Lucille Kersh and others (“Kersh”) appeal the grant of summary judgment in favor of the defendants, the General Council of the Assemblies of God (“General Council”) and the Assemblies of God Northern California and Nevada District Council (“District Council”). The district court held that the defendants were not secondarily liable under section 20(a) of the Securities and Exchange Act of 1934, 15 U.S.C. § 78t (a), for the alleged fraudulent activity of the Christian Life Center (“CLC”), a local Assembly of God church. We affirm.

BACKGROUND

In 1972, CLC raised funds for church construction by selling shares in a “trust fund.” CLC did not register the investment as a security, and in 1978, the California Superintendent of Banking issued a cease and desist order against the fund. The fund filed for bankruptcy in 1979. The plaintiff’s suit against CLC for primary liability under federal securities fraud provisions has been consolidated with the present action, and has not yet been resolved. 1

The district court held that Kersh must show that the defendants “controlled” CLC *548 by establishing: 1) that the defendants had some means of influence or discipline over CLC’s fund raising activity, and 2) that the defendants “participated” in those activities. The court reached only the participation issue, first noting that in cases dealing with broker-dealer secondary liability the breach of a duty to supervise satisfies the participation requirement. The court apparently employed the “flexible duty” standard to determine whether the defendants owed a duty to supervise, and found that the defendants did not owe such a duty. It further noted that the extension of the broker-dealer doctrine to encompass a church “requires a significant leap.” Because Kersh offered no other proof of participation, the court entered summary judgment in favor of the defendants. Kersh timely appeals.

DISCUSSION

This court reviews de novo a grant of summary judgment by the district court. Grigsby v. CMI Corp., 765 F.2d 1369, 1373 (9th Cir.1985).

EVIDENCE OF POWER OR INFLUENCE

The threshold question in a section 20(a) case is whether the defendant is a “controlling person.” It is clear that to establish that the defendant is a “controlling person” the plaintiff must show that the defendant had actual power or influence over the allegedly controlled person. See Christoffel v. E.F. Hutton & Co., 588 F.2d 665, 668 (9th Cir.1978). This circuit has not clearly articulated a standard for determining the existence of power or influence. 2

However, Christoffel did state with approval the conclusion of the Second and Third Circuits that “the term ‘controlling persons’ requires proof ... of a relationship between the controlling and controlled persons which gives the former direct or indirect influence over the policy and decision-making process of the latter.” 3 Id. at 668 (citing Gordon v. Burr, 506 F.2d 1080, 1085 (2d Cir.1974); Lanza v. Drexel & Co., 479 F.2d 1277, 1299 (2d Cir.1973) (en banc); Rochez Brothers, Inc. v. Rhoades, 527 F.2d 880 (3d Cir.1975), cert. denied, 425 U.S. 993 , 96 S.Ct.2205, 48 L.Ed.2d 817 (1976)); see also Pharo v. Smith, 621 F.2d 656, 670 (5th Cir.1980) (citing the SEC definition with approval).

The determination whether a defendant has power or influence over an allegedly controlled person is a complex factual question. See Rochez Brothers, 527 F.2d at *549 890 ; SEC v. Netelkos, 592 F.Supp. 906, 913 (S.D.N.Y.1984). See generally H.R.Rep. No. 1383, 73d Cong., 2d Sess. 26 (1934) (noting that “[i]t would be difficult if not impossible to enumerate or to interpret the many ways in which actual control may be exerted”); Comment, Secondary Liability of Controlling Persons Under the Securities Acts: Toward An Improved Analysis, 126 U.Pa.L.Rev. 1345, 1352-53 (1978) (suggesting that “the factual nature of the control issue has effectively prevented the development of universal standards”). Kersh attempts to show that the bylaws of the defendants and the actual dealings between the defendants and CLC establish at least a genuine issue as to whether the defendants have power or influence over CLC.

