CHANG, District Judge.

The plaintiff, Ernest Gibson, filed suit in Wisconsin state court against former manufacturers of white lead carbonate pigments.1 This pigment was used, before the federal government banned it in the 1970s, in paints, including paints applied to residences. Gibson brings negligence and strict liability claims against the pigment manufacturers, but because he cannot identify which manufacturer made the white lead carbonate pigment that injured him, he relies on the "risk contribution" theory of tort liability fashioned by the Wisconsin Supreme Court. Thomas v. Mallett, 285 Wis.2d 236, 701 N.W.2d 523, 564 (2005). Under the risk-contribution theory, plaintiffs are relieved of the traditional requirement to prove that a specific manufacturer caused the plaintiff's injury. The district court held that risk-contribution theory violates the substantive component of the Due Process Clause, and granted summary judgment in favor of the defendants. As we explain below, in light of the broad deference that the Constitution grants to the development of state common law, risk-contribution theory survives substantive Due Process scrutiny, as well as the manufacturers' other constitutional challenges. We thus reverse the judgment and reinstate the plaintiff's case.

I.

Because this is an appeal from the grant of summary judgment, we review the district court's decision de novo, meaning independently, and draw all reasonable inferences of fact in the non-movant's favor (here, Gibson). Bennett v. Roberts, 295 F.3d 687, 694 (7th Cir.2002). As it turns out, the genuinely disputed facts are not material to the legal question presented by the appeal.

In 1997, Gibson and his family moved into a house in Milwaukee, Wisconsin. The house was built in 1919. Unfortunately, the paint applied to that house contained white lead carbonate pigment. In the late 1800s and in the 1900s, paint manufacturers valued white lead carbonate pigments for several reasons, including their strength, durability, flexibility, washability, brushability, and brightness. The white lead carbonate pigment poisoned Gibson, causing neurological defects, among other injuries. The paint was applied to Gibson's home sometime before 1978, which is when the Consumer Products Safety Commission banned paint makers from intentionally adding lead into residential paint.

Gibson is not able to identify which specific manufacturer made the white lead carbonate pigment that poisoned him. In Wisconsin state court, Gibson sued seven companies that either made white lead carbonate pigment or were successors-in-interest to companies that had made that type of pigment.2 Gibson alleged that he had been injured by the makers' negligence and their failure to warn about the dangers of white lead carbonate pigment. Those seven companies were not the only possible makers of white lead carbonate pigment, although they, along with a no-longer-in-business company, Eagle-Picher Industries, did comprise the primary producers of the pigment.

On the basis of diversity jurisdiction, the case was removed to federal court. The district court initially remanded the case back to state court because of a question over whether the amount-in-controversy minimum had been met. In state court, the parties engaged in discovery on the controversy-amount issue; afterwards, once again the case was removed to federal court. One manufacturer, Millennium Holdings LLC, was dismissed from the case after that defendant filed for bankruptcy (more on this below).

The remaining six pigment manufacturers are:

Because Gibson could not identify which of these manufacturers made the white lead carbonate pigment that poisoned him, he had to rely on a theory of tort liability fashioned by the Wisconsin Supreme Court in Thomas v. Mallett, 285 Wis.2d 236, 701 N.W.2d 523, 564 (2005). As discussed in more detail below, Thomas held that a plaintiff who brings a white lead carbonate pigment case does not bear the traditional burden of proving that a particular lead-pigment manufacturer caused the plaintiff's injury. Instead, so long as a plaintiff makes a prima facie showing that the manufacturer produced or marketed white lead carbonate pigment sometime during the house's existence, then the burden is on each manufacturer to prove that it did not produce or market white lead carbonate pigment either during the house's existence or in the geographical market where the house is located. If there are no records (or no longer any records) to prove the manufacturer's defense, then the defense fails.

Atlantic Richfield Corporation (better known as ARCO) moved for summary judgment, arguing that Thomas's liability framework violates the Constitution. ARCO presented various constitutional arguments, including that the risk-contribution theory of liability violates the Due Process Clause. The district court granted summary judgment for ARCO, and then followed-up with summary judgment for the other five remaining defendants. R.39, R. 107. Gibson appeals.

II.

A.

Before addressing the merits of the dispute, first we must ensure, as in all cases, that there is subject matter jurisdiction over the case in the district court, as well as appellate jurisdiction over the appeal. On the question of subject matter jurisdiction, Gibson's opening brief disclaimed knowledge about the citizenship of one of the former defendants in the case, Millennium Holdings LLC. As discussed in the next section, Millennium Holdings has been dismissed from the case in the district court. But at the time of the complaint's removal (the second time around) to federal court, Millennium Holdings was a named defendant and its citizenship had to be evaluated for diversity of citizenship. So we ordered the parties to file jurisdictional memoranda.

In response, the manufacturer-defendants filed an affidavit executed by a Millennium Holdings officer, Regina Lee. Lee was the Secretary and Treasurer of Millennium Holdings. In the affidavit, Lee averred that Millennium Holdings is a Delaware limited liability company, with only one member, Millennium America, Inc. That corporation was incorporated in Delaware and had its principal place of business there. So Millennium Holdings LLC was, for purposes of diversity jurisdiction, a citizen of Delaware. The plaintiffs (Gibson and his guardian) were citizens of Wisconsin, as was Milwaukee County, a party that had been realigned to be a plaintiff. Accordingly, there was complete diversity at the time of the filing of the notice of removal.

Against this, Gibson argues that Lee's affidavit should not be considered because Millennium Holdings had filed an answer to the complaint, and the answer had stated that Millennium Holdings was a Delaware corporation with its principal place of business in Texas. But the answer does not undermine diversity jurisdiction. First, even if Millennium Holdings was bound by the characterization of citizenship in the answer, then there still would be complete diversity, with only Wisconsin citizens on the plaintiffs' side of the litigation and only non-Wisconsin citizens on the other side. More importantly, where subject matter jurisdiction turns on actual facts, the pleadings are not the end-all of determining the facts. Indeed, "[d]efective allegations of jurisdiction may be amended, upon terms, in the trial or appellate courts." 28 U.S.C. § 1653. We have previously permitted jurisdictional statements to be filed on appeal to fix defective allegations. E.g., Thomas v. Guardsmark, LLC, 487 F.3d 531, 533-34 (7th Cir.2007); Tylka v. Gerber Products Co., 211 F.3d 445, 448 (7th Cir.2000). Accordingly, we consider the notice of removal to be amended by the defendants' filing of Lee's affidavit, which establishes that Millennium Holdings' citizenship does not undermine complete diversity. Diversity jurisdiction was the proper basis for subject matter jurisdiction over the case.

B.

In addition to subject matter jurisdiction over the case, we also must ensure that there is jurisdiction over the appeal, whether or not the parties raise the issue. Wingerter v. Chester Quarry Co., 185 F.3d 657, 660 (7th Cir.1998) (per curiam). Here, the only question is whether there is a final, appealable decision in the district court in light of the fact that Millennium Holdings was dismissed from the case "without prejudice." Specifically, after Millennium Holdings filed for bankruptcy in the Southern District of New York, the district court and the parties treated Millennium Holdings as if it was no longer a party to the case. When the district court entered a final judgment under Federal Rule of Civil Procedure 58(a), the district court stated that the "claims against Millennium Holdings LLC are dismissed without prejudice because it is in Chapter 11 bankruptcy." The district court then stated, on the judgment, "This action is hereby dismissed." By the time of the entry of the judgment in the district court, the bankruptcy court had already discharged Gibson's claim in the bankruptcy proceeding by confirming a plan of reorganization.

This procedural posture renders the judgment entered by the district court a final, appealable decision under 28 U.S.C. § 1291. When the district court issued its summary judgment decisions, there was nothing more for the district court to do with the lawsuit, which is the hallmark of a final decision. "A district court's decision is final if `the district court has finished with the case.'" Minnesota Life Ins. Co. v. Kagan, 724 F.3d 843, 847 (7th Cir.2013) (quoting Chase Manhattan Mortg. Corp. v. Moore, 446 F.3d 725, 726 (7th Cir.2006)). In a similar prior decision, we concluded that, in a case where two of the four defendants had filed for bankruptcy but had not been formally dismissed from the case in the district court, the judgment of the district court with regard to the remaining defendants was still a final decision, for purposes of appellate jurisdiction, because any pursuit of the particular claims "will be pursued if at all in the bankruptcy court." Dimmitt & Owens Financial, Inc. v. United States, 787 F.2d 1186, 1190 (1986).

