In the
United States Court of Appeals
               For the Seventh Circuit
                          ____________

No. 02-1605
EQUAL EMPLOYMENT OPPORTUNITY COMMISSION,
                                                  Applicant-Appellee,
                                  v.

SIDLEY AUSTIN BROWN & WOOD,
                                              Respondent-Appellant.
                          ____________
             Appeal from the United States District Court
        for the Northern District of Illinois, Eastern Division.
          No. 01-C-9635—Joan Humphrey Lefkow, Judge.
                          ____________
    ARGUED SEPTEMBER 6, 2002—DECIDED OCTOBER 24, 2002
                          ____________


  Before POSNER, EASTERBROOK, and DIANE P. WOOD, Circuit
Judges.
  POSNER, Circuit Judge. In 1999, Sidley & Austin (as it
then was) demoted 32 of its equity partners to “counsel”
or “senior counsel.” The significance of these terms is un-
clear, but Sidley does not deny that they signify demotion
and constitute adverse personnel action within the mean-
ing of the antidiscrimination laws. The EEOC began an
investigation to determine whether the demotions might
have violated the Age Discrimination in Employment Act.
After failing to obtain all the information it wanted without
2                                               No. 02-1605

recourse to process, the Commission issued a subpoena
duces tecum to the firm, seeking a variety of documenta-
tion bearing on two distinct areas of inquiry: coverage and
discrimination. The reason for the inquiry about coverage
is that the ADEA protects employees but not employers.
E.g., Simpson v. Ernst & Young, 

, 443 (6th Cir.
1996); see 29 U.S.C. §§ 623(a)(2), (a)(3), 630(f). To be able
to establish that the firm had violated the ADEA, there-
fore, the Commission would have to show that the 32
partners were employees before their demotion.
   Sidley provided most of the information sought in the
subpoena that related to coverage (but no information
relating to discrimination, though Sidley claims that the
demotions were due to shortcomings in performance
rather than to age), but not all. It contended that it had
given the Commission enough information to show that
before their demotion the 32 had been “real” partners and
so there was no basis for the Commission to continue
its investigation. The Commission applied to the district
court for an order enforcing the subpoena. The court or-
dered the firm to comply in full, and the firm appeals. The
order to comply was a final order appealable under 28
U.S.C. § 1291 because it terminated the judicial proceed-
ing. The only relief sought was enforcement of the sub-
poena, and so when enforcement was ordered the EEOC
had gotten everything it wanted. CFTC v. Collins, 

, 1232 (7th Cir. 1993); United States v. Construction
Products Research, Inc., 

, 469 (2d Cir. 1996).
  The Commission also appears to be seeking informa-
tion on whether Sidley may be forcing other partners
whom the Commission suspects may also be employees
within the meaning of the age discrimination law to retire
on account of their age, contrary to the abolition of man-
datory retirement by the age discrimination law. But the
No. 02-1605                                                  3

parties appear to have assumed that if the 32 are (as Sidley
contends) employers, so are all of Sidley’s other partners.
So we need not consider the mandatory-retirement issue
separately.
  The law firm’s argument proceeds in three steps: (1) the
question whether the 32 demoted partners are within the
ADEA’s coverage is a jurisdictional question, which once
answered against the Commission requires that the inves-
tigation cease; (2) the target of a subpoena need comply
only up to the point at which it has produced evidence
that establishes that there is no jurisdiction; (3) the Com-
mission has no jurisdiction in this case because a partner
is an employer within the meaning of the federal anti-
discrimination laws if (a) his income included a share of
the firm’s profits, (b) he made a contribution to the capital
of the firm, (c) he was liable for the firm’s debts, and (d) he
had some administrative or managerial responsibilities—
and all these things, the firm argues, have been proved.
  The facts as developed so far reveal the following:
  The firm is controlled by a self-perpetuating executive
committee. Partners who are not members of the commit-
tee have some powers delegated to them by it with re-
spect to the hiring, firing, promotion, and compensation
of their subordinates, but so far as their own status is con-
cerned they are at the committee’s mercy. It can fire them,
promote them, demote them (as it did to the 32), raise
their pay, lower their pay, and so forth. The only firm-
wide issue on which all partners have voted in the last
quarter century was the merger with Brown & Wood
and that vote took place after the EEOC began its inves-
tigation. Each of the 32 partners at the time of their demo-
tion by the executive committee had a capital account
with the firm, averaging about $400,000. Under the firm’s
rules, each was liable for the firm’s liabilities in propor-
4                                              No. 02-1605

tion to his capital in the firm. Their income, however, was
determined by the number of percentage points of the
firm’s overall profits that the executive committee as-
signed to each of them. Each served on one or more of the
firm’s committees, but all these committees are subject to
control by the executive committee.
  Sidley can obtain no mileage by characterizing the cov-
erage issue as “jurisdictional.” It is the law that the EEOC
cannot protect employers; and it is also the law that like
any agency with subpoena powers the EEOC is entitled
to obtain the facts necessary to determine whether it can
proceed to the enforcement stage. EEOC v. United Air
Lines, Inc., 

, 651 (7th Cir. 2002); Commodity
Trend Service, Inc. v. CFTC, 

, 986-87 (7th Cir.
2000); SEC v. Brigadoon Scotch Distributing Co., 

, 1052-53 (2d Cir. 1973). Among these are facts bearing
on whether the 32 demoted partners were employees with-
in the meaning of the age discrimination law. The Com-
mission is entitled to the information that it thinks it
needs in order to be able to formulate its theory of cov-
erage before the court is asked to choose between the
Commission’s theory and that of the subpoenaed firm.
Only if, as in Reich v. Great Lakes Indian Fish & Wildlife
Comm’n, 