Kersh appears to have raised sufficient evidence with respect to the District Council, contending that the power to approve the project constitutes direct influence over the policy and decisionmaking process of CLC with respect to that project. According to the church bylaws, the District Council must approve fund raising projects which have effect beyond the local church and community. The District Council argues that no approval was required because CLC intended to use the funds for local church construction. However, the general treasurer of the General Council testified that under this provision the solicitation of contributions from non-church members requires District Council approval. Because it is not disputed that CLC’s investors came from outside the local church and community, it would appear there is at least a material issue as to whether District Council approval was required for the project.

However, we find the evidence that Kersh presents regarding the power or influence of the General Council insufficient to establish “control” under section 20(a). Kersh argues that the General Council maintains control by licensing ' ministers; however, there is no evidence indicating how such control is exercised once a minister has been licensed. Kersh contends that the General Council exercises control over a minister’s “promotion” to larger congregations; however, each local church has independent power to select its ministers. Moreover, the General Council was not required to approve the CLC fundraiser. Indeed, there are no facts at all suggesting a nexus between CLC and the General Council regarding the transaction. See, e.g., In re Catanella and E.F. Hutton & Co. Securities Litigation, 583 F.Supp. 1388, 1421 (E.D.Pa.1984) (stock market adviser held not to “control” brokerage firm which sometimes used his advice, because brokerage firm exercised independent judgment as to whether to employ that advice); Wright v. Schock, 571 F.Supp. 642, 664 (N.D.Cal. 1983) (defendant banks and title companies held not to “control” a mortgage loan broker simply because they provide indispensable services to the broker), aff’d, 742 F.2d 541 (9th Cir.1984); In re Commonwealth Oil/Tesoro Petroleum Securities Litigation, 484 F.Supp. 253, 268-69 (W.D.Tex. 1979) (in determining liability for unlawful financial reporting, independent accounting firm that monitors a corporation’s financial condition held not to “control” that corporation, because it did not “direct or cause the direction of management and policies” of the corporation).

The district court did not discuss the control issue. Nevertheless, we affirm the grant of summary judgment in favor of the General Council for lack of evidence indicating that the General Council had power over CLC with regard to the unlawful transaction. However, Kersh presents sufficient evidence raising an issue as to the District Council’s power or influence over the transaction.

CULPABLE PARTICIPATION

To establish a prima facie case that the defendant was a “controlling person,” a plaintiff must show not only that the defendant had actual power or influence, but also that he was a “culpable participant” in the alleged illegal activity. Christoffel, 588 F.2d at 668 . Kersh cannot show that the District Council or the General Council affirmatively participated in the alleged *550 fraudulent securities transaction. However, Kersh attempts to prove that the defendants’ failure to supervise CLC constituted indirect participation in CLC’s scheme.

We have held that a brokerage house’s failure to supervise its agent constituted indirect participation in the agent’s illegal acts, subjecting the brokerage house to secondary liability under section 20(a). See Hecht v. Harris, Upham & Co., 283 F.Supp. 417, 438-39 (N.D.Cal.1968), modified on other grounds, 430 F.2d 1202 (9th Cir.1970). Thus, under certain circumstances inaction in the form of a failure to supervise can nevertheless result in secondary liability. This “rule” or “exception” which allows inaction to be characterized as “indirect participation” has been described by the parties as the “broker dealer rule” or “exception” because it has for the most part been applied in the broker-dealer context.

A. Should the Rule Be Applied to Non-Brokers?

The defendants argue that secondary liability for breach of a duty to supervise is applicable only in the broker-dealer context. They correctly point out that this circuit has applied the rule only in Hecht, a broker-dealer case.