The finality of the judgment in this case distinguishes our situation from Willhelm v. Eastern Airlines, Inc., 927 F.2d 971, 972 (7th Cir.1991). There, the plaintiff filed suit against two defendants; one of the defendants filed for bankruptcy, and the other defendant won a motion to dismiss with prejudice for failure to state a claim. Id. at 972. The plaintiff sought to appeal the dismissal for failure to state a claim, but the district court had not dismissed the entirety of the action. Instead, the district court had entered an order stating that the plaintiff could either file his claim in the bankruptcy proceeding or move the bankruptcy court to lift the automatic stay and thereby reopen the district-court case. Id. We held that it was possible for the plaintiff to reopen the case, so the case was not entirely finished in the district court. Id. Accordingly, if it is possible for the automatic stay to be lifted, thereby allowing the district-court litigation to resume, then there is no final decision under 28 U.S.C. § 1291. Kimbrell v. Brown, 651 F.3d 752, 756 (7th Cir.2011). In contrast, as we discussed above, the discharge of Gibson's claim in Millennium Holdings' bankruptcy proceeding renders the judgment entered in the district court a final, appealable decision. Our appellate jurisdiction is secure.

III.

A.

There is yet another issue that we must address before getting to the merits of the appeal. During the appeal's pendency, the Wisconsin state legislature enacted Wisconsin Statute 895.046, which purports to extinguish risk-contribution theory in Wisconsin state courts, including for cases that were already pending at the time of the statute's enactment, like Gibson's case. Plaintiffs in already-filed lead-pigment cases have challenged the constitutionality of Section 895.046, primarily on the ground that retroactive application of the statute violates the Wisconsin Constitution's guarantee of due process. Wisc. Const. art. I, § 1. At our request, the parties filed supplemental briefs on the impact of Section 895.046, including on the question of whether we must (or should) address the constitutional challenge to the statute, and on the merits of that challenge.3

We conclude that we have no choice but to address the challenge under the Wisconsin Constitution to the state legislature's attempt to extinguish risk-contribution theory in already-pending cases. This conclusion arises from our general duty to avoid federal constitutional issues if the matter can be resolved on other grounds — including state constitutional grounds. See, e.g., RAR, Inc. v. Turner Diesel, Ltd., 107 F.3d 1272, 1276 (7th Cir.1997) ("And we must at least try to address the state constitutional issue first because the doctrine of constitutional avoidance counsels that federal courts should avoid addressing federal constitutional issues when it is possible to dispose of a case on pendent state grounds."); Allstate Ins. Co. v. Serio, 261 F.3d 143, 150 (2d Cir.2001) ("[W]here possible, courts will render decisions on federal constitutional questions unnecessary by resolving cases on the basis of state law (whether statutory or constitutional)."). If Section 895.046 has indeed successfully (meaning, constitutionally) extinguished risk-contribution theory in this and other already-pending cases, then discussing the federal constitutional challenges to risk-contribution theory would amount to issuing an advisory opinion.

So, in light of our duty to avoid opining on federal constitutional issues if possible, we must apply the Wisconsin Supreme Court's precedent on the retroactive application of state legislation. As noted above, plaintiffs in already-pending lead-pigment cases have already brought Wisconsin-Constitution based challenges to Section 895.046.4 In Clark ex rel. Gramling v. American Cyanamid Co., 2014 WL 1257118, Case No. 06-CV-12653 (Wis. Circuit Ct. March 25, 2014), the Circuit Court of Milwaukee County struck down the statute as violating Wisconsin's constitutional guarantee of due process.

We agree with Clark that Wisconsin Supreme Court precedent demands holding that Section 895.046 violates state due-process principles by trying to extinguish Gibson's vested right in his negligence and strict-liability causes of action. The state high court tests the due-process constitutionality of the retroactive application of state statutes by asking, first, whether the statute is taking away a "vested right" of the challenger. Matthies v. Positive Safety Mfg. Co., 628 N.W.2d 842, 852-53, 244 Wis.2d 720, 737-38 (Wis.2001); see Martin by Scoptur v. Richards, 531 N.W.2d 70, 90, 192 Wis.2d 156, 206 (1995). If the answer is that no vested right is at stake, then the statute satisfies due process and the inquiry ends. If, however, the challenger is losing a vested right, then the second step of the inquiry asks whether retroactive application has a rational basis, which is discerned by balancing the public interest served by retroactive application against the private interest impacted by the statute. Matthies, 628 N.W.2d at 855, Martin, 531 N.W.2d at 93.

Under Wisconsin law, Gibson did have a "vested right" in his claims under Thomas's risk-contribution theory. The Wisconsin Supreme Court decisions in Matthies and in Martin both dictate that a plaintiff's interest in a common-law claim is a protected vested interest. In Matthies, the interest was the plaintiff's previously existing common-law negligence claim, specifically, his right to hold a defendant jointly and severally liable without having to prove that the defendant was more than 50% negligent (the statute at issue extinguished joint and several liability unless that threshold of comparative negligence was met). 628 N.W.2d at 852-53. Matthies explained that an "existing right of action which has accrued under the rules of the common law or in accordance with its principles is a vested property right." Id. at 852 (quotation and citation omitted). Similarly, in Martin, the statute at issue imposed caps on non-economic damages, and the Wisconsin Supreme Court held that, when the plaintiffs' claim accrued, they "had a substantive right to recover, in full, the non[-]economic damages awarded by the jury," and that right was vested for due-process purposes. 531 N.W.2d at 91-92. Just so here, where Gibson's right to pursue the risk-contribution theory of liability on his negligence and strict-liability claims had already accrued by the time Section 895.046 tried to extinguish that right in June 2013.

On the second step of the analysis — the balancing of the public interest and the private interest — it is true that Wisconsin case law grants Section 895.046 a presumption of constitutionality, even in retroactive application. But here again Wisconsin Supreme Court precedent dictates that Section 895.046 cannot be retroactively applied in light of the state constitution's guarantee of due process. In Martin, the state high court rejected retroactive application of the cap on non-economic damages in medical malpractice cases, because the legislature burdened — retroactively — "severely injured litigants" at the expense of trying to bring down the costs of medical-malpractice suits and of health care overall. 531 N.W.2d at 93.

Similarly, in Matthies, despite the public interest in modifying joint and several liability so that a defendant who is less than 51% negligent would not have to pay the entirety of the damages award, the Wisconsin Supreme Court emphasized that retroactive application of the statute on the plaintiff would deprive him of his ability to recover full compensation for his injury. 628 N.W.2d at 860-61. Moreover, the public interest in retroactive application did not outweigh the plaintiff's private interest because there already existed contribution claims among tortfeasors to apportion liability. Id. at 857.

In our case, Section 895.046 was enacted to serve the public interest in permitting businesses to operate in Wisconsin without fear of products-liability litigation in the indefinite future based on risk-contribution theory. Wis. Stat. § 895.046(1g) (describing legislative findings and intent). And, of course, Section 895.046 serves this purpose by extinguishing risk-contribution theory altogether. But the competing private interest is significant, even more so than in Martin and Matthies. Without risk-contribution theory, Gibson (and similarly situated plaintiffs) cannot prove causation-in-fact as to a particular manufacturer and thus will likely recover nothing, even though Gibson can show (if he proves his prima facie case) that the pigment manufacturers contributed to the risk of injuring him. In Martin and Matthies, the statutes at issue did not entirely extinguish the plaintiffs' vested right to recover some amount of damages, whether it was from another negligent defendant (Martin) or under the statutory cap of damages (Matthies), and yet even there the Wisconsin Supreme Court rejected retroactive application. Here, Gibson would likely have no remedy at all. As interpreted by the Wisconsin Supreme Court, the state constitution's due-process guarantee prohibits retroactive application of Section 895.046. Gibson's claims remain viable, so we therefore cannot avoid the federal constitutional issues.

B.

Turning now to the merits of this appeal, the manufacturers challenge the constitutionality of the liability framework created by Thomas v. Mallett, 285 Wis.2d 236, 701 N.W.2d 523, 564 (2005). Before we get to Thomas, however, we first need to discuss the case on which Thomas was built. That building-block case is another Wisconsin Supreme Court case, Collins v. Eli Lilly Co., 116 Wis.2d 166, 342 N.W.2d 37 (1984).

In that case, the plaintiff was Therese Collins, and her mother was Roseann Collins. In 1957, during Roseann Collins's pregnancy, her doctor prescribed diethylstilbestrol, known as "DES," a drug that would ostensibly prevent miscarriages by keeping hormonal levels constant. Id. at 43. Therese Collins was born without apparent incident in 1958. But seventeen years later, in 1975, Therese Collins was diagnosed with vaginal cancer. Id. at 41. Therese Collins filed suit against twelve drug companies that allegedly produced or marketed DES. The trial court granted the drug companies' motion for summary judgment because Collins could not prove which specific drug company manufactured the DES that her mother used. Id. at 42.