(7th Cir. 1993), the information that the
subpoenaed firm resists furnishing is not even arguably
relevant, because it is evident at the outset that whether
the agency has any business conducting the investiga-
tion depends on a pure issue of statutory interpretation,
can the court resolve the issue then and there without
insisting on further compliance with the subpoena. See
also EEOC v. Shell Oil Co., 

, 64-65 (1984); FTC
v. Miller, 

, 460-61 (7th Cir. 1977); FTC v. Ken
Roberts Co., 

, 586-87 (D.C. Cir. 2001); EEOC
v. Karuk Tribe Housing Authority, 

, 1076-77
(9th Cir. 2001); EEOC v. Ocean City Police Dept., 820 F.2d
No. 02-1605                                                   5

1378, 1380 (4th Cir. 1987) (en banc), vacated on other
grounds, 

(1988). The issue in Great Lakes was
whether the Fair Labor Standards Act applies to game
wardens on Indian reservations. If, as we held, it did not,
the continued investigation of the wardens’ employer was
all burden and no benefit, making insistence on compli-
ance with the Labor Department’s subpoena unreason-
able. See EEOC v. United Air Lines, 

, 287 F.3d
at 653.
   Great Lakes does not hold, as Sidley argues, that charac-
terizing a threshold issue as “jurisdictional” takes a case
out of the general rule (on which see Oklahoma Press Pub-
lishing Co. v. Walling, 

, 212-14 (1946); Endicott
Johnson Corp. v. Perkins, 

, 508-09 (1943); EEOC
v. Peat, Marwick, Mitchell & Co., 

, 930-31 (8th
Cir. 1985); FTC v. Ken Roberts 

, 276 F.3d at 585-
87) that enforcement of a subpoena cannot be resisted on the
ground that the information the agency is seeking would
not justify an enforcement action. The cases are legion that
there is no general exception to the rule for issues going to
the agency’s jurisdiction. See, e.g., FTC v. Feldman, 

, 1095-96 (7th Cir. 1976); FTC v. Ken Roberts 

,
276 F.3d at 585-87; United States v. Sturm, Ruger & Co., 

, 5-6 (1st Cir. 1996). As explained in United States v.
Construction Products Research, 

, 73 F.3d at 470, “at
the subpoena enforcement stage, courts need not determine
whether the subpoenaed party is within the agency’s
jurisdiction or covered by the statute it administers; rather
the coverage determination should wait until an enforce-
ment action is brought against the subpoenaed party.”
Sidley gains nothing, therefore, from characterizing the
coverage issue as jurisdictional, and so we need not decide
whether the characterization is correct.
 But the cases leave intact the principle that a subpoena
may be challenged as unreasonable. And one basis on
6                                                No. 02-1605

which it may be found unreasonable is that, as in Great
Lakes, the agency clearly is ranging far beyond the bound-
aries of its statutory authority. As the Supreme Court
explained in United States v. Morton Salt Co., 

,
652 (1950), “of course a governmental investigation into
corporate matters may be of such a sweeping nature and
so unrelated to the matter properly under inquiry as to
exceed the investigatory power. But it is sufficient if the
inquiry is within the authority of the agency, the demand
is not too indefinite and the information sought is rea-
sonably relevant.” In Endicott Johnson Corp. v. 

, 317 U.S. at 509, the Court noted that “the evidence
sought by the subpoena was not plainly incompetent or
irrelevant to any lawful purpose of the Secretary in the
discharge of her duties under the Act, and it was the duty
of the District Court to order its production for the Secre-
tary’s consideration”—implying therefore that had the
evidence sought by the subpoena been “plainly incompe-
tent or irrelevant to any lawful purpose,” enforcement
would have been denied. See also United States v. Con-
struction Products Research, 

, 73 F.3d at 471.
  Great Lakes was such a case; this one is not (not yet
anyway, though it may become one, as we shall point out
later). But the difference is not, as implied in EEOC v. Karuk
Tribe Housing 

, 260 F.3d at 1077-78, that
Great Lakes involved an issue of jurisdiction and this case
a “mere” issue of coverage and that there should be a
special rule if not for jurisdictional defects then at least
for “patent” jurisdictional defects. “Patent” is important
to the issue of reasonableness, but not “jurisdictional.” The
Commission has the same right to obtain information
bearing on its jurisdiction as to obtain any other informa-
tion that it needs in order to decide whether there has
been a violation of one of the laws that it enforces, while
the recipient of the subpoena has the same right to chal-
No. 02-1605                                                 7

lenge the subpoena as unreasonable because of lack of
coverage as it does to complain that the subpoena is unrea-
sonable because the recipient is outside the agency’s juris-
diction. Suppose Sidley were conceded to have only eight
employees, when an employer must have at least 20 em-
ployees in order to be subject to the ADEA. 29 U.S.C.
§ 630(b). It would not follow that Sidley could not com-
plain about the subpoena unless the issue were char-
acterized as jurisdictional. The distinction for which Karuk
contended would complicate litigation pointlessly by forc-
ing judges to distinguish between jurisdictional and non-
jurisdictional limitations on agencies’ powers and to de-
cide on which side of the line coverage issues belong. So,
to repeat, whether the coverage issue in this case should
be characterized as jurisdictional is irrelevant.
   The case law dealing with challenges to subpoenas
contains the statement that the EEOC’s “investigative
authority is tied to charges filed with the Commission;
unlike other federal agencies that possess plenary author-
ity to demand to see records relevant to matters within
their jurisdiction, the EEOC is entitled to access only to
evidence ‘relevant to the charge under investigation.’ ”
EEOC v. Shell Oil 

