PER CURIAM:

This interlocutory appeal presents an issue of first impression in the circuit courts: whether, and to what extent, the proportionate liability scheme of section 21(D)(f) of the Securities Exchange Act of 1934 (the "Act"), enacted as part of the Private Securities Litigation Reform Act of 1995 (the "PSLRA"), amends section 20(a) of the Act, under which a person who controls a violator of the Act is "liable jointly and severally with and to the same extent" as that violator.

In 1998, Appellants, a group of investors in publicly traded securities, filed a securities class action against three groups of defendants: (1) Vesta Insurance Group, Inc. ("Vesta") and certain of its officers and directors; (2) KPMG Peat Marwick, LLP, Vesta's outside auditor; and (3) Appellee, Torchmark Corporation, the former parent company of Vesta. After completing discovery, obtaining class certification, and surviving various motions to dismiss by defendants, Appellants reached court approved settlements with Vesta and KPMG. Torchmark, whose motion for summary judgment was denied by the district court, is the only remaining defendant in the action.

In 2003, Appellants filed a motion to strike two of Torchmark's affirmative defenses to the extent those defenses improperly sought to graft the PSLRA's scheme of "proportionate liability" onto the joint and several liability existing between a controlling person and a controlled person under section 20(a). In 2004, the district court entered an order denying the motion to strike, concluding, as a matter of first impression, that the proportionate liability regime set out in section 21(D)(f) of the Act "trumps" section 20(a). In 2006, the district court granted Appellants' motion to file an interlocutory appeal and certified the present issue as one "involv[ing] a controlling question of law as to which there is substantial ground for difference of opinion." 28 U.S.C. § 1292(b) (West 2007).

Under section 21(D)(f), a controlling person is liable jointly and severally for the entirety of plaintiffs' damages only if it commits a knowing violation of the Act. Section 20(a) exposes a controlling person who cannot prove the affirmative defense provided in that section to derivative liability for the acts of its controlled person. For the reasons explained below, we do not interpret section 21(D)(f) as "trumping" a controlling person's derivative liability under section 20(a). Recognizing that implicit repeals of statutory provisions are disfavored, we hold that section 21(D)(f) and section 20(a) should be read in harmony to preserve both the PSLRA's proportionate liability scheme and a controlling person's derivative liability under section 20(a).

BACKGROUND A. The PSLRA and Proportionate Liability

In 1995, "motivated in large part by a perceived need to deter strike suits by opportunistic private plaintiffs that filed securities fraud claims of dubious merit in order to exact large settlement recoveries," Congress passed the PSLRA. Novak v. Kasaks, 216 F.3d 300, 306 (2d Cir. 2000). The PSLRA was a reaction to the "significant evidence of abuse in private securities lawsuits," including "the routine filing of lawsuits against issuers of securities and others whenever there is a significant change in an issuer's stock price, without regard to any underlying culpability of the issuer." H.R. Conf. Rep. No. 104-369, at 31 (1995), reprinted in 1995 U.S.C.C.A.N. 730, 730. Congress also hoped to put an end to "the abuse of the discovery process to impose costs so burdensome that it is often economical for the victimized party to settle." Id.

Before the PSLRA, the general rule in most securities law actions was that defendants found to have violated the Act were jointly and severally liable for all the plaintiffs damages. See Musick, Peeler Garrett, 508 U.S. at 292, 113 S.Ct. 2085 (noting that violators "share joint liability for that wrong under a remedial scheme established by the federal courts."); G.A. Thompson Co., Inc., 636 F.2d at 963; TBG, Inc. v. Bendis, 36 F.3d 916, 927 (10th Cir. 1994). In addition to strike suits, the legislative history of the PSLRA suggests Congress was concerned about the many cases in which the application of traditional joint and several liability unfairly resulted in defendants having to pay for damages caused by other defendants. See H.R. Rep. No. 104-369, at 37 (1995) (Conf.Rep.), reprinted in 1995 U.S.C.C.A.N. 730, 736 ("Under current law, a single defendant who has been found to be 1% liable may be forced to pay 100% of the damages in the case."); S.Rep. No. 104-98, at 20 (1995), reprinted in 1995 U.S.C.C.A.N. 679, 699 ("Under joint and several liability, each defendant is liable for all of the damages awarded to the plaintiff. Thus, a defendant found responsible for only 1% of the harm could be required to pay 100% of the damages.").