In reversing the trial court, the Wisconsin Supreme Court identified the problems of proof faced by Collins in trying to prove which specific DES maker caused her injuries. First, DES was produced in generic form, and the drug itself did not have any clearly identifiable characteristics that could distinguish one maker's version of the drug from any other maker. 342 N.W.2d at 44. Second, during the twenty-four-year period that DES was on the market, over 300 companies produced or marketed DES. Id. Third, many drug companies did not have access to accurate records as to where, when, and what type of DES they produced or marketed. Id. In light of these proof problems, the Wisconsin Supreme Court observed that the choice it faced was either to fashion a novel method for recovery for DES plaintiffs, or to permit possibly negligent defendants to escape liability. Id. at 45. The state high court refused to allow liability to go unaddressed, relying on Article I, Section 9 of the Wisconsin Constitution to make the choice. That section provides that "every person is entitled to a certain remedy in the laws for all injuries, or wrongs, which he may receive in his person, property, or character." Id. at 45.

In deciding what form the remedy would take, the Wisconsin Supreme Court chose to fashion the risk contribution theory of liability, and rejected other potential theories. First among the rejected theories was the "alternative" liability theory embodied in Summers v. Tice, 33 Cal.2d 80, 199 P.2d 1, 4-5 (1948). In Summers, there were only two potentially liable defendants, but for DES cases, there were hundreds of drug makers that might be liable. Therefore, alternative liability would not be a fair way to apportion damages among the defendants. Collins, 342 N.W.2d at 46.

The Wisconsin Supreme Court also rejected enterprise liability, which is a framework that allows a plaintiff to hold defendants liable for industry-wide practices that created a risk of harm. 342 N.W.2d at 47. The state high court observed that DES manufacturers did not jointly control the risk of injury to plaintiffs because so many different companies entered and exited the market over twenty-four years. Id. Collins also rejected the plaintiff's theory that the drug companies conspired to misrepresent DES's safety. Id. at 47-48.

Finally, Collins decided not to adopt, in its entirety, the market share theory adopted by the California Supreme Court in Sindell v. Abbott Labs., 26 Cal.3d 588, 613, 163 Cal.Rptr. 132, 607 P.2d 924, 937 (1980). Sindell also involved DES, and the California Supreme Court reasoned that it was fair to shift the burden of causation to the defendants. Based on the market share theory, the defendants would be liable for the percentage of damages that approximated their share of the market. Collins, 342 N.W.2d at 48. The Wisconsin Supreme Court rejected the market-share theory, however, because of the practical difficulty of defining and proving market share. Id. But in rejecting market-share liability, Collins still noted that market share, if it could be determined, is a relevant "factor" in deciding how to apportion liability among defendants. Id. at 49.

After rejecting these other theories of liability, Collins adopted the "risk contribution" theory of liability. The premise of this form of liability is that each defendant "contributed to the risk of injury to the public and consequently, the risk of injury to individual plaintiffs." 342 N.W.2d at 49. The Wisconsin Supreme Court explained that it was better for drug companies to share the cost of injury than to place the entire burden on the innocent plaintiff. So instead of having to prove that a particular defendant produced or marketed the DES taken by her mother, Collins simply had to prove — by a preponderance of the evidence — that the defendant produced or marketed the "type" of DES taken by her mother, "type" meaning the color, shape, markings, size, or other characteristics of the DES. 342 N.W.2d at 50. Practically speaking, that burden of proof would not narrow the possible defendants by much, because "DES was, for the most part, produced in a `generic' form which did not contain any clearly identifiable shape, color, or markings," Id. at 37. Collins would not have to prove any facts related to the time period during which the defendant made or marketed DES, nor what the geographic area was in which the defendant distributed DES. Id. at 50. On the issues of time period of distribution and geographic area of distribution, Collins decided that "it is appropriate to shift the burden of proof on time and geographic distribution to the defendant drug companies because they will have better access to relevant records than the plaintiff." Id. at 53. But even if relevant records did not exist any longer, the Wisconsin Supreme Court opined, "we believe that the equities of DES cases favor placing the consequences on the defendants." Id. at 53.

C.

This brings us to the heart of this appeal, the Wisconsin Supreme Court's extension of Collins's risk-contribution theory of liability to white lead carbonate pigment cases, as held by Thomas v. Mallett, 285 Wis.2d 236, 701 N.W.2d 523 (2005). Thomas compared DES cases with white lead carbonate pigment cases, and concluded, over two dissenting opinions, that the "main policy reasons identified in Collins warrant extension of the risk-contribution theory here." Id. at 558. Primary among those reasons was, as the Wisconsin Supreme Court put it, the widespread health problem posed by white lead carbonate poisoning, a problem so significant that Thomas described it as "a public health catastrophe that is poised to linger for quite some time." Id.

Thomas went on to explain that the blame for this public-health problem is on the defendants, each of which contributed to the risk of injury. 701 N.W.2d at 558. Indeed, the blame was deeper than negligence: "Many of the individual defendants or their predecessors-in-interest did more than simply contribute to a risk; they knew of the harm white lead carbonate pigments caused and continued production and promotion of the pigment notwithstanding that knowledge." Id. In addition to the culpability of the defendants and the innocence of the plaintiff, Thomas also relied on the view that the defendants are "in a better position to absorb the cost of the injury," because the defendants "can insure themselves against liability, absorb the damage award, or pass the cost along to the consuming public as a cost of doing business." Id.

In extending the risk contribution theory of liability to white lead carbonate pigment, Thomas also rejected the manufacturers' attempts to distinguish lead pigment from DES. First, the manufacturers argued that plaintiffs in white lead carbonate pigment cases do have alternative remedies, because plaintiffs can sue their landlords, so it is unnecessary to apply risk contribution liability. In contrast, the manufacturers contended, DES plaintiffs had no other remedies. The Wisconsin Supreme Court disagreed, rejecting the manufacturers' argument that suing landlords would provide an adequate remedy. 701 N.W.2d at 552. Thomas explained that the landlords' insurers could rely on a "pollution exclusion" in commercial general liability insurance policies to avoid a duty to indemnify landlords. Id. So, although Thomas himself had been able to obtain a settlement from two of the landlord's insurers, other plaintiffs might not be so lucky. Id. at 552-53. Moreover, under a Wisconsin statute (which has since been repealed), landlords could immunize themselves from liability if they had received a certificate (from a certified lead-risk assessor) that the housing units were lead-free. Id. at 553. And, in any event, it did not matter that plaintiffs could seek remedies against other wrongdoers, such as landlords, because those remedies did not absolve other wrongdoers, such as the manufacturers, from liability. Id. at 553-54.

In addition to the alternative-remedies argument, white lead carbonate pigment makers also tried to distinguish their product from DES by pointing out that white lead carbonate came in three different chemical formulas, whereas DES was a fungible drug produced with a chemically identical formula. 701 N.W.2d at 559. But that distinction did not make a difference, Thomas concluded. Even though there are three different chemical compositions, the bottom line was that all forms of white lead carbonate pigment contained an inherently hazardous element: lead. Id. at 550-60. What's more, the various forms of white lead carbonate pigment still all performed the same function, were physically indistinguishable, and presented the same risk to health. Id. at 560-61. Thomas concluded that the different forms of white lead carbonate pigment were sufficiently similar to warrant the same risk-contribution treatment.

Thomas also rejected the manufacturers' argument that, unlike the nine-month pregnancy period in DES cases, the time period during which the white lead carbonate pigment could have been applied was, in some cases, on the order of decades. The plaintiff in Thomas lived in houses built in 1900 and 1905, so the white lead carbonate pigment could have been applied any time between then and the 1978 lead-paint ban. 701 N.W.2d at 562. The Wisconsin Supreme Court acknowledged that the time period was "drastically larger" than the nine-month window in DES cases. Id. In response, however, Thomas reasoned that "the window will not always be potentially as large as appears in this case," but even if the time window would "routinely" be that long, "the Pigment Manufacturers' argument must be put into perspective: they are essentially arguing that their negligent conduct should be excused because they got away with it for too long." Id. Ultimately, Thomas again invoked the "equities" of the situation, and concluded that the time window, although potentially long, did not justify putting the causation burden back on the innocent plaintiff. Id. at 563.