, 466 U.S. at 64; see also EEOC
v. United Air Lines, 

, 287 F.3d at 650. But this
was said with reference to Title VII; the ADEA’s grant
of investigative authority to the Commission is not
cabined by any reference to charges. 29 U.S.C. § 626(a); 29
C.F.R. § 1626. Anyway the Supreme Court has held that
the Commission can file its own charges of violation of
Title VII. EEOC v. Shell Oil 

, 466 U.S. at 69-70. So
it is doubly irrelevant that none of the 32 demoted part-
ners has filed a charge.
  A remarkable feature of the way the case has been ar-
gued is that neither party has addressed the question why
8                                                 No. 02-1605

some or all members of partnerships should for purposes
of the federal antidiscrimination laws be deemed employ-
ers and so placed outside the protection of these laws.
That question might be avoidable if the laws contained
an exemption for discrimination against partners; we
might then simply look to the definition of the term in
federal or state law. And if we looked there, we would find
that Sidley was indeed a partnership and the 32 demoted
partners were indeed partners before their demotion.
Sidley has complied with all the formalities required by
Illinois law to establish and maintain a partnership; the
32 were partners within the meaning of the applicable
partnership law.
  Although the EEOC does not concede that the 32 are
bona fide partners even under state law, it is emphatic
that their classification under state law is not dispositive
of their status under federal antidiscrimination law. The
antidiscrimination laws do not exempt partnerships from
coverage (Sidley concedes that) or deny partners, as such,
the protection of the laws. Employers are not protected
by discrimination laws such as Title VII and the ADEA,
but are partners employers? Always? Always for purposes
of Title VII or the ADEA, or the other federal laws
that prohibit employment discrimination? Statutory pur-
pose is relevant. When the Supreme Court in Robinson v.
Shell Oil Co., 

, 346 (1997), was faced with the
question whether “employee” in Title VII includes a for-
mer employee, it looked to “consistency with a primary
purpose of antiretaliation provisions: Maintaining unfet-
tered access to statutory remedial mechanisms. The
EEOC quite persuasively maintains that it would be de-
structive of this purpose of the antiretaliation provision
for an employer to be able to retaliate with impunity against
an entire class of acts under Title VII.” And when in Papa
v. Katy Industries, Inc., 

(7th Cir. 1999), we held
No. 02-1605                                               9

that, in deciding whether a firm has the minimum num-
ber of employees required for it to be covered by federal
antidiscrimination law we would not pierce the corpo-
rate veil, we did so on the basis of “the policy behind the
exemption for employers that have very few employees.”

   An individual who was classified as a partner-employer
under state partnership law might be classified as an
employee for other purposes, including the purpose for
which federal antidiscrimination law extends protection
to employees but not employers. Against this conclusion
it can be argued that partners should be classified as em-
ployers rather than employees for purposes of the age
discrimination law because partnership law gives them
effective remedies against oppression by their fellow
partners, because partnership relations would be poi-
soned if partners could sue each other for unlawful dis-
crimination, and because the relation among partners
is so intimate that they should be allowed to discrim-
inate, just as individuals are allowed to discriminate in
their purely personal relations. This is not the occasion
on which to come down on one side or the other of the
issue, though we note that in Hishon v. King & Spalding,

, 78 (1984), the Supreme Court rejected the
argument that the intimate nature of the partnership
relation precludes a challenge under Title VII to a discrim-
inatory refusal to promote an employee to partner.
  But we do not understand how Sidley, without address-
ing the purpose of the employer exemption, can be so
certain that it has proved that the 32 are employers with-
in the meaning of the ADEA. They are, or rather were,
partners, but it does not follow that they were employers.
A firm that under pursuit by the EEOC on suspicion of
discrimination redesignated its employees “partners” with-
out changing the preexisting employment relation an iota
10                                              No. 02-1605

would not by doing this necessarily buy immunity, even
if the redesignation sufficed to make them partners un-
der state law.
  This case is not as extreme; it does not involve relabel-
ing. Yet it involves a partnership of more than 500 part-
ners in which all power resides in a small, unelected
committee (it has 36 members). The partnership does not
elect the members of the executive committee; the com-
mittee elects them, like the self-perpetuating board of
trustees of a private university or other charitable founda-
tion. It is true that the partners can commit the firm, for
example by writing opinion letters; but employees of a
corporation, when acting within the scope of their employ-
ment, regularly commit the corporation to contractual
undertakings, not to mention to tort liability. Partners who
are not members of the executive committee share in the
profits of the firm; but many corporations base their em-
ployees’ compensation in part anyway, but sometimes in
very large part, on the corporation’s profits, without any-
one supposing them employers. The participation of the
32 demoted partners in committees that have, so far as
appears, merely administrative functions does not distin-
guish them from executive employees in corporations.
Corporations have committees and the members of the
committees are employees; this does not make them em-
ployers. Nor are the members of the committees on
which the 32 served elected; they are appointed by the
executive committee. The 32 owned some of the firm’s
capital, but executive-level employees often own stock in
their corporations. We shall see that there is authority that
employee shareholders of a professional corporation are
still employees, not employers, for purposes of federal
antidiscrimination law.
  Particularly unconvincing is Sidley’s contention that
since the executive committee exercises its absolute pow-
No. 02-1605                                                11

er by virtue of delegation by the entire partnership in
the partnership agreement, we should treat the entire
partnership as if it rather than the executive committee
were directing the firm. That would be like saying that if
the people elect a person to be dictator for life, the gov-
ernment is a democracy rather than a dictatorship. The
partners do not even elect the members of the commit-
tee. They have no control, direct or indirect, over its com-
position.
   Perhaps the most partneresque feature of the 32 part-
ners’ relation to the firm is their personal liability for the
firm’s debts: not because unlimited liability is a sine qua
non of partnership (there can be limited partnerships,
and there are other business entities besides partnership
that have unlimited liability—a sole proprietorship, for
example), but because it is the most salient practical dif-
ference between the standard partnership and a corpora-
tion. Sidley does not have limited liability, and this
means, by the way, that although under the firm’s rules
each partner is liable for the firm’s debts only in propor-
tion to his capital, a creditor of the firm could sue any
partner for the entire debt owed it. Is this enough to pin
the partner tail on the donkey? Wheeler v. Hurdman, 

