In the
United States Court of Appeals
For the Seventh Circuit
____________________
No. 14-‐‑1532
IN THE MATTER OF:
IKO ROOFING SHINGLE PRODUCTS LIABILITY LITIGATION
APPEAL OF:
DEBRA ZANETTI, et al., on behalf of a proposed class
____________________
Appeal from the United States District Court
for the Central District of Illinois.
No. 09-‐‑md-‐‑2104 — Harold A. Baker, Judge.
____________________
ARGUED JUNE 5, 2014 — DECIDED JULY 2, 2014
____________________
Before WOOD, Chief Judge, and EASTERBROOK and KANNE,
Circuit Judges.
EASTERBROOK, Circuit Judge. Purchasers of organic asphalt
roofing shingles in many states have filed suits against IKO
Manufacturing and affiliated firms, contending that it falsely
told customers that these shingles met an industry standard
known as ASTM D225, and that compliance had been ascer-‐‑
tained by use of a testing protocol known as ASTM D228.
What distinguishes an “organic” asphalt tile is the inclusion
2 No. 14-‐‑1532
of a layer made from felt or paper; tiles that include a fiber-‐‑
glass layer are not called organic, even though asphalt itself
has organic components. We omit other details about the
tiles and the standards, for the merits of plaintiffs’ claims are
not material to the current dispute.
In 2009 the Panel on Multidistrict Litigation transferred
all of the federal suits to the Central District of Illinois for
consolidated pretrial proceedings under 28 U.S.C. §1407.
Plaintiffs asked the court to certify a class that would cover
IKO’s sales in eight states since 1979. The court declined,
2014 U.S. Dist. LEXIS 80243 (C.D. Ill. Jan. 28, 2014), and we
granted the plaintiffs’ request for interlocutory review under
Fed. R. Civ. P. 23(f).
Before addressing plaintiffs’ arguments about the class-‐‑
certification decision, we must consider whether the judge
who denied plaintiffs’ motion had authority to preside over
the litigation. Section 1407(b) provides that the “pretrial pro-‐‑
ceedings shall be conducted by a judge or judges to whom
such actions are assigned by the judicial panel on multidis-‐‑
trict litigation.” The Panel assigned the cases to Judge
McCuskey, who had agreed to accept them. In his capacity
as the district’s Chief Judge at the time, he reassigned them
to Judge Baker early in 2010. Unfortunately, Judges McCus-‐‑
key and Baker failed to recognize that §1407(b) gives the
Panel exclusive power to select the judge. Its rules provide
that, “[i]f for any reason the transferee judge is unable to
continue those responsibilities, the Panel shall make the re-‐‑
assignment of a new transferee judge.” Rules of Procedure of
the United States Panel on Multidistrict Litigation 2.1(e).
Neither judge asked the Panel to change the assignment. The
Panel proceeded to transfer follow-‐‑on cases to Judge
No. 14-‐‑1532 3
McCuskey through 2012, without appreciating that all he
did was relay them to Judge Baker. Eventually the Panel got
wind of the situation (the record does not reveal how) and
on February 12, 2014, issued an order transferring all of the
cases to Judge Baker. But he had denied plaintiffs’ motion
for class certification two weeks earlier. In legal jargon, he
was acting ultra vires.
If the problem deprived the court of subject-‐‑matter juris-‐‑
diction, then there is nothing for us to do but vacate the or-‐‑
der of January 28—and every other order Judge Baker en-‐‑
tered during the preceding four years. We do not think,
however, that §1407 affects subject-‐‑matter jurisdiction, a
term that deals with the tribunal’s adjudicatory competence.
See, e.g., Gonzalez v. Thaler, 132 S. Ct. 641 (2012); Henderson v.
Shinseki, 131 S. Ct. 1197 (2011); Reed Elsevier, Inc. v. Muchnick,
559 U.S. 154 (2010); Minn-‐‑Chem, Inc. v. Agrium, Inc., 683 F.3d
845 (7th Cir. 2012) (en banc).