The next unsuccessful attempt to distinguish white lead carbonate pigment from DES was the lack of a "signature" injury arising from white lead carbonate pigment. The Wisconsin Supreme Court acknowledged that the records showed that lead poisoning could be caused by many different sources, including "ambient air, many foods, drinking water, soil, and dust." 701 N.W.2d at 563. And the injuries themselves (cognitive defects) could have causes other than lead poisoning, such as genetics or complications during birth. Id. In rejecting this purported distinction, Thomas reasoned that, "Harm is harm, whether `signature' or otherwise." Id. The important thing is that a white lead carbonate pigment plaintiff still must prove that the pigment caused his or her injuries. Id. "[T]hat merely means that Thomas may have a harder case to make to his jury." Id. It did not mean, Thomas held, that risk-contribution theory should not apply. Id.

The final attempt by the pigment manufacturers to distinguish DES cases also failed. Specifically, the pigment manufacturers argued that they did not have exclusive control of the risk posed by their white lead carbonate pigment; for example, paint manufacturers took control of the pigment when making paint. But Thomas responded that the level of control over the product was no different from DES cases, where "doctors were the ones who prescribed the dosage of DES" or pharmacists filled prescriptions. 701 N.W.2d at 563. And, in any event, the paint manufacturers' exertion of control over white lead carbonate pigment diluted (if it did anything) the toxicity of the white lead carbonate from the time it left the pigment manufacturers' hands. Id. Worse, the manufacturers "actually magnified the risk through their aggressive promotion of white lead carbonate, even despite the awareness of the toxicity of the lead." Id. at 564. Thomas concluded that the level of control over white lead carbonate pigment made no difference. Id.

With all the proffered differences from DES cases rejected,5 Thomas extended the risk-contribution theory of liability to white lead carbonate pigment cases. That means, for negligence claims, that the plaintiff must prove duty, breach of duty, and injury caused by white lead carbonate ingestion, but with regard to imposing liability on a particular manufacturer, the plaintiff "need only prove that the Pigment Manufacturers produced or marketed white lead carbonate for use during the relevant time period: the duration of the houses' existence." 701 N.W.2d at 564. For strict liability claims, the identification of the manufacturer on which liability can be imposed is proven by showing "[t]hat the pigment manufacturer engaged in the business of producing or marketing white lead carbonate or, put negatively, that this is not an isolated or infrequent transaction not related to the principal business of the pigment manufacturer." Id.

As Thomas describes it, the actual implementation of the risk-contribution theory comprises the following: the plaintiff makes a prima facie case for either a negligence or strict liability claim (or both), and then "the burden of proof shifts to each defendant[-manufacturer] to prove by a preponderance of the evidence that it did not produce or market white lead carbonate either during the relevant time period or in the geographical market where the house is located." 701 N.W.2d at 564. Thomas also specifically instructs trial courts what to do if there are no records (or no longer any records) to prove the defense: "if relevant records do not exist that can substantiate either defense, we believe that the equities of [white lead carbonate] cases favor placing the consequences on the [Pigment Manufacturers]." Id. (internal quotation omitted).

With risk-contribution theory in place for white lead carbonate pigment claims, the Wisconsin Supreme Court remanded the case to the trial court for a jury trial. At the trial, the jury decided that Thomas had failed to prove that his injuries were caused by white lead carbonate pigment, so the jury did not end up applying the risk contribution theory. Thomas v. Mallett, 331 Wis.2d 486, 795 N.W.2d 62 (Wisc.Ct.App.2010) (unpublished order).

D.

In considering the manufacturers' constitutional challenges against applying risk-contribution theory to white lead carbonate pigment cases, we start with the proposition that the federal Constitution gives a wide berth to state (and local) laws, allowing state legislatures to enact laws unless a specific constitutional bar prevents it. In this case, the manufacturers' primary challenge to risk contribution theory is that it violates the substantive component of the Due Process Clause, and this is the argument with which the district court agreed.

Generally speaking, state laws need only be rational and non-arbitrary in order to satisfy the right to substantive due process. Usery v. Turner Elkhorn Mining Co., 428 U.S. 1, 15, 96 S.Ct. 2882, 49 L.Ed.2d 752 (1976); see also Pension Ben. Guar. Corp. v. R.A. Gray & Co., 467 U.S. 717, 730, 104 S.Ct. 2709, 81 L.Ed.2d 601 (1984). The reason for this deference is that other parts of the Constitution contain more specific guarantees of rights, and judicial self-restraint requires caution when invoking the "more generalized notion of `substantive due process.'" See Graham v. Connor, 490 U.S. 386, 395, 109 S.Ct. 1865, 104 L.Ed.2d 443 (1989). "As a general matter, the [Supreme] Court has always been reluctant to expand the concept of substantive due process because guideposts for responsible decisionmaking in this unchartered area are scarce and open-ended." Collins v. Harker Heights, 503 U.S. 115, 125, 112 S.Ct. 1061, 117 L.Ed.2d 261 (1992) (citing Regents of Univ. of Mich. v. Ewing, 474 U.S. 214, 225-26, 106 S.Ct. 507, 88 L.Ed.2d 523 (1985)). Substantive due process protections "have for the most part been accorded to matters relating to marriage, family, procreation, and the right to bodily integrity." Albright v. Oliver, 510 U.S. 266, 272, 114 S.Ct. 807, 127 L.Ed.2d 114 (1994) (plurality opinion).

Of course, none of those concerns are at stake in risk-contribution theory, but the manufacturers argue that the theory still violates substantive due process. In particular, the manufacturers argue that the district court correctly held that a combination of the United States Supreme Court's plurality, concurring, and dissenting opinions in Eastern Enterprises v. Apfel, 524 U.S. 498, 118 S.Ct. 2131, 141 L.Ed.2d 451 (1998), dictates a substantive due process analysis that renders risk-contribution theory unconstitutional. Relying on the combined opinions, the district court concluded that risk-contribution theory "imposes severe retroactive liability on a limited class of parties that could not have anticipated the liability, and the extent of that liability is substantially disproportionate to the parties' experience." R. 39 at 29 (quoting Eastern Enterprises, 524 U.S. at 528-29, 118 S.Ct. 2131 (plurality opinion)).

Eastern Enterprises, however, did not produce a binding precedent (other than its specific result) because no controlling principle can be gleaned from the plurality, concurrence (which was a concurrence in the judgment only), and the dissenting opinions. In order to understand why Eastern Enterprises cannot be said to have produced binding precedent, it is necessary to examine the opinions in detail and the history and context of the federal statutory scheme at issue there. It starts in 1946, when the United Mine Workers of America and various coal companies reached an agreement that led to the creation of other funds that would provide health benefits, survivors' benefits, and pension-type benefits for miners and their dependents. Eastern Enterprises, 524 U.S. at 505, 118 S.Ct. 2131. These funds served as the basis for the creation of funds in later years that operated as trusts: a portion of coal-production profits were placed into the funds, which would provide benefits to miners and their families. Id. But no specific amount of benefits was promised by the funds. Id. In 1950, the United Mine Workers and the coal operators entered into an agreement that increased the royalty payments into another fund (which again took the form of a trust). Id. at 506, 118 S.Ct. 2131. But, again, the fund did not promise miners and their dependents a specific amount of benefits. Id. at 506, 507, 118 S.Ct. 2131. This included no promise as to providing lifetime health benefits for coal miners and their dependents. Id. at 508, 118 S.Ct. 2131.

In 1974, the Employment Retirement Income Security Act (ERISA) created new requirements for pension plans, and to comply with ERISA, the United Mine Workers and the Bituminous Coal Operators Association entered into an agreement. Id. at 509, 118 S.Ct. 2131. The new agreement stated that miners who retired before 1976 would be covered by the 1950 plan, and miners who retired after 1975 would be covered by the 1974 plan. Id.

Soon, with declining coal operator profits, along with the acceleration of health care costs, both the 1950 and 1974 plans ran into financial trouble. Id. at 510, 118 S.Ct. 2131. Coal companies began to withdraw from the plans, leaving the remaining employers to absorb the increasing cost of covering retirees. Id. In an attempt to deal with the problem, eventually Congress passed the Coal Act in 1992. Id. at 514, 118 S.Ct. 2131. The Coal Act combined the 1950 and 1974 plans (the "Combined Fund"), and provided substantially the same health benefits to retirees and their dependents as they were receiving under the prior plans. The Combined Fund was financed by "annual premiums assessed against `signatory coal operators,' i.e., coal operators that signed any [National Bituminous Coal Wage Agreement] [NBCWA] or any other agreement requiring contributions to the 1950 or 1974 Benefit Plans." Id. Any signatory operator who "conducts or derives revenue from any business activity, whether or not in the coal industry," could be liable for those premiums. Id. (internal quotation omitted). Beyond signatory operators themselves, where a signatory was no longer involved in any business activity, premiums could be levied against "related persons," including successors-in-interest and businesses or corporations under common control. Id. The Act gave the Commissioner of Social Security the duty to assign retirees to employers who would be responsible for paying for the retirees' benefits. Id. at 515, 118 S.Ct. 2131.