, 274-75 (10th Cir. 1987), comes close to saying
it is; see also Fountain v. Metcalf, Zima & Co., P.A., 

, 1400-01 (11th Cir. 1991). Yet it does not quite
deny the necessity of considering other factors. And
tugging the other way are Strother v. Southern Califor-
nia Permanente Medical Group, 

, 866-68 (9th
Cir. 1996) (interpreting a similar provision of state anti-
discrimination law), and Simpson v. Ernst & 

,
100 F.3d at 441-42. Simpson classified partners as em-
ployees in circumstances broadly similar to, though dis-
tinguishable from, those of the present case:
12                                              No. 02-1605

     Simpson had no authority to direct or participate in
     the admission or discharge of partners or other firm
     personnel; participate in determining partners or other
     personnel compensation predicated upon perform-
     ance levels, responsibility, and years of service with
     the firm, including his own; participate in the vote
     for the chairman or the members of the Management
     Committee; or participate in the firm’s profits and
     losses or share in unbilled uncollected client accounts
     (“UBT’s”). Simpson had no right to examine the
     books and records of the firm except to the extent
     permitted by the Management Committee. He was
     required to execute a will which mandated that his
     heirs accept as binding the accounts provided by Ernst
     & Young with no right of inspection or verification.
     He had no authority to sign promissory notes on behalf
     of the firm, or pledge, transfer, or otherwise assign
     his interest in the firm. He was refused access to
     data concerning various client accounts. He was de-
     nied participation in annual performance reviews and
     other indicia of partnership status.
        In sum, the firm’s business, assets, and affairs were
     directed exclusively by a 10 to 14 member Management
     Committee and its chairman. The Management Com-
     mittee exercised exclusive control over the admission
     and discharge of all personnel, including Simpson.
     It could terminate employees without cause and with-
     out a right to appeal such decisions. The Management
     Committee unilaterally determined the compensation
     of all personnel, which authority was exercised and
     executed in total secrecy. Simpson and those similarly
     situated had no vote for the chairman or the members
     of the Management Committee. Instead, the Manage-
     ment Committee appointed its chairman and its chair-
     man appointed the members of the Management Com-
No. 02-1605                                                 13

    mittee. “For all practical purposes [the court is quot-
    ing here from the district court’s opinion], he was an
    employee with the additional detriment of having
    promised to be liable for the firm’s losses. Ernst &
    Young was free to draft its Partnership Agreement
    and U.S. Agreement in such a way as to generate the
    belief in its employees that they enjoyed partnership
    status and to permit them to represent themselves as
    partners. However, because these individuals actual-
    ly had no bona fide ownership interest, no fiduciary
    relationship, no share in the profits and losses, no
    significant management control, no meaningful voting
    rights, no meaningful vote in firm decisions, and no job
    security, they were not bona fide partners. Therefore
    Ernst & Young was obligated not to discriminate
    against them because of their age, sex, race, religion,
    national origin, or handicap.”
  The matter of liability for partnership debts illustrates
the importance of referring the question whether a part-
ner in a particular firm is an employer or an employee
to statutory purpose. If implicit in the ADEA’s exemption
for employers is recognition that partners ordinarily have
adequate remedies under partnership law to protect them-
selves against oppression (including age or other forms of
invidious discrimination) by the partnership, then ex-
posure to liability can hardly be decisive. These 32 part-
ners were not empowered by virtue of bearing large po-
tential liabilities! The 32 were defenseless; they had no
power over their fate. If other partners shirked and as a
result imposed liability on the 32, the 32 could not, as part-
ners in a conventional partnership could do, vote to
expel them. They had no voting power. What could be
argued but is not is that because the other partners are
potentially liable for the pratfalls of the 32, the partnership
should have greater power over their employment than
14                                                 No. 02-1605

if the firm were a corporation and so had limited liability.
To repeat, the issue is not whether the 32 before their
demotion were partners, an issue to which their liability
for the firm’s debts is germane; the issue is whether
they were employers. The two classes, partners under state
law and employers under federal antidiscrimination law,
may not coincide.
   The problem of line drawing presented by this case is
not unique to employment. It arises whenever legal con-
sequences turn on classification as partner versus em-
ployee, whether in tax and tort cases or in discrimina-
tion cases. See, e.g., Armstrong v. Phinney, 

, 663-
64 (5th Cir. 1968) (tax); Davis v. Loftus, 

(Ill.
App. Sept. 9, 2002) (tort liability). The law does not allow
firms to obtain the benefits or avoid the costs associated
with particular forms of doing business by simple re-
designation. Of course firms have broad freedom of
election among the different forms of doing business, such
as the corporate, partnership, LLC, and so forth. Their
freedom is not unlimited; there is, for example, the “sub-
stance over form” rule of tax law. See, e.g., Yosha v. Commis-
sioner, 