All of these suits are properly in federal court, and for
that matter properly in the Central District of Illinois for
consolidated pretrial proceedings. A district court’s error in
complying with §1407(b), which gives the Panel exclusive
authority to select the transferee judge, does not vitiate the
Panel’s selection of the appropriate district court, the subject
of a separate clause in §1407(a). In the Supreme Court’s cur-‐‑
rent terminology, §1407(b) creates a case-‐‑processing rule ra-‐‑
ther than a jurisdictional one. One vital difference between
the two is that the litigants may waive or forfeit the benefits
of case-‐‑processing rules, while jurisdictional rules must be
enforced by the judiciary on its own initiative, even if the lit-‐‑
igants are content. No one protested Judge Baker’s role, so
4 No. 14-‐‑1532
whatever benefit the original selection of Judge McCuskey
may have offered to any litigant has been forfeited.
Lexecon Inc. v. Milberg Weiss Bershad Hynes & Lerach, 523
U.S. 26 (1998), which thoroughly analyzes §1407, supports
this understanding. The Court held in Lexecon that transferee
courts’ authority depends on §1407, and that they cannot ex-‐‑
ercise greater power than the statute provides. This means
that transferee judges and the Panel must implement the fi-‐‑
nal sentence of §1407(a), which reads: “Each action so trans-‐‑
ferred shall be remanded by the panel at or before the con-‐‑
clusion of such pretrial proceedings to the district from
which it was transferred unless it shall have been previously
terminated”. That sentence prevents transferee courts from
seizing control for all purposes, including trial, of litigation
that is supposed to be in their charge only for pretrial coor-‐‑
dination. The Court analogized this provision to the venue
rules in 28 U.S.C. §1391 and observed that it, like other ven-‐‑
ue rules, gives plaintiffs a right to proceed in their chosen
forum. 523 U.S. at 41–42. But venue differs from subject-‐‑
matter jurisdiction, which is why Lexecon was careful to ob-‐‑
serve, id. at 43, that its holding applies to a “party who con-‐‑
tinuously objected to an uncorrected categorical violation of
the mandate [in §1407(a)]”. The need for objection puts the
requirements of §1407 on the “case-‐‑processing” rather than
the “jurisdictional” side of the Supreme Court’s line.
Statutes in addition to §1407 allow federal judges to be
designated to hear cases that otherwise would be in the bail-‐‑
iwick of a different judge, and this is not the first time there
has been a foul-‐‑up in the process. Both McDowell v. United
States, 159 U.S. 596, 601–02 (1895), and Ball v. United States,
140 U.S. 118, 128–29 (1891), hold that errors in designation
No. 14-‐‑1532 5
do not spoil the action of the judge who actually renders de-‐‑
cision. Nguyen v. United States, 539 U.S. 69, 77–83 (2003), de-‐‑
clines to extend the principle of these opinions to judges
who are ineligible to be designated. The chief judge of a court
of appeals had designated a judge from the Northern Maria-‐‑
na Islands to sit on an appellate panel. Because judges of that
territory serve under Article IV rather than Article III, the
designated judge was ineligible under both the statute and
the Constitution. Nguyen holds that the participation of an
ineligible decisionmaker requires vacatur of the judgment,
even though two Article III judges, a quorum of the panel,
had joined it, and no one objected to the panel’s composition
until after it had released its decision.
The transfer of these proceedings to Judge Baker is on the
technical side of Nguyen’s line. No one thinks him ineligible
to be a transferee judge. The Panel now has designated him
to serve in that capacity. Given the technical nature of this
problem and the lack of objection by any litigant, we need
not decide whether the Panel’s order would have had retro-‐‑
active effect and thus overridden a litigant’s objection to
Judge Baker’s role.
On to the class-‐‑certification issue. One vital question un-‐‑
der Fed. R. Civ. P. 23(b)(3) is whether “the questions of law
or fact common to class members predominate over any
questions affecting only individual members”. IKO’s tiles
are exposed to the elements, where they may fail no matter
how well constructed. Tornados and hurricanes may rip
them off; storms may lift them up so that water gets under
them. Poor installation may shorten their expected life. At
the same time, some tiles that flunk the D225 standard will
last indefinitely. The district court thought that the inevitable
6 No. 14-‐‑1532
differences in consumers’ experiences with IKO’s tiles pre-‐‑
vent class certification under the language we have quoted.