Under the Coal Act, the Social Security Commissioner assigned Eastern Enterprises an obligation for premiums covering over 1,000 retired miners who had worked for the company before 1966. But Eastern had, back in 1965, transferred all of its coal-related operations to a subsidiary. At that time (more precisely, in 1966), the 1950 fund had a positive balance topping $145 million. In 1987, Eastern had sold its ownership interest in the subsidiary to another corporation. This was five years before the passage of the Coal Act, and the two not-yet-combined funds still had a total positive balance of over $33 million.

After Eastern was assigned the 1,000+ retired miners, which represented more than a $5 million liability to the Combined Fund, Eastern filed suit, arguing that the Coal Act violated substantive due process (as applied to Eastern) and was a "taking" that violated the Fifth Amendment's Takings Clause. The Supreme Court held that this aspect of the Coal Act was unconstitutional as applied to Eastern, but the four-Justice plurality opinion and the one-Justice concurring-in-the-judgment opinion invoked different grounds for the decision. The four dissenters would have upheld the constitutionality of the Coal Act.

According to the four-Justice plurality opinion, the Coal Act was an unconstitutional taking of Eastern's property in violation of the Fifth Amendment; the plurality decided not to reach the substantive due process issue. Specifically, the plurality, in an opinion authored by Justice O'Connor, analogized the Coal Act's assignment to an economic regulation that amounted to a taking. Eastern Enterprises, 524 U.S. at 522-23, 118 S.Ct. 2131. To be sure, "Congress had considerable leeway to fashion economic legislation," including "impos[ing] retroactive liability to some degree." Id. at 528, 118 S.Ct. 2131. But, the plurality explained, economic legislation can amount to an unconstitutional taking "if it imposes severe retroactive liability on a limited class of parties that could not have anticipated the liability, and the extent of that liability is substantially disproportionate to the parties' experience." Id. at 528-29, 118 S.Ct. 2131.

The plurality explained that the Coal Act did amount to an unconstitutional taking. The assignment of liability to Eastern was retroactive because it "substantially interfere[d] with Eastern's reasonable investment-backed expectations" by reaching back "30 to 50 years to impose liability... based on the company's activities between 1946 and 1965." 524 U.S. at 532, 118 S.Ct. 2131. The Coal Act operated "retroactively, divesting Eastern of property long after the company believed its liabilities under the 1950 W & R Fund to have been settled." Id. at 534, 118 S.Ct. 2131. Indeed, the 1974 fund had not even been created at the time that Eastern transferred its coal-related operations to a subsidiary.

According to the plurality, the assignment of liability was not only retroactive, it would be severe. Although the parties provided different estimates for what Eastern's total payments would be under the Act, the range was between $50 to $100 million (the $5 million liability was only for the first year of the Coal Act's operation). Id. at 529, 118 S.Ct. 2131. And the severe financial impact was not proportionate to Eastern's experience with the funds. Id. at 529-30, 118 S.Ct. 2131. During the time that Eastern actually employed minors, the benefits were "far less extensive," and indeed there were no promises to pay specific amounts. Id. at 530-31, 118 S.Ct. 2131. Combining the retroactive nature of the liability with the enormous financial impact resulted in substantial interference with Eastern's reasonable investment-backed expectations. Id. at 532-33, 118 S.Ct. 2131.

Finally, the plurality acknowledged that "analysis of legislation under the Takings and Due Process Clauses is correlated to some extent." Id. at 537, 118 S.Ct. 2131. But, in line with prior cases, the plurality expressed hesitation about using the Due Process clause to invalidate economic legislation. Id. Ultimately, the plurality expressly declined to address Eastern's substantive due process claim. Id. at 538, 118 S.Ct. 2131.6

Justice Kennedy provided the fifth vote to invalidate the Coal Act's assignment of liability to Eastern. But his concurrence rejected that a taking had occurred, because no specific property or assets were identified to be taken. Id. at 540, 118 S.Ct. 2131. Justice Kennedy acknowledged that, in some prior cases, economic regulations had such a broad reach that the Court had deemed those regulations to be a taking of property, but always there was a "specific property right or interest... at stake." Id. at 540-41, 118 S.Ct. 2131. In contrast, the Coal Act's assignment did not destroy or take a specific asset or property interest, so it was imprecise to interpret the Coal Act as amounting to a taking. Id. at 541-42, 118 S.Ct. 2131.

Instead, Justice Kennedy reasoned, substantive due process was the correct way to analyze the Coal Act's constitutionality, a test that, in his view, the Act failed as applied to Eastern. Id. at 547, 118 S.Ct. 2131. Like the plurality, Justice Kennedy acknowledged the general hesitancy with scrutinizing economic legislation under the Due Process Clause. Id. But the concurrence stated that retroactive laws had long invoked "a singular distrust," "requir[ing] an inquiry into whether in enacting the retroactive law the legislature acted in an arbitrary and irrational way." Id. One reason for this concern over retroactive laws is the "`tempt[ation] to use retroactive legislation as a means of retribution against unpopular groups or individuals.'" Id. at 548, 118 S.Ct. 2131 (quoting Landgraf v. USI Film Prods., 511 U.S. 244, 266, 114 S.Ct. 1483, 128 L.Ed.2d 229 (1994)). Without due-process scrutiny of retroactive laws, the stability of property ownership would be vulnerable to government action: "If retroactive laws change the legal consequences of transactions long closed, the change can destroy the reasonable certainty and security which are the very objects of property ownership." Id. at 548, 118 S.Ct. 2131.

With regard to the Coal Act, Justice Kennedy concluded that this was one of the "rare instances" where the legislature exceeded due process limits. Id. at 549, 118 S.Ct. 2131. The concurring opinion referred to the plurality opinion's "convincing" demonstration that "in creating liability for events which occurred 35 years ago the Coal Act has a retroactive effect of unprecedented scope." Id. "As the plurality opinion discusses in detail, the expectation was created by promises and agreements made long after Eastern left the coal business. Eastern was not responsible for the resulting chaos in the funding mechanism caused by other coal companies leaving" the prior funding agreement. Id. at 550, 118 S.Ct. 2131. Thus, Justice Kennedy concluded, the Coal Act's assignment of liability to Eastern exceeded even the "permissive standard" of substantive due process. Id. at 550, 118 S.Ct. 2131.

Justice Breyer, writing for the four dissenting Justices, agreed with Justice Kennedy that the Coal Act did not present a takings issue.7 Along the same lines as Justice Kennedy's concurring opinion, the dissent explained that although some economic regulations amount to a taking, those regulations identified specific physical property or specific assets, rather than a general liability. Id. at 555, 118 S.Ct. 2131.

Like Justice Kennedy, the dissent identified the Due Process Clause as the "natural home" for scrutinizing "the potential unfairness of retroactive liability." Id. at 557, 118 S.Ct. 2131. Due Process protects against arbitrary and irrational legislation, and if a law is fundamentally unfair because of its retroactivity, then it is arbitrary. Id. The question presented by the Coal Act is "whether or not it is fundamentally unfair to require Eastern to make future payments for health care costs of retired miners and their families, on the basis of Eastern's past association with those miners." Id. at 558-59, 118 S.Ct. 2131.

The dissent concluded that the Coal Act did not violate due process, because the Act only made assignments of miners whom Eastern had employed in the past. Id. at 560, 118 S.Ct. 2131. Moreover, even though Eastern had not made contractually enforceable promises to the miners, coal companies and the federal government had acted in a way that led the miners to reasonably expect that they would continue to receive medical benefits. Id. at 560-63, 118 S.Ct. 2131. The dissent viewed the historical record as showing that, in reaction to the conduct of the coal companies and the government, the United Mine Workers had acted in a way (by giving layoff and work-force concessions) that was based on assurances of continued benefits. Id. at 563-64, 118 S.Ct. 2131. On top of this, Eastern continued to receive, through its subsidiary, profits from the coal mining industry. Id. at 565-66, 118 S.Ct. 2131. Under these circumstances, it was not fundamentally unfair to impose the liability on Eastern. Id. at 566, 118 S.Ct. 2131.