To combat these perceived injustices, Congress enacted section 21(D)(f) of the PSLRA, which replaced the existing joint and several liability regime with a proportionate liability scheme that restricts joint and several liability to persons who knowingly violate the Act. Section 21(D)(f) provides in relevant part:

(f) Proportionate liability

(1) Applicability

Nothing in this subsection shall be construed to create, affect, or in any manner modify, the standard of liability associated with any action arising under the securities laws.

(2) Liability for damages

(A) Joint and several liability

Any covered person against whom a final judgment is entered in a private action shall be liable for damages jointly and severally only if the trier of fact specifically determines that such covered person knowingly committed a violation of the securities laws.

(B) Proportionate liability

(i) In general

Except as provided in subparagraph (A), a covered person against whom a final judgment is entered in a private action shall be liable solely for the portion of the judgment that corresponds to the percentage of responsibility of that covered person, as determined [by the fact finder].

15 U.S.C. § 78u-4(f)(2) (West 2007).

Importantly, Congress clarified that section 21(D)(f) affects only the allocation of damages between liable defendants and must not "be construed to create, affect, or in any manner modify, the standard of liability associated with any action" arising under the Act. 15 U.S.C. § 78u-4(f)(1).

Under Section 21(D)(f)'s proportionate liability scheme, there is a three step process to be followed by the fact finder determining a liable defendant's share of responsibility under the Act. The first step is for the fact finder to determine whether the "covered person [or] other persons claimed by any of the parties to have caused or contributed to the loss incurred by the plaintiff' violated the securities laws. Second, the fact finder determines the percentage of responsibility of each person "measured as a percentage of the total fault of all persons who caused or contributed to the loss incurred by the plaintiff." Lastly, the fact finder determines whether such person knowingly committed a violation of the securities laws.

A person "knowingly commits a violation of the securities laws" and thus is responsible for damages jointly and severally if it (1) "makes an untrue statement of a material fact, with actual knowledge that the representation is false" or (2) "omits to state a fact necessary in order to make the statement made not misleading, with actual knowledge that . . . one of the material representations of the covered person is false" or (3) "engages in . . . conduct with actual knowledge of the facts and circumstances that make the conduct of that covered person a violation of the securities laws." 15 U.S.C. § 78u-4(10) (West 2007).

B. Section 20(a) of the Act and Derivative Liability

The Act imposes liability not only on the person who actually commits a securities law violation, but also on an entity or individual that controls the violator. Section 20(a) provides:

Every person who, directly or indirectly, controls any person liable under any provision of this chapter or of any rule or regulation thereunder shall also be liable jointly and severally with and to the same extent as such controlled person to any person to whom such controlled person is liable, unless the controlling person acted in good faith and did not directly or indirectly induce the act or acts constituting the violation or cause of action.

15 U.S.C. § 78t(a) (emphasis added).

The text of section 20(a) unambiguously imposes derivative liability on persons that control primary violators of the Act. Under section 20(a), a controlling person is liable to the plaintiff jointly and severally with and to the same extent as a controlled person for the controlled person's acts, unless the controlling person can establish the affirmative defense of good faith and non-inducement.

The legislative purpose in enacting a control person liability provision was to prevent people and entities from using straw parties, subsidiaries, or other agents acting on their behalf to accomplish ends that would be forbidden directly by the securities laws. In congressional hearings preceding the passage of the Act, Congress referred to correcting the "dangerous and unreliable system of depending upon dummy directors" that lacked any accountability or responsibility. The House of Representatives Report accompanying the Act summarized section 20(a) and clarified that Congress intended to achieve its purpose by making "a person who controls a person subject to the act . . . liable to the same extent as the person controlled unless the controlling person acted in good faith."

Under traditional legal principles in effect prior to the Act, nonfeasance was a basis for fraud liability only when the defendant owed a "special duty" to the plaintiff. Congress enacted section 20(a) because it understood that many control relationships are not within the special duty exceptions recognized at common law. Further, many controlling persons could have avoided traditional secondary liability under the common law to the extent such secondary liability relied on agency principles.