, 497-98 (7th Cir. 1988). But the ques-
tion, as we have been at pains to emphasize throughout
this opinion, is not whether Sidley is a partnership; it is.
The question is whether, when a firm employs the lati-
tude allowed to it by state law to reconfigure a partnership
in the direction of making it a de facto corporation, a fed-
eral agency enforcing federal antidiscrimination law is
compelled to treat all the “partners” as employers.
  The same problem of the tyranny of labels arose when
the Supreme Court had to draw the line in ERISA be-
tween an “employee,” defined unhelpfully as in the
ADEA as “any individual employed by an employer,” 29
U.S.C. § 1002(6), and an independent contractor. The
No. 02-1605                                               15

Court could have said that an employee is anyone who
is called an employee. Instead it said:
    In determining whether a hired party is an employee
    under the general common law of agency, we con-
    sider the hiring party’s right to control the manner and
    means by which the project is accomplished. Among
    the other factors relevant to this inquiry are the
    skill required; the source of the instrumentalities and
    tools; the location of the work; the duration of the
    relationship between the parties; whether the hiring
    party has the right to assign additional projects to the
    hired party; the extent of the hired party’s discre-
    tion over when and how long to work; the method of
    payment; the hired party’s role in hiring and paying
    assistants; whether the work is part of the regular
    business of the hiring party; whether the hiring party is
    in business; the provision of employee benefits; and
    the tax treatment of the hired party. Since the common-
    law test contains no shorthand formula or magic
    phrase that can be applied to find the answer, . . . all
    of the incidents of the relationship must be assessed
    and weighed with no one factor being decisive.
Nationwide Mutual Ins. Co. v. Darden, 

, 323-25
(1992) (citations and internal quotation marks omitted). In
a subsequent case, we said the most important factor
in deciding whether a worker was an employee or an
independent contractor was the employer’s right to con-
trol the worker’s work. Ost v. West Suburban Travelers
Limousine, Inc., 

, 438 (7th Cir. 1996). That is a
potentially important factor here as well. Both decisions
reject mechanical tests.
  We can get a little help on the question in our case from
Justice Powell’s concurring opinion in Hishon v. King &
16                                                  No. 02-1605

, 467 U.S. at 79-81, one of the few discus-
sions of the applicability of Title VII to partnerships. Here
is what he said (id. at 79-80; record reference omitted):
     I write to make clear my understanding that the Court’s
     opinion [holding discriminatory refusal to promote
     an associate to partner actionable] should not be read
     as extending Title VII to the management of a law
     firm by its partners. The reasoning of the Court’s opin-
     ion does not require that the relationship among part-
     ners be characterized as an “employment” relation-
     ship to which Title VII would apply. The relationship
     among law partners differs markedly from that be-
     tween employer and employee—including that be-
                                                 2
     tween the partnership and its associates. The judg-
     mental and sensitive decisions that must be made
                                                            3
     among the partners embrace a wide range of subjects.
     The essence of the law partnership is the common con-
     duct of a shared enterprise. The relationship among
     law partners contemplates that decisions important
     to the partnership normally will be made by common
     agreement or consent among the partners.
     _____________________
     2. Of course, an employer may not evade the strictures of
     Title VII simply by labeling its employees as “partners.” Law
     partnerships usually have many of the characteristics that
     I describe generally here.
     3. These decisions concern such matters as participation
     in profits and other types of compensation; work assign-
     ments; approval of commitments in bar association, civic,
     or political activities; questions of billing; acceptance of
     new clients; questions of conflicts of interest; retirement
     programs; and expansion policies. Such decisions may affect
     each partner of the firm. Divisions of partnership profits,
     unlike shareholders’ rights to dividends, involve judgments
No. 02-1605                                                     17

    as to each partner’s contribution to the reputation and
    success of the firm. This is true whether the partner’s partic-
    ipation in profits is measured in terms of points or percent-
    ages, combinations of salaries and points, salaries and
    bonuses, and possibly in other ways.
Justice Powell was saying that a traditional law partner-
ship, involving “the common conduct of a shared enter-
prise” and a relationship among the partners that “contem-
plates that decisions will be made by common agreement
or consent among the partners,” has a governance struc-
ture different from the one contemplated or assumed by
Title VII. At the same time he was making clear that label-
ing an enterprise that does not have the structure, the
character, of the traditional partnership will not immunize
it from the statute. In a case in which we held that part-
ners were employers for purposes of Title VII, the partner-
ship was, so far as appears, an equal partnership of four
partners. See Burke v. Friedman, 

, 868 (7th
Cir. 1977).
  A functional approach need not always lead to an expan-
sion in coverage. When physicians who were both the
shareholders of a professional medical corporation and
employees of the corporation argued that they were “really”
employers for purposes of deciding whether the corpora-
tion had enough employees to come within the reach of
the Americans with Disabilities Act, a panel of the Ninth
Circuit, rejecting the “economic realities” test proposed
by the dissent, refused to look behind the formal status
of the medical personnel as employees and concluded
that the corporation was indeed covered by the Act. Wells
v. Clackamas Gastroenterology Associates, P.C., 

,
905-06 (9th Cir. 2001), cert. granted, 

(U.S.
Oct. 1, 2002); see also Hyland v. New Haven Radiology Associ-
ates, P.C., 

, 796-98 (2d Cir. 1986). Like the dis-
sent in Wells, we had employed an “economic realities” test
18                                                 No. 02-1605

to hold that a professional corporation could be treated
as a partnership, EEOC v. Dowd & Dowd, Ltd., 