Returning to our case, the district court here concluded, as noted above, that the opinions in Eastern Enterprises established a substantive due process right that invalidates state law when the law "imposes severe retroactive liability on a limited class of parties that could not have anticipated the liability, and the extent of that liability is substantially disproportionate to the parties' experience." R. 39 at 29 (quoting 524 U.S. at 528-29, 118 S.Ct. 2131 (plurality opinion)). On appeal, the parties debate whether Eastern Enterprises establishes a rule of decision.

Generally put — and more easily stated than applied — when the Supreme Court issues divided opinions with no single opinion commanding a majority, the holding of the case "may be viewed as that position taken by those Members who concurred in the judgments on the narrowest grounds." Marks v. United States, 430 U.S. 188, 193, 97 S.Ct. 990, 51 L.Ed.2d 260 (1977). "When, however, a concurrence that provides the fifth vote necessary to reach a majority does not provide a `common denominator' for the judgment, the Marks rule does not help to resolve the ultimate question." United States v. Heron, 564 F.3d 879, 884 (7th Cir.2009) (collecting cases). This means that Marks applies "only when one opinion is a logical subset of other, broader opinions." King v. Palmer, 950 F.2d 771, 781 (D.C.Cir. 1991) (en banc). "[W]hen it is not possible to discover a single standard that legitimately constitutes the narrowest ground for a decision on that issue, there is then no law of the land because no one standard commands the support of a majority of the Supreme Court." United States v. Alcan Aluminum Corp., 315 F.3d 179, 189 (2d Cir.2003).

There is no narrow-grounds rationale that supplies the rule of decision in Eastern Enterprises. The five Justices "who concurred in the judgments," Marks, 430 U.S. at 193, 97 S.Ct. 990, did not even agree on which constitutional provision applied to the Coal Act to render it invalid. The four-justice plurality based its decision on the Takings Clause, whereas Justice Kennedy — who concurred in the judgment only and was the necessary fifth vote for the case's result — concluded that the Coal Act violated substantive due process. Unlike Marks, which examined a prior set of opinions where at least the same constitutional provision was the basis for the rule of decision (the First Amendment), in Eastern Enterprises the plurality and the concurrence were not even interpreting the same constitutional right, so neither one of those opinions could be said to be the narrower of the other. Asking which opinion is narrower than the other would be like examining a square with a width that is the same length as the diameter of a circle, and futilely asking which is narrower, the square or the circle.

It is true that, at times, the plurality opinion and Justice Kennedy's concurring opinion refer to one another in a way that suggests some level of overlap in their respective analysis. 524 U.S. at 537, 118 S.Ct. 2131 (plurality opinion) (the "analysis of legislation under the Takings and Due Process Clauses is correlated to some extent"); id. at 549, 118 S.Ct. 2131 (Kennedy, J., concurring) ("The plurality opinion demonstrates in convincing fashion that the remedy created by the Coal Act bears no legitimate relation to the interest which the Government asserts in support of the statute."); id. at 530, 118 S.Ct. 2131 (Kennedy, J., concurring) ("As the plurality opinion discusses in detail, the expectation was created by promises and agreements made long after Eastern left the coal business."). But it is not possible to take these isolated references and translate the plurality opinion's Takings Clause analysis into Justice Kennedy's substantive due process analysis, and vice-versa. The opinions themselves do not purport to decode how the respective analyses relate to one another. The plurality opinion declined to address the substantive due process argument. And the ability to compare the two opinions for Marks purposes is undermined even more by Justice Kennedy's concurring opinion, which expressly rejected the plurality's reasoning and concluded that the Takings Clause did not apply at all to the Coal Act. In Justice Kennedy's view, no taking occurs if no "specific property interest" is invaded. Id. at 543, 118 S.Ct. 2131. Without that limitation on the Takings Clause, the "plurality opinion would throw one of the most difficult and litigated areas of the law into confusion...." Id. at 542, 118 S.Ct. 2131. The concurring opinion did not try to fit the substantive due process analysis into the "already difficult and uncertain" jurisprudence on regulatory takings. Id.

In light of the different provisions and different approaches of the plurality opinion and Justice Kennedy's opinion, neither can be characterized as narrower or broader than the other. Our colleagues in other Circuits agree that no governing holding emerged from Eastern Enterprises. See Alcan, 315 F.3d at 189; A.T. Massey Coal Co. v. Massanari, 305 F.3d 226, 240-41 (4th Cir.2002); Anker Energy Corp. v. Consolidation Coal Co., 177 F.3d 161, 172 (3d Cir.1999); Association of Bituminous Contractors, Inc. v. Apfel, 156 F.3d 1246, 1256 (D.C.Cir.1998). The specific result of Eastern Enterprises — the Coal Act's unconstitutionality, on a combination of the plurality and the concurrence's votes — is the only binding precedent that arises from the case.

In deciding that Eastern Enterprises articulated a governing substantive due process standard applicable to risk-contribution theory, the district court reasoned that Justice Kennedy and the four dissenting Justices agreed to apply substantive due process to the Coal Act, and they combined for a majority of the Court on that proposition. But the problem with that approach is that Marks itself instructs that "the holding is the narrowest position taken by those members who concurred in the judgment." 430 U.S. at 193, 97 S.Ct. 990 (emphasis added). So, under Marks, the positions of those Justices who dissented from the judgment are not counted in trying to discern a governing holding from divided opinions. It makes sense to exclude the dissenting opinions: by definition, the dissenters have disagreed with both the plurality and any concurring Justice on the outcome of the case, so by definition, the dissenters have disagreed with the plurality and the concurrence on how the governing standard applies to the facts and issues at hand (even if there is agreement on what constitutional provision is being interpreted). It is very likely that if the dissenters disagree with the outcome of the case, then lower courts and (more importantly) litigants will not have a clear idea on the contours of the standard and how to apply it in future cases. This is not the way to make binding precedent.

Eastern Enterprises is itself an example of the difficulty in combining a concurring opinion and a dissenting opinion to arrive at binding precedent. Justice Kennedy reasoned that Eastern Enterprises was "not responsible for their [the former miners and their beneficiaries] expectation of lifetime health benefits or for the perilous financial condition" of the benefits plans. Id. at 550, 118 S.Ct. 2131. But the dissenting opinion concluded otherwise, relying on the significance of the employer-employee relationship and, more importantly, on the statements and conduct of Eastern Enterprises, the coal industry, and even the federal government, in creating an expectation of lifetime benefits. Id. at 560-64, 118 S.Ct. 2131. So while both Justice Kennedy's concurrence and the dissenting opinion applied the Due Process Clause, their application of substantive due process was starkly different and provides little guidance for future applications to future cases. Eastern Enterprises demonstrates why dissenting opinions cannot be counted under Marks to create binding precedent.8

It is possible to argue that, in two prior cases, we have at least made reference to dissenting opinions when discussing Marks. See United States v. Gerke Excavating, Inc., 464 F.3d 723, 724-25 (7th Cir.2006); United States v. Hodge, 558 F.3d 630, 634 (7th Cir.2009). Of course, Marks itself is binding on us, and instructs that only those positions of the Justices concurring in the outcome count in the analysis. And, in any event, in neither of our prior cases was inclusion of the dissenting opinion necessary to the outcome of the appeal; in other words, the references to the dissenting opinion were dicta. In Gerke Excavating, we examined a Supreme Court opinion, Rapanos v. United States, 547 U.S. 715, 126 S.Ct. 2208, 165 L.Ed.2d 159 (2006), that had been decided on a four-Justice plurality opinion, a concurring opinion by Justice Kennedy, and a four-Justice dissent. We observed that Justice Kennedy's standard for testing the scope of federal authority over wetlands under the Clean Water Act was narrower that the plurality's (in the sense that it imposed greater restrictions on federal authority) in most cases, so we went ahead and applied the rationale in Justice Kennedy's concurrence. Gerke Excavating, 464 F.3d at 724-25. It is true that we made the same narrower-grounds point in comparing the concurrence with the dissenting opinion, id. at 725, but that comparison was not necessary to resolving the appeal, so it was dicta.

The same is true with the mention of a dissenting opinion in United States v. Hodge. There, we discussed the potential combination of a one-Justice concurrence and four-Justice dissent in United States v. Santos, 553 U.S. 507, 128 S.Ct. 2020, 170 L.Ed.2d 912 (2008). But there was no need to decide what impact the combination would have, if anything, on the appeal because the government conceded what standard should control. 558 F.3d at 633-34. We therefore expressly declined to decide the issue. Id. at 633 ("Whether the concession was appropriate is a difficult question, which we need not answer....") So the discussion of the dissenting opinion was dicta. The bottom line is that neither Gerke Excavating nor Hodge provides support for counting dissenting opinions in a Marks analysis.