Prior to section 20(a), the conduct of a corporation's agents in violation of the securities laws was not attributable to those who were part of the corporate hierarchy, such as shareholders, directors, and officers, because under the common law of agency the principal in the agency relationship was the corporation. Congress intended section 20(a) to expand upon the common law of vicarious liability and crafted a statute that encompassed relationships broader than those associated with master and servant, or principal and agent.

Pure vicarious liability, such as respondeat superior liability, attributes liability to one party based on the actions of the other party regardless of any allegation of culpability on the party held vicariously liable. Section 20(a) is a derivative liability statute because it in effect requires, through the affirmative defense, a defendant to disprove certain wrongful conduct on its part. See Musick, Peeler Garrett v. Employers Ins. of Wausau, 508 U.S. 286, 296, 113 S.Ct. 2085, 2091, 124 L.Ed.2d 194 (1993) ("§ 20 of the 1934 Act . . . impose[s] derivative liability."). Specifically, a defendant must prove that it did not act in bad faith or with a recklessness that equates to inducing the acts constituting a securities law violation. See G.A. Thompson Co., Inc. v. Partridge, 636 F.2d 945, 963 (5th Cir. 1981). Notwithstanding Section 20(a) liability involves culpable conduct both by the person who is derivatively liable and the person whose wrongful conduct was the direct cause of the securities law violation and injury to the plaintiff, a controlling person's liability under section 20(a) is derived from the acts of its controlled person.

A plaintiff must show that the controlled person (not the controlling person) violated the federal securities laws. A plaintiff must also show that the controlling person exercised control over the controlled person. Congress recognized that it would be difficult, if not impossible, to enumerate or anticipate the many ways in which actual control may be exercised and expressly declined to define the term "control," leaving courts free to decide issues of control status on a case by case basis. The Securities and Exchange Commission promulgated a more specific definition of control under the Act, defining "control" as "the possession, direct or indirect, of the power to direct or cause the direction of the management and policies of a person, whether through ownership of voting securities, by contract, or otherwise." 17 C.F.R. § 230.405 (West 2007). Circuit courts have recognized, however, that the control regulation, like the statute, does not attempt to formulate a precise definition of "control" applicable to all cases, but is intended only to provide some guidance, leaving a determination as to whether control exists dependent on the particular factual circumstances of each case. See, e.g., G.A. Thompson Co., Inc., 636 F.2d at 945; Sheinkopf v. Stone, 927 F.2d 1259, 1270 (1st Cir. 1991); Adams v. Kinder-Morgan, Inc., 340 F.3d 1083, 1108 (10th Cir. 2003); SEC v. First Jersey Sec., Inc., 101 F.3d 1450, 1472-73 (2d Cir. 1996); Maher v. Durango Metals, Inc., 144 F.3d 1302, 1305 (10th Cir. 1998).

The derivative nature of section 20(a) liability has caused courts to disagree over the extent and nature of the burden that the plaintiff must bear to prove section 20(a) liability. In Brown v. Enstar Group, Inc., 84 F.3d 393, 396 (11th Cir. 1996), we held that a plaintiff alleging controlling person liability under section 20(a) must allege that (1) the defendant had the power to control the general affairs of the primary violator, and (2) the defendant had the power to control the specific corporate policy that resulted in the primary violation.

Some of our sister circuits have required a plaintiff seeking to hold a controlling person liable under section 20(a) to prove that the controlling person was a "culpable participant" in the primary violation. However, a careful review of the decisions applying the "culpable participant" standard reveals that courts stop short of requiring, the controlling person to knowingly participate in or independently commit a violation of the Act. Despite the varying standards, courts for good reason appear to be in agreement that a controlling person need not commit an intentional violation of the Act to be liable under section 20(a). As the former Fifth Circuit observed, "[w]hat would be the purpose of the controlling person provision if intent were required-the provision would hardly make anyone liable who would not be so otherwise."