The court read Comcast Corp. v. Behrend, 133 S. Ct. 1426
(2013), and Wal-‐‑Mart Stores, Inc. v. Dukes, 131 S. Ct. 2541
(2011), to require proof “that the plaintiffs will experience a
common damage and that their claimed damages are not
disparate.” 2014 U.S. Dist. LEXIS 80243 at *9. Elsewhere the
district court wrote that “commonality of damages” is essen-‐‑
tial. Id. at *13, 20, 23. If this is right, then class actions about
consumer products are impossible, and our post-‐‑Comcast de-‐‑
cision in Butler v. Sears, Roebuck & Co., 727 F.3d 796 (7th Cir.
2013), must be wrong. Pella Corp. v. Saltzman, 606 F.3d 391
(7th Cir. 2010), which approved class treatment of litigation
asserting defects in casement windows, another product
used in home construction or renovation, also would be
wrongly decided under the district court’s reading of Com-‐‑
cast and Wal-‐‑Mart. Indeed, we could not affirm the district
court’s decision in this case without overruling Pella, so close
are the circumstances of the two home-‐‑product-‐‑defect suits.
Yet Wal-‐‑Mart has nothing to do with commonality of
damages. It dealt instead with the need for conduct common
to members of the class, and it concerned Rule 23(a)(2) rather
than Rule 23(b)(3). Plaintiffs in Wal-‐‑Mart contended that dis-‐‑
cretionary acts by managers of more than 2,000 local stores
produced discriminatory effects. When writing that com-‐‑
monality under Rule 23(a)(2) requires proof of the same inju-‐‑
ry, the Court observed that each store was managed inde-‐‑
pendently; it held that when multiple managers exercise dis-‐‑
cretion, conditions at different stores do not present a com-‐‑
mon question. See also Bolden v. Walsh Construction Co., 688
F.3d 893 (7th Cir. 2012). In that situation damages differ, to
No. 14-‐‑1532 7
be sure, but only because the underlying conduct differs. In
a suit alleging a defect common to all instances of a consum-‐‑
er product, however, the conduct does not differ.
Comcast, by contrast, does discuss the role of injury under
Rule 23(b)(3), though not in the way the district court
thought. Plaintiffs filed an antitrust suit and specified four
theories of liability. The district judge certified a class limited
to one of these four. The plaintiffs’ damages expert, howev-‐‑
er, estimated harm starting with the assumption that all four
theories had been established. The Court held that this made
class treatment inappropriate: without a theory of loss that
matched the theory of liability, the class could not get any-‐‑
where.
That would be equally true in a suit with just one plain-‐‑
tiff. In antitrust law, damages are limited to the sort of injury
that flows from unlawful conduct. See, e.g., Brunswick Corp.
v. Pueblo Bowl-‐‑O-‐‑Mat, Inc., 429 U.S. 477 (1977). Competition
creates benefits for consumers and harm for producers at the
same time, while monopoly causes harm to consumers and
some producers. It is essential to distinguish the encouraged
injuries (to producers, from competition) from the forbidden
ones (to consumers, from monopoly). That requires match-‐‑
ing the theory of liability to the theory of damages. Comcast
explained: “The first step in a damages study is the transla-‐‑
tion of the legal theory of the harmful event into an analysis of
the economic impact of that event.” 133 S. Ct. at 1435, quoting
from Federal Judicial Center, Reference Manual on Scientific
Evidence 432 (3d ed. 2011) (emphasis added by Comcast).
Plaintiffs in our litigation have two theories of damages
that match their theory of liability. The first is that every
purchaser of a tile is injured (and in the same amount per
8 No. 14-‐‑1532
tile) by delivery of a tile that does not meet the quality
standard represented by the manufacturer. Damages reflect
the difference in market price between a tile as represented
and a tile that does not satisfy the D225 standard. See Uni-‐‑
form Commercial Code §2–714(2). This remedy could be ap-‐‑
plied to every member of the class.