Without a controlling test from Eastern Enterprises, then, we are back to where we started: economic legislation does not violate substantive due process unless the law is arbitrary and irrational. The question is not whether a law is wise or not; we test only whether the law is arbitrary or irrational. As put by the Supreme Court:

Turner Elkhorn Mining, 428 U.S. at 15, 96 S.Ct. 2882; see also Goodpaster v. City of Indianapolis, 736 F.3d 1060, 1071 (7th Cir. 2013). This rational-basis review applies "even though the effect of the legislation is to impose a new duty or liability based on past acts." Concrete Pipe and Prods. of Cal., Inc. v. Construction Laborers Pension Trust for S. Cal., 508 U.S. 602, 637, 113 S.Ct. 2264, 124 L.Ed.2d 539 (1993) (quoting Pension Ben. Guar. Corp. v. R.A. Gray & Co., 467 U.S. 717, 730, 104 S.Ct. 2709, 81 L.Ed.2d 601 (1984)). Legislation "is not unlawful solely because it upsets otherwise settled expectations." Concrete Pipe and Prods., 508 U.S. at 637, 113 S.Ct. 2264 (quoting Gray, 467 U.S. at 729, 104 S.Ct. 2709).

The manufacturers primarily argue that two characteristics of risk-contribution theory render it arbitrary and irrational: first, that the theory dispenses with the traditional tort requirement that the plaintiff prove that the defendant caused the injury at issue; and second, that the theory imposes liability retroactively, that is, the rule of liability has changed after the defendants engaged in the conduct that is the subject of the suit.

On the latter point, although "retroactive legislation does have to meet a burden not faced by legislation that has only future effects," Gray, 467 U.S. at 730, 104 S.Ct. 2709, "that burden is met simply by showing that the retroactive application of the legislation is itself justified by a rational legislative purpose," id. Indeed, while we have been setting out the deferential standard for reviewing state legislation, even more deference is owed to judicial common-law developments, which by their nature must operate retroactively on the parties in the case.

The development of state common law is a fundamental feature of our legal system. And, in turn, "the foundation of the common law system" is "the incremental and reasoned development of precedent." See Rogers v. Tennessee, 532 U.S. 451, 461, 121 S.Ct. 1693, 149 L.Ed.2d 697 (2001). If strict constraints on retroactivity applied to state-court common-law decisions, then the development of common law would be impaired. Rogers explained this point in the context of deciding whether to extend Ex Post Facto Clause protection (which applies against legislatures only) through the Due Process Clause against the courts. Rogers was a criminal defendant whose victim died 15 months after Rogers stabbed him. Rogers was convicted of murder in Tennessee state court, despite the fact that, at the time of his trial, Tennessee common law held that no defendant could be convicted of murder if his victim died more than a year and a day after the fatal act. 532 U.S. at 455, 121 S.Ct. 1693. On appeal, the Tennessee Supreme Court acknowledged that the year-and-a-day rule was indeed part of the state's common law, but the state high court abolished the rule in Rogers's appeal and affirmed his conviction. In Rogers's appeal to the United States Supreme Court, he argued that Ex Post Facto Clause protection should apply against state common-law decision-making.

Even though the direct question presented in Rogers was whether to incorporate the Ex Post Facto Clause into the Due Process Clause, which is not the issue in this case, the rationale of Rogers helps in evaluating the retroactivity concerns raised by the pigment manufacturers here. There are indeed Due Process limits on the retroactive application of a judicial decision, but only if the judicial decision "is unexpected and indefensible by reference to the law which had been expressed prior to the conduct in issue." 532 U.S. at 457, 462, 121 S.Ct. 1693 (quoting Bouie v. City of Columbia, 378 U.S. 347, 352, 84 S.Ct. 1697, 12 L.Ed.2d 894 (1964)). Rogers declined to apply to courts the same ex post facto standard applicable to legislatures, because of the lesser danger presented by judicial interpretations and the need to allow for common-law developments. 532 U.S. at 461-62, 121 S.Ct. 1693. On the point about danger, a "court's opportunity for discrimination ... is more limited than a legislature's in that it can only act in construing existing law in actual litigation." 532 U.S. at 460-61, 121 S.Ct. 1693 (quoting James v. United States, 366 U.S. 213, 247 n. 3, 81 S.Ct. 1052, 6 L.Ed.2d 246 (1961)) (Harlan, J., concurring in part and dissenting in part). Unlike legislatures, which have the general freedom to explore whatever subject area at whatever time, courts can only decide issues in cases brought them to by litigants.

On the second point — the breathing space that commonlaw development requires — Rogers explained that "[i]n the context of common law doctrines ..., there often arises a need to clarify or even to reevaluate prior opinions as new circumstances and fact patterns present themselves." 532 U.S. at 461, 121 S.Ct. 1693. The need to adjust the common law as new cases are presented is the reason why common law courts are granted "substantial leeway ... [in] reevaluating and refining [doctrines] as may be necessary to bring the common law into conformity with logic and common sense." Id. To wrap any greater straitjacket on common-law development "would place an unworkable and unacceptable restraint on normal judicial processes and would be incompatible with the resolution of uncertainty that marks any evolving legal system." Id. at 461, 121 S.Ct. 1693. Thus, judicial decisions that retroactively change the common law do not violate due process unless they are "unexpected and indefensible by reference to the law which has been expressed prior to the conduct in issue." Id. at 462, 121 S.Ct. 1693.9

With these principles in mind, we conclude that risk-contribution theory is not arbitrary and irrational, nor is it unexpected and indefensible. In developing the common-law torts of negligence and strict liability by adopting risk-contribution theory, the Wisconsin Supreme Court balanced the tortious conduct of pigment manufacturers in distributing an unreasonably dangerous product with the possibility of leaving the non-culpable plaintiff without a sufficient remedy, while recognizing that the state high court was relaxing the traditional standard of causation. Thomas, 701 N.W.2d at 558. Thomas rationally relied on the wide scope of the health dangers posed by white carbonate lead pigment. The lead poisoning caused by the pigment is not only widespread in terms of the number of individuals affected, but just as problematic, in the Wisconsin Supreme Court's view, is the ongoing exposure to lead pigment that would continue to cause injuries in the foreseeable future. Id. (describing the hazard as "a public health catastrophe that is poised to linger for quite some time"). At the same time, victims of pigment poisoning face difficult problems of proof, in part because the pigment was so unreasonably dangerous that it remains a health danger even decades later.

To address the problem of compensating victims and the problem of proof, neither problem of which could be blamed on plaintiffs in pigment cases, Thomas extended risk-contribution theory to the pigment manufacturers, each of which contributed to the risk of injury, either directly or via their predecessors-in-interest. 701 N.W.2d at 558. The manufacturers either knew or should have known of the harm that they were causing, so culpability was laid at the feet of the manufacturers. Id. Relaxing the standard of causation was justified in favor of the innocent plaintiff and against the risk-creating manufacturers. Id. In addition to the culpability of the manufacturers and the innocence of the plaintiff, Thomas also reasoned that the manufacturers are "in a better position to absorb the cost of the injury," because they "can insure themselves against liability, absorb the damage award, or pass the cost along to the consuming public as a cost of doing business." Id. In sum, the Wisconsin Supreme Court rationally concluded that, under the state's common law, "it is better to have the Pigment Manufacturers or consumers share the cost of injury rather than place the burden on the innocent plaintiff." Id. There is nothing irrational about developing the state's common law to prevent the manufacturers from avoiding liability for injuries caused by risks to which they contributed.

It is important to understand that, in fashioning risk-contribution theory and relaxing the traditional cause-in-fact requirement, Thomas did not entirely eliminate causation. In order to invoke the risk-contribution theory against a particular manufacturer, the plaintiff still must "prove that the Pigment Manufacturers produced or marketed white lead carbonate for use during the relevant time period: the duration of the houses' existence." 701 N.W.2d at 564. And even under the relaxed causation-in-fact standard of risk-contribution theory, liability is far from automatic: the plaintiff still must prove that white carbonate lead pigment was the cause of the lead poisoning. This poses a substantial causation question because there are other sources of lead poisoning (such as the ambient air, drinking water, soil, and dust), and there is no "signature" injury for lead poisoning specifically from white carbonate lead pigment, 701 N.W.2d at 563. In other words, causation is not entirely eliminated: plaintiffs still must prove that lead pigment caused their injuries, and only then do the manufacturers face liability for having contributed to the risk.