Even though section 20(a) should be construed liberally, it is not applicable wherever the fact of control exists, as Congress rejected an "insurer's liability" for controlling persons by establishing a "good faith defense" not available under common law principles of agency. Courts have also interpreted the duty imposed on a defendant by section 20(a) differently in varying contexts. In G.A Thompson Co., Inc., the former Fifth Circuit held that a controlling person attempting to invoke the affirmative defense has the burden of establishing that "he did not act recklessly in inducing, either by his action or his inaction, the act or acts constituting the violation." G.A. Thompson Co., Inc., 636 F.2d at 960. Accordingly, we interpret section 20(a) as requiring a controlling person to prove more than a lack of participation in the primary violation. A controlling person is derivatively liable under section 20(a) for the violations of its controlled person if the controlling person "acted recklessly in failing to do what he could have done to prevent the violation." Id.

From our prior cases it is clear that section 20(a) was founded and continues until today as a derivative liability statute, imputing liability to a controlling person based on its relationship to a primary violator and not based on any independent violation of the securities laws by the controlling person. We note that secondary actors may be held liable under the Act by means other than section 20(a). There is a consensus in the circuit courts that section 20(a) was intended to supplement, not to supplant the common law theory of respondeat superior as a basis for vicarious liability under the Act and thus, section 20(a) does not constitute an exclusive substitute for the vicarious liability of secondary actors that might otherwise exist under common law agency principles. See, e.g., In re Atlantic Fin. Management, Inc., 784 F.2d 29, 32-34 (1st Cir. 1986), cert. denied, 481 U.S. 1072, 107 S.Ct. 2469, 95 L.Ed.2d 877 (1987); Commerford v. Olson, 794 F.2d 1319, 1322-23 (8th Cir. 1986); Marbury Management, Inc. v. Kohn, 629 F.2d 705, 712-16 (2d Cir. 1980); Paul F. Newton Co. v. Texas Commerce Bank, 630 F.2d 1111, 1115-19 (5th Cir. 1980); Holloway v. Howerdd, 536 F.2d 690, 694-95 (6th Cir. 1976); Hollinger v. Titan Capital Corp., 914 F.2d 1564 (9th Cir. 1990); Kerbs v. Fall River Indus., Inc., 502 F.2d 731, 740-41 (10th Cir. 1974); but see Rochez Bros., Inc., 527 F.2d at 884-86 (concluding that the principles of agency, i.e., respondeat superior, are inappropriate to impose secondary liability in a securities violation case). However, because Congress intended for persons "presumptively beyond the reach of respondeat superior . . . [to] be caught in the net of § 20(a)," In re Villa, 261 F.3d 1148, 1153 (11th Cir. 2001), common law agency principles alone are insufficient to capture the full set of secondary actors that Congress expressly intended to be liable under section 20(a) of the Act.

DISCUSSION

With this background in mind, we turn to the question asked of us by the district court: "whether, and to what extent, the proportionate liability provisions of Section 21D(g) of the Securities Act of 1933, 15 U.S.C. § 78u-4(f), enacted as part of the Private Securities Litigation Reform Act of 1995, amended the joint and several liability provisions of Section 20(a), 15 U.S.C. § 78t(a)." Before we answer this question, however, we must first ask whether the proportionate liability provisions of section 21(D)(f) apply to section 20(a) claims against controlling persons at all. We find that they do.

The PSLRA provides that: "Any covered person against whom a final judgment is entered in a private action shall be liable for damages jointly and severally only if the fact finder specifically determines that such covered person knowingly committed a violation of the securities laws." 15 U.S.C. § 78u-4(f)(2)(A). If the fact finder does not specifically determine that the covered person knowingly committed a violation of the securities laws, then the "covered person against whom a final judgment is entered in a private action shall be liable solely for the portion of the judgment that corresponds to the percentage of responsibility of that covered person." 15 U.S.C. § 78u-4(f)(2)(B)(i).

By its plain language, then, the PSLRA applies solely to "covered persons." If a party to a securities fraud action is not a covered person, then the proportionate liability provisions of the PSLRA do not apply.