However, in at least one instance, a circuit court has found a non-broker secondarily liable for failure to supervise. See G. A. Thompson & Co. v. Partridge, 636 F.2d 945, 959 (5th Cir.1981) (corporate director). 4 More importantly, we have at least contemplated applying the rule in a non-broker context. See Zweig v. Hearst Corp., 521 F.2d 1129,1134-36 (9th Cir.) (newspaper publisher), cert. denied, 423 U.S. 1025 , 96 S.Ct. 469 , 46 L.Ed.2d 399 (1975). Thus, application of the rule cannot arbitrarily be limited to the broker-dealer context.

Nevertheless, in Zweig , we viewed the broker-dealer context as a special case, where facts and policy merged to create a requirement that broker-dealers maintain and enforce a reasonable and proper system of supervision and internal control over controlled persons. In determining whether to impose a similar requirement on controlling persons other than broker-dealers, we considered several factors: a) whether the controlling person derives direct financial gain from the activity of the controlled person, b) the extent to which the controlled person is tempted to act unlawfully because of the controlling person’s policies (e.g., compensation system), c) the extent to which statutory or regulatory law or the defendant’s own policies require supervision, d) the relationship between the plaintiff and the controlling person, and e) the demonstration of some public policy need to impose such a requirement. See Zweig, 521 F.2d at 1135 .

Kersh attempts to characterize the defendants as so similar to brokers that imposition of a requirement to maintain a system of control and supervision is warranted. However, we agree with the district court that the relationship between the national church and CLC cannot be analogized to that between a brokerage house and its employees. CLC did not act as the agent of the national church in this transaction. CLC did not receive compensation from the national church for the sale. More importantly, the sale was solely for the benefit of CLC with no direct benefit going to the national church.

Kersh points to evidence of the defendants’ substantial, nationwide securities activities. However, this evidence is not dis-positive. Like any corporation, the national church may sell its own securities, and would be fully responsible for any fraudulent activity in that sale. The fact that the *551 national church generates its own securities for sale does not imply that it, like a broker, is in the business of recommending the sale of securities in general, or the securities of CLC in particular.

Thus, we affirm the district court’s conclusion that to consider the defendant church in the same position as a broker-dealer requires too significant a leap. The defendants cannot be held to the broker-dealer standard with respect to CLC’s independent fund raising activity.

B. Liability for Inaction by Non-Brokers Generally

Nevertheless, it is an open question in this circuit whether inaction outside the broker-dealer context as a general matter can constitute sufficient participation in the unlawful activity to warrant imposition of liability. This is a slightly more general question than whether the defendants should be considered broker-dealers for the purpose of imposing secondary liability for failing to supervise.

We confronted this issue in Zweig , after rejecting application of the broker-dealer rule. Without clearly articulating a test for determining whether inaction would constitute participation, we concluded that a newspaper did not participate in the unlawful acts of its employee by failing to maintain a reasonable system of supervision. See Zweig, 521 F.2d at 1135 . Lower courts in this circuit have applied the “flexible duty” standard to determine whether a party should be held secondarily liable for failing to act. See, e.g., Bradshaw v. Van Houten, 601 F.Supp. 983, 985-86 (D.Ariz. 1985). In the instant case, the district court stated that the flexible duty standard was appropriate, though it is not clear from the disposition whether that standard or some other more fact-based test was actually applied.

To the extent that the district court applied the flexible duty standard it was in error. The flexible duty standard was developed in order to determine whether a defendant owed a duty to disclose material information for the purpose of assessing primary liability under Rule 10b-5. See White v. Abrams, 495 F.2d 724, 730-36 (9th Cir.1974). It had nothing' to do with the duty to supervise. Arguably, the introduction of such a general standard would allow future plaintiffs to establish the existence of other types of duties (e.g., duty to investigate, duty to minimize possibility of fraud), the breach of which would show indirect participation in the fraudulent activity.