The second theory is that purchasers whose tiles actually
failed are entitled to recover damages, if nonconformity to
the D225 standard caused the failure. That sort of remedy
would require buyer-‐‑specific hearings along the lines dis-‐‑
cussed in Butler, 727 F.3d 798. The problems encountered in
an effort to settle Pella (see Eubank v. Pella Corp., No. 13-‐‑2091
(7th Cir. June 2, 2014)) imply the wisdom of using the first
approach to damages. (The first and second are alternatives:
they imply the same gross compensation to the class as a
whole, and to award both remedies would be double count-‐‑
ing.) But neither approach runs afoul of Comcast: both the
uniform and the buyer-‐‑specific remedies match the theory of
liability, as Comcast requires.
A buyer-‐‑specific remedial approach would require con-‐‑
fining any class certification to questions of liability, as in
Pella I and Butler. It is not hard to frame liability issues suited
to class-‐‑wide resolution. (1) Did IKO’s tiles generally con-‐‑
form to the substantive aspects of the D225 specification?
(“Generally” is an important qualifier. In any mass produc-‐‑
tion operation, defects or deviations are inevitable.) (2) Did
IKO test the tiles according to the D228 procedure? (3) If IKO
failed to use the D228 procedure, did it nonetheless test its
output in a way that would verify compliance? (If yes, this
implies that a negative answer to Question 2 would not play
a causal role in buyers’ losses.) (4) Did any failure to satisfy
No. 14-‐‑1532 9
the D225 standard cause the sorts of problems plaintiffs re-‐‑
ported, or would roughly the same failure rate have been
experienced with complying tiles? (The answer could be “no
causation” because the difference between D225-‐‑compliant
tiles and other organic tiles is sufficiently small that it would
not lead to a materially different failure rate.)
We do not say that a district court is required to certify a
class action on all four of these subjects, or on a class-‐‑wide
damages theory, or indeed on any issue. A district judge has
discretion to evaluate practical considerations that may
make class treatment unwieldy despite the apparently com-‐‑
mon issues. On occasion the problems are so grave that it is
an abuse of discretion to certify a class. See, e.g., Parko v.
Shell Oil Co., 739 F.3d 1083 (7th Cir. 2014). But when exercis-‐‑
ing its discretion the court must apply the correct legal
standards. The district court denied plaintiffs’ motion to cer-‐‑
tify under a mistaken belief that “commonality of damages”
is legally indispensible. With that error corrected, the district
court can proceed using the proper standards.
The papers filed in this court address a number of addi-‐‑
tional issues that are outside the scope of a Rule 23(f) appeal.
Whether particular experts’ reports satisfy the criteria of Fed.
R. Evid. 702, for example, is a subject addressed to the dis-‐‑
trict court in the first instance (the judge has not ruled yet)
and reviewable on appeal from a final decision. Likewise we
do not express any view on the question whether the evi-‐‑
dence compiled so far supports the plaintiffs’ contention that
the tiles are defective. That some or even many tiles exam-‐‑
ined 15 years after their manufacture flunk the D225 stand-‐‑
ard does not necessarily imply that they did so when they
left the factory; it is essential to understand how organic tiles
10 No. 14-‐‑1532
change over time and extrapolate from current data to the
tiles’ likely condition when sold. That the average age of tiles
that have already failed is 12 years does not establish that all
tiles had an expected 12-‐‑year life when made; IKO repre-‐‑
sents that less than 1% of its tiles have failed (or, perhaps
more accurately, that less than 1% are reported as having
failed). It may require sophisticated analysis to determine
what failure data imply about the durability of tiles as man-‐‑
ufactured. The simplistic pro-‐‑and-‐‑con presentations in the
papers filed here do not make much progress in establishing
the tiles’ mean time between failures or their half-‐‑life in use.
Getting these numbers right will be a central task for the
district court whether or not a class is certified on remand.
Development of facts also may influence the ultimate deci-‐‑
sion about class certification—and, to repeat, we do not say
that certification is required (or necessarily would be pru-‐‑
dent), or what if any issues should be certified. Those mat-‐‑
ters are for the district court.
The decision declining to certify a class is vacated, and
the case is remanded for proceedings consistent with this
opinion.