We have consolidated for decision two appeals that raise the same issue. That issue is what test to use to determine whether an employer that has fewer than 15 or 20 employees, and thus falls below the threshold for coverage by the major federal antidis-crimination laws, 42 U.S.C. § 2000e(b) (Title VII of Civil Rights Act of 1964) (15 employees); 42 U.S.C. § 12111(5)(A) (Americans with Disabilities Act) (15 after 1994); 29 U.S.C. § 630(b) (Age Discrimination in Employment Act) (20), should be deemed covered because it is part of an affiliated group of corporations that has in the aggregate the minimum number of employees. The parties treat the issue, as have previous cases to address it, as one of federal common law.

In the first case that we review today, James Papa brought suit under the age-discrimination and disabilities laws against Katy Industries, Inc. and its wholly owned subsidiary Walsh Press Company, Inc. Papa was employed by Walsh, which had too few employees to be covered by the laws under which Papa is suing. Katy, however, the parent, has numerous subsidiaries, employing in the aggregate more than a thousand employees. Katy ordered Walsh’s president to discontinue one of its production lines. To comply with this directive, the president had to lay off some of Walsh’s employees. One of the ones he laid off was Papa. The question is whether Papa’s “real” employer was Katy, or (it makes no difference in this case) the entire consolidated group, rather than Walsh.

Walsh complies with all the formalities required of a subsidiary that wants to have a corporate identity different from'that of its parent. Yet there is — unsurprisingly, considering how small Walsh is — a degree of integration between it and Katy. It is illustrated by the latter’s command to Walsh to abandon a part of Walsh’s business, requiring layoffs. And there is more. The salaries of Walsh’s employees are fixed by Katy, and the employees participate in Katy’s pension plan; Katy funds Walsh; their computer operations are integrated; Walsh had the use of certain subaccounts in Katy’s checking account rather than having its own bank account; and Walsh needs Katy’s approval to write checks of more than $5,000. The district court held, nevertheless, that Katy was not Papa’s employer and hence that the court lacked jurisdiction over Papa’s suit.

The facts in the second case are quite similar. The plaintiff — the EEOC suing on behalf of Richard Mueser, a former regional manager of defendant GJHSRT — charged that the company had violated the antiretali-ation provision of Title VII by firing Mueser. GJHSRT didn’t have 15 employees but is a part of an affiliated group known as the “Frederick Group of Companies,” which like Katy has many more employees than is necessary to trigger the coverage of the antidis-crimination laws. Each of the Frederick companies is engaged in a different phase of the trucking business and at a different location, but as with Katy there is a degree of integration. Payroll and benefits are centralized, as are computer operations; the membership of the boards of directors of the two companies overlaps; and employees are moved back and forth among affiliates. The district court, nevertheless, granted summary judgment for the defendants, holding that only GJHSRT was Mueser’s employer.

The briefs in these two cases examine four factors, treating none as entitled to more weight than any of the others, to determine whether the nominal employer is part of an “integrated enterprise” and in consequence not allowed to invoke the few-employees exemption. These factors are interrelation of operations, common management, common ownership, and centralized control of labor relations and personnel. We cannot blame the lawyers for structuring their analysis this way, because we and other courts of appeals have often done likewise, Sharpe v. Jefferson Distributing Co., 148 F.3d 676, 678 (7th Cir.1998); Rogers v. Sugar Tree Products, Inc., 7 F.3d 577, 582 (7th Cir.1994); Lockard v. Piz [*940] za Hut, Inc., 162 F.3d 1062, 1069-70 (10th Cir.1998); Artis v. Francis Howell North Band Booster Ass’n, Inc., 161 F.3d 1178, 1184 (8th Cir.1998); Swallows v. Barnes & Noble Book Stores, Inc., 128 F.3d 990, 993-94 (6th Cir.1997); Cook v. Arrowsmith Shelburne, Inc., 69 F.3d 1235, 1240-41 (2d Cir.1995); Herman v. United Brotherhood of Carpenters & Joiners, 60 F.3d 1375, 1383 (9th Cir.1995) — though not always. In EEOC v. Illinois, 69 F.3d 167, 171-72 (7th Cir.1995); Lusk v. Foxmeyer Health Corp., 129 F.3d 773, 777 (5th Cir.1997); Schweitzer v. Advanced Telemarketing Corp., 104 F.3d 761, 765 (5th Cir.1997), and Johnson v. Flowers Industries, Inc., 814 F.2d 978, 981 n. * (4th Cir.1987), although the four-factor test is cited or recited, the focus of the opinions is on whether the parent corporation made the personnel decision' — committed the discriminatory act — of which the plaintiff was complaining. There is no necessary incompatibility between the two lines of case. The opinion in Lusk explains that the four factors “are examined only as they bear on this precise issue,” 129 F.3d at 777, that is, the issue of whether the parent was the real decision maker. In a number of other cases, such as our Sharpe decision, the four-factor test is applied because the parties agreed that it is the right test, but is not endorsed by the court.