Extended to the logical extreme, the rule would hold that once a plaintiff proved the existence of power or influence over the controlled person, a court would then determine whether the defendant behaved “culpably” in failing to exercise that power. Thus, application of the flexible duty standard would transform what was once an extension of the participation requirement (i.e., that a broker-dealer’s failure to supervise, coupled with additional facts linking the broker-dealer to the fraudulent transaction, constituted indirect participation) into a new rule that inaction alone can result in secondary liability as long as the defendant owed a duty to the plaintiff to act. 5 As we have previously held, we are “unwilling to extend vicarious liability where the secondary defendant’s activity has been mere inaction.” Christoffel, 588 F.2d at 669 (quoting Rochez, 527 F.2d at 889 ).

Moreover, by applying the flexible duty standard, a court would be determining whether the controlling person owed a duty to the plaintiff to exercise power over a third person (the controlled person). This *552 shifts the focus from the relationship between the controlling and controlled persons to that between the controlling person and the plaintiff. Application of a flexible duty analysis would in effect merge secondary liability into primary liability. Indeed, the flexible duty standard was devised as a test for determining whether a plaintiff had successfully made out a prima facie case of primary 10b-5 liability. See White, 495 F.2d at 730-36 .

However, this is not to say that inaction can never constitute indirect participation outside the broker-dealer context. A plaintiff may establish a set of facts indicating that the defendant culpably participated in the fraudulent transaction through inaction without attempting to establish breach of a duty to act. Indeed, the Hecht case, which established the broker-dealer rule in this circuit, did not discuss the broker’s failure to supervise in terms of the breach of a duty owed to the plaintiff to supervise employees. Instead, the court relied on numerous facts: the existence of federal rules and internal policies requiring supervision, the direct financial gain to the broker from sales by its representatives, the nature of the relationship in that the broker gives advice to the public and its representatives simply effect the sale, and the temptation of the representative to cheat since representatives are compensated by commission in direct proportion to sales. Moreover, the court found that the broker knew that the transactions were occurring (but not that they were unlawful), and that the broker approved of and accepted the benefits from those transactions. Under those factual circumstances, the broker could not claim that it did not participate in or induce the unlawful sale. See Hecht, 283 F.Supp. at 439 .

Moreover, in Zweig , we did not apply the flexible duty standard in determining whether a newspaper could be held liable for failure to supervise its writers. After rejecting application of the broker-dealer rule, we examined the facts of the case and determined that the controlling person acted in good faith and did not directly or indirectly induce the unlawful act of its writer. We held that in the absence of notice to the newspaper “that a particular writer has violated or may violate the requirement of fairness and accuracy, the newspaper may safely anticipate compliance.” Zweig, 521 F.2d at 1135 . Moreover, we noted that it was not practicable to expect newspaper editors to recheck each item of news, especially where the writer in question had done nothing to warrant that treatment during his thirty-year relationship with the newspaper. Id.

Kersh does not argue that the facts of the case and the nature of the relationship between GLC and the defendants indicate that the defendants’ failure to supervise constituted indirect participation in CLC’s allegedly fraudulent activity. There is no evidence that CLC had acted unlawfully before, or that the defendants knew that CLC was engaged in any type of unlawful activity. Moreover, while there is some evidence that the District Council knew that CLC was selling securities, there is no evidence that the District Council knew that it even had the power to approve the transactions (i.e., that sales were being made to persons outside the local community), or that it encouraged or directly benefited from the sale of those securities. In short, there are no facts indicating that the defendants indirectly participated in CLC’s unlawful fund raising activity.

Rather, Kersh’s arguments appear to revolve solely around application of the broker-dealer rule. Kersh in essence states that the defendants are so much like broker-dealers that they also should be required to maintain a reasonable and proper system of supervision and internal control. As we have said above, we affirm the district court’s conclusion that the relationship between CLC and the defendants should not be analogized to that between a broker-dealer and its employee/salesmen.