(7th Cir. 1984), and it could be argued that consis-
tency precludes our rejecting the Commission’s “economic
realities” test in this case, at least at this early stage in the
Commission’s investigation. And even the majority in
Wells stated that a firm could not, by affixing the label
of “partner” to someone who was functionally an em-
ployee, avoid federal antidiscrimination law. Wells v.
Clackamas Gastroenterology Associates, 

, 271 F.3d
at 905.
  All that is clear amidst this welter of cases is that the
coverage issue in the present case remains murky despite
Sidley’s partial compliance with the subpoena. The Com-
mission is therefore entitled to full compliance, at least
with regard to coverage, unless the additional documents
the Commission is seeking are obviously irrelevant. What
the Commission particularly wants to know is how un-
evenly the profits are spread across the entire firm. Are
profits so concentrated in members of the executive com-
mittee, or in some smaller or larger set of partners, in
relation to the profits that the executive committee al-
located to the 32, that the latter occupied the same position
they would have if they had been working at a compar-
able rank for one of the investment banks that once were
partnerships but now are corporations? This might not
be decisive but it would bear on the unavoidably multi-
factored determination of whether this large law firm—
which in recognition that conventional partnership is
designed for much smaller and simpler firms has contractu-
ally altered the structure of the firm in the direction of
the corporate form—should for purposes of antidiscrim-
ination law be deemed the employer of some at least of
the individuals whom it designates as partners.
No. 02-1605                                             19

   But we think the district court was premature to order
the subpoena complied with in its entirety. It is not only
the law firm that has failed to argue the purpose of the
exclusion of employers from the protection of the statute;
it is also the EEOC. Without having proposed a standard
or criterion to guide the determination, the Commission
has not earned the right to force the law firm not merely
to finish complying with the coverage portion of the sub-
poena but to go on and produce the voluminous and
sensitive documentation sought relating to the question
whether, if these 32 partners were employees, they were
demoted on account of their age and therefore in viola-
tion of the age discrimination law. We are therefore vacat-
ing the district court’s order and remanding the case with
directions to order the law firm to comply fully with the
part of the subpoena that requests documents relating to
coverage, but upon completion of those submissions to
make a determination whether the 32 demoted partners
are arguably covered by the ADEA. If it is plain on the
basis of uncontested facts that before their demotion the
32 were not employees and therefore were not protected
by the Act, which would place the case under the prin-
ciple of our Great Lakes decision, then, barring circum-
stances that we do not at present foresee, the court should
excuse the firm from compliance with the part of the
subpoena that requests documents relating to the merits.
  We are not ruling that the 32 demoted partners were
in fact employees within the meaning of the age discrim-
ination law. Such a ruling would be premature. Sidley has
respectable arguments on its side, not least that the func-
tional test of employer status toward which the EEOC is
leaning is too uncertain to enable law firms and other
partnerships to determine in advance their exposure to
discrimination suits—that it would be better if the courts
and the Commission interpreted the employer exclusion
20                                             No. 02-1605

to require treating all partners as employers, with per-
haps a narrow sham exception. These issues will become
ripe when Sidley finishes complying with the coverage
part of the subpoena. We hold only that there is enough
doubt about whether the 32 demoted partners are cov-
ered by the age discrimination law to entitle the EEOC to
full compliance with that part, at least, of its subpoena.
              VACATED AND REMANDED WITH DIRECTIONS.




   EASTERBROOK, Circuit Judge, concurring in part and
concurring in the judgment. I join my colleagues’ exem-
plary discussion of the law governing agency subpoenas
(slip op. 3-7) but otherwise concur only in the judgment.
I do not think that the scope of the ADEA’S coverage is
as unfathomable as the majority makes out, nor do I be-
lieve that if the law were so ambulatory we should punt
the legal question to the district court. Instead we should
do our best to reduce uncertainty. Sidley and other large
partnerships need to plan their affairs; their members
also need to know their legal status. Can large law firms
adopt mandatory-retirement rules? It is disappointing
that the EEOC should profess, some 30 years after the
ADEA’s enactment, that it hasn’t a clue about the answer.
My colleagues’ opinion does not help matters, and this
is a missed opportunity.
  The ADEA’s definition of “employee” has a circular qual-
ity:
     The term “employee” means an individual em-
     ployed by any employer except that the term
No. 02-1605                                                 21

   “employee” shall not include any person elected
   to public office in any State or political subdivi-
   sion of any State by the qualified voters thereof,
   or any person chosen by such officer to be on such
   officer’s personal staff, or an appointee on the
   policymaking level or an immediate adviser with
   respect to the exercise of the constitutional or legal
   powers of the office. The exemption set forth in the
   preceding sentence shall not include employees
   subject to the civil service laws of a State govern-
   ment, governmental agency, or political subdivi-
   sion. The term “employee” includes any individual
   who is a citizen of the United States employed by
   an employer in a workplace in a foreign country.
29 U.S.C. §630(f). Yet this does not condemn us to wander-
ing forever through the mist like the Flying Dutchman.
The ADEA was interpolated into the Fair Labor Standards
Act, and its definition of employee tracks the FLSA’s. 29
U.S.C. §203(e). It turns out to be a definition in wide
use. Language essentially identical to the first clause of
§630(f) appears in the National Labor Relations Act, 29
U.S.C. §158(b)(4)(i); the Labor-Management Reporting
and Disclosure Act, 29 U.S.C. §402(f); the Employee Retire-
ment Income Security Act, 29 U.S.C. §1002(6); the Family
and Medical Leave Act, 29 U.S.C. §2611(3) (incorporat-
ing §203(e)); Title VII of the Civil Rights Act of 1964,
42 U.S.C. §2000e(f); and the Americans with Disabilities
Act, 42 U.S.C. §12111(4). This means on the one hand
that a search for legislative purpose is futile—Congress
took off the rack language devised, and often used,
for subjects other than employment discrimination—and
on the other hand that a definition may be secured from
opinions that have addressed these other statutes. For
example, in Hishon v. King & Spalding, 