There is enough uncertainty about the standard to warrant a fresh look. This is especially appropriate because of the vagueness of three of the four factors (all but “common ownership” and it, as we shall see, is useless); because, being unweighted, the four factors do not yield a decision when, as in the two cases before us, they point in opposite directions; and because the test was not custom-designed for answering exemption questions under the antidiscrimination laws, but instead was copied verbatim from the test used by the National Labor Relations Board to resolve issues of affiliate liability under the laws administered by the Board. Rogers v. Sugar Tree Products, Inc., supra, 7 F.3d at 582; Armbruster v. Quinn, 711 F.2d 1332, 1336-38 (6th Cir.1983); Baker v. Stuart Broadcasting Co., 560 F.2d 389, 392 (8th Cir.1977).

The place to start in rethinking the proper standard is with the purpose, so far as it can be discerned, of exempting tiny employers from the antidiscrimination laws. The purpose is not to encourage or condone discrimination; and Congress must realize that the cumulative effect of discrimination by many small firms could be substantial. The purpose is to spare very small firms from the potentially crushing expense of mastering the intricacies of the antidiscrimi-nation laws, establishing procedures to assure compliance, and defending against suits when efforts at compliance fail. See Tomka v. Seiler Corp., 66 F.3d 1295, 1314 (2d Cir.1995) (reviewing legislative history); Miller v. Maxwell’s International Inc., 991 F.2d 583, 587 (9th Cir.1993). This purpose or policy is unaffected by whether the tiny firm is owned by a rich person or a poor one, or by individuals or another corporation. If a firm is too small to be able economically to cope with the antidiscrimination laws, the owner will not keep it afloat merely because he is rich; rich people aren’t famous for wanting to throw good money after bad. So an approach actually hinted at in the EEOC’s brief of treating any affiliated group of corporations as a single employer of all the employees of all the corporations in the group would lead as rapidly to the destruction of tiny firms as the approach obviously rejected by Congress of applying the antidis-crimination laws to every employer, no matter how few employees he has.

There are three situations in which the policy behind the exemption of the tiny employer is vitiated by the presence of an affiliated corporation; the exemption should be construed to exclude them. The first situation is where, the traditional conditions being present for “piercing the veil” to allow a creditor, voluntary or involuntary, of one corporation to sue a parent or other affiliate, e.g., United States v. Bestfoods, 524 U.S. 51, 118 S.Ct. 1876, 1885, 141 L.Ed.2d 43 (1998); Anderson v. Abbott, 321 U.S. 349, 362-63, 64 S.Ct. 531, 88 L.Ed. 793 (1944); Secon Service System, Inc. v. St. Joseph Bank & Trust Co., 855 F.2d 406, 413-16 (7th Cir.1988); In re Kaiser, 791 F.2d 73, 75-77 (7th Cir.1986); Fletcher v. Atex, Inc., 68 F.3d 1451, 1458-61[*941] (2d Cir.1995), the parent or affiliates of the plaintiffs employer would be liable for the employer’s debts. If because of neglect of corporate formalities, or a holding out of the parent as the real party with whom a creditor nominally of a subsidiary is dealing, a parent (or other affiliate) would be liable for the torts or breaches of contract of its subsidiary, it ought equally to be liable for the statutory torts created by federal antidis-crimination law. In such a case the parent by its actions has forfeited its limited liability. This approach is conventional in discrimination cases where the employee of a subsidiary seeks to affix liability on the parent for reasons unrelated to the subsidiary’s being within the exemption for employers who have only a few employees. Harrington v. Aetna-Bearing Co., 921 F.2d 717, 718 (7th Cir.1991); Marzano v. Computer Science Corp., 91 F.3d 497, 513-14 (3d Cir.1996); Johnson v. Flowers Industries, Inc., supra; Watson v. Gulf & Western Industries, 650 F.2d 990, 993 (9th Cir.1981).

