. This

evidence is no doubt worthwhile, but standing alone, it is

insufficiently probative to determine, based on “empirical facts,”

14 see Cammer,

, that a causal connection exists.

In short, the court incorrectly used all three factors it found in

favor of market efficiency as a checklist rather than an analytical

tool.

Similarly, the court failed to evaluate the significance

of the market-efficiency factors lacking in the instant case. For

instance, the number of securities analysts following the stock is

an important factor. See, e.g., Krogman,

;

Cammer,

. Hence, the fact that no analyst

was reporting on Amedisys stock at the time in question should have

been weighed against the rather scant utility of, for example, the

number of “market makers.” Further, the court did not address the

effect on the market efficiency determination of Amedisys’s

ineligibility to file an SEC Form S-3 at the time in question (the

other factor absent in this case).7 Because Rule 23 mandates a

complete analysis of “fraud on the market” indicators, district

courts must address and weigh factors both for and against market

efficiency.

CONCLUSION

Although we owe considerable deference to district courts

in reviewing certification decisions, we cannot affirm the order as

it is presently supported. After a more thorough inquiry, however,

certification may ultimately prove correct. When a court considers

7 The court also failed to refer to Amedisys’s market capitalization, the bid-ask spread in its stock, and the float.

15 class certification based on the fraud on the market theory, it

must engage in thorough analysis, weigh the relevant factors,

require both parties to justify their allegations, and base its

ruling on admissible evidence. Questions of market efficiency

cannot be treated differently from other preliminary certification

issues. Courts cannot make an informed decision based on bare

allegations, one-sided affidavits, and unexplained Internet

printouts.

For the foregoing reasons, the class certification order

is VACATED and REMANDED for further proceedings consistent

herewith.

VACATED AND REMANDED.

16 JAMES L. DENNIS, Circuit Judge, specially concurring:

Although I concur in the outcome, I disagree with the

majority opinion’s statement that “[c]ourts have likened the degree

of proof required [in determining market efficiency] to the

standards used in preliminary injunction hearings ... and 12(b)(2)

jurisdictional contests.”8 Contrary to the majority’s reading, the

Fourth Circuit’s opinion in Gariety v. Grant Thorton, LLP,

(4th Cir. 2004), and the Seventh Circuit’s opinion in

Szabo v. Bridgeport Machines, Inc.,

(7th Cir.

2001), do not liken or compare standards or degrees of proof from

other proceedings at all. Instead, the Fourth and Seventh Circuits

simply referred to those inquiries as models or analogs of how

district courts can “probe behind the pleadings in resolving class

action certifications”9 without disobeying the Supreme Court’s

admonishment in Eisen against “expanding the...certification

analysis to include consideration of whether the proposed class is

likely to prevail ultimately on the merits.”10 For example, Gariety

simply says:

A model for [the certification] process can be observed in the context of the preliminary injunction practice.

8 Op. Pg. 10. 9 Gariety,

(quoting General Telephone Co. of Southwest v. Falcon,

(1982)). 10

(citing Castano v. American Tobacco Co.,

(5th Cir. 1996)).

17 Courts make factual findings in determining whether a preliminary injunction should issue, but those findings do not bind the jury..., and the jury’s findings on the merits govern the judgment to be entered in the case.11

And Szabo in the same vein observes that “[c]ourts make similar

inquiries routinely ... before deciding whether [courts] possess

jurisdiction over the subject matter of the case and the persons of

the defendants, the location of the proper venue, application of

forum non conveniens, and other preliminary issues.”12

The only “standards” that have ever been required in

class certifications are more open textured: e.g., “close look,”

Achem Products, Inc. v. Windsor,

(1997);

“rigorous analysis,” Falcon,

; Spence v. Glock,

Ges.m.b.H,

(5th Cir. 2000); Castano v. American

Tobacco Co.,

(5th Cir. 1996). On the other hand,

the Supreme Court in Eisen v. Carlisle and Jaquelin,

(1974), admonished that a “more than likely to prevail”

standard is inappropriate in a Rule 23 certification analysis. In

fact, we recently held that a court must conduct an “intense

factual investigation” while at the same time “tak[ing] care to

inquire into the substance and structure of the underlying claims

without passing on their merits.” Robinson v. Texas Auto. Dealers

Ass’n,

(5th Cir. 2004).

11

at 366 (citing Univ. of Texas v. Camenisch,

(1981)). 12 Szabo,

.

18 Thus, although I agree with the majority’s holding that

the district court did not adequately weigh the factors for and

against a finding of market efficiency, I strongly disagree with

the majority’s reading of Gariety and Szabo. Contrary to the

majority’s impression, these cases do not support or suggest the

adoption or application of degrees or standards of proof in

efficient market determinations for the purposes of class

certification.

19