(1984),
all of the Justices assumed—and Justice Powell in con-
22                                               No. 02-1605

currence was explicit—that a bona fide partner of a large
law firm is not an “employee” for purposes of Title VII.
More recently, when dealing with ERISA, the Court held
unanimously that the definition’s circularity should be
fixed by incorporating into federal law the traditional
state agency-law criteria for identifying master-servant
relations. Nationwide Mutual Insurance Co. v. Darden, 

, 322-27 (1992).
  Darden turned on the distinction between an employee
and an independent contractor. As they had done when
resolving a similar problem in copyright law, see Com-
munity for Creative Non-Violence v. Reid, 

(1989),
the Justices looked to the approach in the Restatement
(Second) of Agency §220(2) (1958). We have done the same
in cases arising under other federal statutes that have
circular definitions of “employee.” See, e.g., Ost v. West
Suburban Travelers Limousine, Inc., 

, 437-39
(7th Cir. 1996). Likewise we have drawn from state-law
principles—such as the rule that corporate form must
be respected—to determine whether an “employer” has
enough “employees” to come under a federal statute.
See, e.g., Papa v. Katy Industries, Inc., 

(7th
Cir. 1999). As Darden recognized, these bodies of law
contain some flexible elements but give ready answers
for the great majority of situations.
  So too with partnership law. No one believes that a
bona fide partner is in a master-servant relation with the
partnership, or that the partner “is employed by” the
partnership. The qualification “bona fide” is important; as
Justice Powell observed in Hishon, an employer may not
evade obligations under federal law by plastering the
name “partner” on someone whose legal and economic
characteristics are those of an employee. See also Restate-
ment (Third) of Agency §1.02 (T.D. 2, 2001) (parties’ labels
No. 02-1605                                                23

do not control). But if a person has those attributes that
differentiate “partners” from “employees” in normal
legal usage, then Darden classifies that person as a non-
employee for purposes of a federal statute such as
§630(f). It is neither our duty, nor our privilege, to invent
a federal law of employment relations, as my colleagues
appear to believe. The law must be federal (because §630
is a federal statute), but Darden tells us that federal law
tracks ordinary principles of master-servant relations
that come from state law.
  Were the 32 lawyers bona fide partners? The majority all
but concedes that they were. If this had been a suit un-
der the diversity jurisdiction, and we needed to decide
whose citizenship counted for purposes of the rule that
a partnership has every partner’s citizenship, see Carden
v. Arkoma Associates, 

(1990), we would have
acknowledged that all 32 were partners by normal reckon-
ing. We know that all 32 (i) received a percentage of
Sidley’s profits and had to pony up if Sidley incurred a
loss; (ii) had capital accounts that were at risk if the firm
foundered; and (iii) were personally liable for the firm’s
debts and thus put their entire wealth, not just their
capital accounts, on the line. We also know that (iv) no non-
partner has an equity interest in the firm. The most im-
portant of these is the first (which implies the third): under
the Uniform Partnership Act, it is profit-sharing (coupled
with the lack of organization as an entity under some
other law) that defines a partnership and identifies its
partners, all of whom are personally liable for the venture’s
debts. See Uniform Partnership Act §202 (1997 rev.); see
also Daniel S. Kleinberger, Agency, Partnerships, and LLCs
§7.2.1 (2d ed. 2002). Illinois, which has enacted the model
act into positive law, treats participation in profits as the
defining characteristic of a bona fide partner. The court
in Davis v. Loftus, 

(Ill. App. 1st Dist.
24                                                No. 02-1605

Sept. 9, 2002), held that a partner who shares in the profits
or loss is personally liable for the law firm’s debts, while an
“income partner” who receives a salary plus a bonus is an
employee and not liable for the firm’s debts.
   The 32 lawyers were real partners and consequently not
“employees.” My colleagues’ suggestion that one can be
a partner under normal agency principles and still be
an “employee” because of a federal-law override is in-
compatible with Darden. Anyway, it makes both linguis-
tic and economic sense to say that someone who is liable
without limit for the debts of an organization is an entre-
preneur (a principal) rather than an “employee” (an agent).
Unlimited liability and profit-sharing give each partner
an interest in monitoring (and if need be expelling) those
other partners who are shirking or otherwise not carry-
ing their part of the load. Their actions in this respect
are those of owners. Cf. Eugene F. Fama & Michael
Jensen, Separation of Ownership and Control, 26 J.L. & Econ.
301, 315-17 (1983); Fama & Jensen, Agency Problems and
Residual Claims, 

334-36.
  Perhaps each practice group at a large firm is best viewed
as a distinct venture, and the umbrella organization (run
by the Executive and Management Committees at Sidley)
as a partnership of partnerships. The top committees can
make all decisions, but much power is bound to be dele-
gated, just as departments at a university make their own
hiring and salary decisions even though a self-perpetuat-
ing board of trustees holds all the legal authority. Mem-
bership on an academic department’s appointments com-
mittee is a position of real influence and responsibility
even though the trustees formally make all appointments.
See NLRB v. Yeshiva University, 