Second, an enterprise might split itself up into a number of corporations, each with fewer than the statutory minimum number of employees, for the express purpose of avoiding liability under the discrimination laws. The division might be accomplished in such a way as to avoid creating the conditions in which the corporate veil is normally pierced. Each subsidiary might be adequately funded and comply with all requisite formalities for separate corporate status, and the group might make clear to all employees and all creditors that they could look only to the particular corporation with which they had dealt for the enforcement of their contractual entitlements. But if the purpose of this splintered incorporation were to elude liability under the antidiscrimination laws, the corporations should be aggregated to determine how many employees each corporation had. The privilege of separate incorporation is not intended to allow enterprises to duck their statutory duties. See Central States, Southeast & Southwest Areas Pension Fund v. Central Transport, Inc., 85 F.3d 1282, 1288 (7th Cir.1996); Lumpkin v. Envirodyne Industries, Inc., 933 F.2d 449, 461 (7th Cir.1991); UA Local 343 v. Nor-Cal Plumbing, Inc., 38 F.3d 1467, 1474, 1477 (9th Cir.1994); cf. Yosha v. Commissioner, 861 F.2d 494 (7th Cir.1988) (substance over form doctrine in tax law).

Third, the parent corporation might have directed the discriminatory act, practice, or policy of which the employee of its subsidiary was complaining. In that event, the parent, provided that the sum of its employees and those of the subsidiary employing the plaintiff exceeded the statutory minimum, would be the violator. This is the- holding of EEOC v. Illinois; Lusk v. Foxmeyer Health Corp., and Schweitzer v. Advanced Telemarketing Corp., all cited earlier; see also Gorrill v. Icelandair/Flugleidir, 761 F.2d 847, 853-54 (2d Cir.1985); and for the general principle— that limited liability does not protect a parent corporation when the parent is sought to be held liable for its own act, rather than merely as the owner of the subsidiary that acted— see United States v. Bestfoods, supra, 118 S.Ct. at 1886; Spartech Corp. v. Opper, 890 F.2d 949, 953 (7th Cir.1989); Esmark, Inc. v. NLRB, 887 F.2d 739, 756-57 (7th Cir.1989); Kingston Dry Dock Co. v. Lake Champlain Transportation Co., 31 F.2d 265, 267 (2d Cir.1929) (L.Hand, J.). Bestfoods, which we have now cited twice, was not a discrimination case; nor Spartech or any of the other cases we have just cited. But we cannot think of a good reason why the legal principles governing affiliate liability should vary from statute to statute, unless the statute, or the particular policy that animates the statute, ordains a particular- test. (The exception is applicable to the National Labor Relations Act, as we shall see.) The basic principle of affiliate liability is that an affiliate forfeits its limited liability only if it acts to forfeit it — as by failing to comply with statutory conditions of corporate status, or misleading creditors of its affiliate, or configuring the corporate group to defeat statutory jurisdiction, or commanding the affiliate to violate the right of one of the affiliate’s employees. The act requirement is emphasized in our decision in Secon Service System, Inc. v. St. Joseph Bank & Trust Co., supra, 855 F.2d at 413-16, and in numerous other cases across the full range of American law. See, e.g., In re Kaiser, supra, 791 F.2d at 75, 77; Steven v. Roscoe Turner Aeronautical Corp., 324 F.2d[*942] 157, 160 (7th Cir.1963); Lowendahl v. Baltimore & Ohio R.R., 247 A.D. 144, 287 N.Y.S. 62, 76 (App.Div.1936); cf. Phillip I. Blumberg, The Law of Corporate Gmups: Substantive Law § 6.02, p. 114 (1987).

The claim that a group of affiliated corporations is “integrated,” the sort of claim that the four-factor test might be thought to support, not only is vague, but is unrelated to the act requirement just explained or to the policy behind the exemption for employers that have very few employees. Firms too tiny to achieve the realizable economies of scale or scope in their industry will go under unless they can integrate some of their operations with those of other companies, whether by contract or by ownership. The choice between the two modes of integration is unrelated to the exemption. Take contractual integration first. A firm too small to have its own pension plan will join in a multiemployer pension plan or will in effect pool with other employers by buying an insurance policy. David L. Gregory, “Mandatory Arbitration and Wealth Distribution: The Policies and Politics of the Multiemployer Pension Plan Amendments Act,” 24 U.C. Davis L.Rev. 195, 197 (1990); Alicia H. Munnell, The Economics of Private Pensions 219 (1982). It will consult an outside law firm, representing many business firms, rather than having a staff of in-house lawyers. It will hire an accounting firm to do its payroll rather than having its own payroll department. It may ask the Small Business Administration for advice on how to maximize its profits by pruning its least profitable operations. None of these forms of contractual integration would subject tiny employers to the antidis-crimination laws, because the integration is not of affiliated firms. Why should it make a difference if the integration takes the form instead of common ownership, so that the tiny employer gets his pension plan, his legal and financial advice, and his payroll function from his parent corporation without contractual formalities, rather than from independent contractors?