The PSLRA defines the term "covered person" to mean "a defendant in any private action arising under this chapter." 15 U.S.C. § 78u-4(f)(10)(C)(i). "[T]his chapter" refers to chapter 2B of title 15 of the United States Code, where the PSLRA is located. Section 20(a) of the Securities Exchange Act is also located in "this chapter" — chapter 2B of title 15 of the United States Code. Put another way, section 20(a) arises under "this chapter."

The proportionate liability provisions of section 21(D)(f) plainly mean that they apply to "any" covered person — not just to some of them, and not just to any covered person except controlling persons, as Appellants suggest. The Supreme Court has said, and so have we, that the term "any" in a statute has a "broad," "powerful," and "expansive" meaning; "it does not mean `some' or `all but a few,' but instead means `all.'" When section 21(D)(f) refers to " any covered person," it means exactly that and nothing less.

Given that section 21(D)(f) states that it applies to "any covered person," which plainly includes a controlling person covered under section 20(a), we read section 21(D)(f) as applying to section 20(a) controlling person liability. We next address the question of whether the PSLRA amends the joint and several liability provisions of section 20(a). It does not.

We know that section 21(D)(f) does not amend controlling person liability under section 20(a) because the PSLRA tell us so. The text under section 21(D)(f)'s "Applicability" heading, provides that: "Nothing in this subsection shall be construed to create, affect, or in any manner modify, the standard for liability associated with any action arising under the securities laws." 15 U.S.C. § 78u-4(f)(1). In other words, nothing in the proportionate liability provisions of the PSLRA displaces in any way the "standard of liability" created by the Securities Exchange Act, including section 20(a) controlling person liability.

As noted above, section 20(a) provides the standard for holding controlling persons substantively liable for the securities fraud that they do not themselves commit but instead is committed by those they control. Section 20(a) states that controlling persons will be liable — and not just liable, but jointly and severally liable — in those circumstances unless they establish the affirmative defense of good faith and lack of inducement. The "Applicability" provision in the PSLRA states that nothing in section 21(D)(f) modifies "in any manner" "the standard for liability" associated with "any action arising under the securities laws." The plain meaning of that language of the PSLRA compels the conclusion that if controlling person liability existed before the PSLRA added section 21(D)(f), it still exists afterwards.

If the clear and plain language were not enough, and it is, the legislative history of the PSLRA leads to the same conclusion. The Conference Committee that produced the final version of the PSLRA explained in its report that: "[T]he `fair share' rule of proportionate liability does not create any new cause of action or expand, diminish, or otherwise affect the substantive standard for liability in any action under the 1933 Act or the 1934 Act. . . . [T]he standard of liability in any such action should be determined by the pre-existing, unamended statutory provision that creates the cause of action, without regard to this provision, which applies solely to the allocation of damages." H.R. Conf. Rept. No. 104-369, at 38, as reprinted in 1995 U.S.C.C.A.N. 730, 737 (emphasis added).

All of this means that the PSLRA, including the proportionate liability provisions, does not change the rules for determining who is liable for violating the securities laws. If a controlling person would have been substantively liable under section 20(a) before the PSLRA, it still will be afterwards.

Section 21(D)(f) is not superfluous, however. It does have a role to play. As the Conference Committee Report also explained, while the PSLRA did not modify "in any manner" the standard of liability under the securities laws, including section 20(a), it did change the rules for allocating damages among the parties once liability has been established by the fact finder. Before the PSLRA was enacted, if one of those parties was found liable as a controlling person of violating section 20(a), it would have been responsible jointly and severally for the damages to the same extent as the primary violator. Under the proportionate liability provisions of the PSLRA, if a party is found liable as a controlling person under section 20(a), there is a new standard for allocating damages. What has changed is not the standard of liability that applies to controlling persons — the "Applicability" provision states that has not been modified "in any manner" — but their responsibility as liable persons for the damages. The proportionate liability provisions of section 21(D)(f) are applicable only after liability is determined, and liability is governed by the standard set out in section 20(a).