(1980) (hold-
ing that all faculty members are managers for purposes of
federal labor law even though they lack any legal instru-
No. 02-1605                                              25

ments of control). So too within the judiciary: committees
of the Judicial Conference effectively make many of the
most important administrative decisions although com-
mittee members do not sit on the Conference. Doubtless
things work similarly at large law firms, so my colleagues
ought not sneeze at Sidley’s observation that all 32 de-
moted partners served on committees. But the relation
between practice groups and the whole firm, and the
allocation of managerial authority among the lawyers,
do not matter to classification: a member of a large part-
nership including smaller associations remains a partner
rather than an employee in both economic and legal senses.
My colleagues tellingly do not cite a single state case, or
any scholarly commentary, restatement, or model code,
for the proposition that concentration of decision-making
authority within an entity alters the legal status of those
who share profits and bear all residual risk of loss.
  What leads me to concur in the judgment is not any
doubt about the right characterization of the 32 demoted
partners but uncertainty about that of other lawyers.
Sidley has a retirement age for everyone it dubs a partner.
Whether this is lawful can be determined only by classify-
ing, as “employee” or not, every lawyer who carries a
“partner” label. The EEOC is entitled to investigate without
knowing in advance how the inquiry will come out. It
may well be that Sidley designates as “partners” lawyers
who are paid straight salaries plus bonuses rather than
a portion of the profit (or loss) set ex ante. Unlike my
colleagues, I do not read the EEOC’s brief as conceding
that all of Sidley’s members are just like the 32 about
whom Sidley has provided information. How could the
EEOC tell at this stage whether the 32 are a representative
sample? It wants to know whether some “partners” are
paid entirely on the basis of guarantees (that is, salaries)
or have compensation packages slanted so heavily toward
26                                             No. 02-1605

the guarantee that they would not be liable for the firm’s
debts. The EEOC also seeks to learn whether all partners
have capital accounts at risk as the demoted 32 did.
These are things that the EEOC is entitled to find out for
everyone covered by the mandatory-retirement policy.
  What is more, EEOC v. Dowd & Dowd, Ltd., 

,
1178 (7th Cir. 1984), appears to commit this circuit to the
view that state law forms don’t matter, and that a fed-
eral court must assess for itself the question whether
the “economic realities” demonstrate the existence of
an employment relation. I say “appears to” not only be-
cause Dowd precedes Darden but also because any refer-
ence to “economic realities” poses the question which of
many realities will be selected as those that matter. Maybe
all that Dowd shows is that our court seeks to search
out those realities that matter under ordinary agency
law, and to look past veneers that lack legal or economic
significance. That would align Dowd with Justice Powell’s
concurring opinion in Hishon, on which the panel in
Dowd relied. But there would be little point in revisiting
the matter today, because the Supreme Court will decide
this Term whether Dowd is correct. The ninth circuit in
Wells v. Clackamas Gastroenterology Associates, P.C., 

(2001), cert. granted, No. 01-1435 (U.S. Oct. 1,
2002), rejected both the method and the outcome of Dowd,
and the grant of certiorari enables the Supreme Court
to resolve some or all of the problems that govern the
classification of Sidley’s members.
  Dowd held that classification of a person as an “employee”
under state law must be disregarded, for purposes of fed-
eral law, when the corporation is closely held—a profes-
sional corporation that differs from a partnership (beyond
matters of form) only in the extent to which the members
are personally liable for the firm’s debts. Clackamas held,
No. 02-1605                                                27

to the contrary, that any person classified as an employee
for purposes of state law necessarily is an employee for
purposes of federal law. Dowd has at least the virtue of
symmetry; it always looks through state-law forms (though
I confess unease about what Dowd identifies as the legally
important “realities”; professionals who are not person-
ally liable for debts lack one of the principal attributes of
partners). Clackamas, by contrast, states only a rule of
inclusion: one can become an “employee” by virtue of state
law even if one has all of the attributes that agency law
associates with being an independent contractor or a
partner, but a classification as a “partner” or “independent
contractor” under state law is never conclusive in the
employer’s favor. This is not Darden’s approach (nor did
the ninth circuit cite Darden or Reid). If we are to use ordi-
nary agency-law principles to identify an “employee,” we
should be consistent. The ninth circuit, though, has a
thumb on the scale in favor of classification as an “em-
ployee.”
   Darden is not alone in preferring symmetry. For ex-
ample, Robinson v. Shell Oil Co., 

(1997),
holds that an ex-employee is an “employee” for some
purposes under Title VII without regard to classification
under state law, and Walters v. Metropolitan Educational
Enterprises, Inc., 

(1997), holds that for pur-
poses of determining whether the employer exceeds the
statutory size threshold an “employee” is a person on the
firm’s payroll at a given time, whether or not that person
is paid for the date used in measurement. Both Robinson
and Walters adopt rules that attach consequences to par-
ticular attributes, without asking what names state law
(or particular employers) give to those attributes. Likewise
with the question whether a person who shares in a ven-
ture’s profits and losses and has unlimited personal liabil-
ity for the enterprise’s debts is an employee. We should
28                                              No. 02-1605

ask not what the organization, or any given state, calls
this person; we should ask how this set of attributes is
classified under the prevailing law of agency. I think it
very likely that the 32 lawyers Sidley demoted would be
classified as partners rather than employees under this
body of rules, but I do not know how Sidley’s other law-
yers should be classified, so a remand is in order. Enforc-
ing those aspects of the subpoena that call for informa-
tion relevant to the merits would be unduly burdensome
until this task has been completed, and unless the evi-
dence then shows that Sidley has classified as “partners”
some persons who are employees under ordinary agency
principles.

A true Copy:
       Teste:

                          _____________________________
                          Clerk of the United States Court of
                            Appeals for the Seventh Circuit




                   USCA-02-C-0072—10-24-02