That is all that’s involved in the cases before us. There is no suggestion in either one that the business enterprise was splintered into separate corporations in order to defeat the antidiscrimination laws; Walsh, for example, was acquired, not created, by Katy. There is no showing that an ordinaiy creditor of one of the subsidiaries could pierce the corporate veil and sue the parent corporation or any of the other subsidiaries. There is no suggestion that the parent, or any other affiliate of Walsh’s, or the enterprise as a whole formulated or administered the specific personnel policies, or directed, commanded, or undertook the specific personnel actions, of which the plaintiffs are complaining. It is true that Katy in a sense “caused” the firing of Papa by ordering Walsh to curtail its operations; for, had Walsh not curtailed them, Papa would probably not have been laid off, or so soon. But Papa cannot complain of being laid off as such; the antidiscrimination laws do not forbid layoffs. He can complain only if he, was selected for layoff on some forbidden ground. Maybe he was, but there is no suggestion that Katy (or any of Katy’s other subsidiaries) was responsible for his being selected for layoff, let alone for Walsh’s having picked on him for a forbidden reason.

Where a focus on integration makes sense is in the original context of the four-factor test: the determination by the National Labor Relations Board of whether it has jurisdiction over an employer or, even more clearly, what the appropriate bargaining unit is. See, e.g., Radio & Television Broadcast Technicians Local Union 1264 v. Broadcast Service of Mobile, Inc., 380 U.S. 255, 85 S.Ct. 876, 13 L.Ed.2d 789 (1965) (per curiam); South Prairie Construction Co. v. Local No. 627, 425 U.S. 800, 96 S.Ct. 1842, 48 L.Ed.2d 382 (1976) (per curiam). If the work forces of two affiliated corporations are integrated, there is an argument for a single bargaining unit covering both of them, and also an argument that they should be combined for purposes of determining whether the effect on commerce is substantial enough to justify the Board in asserting jurisdiction. But there is no argument for making one affiliate liable for the other’s independent decision to discriminate. Courts that have borrowed the four-factor test for use in the discrimination context have, perhaps, been insufficiently sensitive to the bearing of context on the[*943] proper formulation of rules of affiliate liability.

Both plaintiffs argue, though only in general terms, that the conditions for piercing the veil are present in these cases, implying that an ordinary creditor could go against the parent corporation (or another sub). The argument is unconvincing. The plaintiffs seem to think that unless a corporate group erects a Chinese wall between affiliates, each affiliate is responsible for the other’s debts. That is nonsense. It is true that one corporation will sometimes own another corporation purely as an investment, with no desire to achieve economies of scale or scope by integrating various functions, such as borrowing, legal advice, back-office operations, personnel policies, and higher management. But that is not the usual ease, and is certainly not a condition of limited liability. The corporate veil is pierced, when it is pierced, not because the corporate group is integrated, Blumberg, supra, § 20.05, but (in the most common ease) because it has neglected forms intended to protect creditors from being confused about whom they can look to for the payment of their claims. Secon Service System, Inc. v. St. Joseph Bank & Trust Co., supra, 855 F.2d at 415-16; Steinberg v. Buczynski, 40 F.3d 890 (7th Cir.1994); In re Kaiser, supra, 791 F.2d at 75.

So we think the decisions of the district courts in this pair of cases are right and should be affirmed, but there is one loose end. Although the defendants in Papa’s case had tendered their argument for the few-employees exemption in a motion for summary judgment, the district judge, thinking the issue jurisdictional, converted it to a Rule 12(b)(1) motion (motion to dismiss for want of subject-matter jurisdiction) and proceeded to find facts rather than merely to decide whether there was a genuine issue of material fact, which is the determination called for by a motion for summary judgment. The judge was wrong; the issue is not jurisdictional. Sharpe v. Jefferson Distributing Co., supra, 148 F.3d at 678, and cases cited there. But the error was harmless. The factual issues about which the parties disagreed, and which may or may not have been improperly resolved by the informal method adopted by the judge, are irrelevant to the view that we take of the proper way to approach the determination of whether an employer who has fewer than the statutory minimum number of employees is, nevertheless, covered by the antidiscrimination laws because of his connections to another employer.

AFFIRMED.