We find that the grant of summary judgment in favor of the General Council was proper because there is no evidence that the General Council had the power to direct CLC’s fund raising activity; and in light of *553 the absence of facts indicating that both defendants indirectly participated in the unlawful fund raising activity, the grant of summary judgment is AFFIRMED. 6

. At the time summary judgment was granted, the district court did not certify the case for review pursuant to Fed.R.Civ.P. 54(b). After briefs were filed for this appeal, we issued an order to the parties requiring them to come prepared to oral argument to discuss whether we lacked jurisdiction to review only one of two consolidated cases absent a Rule 54(b) certification, citing Huene v. United States, 743 F.2d 703 (9th Cir. 1984). Subsequently, the appellant obtained a proper Rule 54(b) certification from the district court, which was filed with this court. The judgment having been certified as final and thus reviewable, we have jurisdiction *548 in this appeal. See Freeman v. Hittle, 747 F.2d 1299, 1302 (9th Cir.1984) (absent prejudice to the parties, a proper Rule 54(b) certification obtained after notice of appeal was filed is sufficient to confer appellate jurisdiction).

. While Christoffel states the rule that a "controlling person” must have "power" or "influence” over the controlled person, it does not define "power” or "influence." While in Christoffel the defendant brokerage house had no legal authority to compel a guardian (the company’s employee) to make particular investments on behalf of a trust, the plaintiff argued that the defendant "controlled” the guardian because it could have refused to hire him, and because it could have influenced him by using its authority as his employer to discourage him from making improvident investments. Without discussing the plaintiffs contention, we affirmed judgment in favor of the brokerage house because of the absence of evidence indicating that the brokerage house "culpably participated” in the illegal transactions, a second prong in the definition of “controlling person.” Cf. Zweig v. Hearst Corp., 521 F.2d 1129,1132 (9th Cir.) (holding that newspaper was a “controlling person” and thus potentially liable under section 20(a) for the acts of one of its employees), cert. denied, 423 U.S. 1025 , 96 S.Ct. 469 , 46 L.Ed.2d 399 (1975).

. This standard appears to be derived from the Securities & Exchange Commission ("SEC”) definition of "control”: "the possession, direct or indirect, of the power to direct or cause the direction of the management and policies of a person, whether through the ownership of voting securities, by contract, or otherwise.” 17 C.F.R. § 230.405 (1985). The SEC test has also been applied to determine whether to impose liability on persons who "control” those using the mails to sell unregistered securities. See Safeway Portland Employees’ Federal Credit Union v. C.H. Wagner & Co., 501 F.2d 1120 , 1124 & n. 17 (9th Cir.1974); see also Hokama v. E.F. Hutton & Co., 566 F.Supp. 636, 646 (C.D. Cal. 1983) (applying the Safeway Portland definition of "control" to the determination of liability under section 20(a)).

. Kersh cites Sharp v. Coopers & Lybrand, 649 F.2d 175 (3d Cir.1981), cert. denied, 455 U.S. 938 , 102 S.Ct. 1427 , 71 L.Ed.2d 648 (1982), as another example of a court imposing a duty to supervise on a non-broker (an accountant). However, Sharp held that an accounting firm was liable for breach of a duty to supervise under a respondeat superior theory, not section 20(a). We have clearly taken the position that section 20(a) supplants secondary liability based upon common law agency principles, such as respondeat superior. See Christoffel, 588 F.2d at 667 .

. The Fifth Circuit has essentially adopted this position. See G.A. Thompson, 636 F.2d at 959 . In G.A. Thompson, the court extended the reasoning of the broker-dealer rule to the director-corporation context. It held that a controlling person could be secondarily liable for recklessly failing to exercise his power or influence. "The test of whether the controlling person has done enough to prevent the violation depends on what he could have done under the circumstances.” Id. The court noted that while it was impossible for the defendant to have supervised the controlled person, it was nonetheless liable for reckless failure to do what it could to prevent the fraudulent activity.

. Because we hold that there is no evidence indicating that the defendants directly or indirectly participated in the CLC’s alleged fraudulent activity, we do not reach the question whether the defendants’ conduct was "culpable.”