Despite the pigment manufacturers' argument that they will be held liable in particular cases for injuries that they did not cause, what risk-contribution theory does is reflect the overall liability that the manufacturers should have expected to face from selling lead pigment. In other mass-tort contexts, similar tort-liability theories reflect the same overall compensation framework: "Assuming every injured person will sue, looking at the total number of successful claims, each defendant will, at least theoretically, only be held responsible for that part of the damage that it caused to the community." In re Agent Orange Product Liab. Litig., 597 F.Supp. 740, 823 (E.D.N.Y.1984). Put another way, if for example Sherwin-Williams ends up paying for harm it did not cause in a particular case brought by a particular plaintiff, it will also end up paying less than it should in the next case — where it did cause the harm — when another manufacturer is also found liable for harm caused by Sherwin-Williams.

This reflection of overall liability is consistent with other common-law developments in tort schemes where causation-in-fact is not required for recovery and liability is instead premised in some way on the defendants' contribution to the risk of injury. Whether the liability arises from a simple two-defendant scenario of alternative liability, Summers v. Tice, 33 Cal.2d 80, 199 P.2d 1, 4-5 (1948), or from a multi-defendant market-share approach, Sindell v. Abbott Labs., 26 Cal.3d 588, 613, 163 Cal.Rptr. 132, 607 P.2d 924, 937 (Cal.1980), Hymowitz v. Eli Lilly & Co., 73 N.Y.2d 487, 541 N.Y.S.2d 941, 539 N.E.2d 1069, 1078 (1989), or from Thomas's risk-contribution theory, all of these theories dispatch with the requirement that the plaintiff prove which particular defendant harmed the plaintiff in a particular case, and all permit tortfeasors to be held liable to plaintiffs who they did not actually injure or to be held liable for injuries that they did not cause. See also Conley v. Boyle Drug Co., 570 So.2d 275, 286 (Fla. 1990) (DES); Ray v. Cutter Labs., 754 F.Supp. 193, 195 (M.D.Fla.1991) (blood-clotting product, Factor VIII); Smith v. Cutter Biological, Inc., 72 Haw. 416, 823 P.2d 717, 728 (1991) (Factor VIII); Abel v. Eli Lilly & Co., 418 Mich. 311, 343 N.W.2d 164, 173 (1984) (DES); Martin v. Abbott Labs., 102 Wn.2d 581, 689 P.2d 368, 382-83 (1984)(DES). To be sure, most states for most types of claims continue to apply a strict causation-in-fact requirement, but that does not mean that those states that have chosen to develop their common law to permit recovery on a theory of culpable contribution to the risk of injury have made an irrational or arbitrary choice.

One final point on the Due Process challenge to Thomas. The Wisconsin Supreme Court's decision was not an "unexpected and indefensible" break from Wisconsin's prior common law. As discussed earlier, Thomas's foundation in Wisconsin common law was Collins v. Eli Lilly Co., 116 Wis.2d 166, 342 N.W.2d 37, 52 (1984), which applied risk-contribution theory to DES cases. By the time that Thomas was decided, Collins had been part of the state's common law for twenty years. The Wisconsin Supreme Court applied the same rationale in Collins as in Thomas, recognizing for DES cases the same balancing between the culpable set of defendants and the innocent plaintiff:

Collins, 342 N.W.2d at 52. Operating from this same premise, Thomas rationally rejected the pigment manufacturers' attempts to distinguish lead pigment from DES for purposes of applying risk-contribution theory, as we detailed above. See 701 N.W.2d at 552 (rejecting distinction based on purported existence of other remedies, because those remedies are limited at best and do not absolve the manufacturers); id. at 559-561 (rejecting distinction based on three chemical compositions for lead pigment versus one for DES, because the lead pigment was still physically indistinguishable); id. at 562 (rejecting distinction based on time period of exposure, because expansive time period reflected culpability); id. at 563 (rejecting distinction based on other potential sources of lead poisoning, because the plaintiff still must prove lead-pigment as the source of injury); id. at 563 (rejecting distinction based on paint makers as intervening actor, because doctors were involved in distributing DES). In light of the "substantial leeway" given to state courts to develop the common law, see Rogers, 532 U.S. at 461, 121 S.Ct. 1693, taking the step from Collins to Thomas was reasonably expected under Wisconsin law. Thomas satisfies the test of substantive due process.

E.

In light of our conclusion that the manufacturers' substantive-due-process challenge to Thomas must fail, and that Eastern Enterprises does not support that challenge, we can readily reject the manufacturers' other constitutional challenges. The primary premise of the manufacturers' argument that Thomas amounts to a "taking" of property under the Takings Clause is that the plurality opinion in Eastern Enterprises announces the applicable test, but as we explained earlier, there is no binding precedent arising from that case, whether on substantive due process or on the Takings Clause.10 This leaves the manufacturers without a basis to argue that a statute or regulation (let alone a judicial decision) that imposes a liability on a party rather than take or burden a specific property interest owned by that party amounts to a "taking." Instead, the Supreme Court has recognized that Congress, for example, "may set minimum wages, control prices, or create causes of action that did not previously exist," all without implicating the Takings Clause. Connolly v. Pension Ben. Guaranty Corp., 475 U.S. 211, 223, 106 S.Ct. 1018, 89 L.Ed.2d 166 (1986) (emphasis added). Assessing liability "that adjusts the benefits and burdens of economic life to promote the common good ... does not constitute a taking." Id. at 225, 106 S.Ct. 1018; accord Concrete Pipe and Prods. of Cal., Inc. v. Construction Laborers Pension Trust for S. Cal., 508 U.S. 602, 643, 113 S.Ct. 2264, 124 L.Ed.2d 539 (1993).

Next, the manufacturers argue that procedural due process is violated by risk-contribution theory. Ordinarily, the familiar three-factor balancing test of Mathews v. Eldridge, 424 U.S. 319, 335, 96 S.Ct. 893, 47 L.Ed.2d 18 (1976), governs whether the "risk of erroneous deprivation" is too great in light of the private interest at stake, the government's interest, and the probable value of other procedures. But the manufacturers' challenge is not really any different from the substantive-due-process argument, because what the manufacturers primarily complain about is the risk of being found liable even though one or more of them did not actually cause Gibson's injury. But such a finding would not be a "mistake" against which more procedural safeguards are needed; instead, that finding would be a result of the permissible and rational common-law development that the Wisconsin Supreme Court fashioned. And in individual cases, there is no reason to believe that the manufacturers will not have notice and a meaningful opportunity to be heard in litigating whether the plaintiff has proved his or her prima facie case or in litigating their defenses against the rebuttable presumption created by Thomas. Indeed, as noted above, on remand to the trial court in Thomas, the jury found that the plaintiff there failed to prove that lead pigment caused his injuries, see Thomas v. Mallett, 331 Wis.2d 486, 795 N.W.2d 62 (Wisc.Ct.App.2011) (unpublished order), and thus the defendant prevailed. Liability simply is not automatic, and risk-contribution theory satisfies due process, both substantively and procedurally.

Finally, Sherwin-Williams argues that Thomas discriminates against interstate commerce, in violation of the Commerce Clause, U.S. Const., art. I, § 8, cl. 3. Sherwin-Williams cites no precedent for the proposition that a state court's decision providing for tort compensation of the state's residents amounts to a Commerce Clause violation. It is one thing for a state court, through a jury verdict, to burden interstate commerce by imposing sanctions for out-of-state behavior with no in-state impact, see BMW of N. Am., Inc. v. Gore, 517 U.S. 559, 572-73, 116 S.Ct. 1589, 134 L.Ed.2d 809 (1996) (cited by Sherwin-Williams Br. at 47), but there is no Commerce Clause obstacle to fashioning a tort remedy for in-state residents who suffer in-state injuries. Indeed, the closest case, factually speaking, to ours (where a state common-law remedy is challenged for providing a remedy to an injured in-state resident) that is cited by Sherwin-Williams actually rejected the interstate-commerce argument, concluding that the state's interest in the health and safety of its residents has long justified imposing burdens, in the form of tort liability, on out-of-state entities. See Estate of Stone v. Frontier Airlines, Inc. 256 F.Supp.2d 28, 46-47 (D.Mass.2002). Thomas's adoption of risk-contribution theory does not violate the Commerce Clause.11

IV.

The Wisconsin Supreme Court's decision in Thomas, establishing the risk-contribution theory of liability for lead pigment claims, does not violate the Due Process, Takings, or interstate-commerce Clauses of the Constitution. The judgment in favor of the defendants is reversed, and the case is remanded to reinstate the case and for further proceedings.