Under the new statutory regime, a controlling person is liable to the same extent as the controlled person, unless the controlling person affirmatively establishes that it acted in good faith and did not induce the violation by the controlled person. Unless the controlling person carries the burden of establishing that affirmative defense, it is liable for violating the securities law, specifically section 20(a). Then, for the purpose of determining whether the controlling person is responsible for the entire amount of damages or only its proportionate share, we look to the proportionate liability provisions of the PSLRA, as we do for all "covered persons." Under these provisions joint and several liability for all the damages exists only if the fact finder specifically determines that the controlling person knowingly committed the violation. 15 U.S.C. § 78u-4(f)(2)(A). Where a controlling person fails to affirmatively establish good faith and lack of inducement under section 20(a) but the fact finder does not specifically find a knowing violation, there is liability for the violation but responsibility for the damages is only proportionate, not joint and several. 15 U.S.C. § 78u-4(f)(2)(B)(i).

Section 20(a) provides the standard of liability for controlling persons; if they are liable under that standard, the proportionate liability sections of the PSLRA provide their responsibility for damages; and the "Applicability" section acts as a levy ensuring that each of the two provisions stays within its proper channel, preventing either one from washing out the other. The distinction between substantive liability and allocation of damages is reflected in the statutory language, which states that: "Any covered person against whom a final judgment is entered in a private action shall be liable for damages jointly and severally only if. . . ." 15 U.S.C. § 78u-4(f)(2)(A) (emphasis added). That damages language is different from the liability language — "liable jointly and severally" — which is used in section 20(a).

The legislative history also bears this out. The report of the primary committee that drafted the PSLRA indicates that this part of the Act was intended to address the problem of damages being imposed jointly and severally on those who had not themselves done anything wrong. The committee heard from the president of the United Shareholders Association, who testified that "in cases where there was no proof of actual fraud `[e]liminating joint and several liability . . . will significantly reduce the number of strike suits brought against defendants who have done nothing wrong but are seen as having deep pockets.'" S. Rep. 104-98, at 22 (1995), as reprinted in 1995 U.S.C.C.A.N. 679, 701 (alteration and omission in original). Based on testimony like that, and the need "to eliminate unfairness and to reconcile the conflicting interests of investors in a manner designed to best protect the interests of all investors," the committee "modifie[d] joint and several liability" to "impose[] full joint and several liability, as under current law, on all defendants who engage in knowing securities fraud." Id. (emphasis added). However, for "[d]efendants who are found liable but who did not engage in knowing securities fraud," the committee intended that they would be "liable only for their share of the judgment (based upon the fact finder's apportionment of responsibility)." Id. The committee report is entirely consistent with the plain language of the statute.

Reading section 21(D)(f) to exclude section 20(a) controlling persons from the proportionate liability provisions, as Appellants ask us to do, would lead to an untenable result. It would make it possible in some circumstances for the controlling person to be held responsible for more of the damages caused by the primary violator than the primary violator itself. Consider the situation where both a controlling person and a controlled primary violator acted recklessly instead of knowingly. The former Fifth Circuit in G.A. Thompson Co. held that the controlling person's recklessness makes it ineligible for the affirmative defense of good faith under section 20(a). See G.A. Thompson Co., 636 F.2d at 959-60. Reading section 21(D)(f) as Appellants do to exclude section 20(a) controlling persons from the proportionate liability provisions, the reckless controlling person would be responsible jointly and severally for damages even though it did not act knowingly. The primary violator, however, would escape joint and several liability for damages under section 21(D)(f) because it did not act knowingly.

We ought to avoid any interpretation of the statute that would treat controlling persons more harshly than the primary violator — that would put derivatively liable controlling persons on the hook for all damages, but let primary violators off the hook for any damages that their own action did not cause. That result would be contrary to common sense, to what the committee that drafted the PSLRA said it intended to do, and to what Congress actually did in the plain language of the PSLRA.

CONCLUSION

We conclude that section 20(a) controlling person liability survives section 21(D)(f)'s proportionate liability scheme. Those who would have been substantively liable as controlling person under section 20(a) before the PSLRA was enacted will be substantively liable after its enactment. All that the PSLRA has changed for controlling persons is the standard for deciding whether their responsibility for damages is joint and several or proportionate. Damages are now allocated based on the proportionate liability provisions in the PSLRA, including the provision that knowing violators of the securities laws are "liable for damages jointly and severally." The district court's order denying Appellants' motion to strike Torchmark's PSLRA-based affirmative defenses is AFFIRMED.