Michael Yarberry v. Supervalu Incorporated

9 F.4th 455

• Court: Court of Appeals for the Seventh Circuit
• Decided: August 12, 2021
• Docket: 20-2241
• Status: Published

Opinion

In the

United States Court of Appeals For the Seventh Circuit ____________________ No. 20-2241 UNITED STATES OF AMERICA ex rel. TRACY SCHUTTE, et al., Relators-Appellants, v.

SUPERVALU INC., et al., Defendants-Appellees. ____________________

Appeal from the United States District Court for the Central District of Illinois. No. 11-cv-3290 — Richard Mills, Judge. ____________________

ARGUED JANUARY 19, 2021 — DECIDED AUGUST 12, 2021 ____________________

Before ROVNER, HAMILTON, and ST. EVE, Circuit Judges. ST. EVE, Circuit Judge. This Court is no stranger to False Claims Act qui tam actions. The present appeal, however, con- tains a novel question for this Circuit: does the Supreme Court’s interpretation of the Fair Credit Reporting Act’s sci- enter provision in Safeco Insurance Company of America v. Burr, 551 U.S. 47 (2007), apply with equal force to the False Claims Act’s scienter provision? We join the four circuits that have 2 No. 20-2241

answered that question in the affirmative and hold that it does. This issue comes to us in a lawsuit against Defendants (collectively, “SuperValu”), which claims that SuperValu knowingly filed false reports of its pharmacies’ “usual and customary” (“U&C”) drug prices when it sought reimburse- ments under Medicare and Medicaid. SuperValu listed its re- tail cash prices as its U&C drug prices rather than the lower, price-matched amounts that it charged qualifying customers under its discount program. Medicaid regulations define “usual and customary price” as the price charged to the gen- eral public. Based on our decision in U.S. ex rel. Garbe v. Kmart Corporation, 824 F.3d 632 (7th Cir. 2016), the district court held that SuperValu’s discounted prices fell within the definition of U&C price and that SuperValu should have reported them. Relators Tracy Schutte and Michael Yarberry (the “Relators”) thus established falsity, the first prong of their False Claims Act (“FCA” or “the Act”) claims. On the scienter prong, how- ever, the court applied the Safeco standard to the FCA and held that SuperValu did not meet it. We agree that the scienter standard articulated in Safeco applies to the FCA. Here, as with the Fair Credit Reporting Act (“FCRA”), there is no statutory indication that Congress meant its usage of “knowingly,” or the scienter definitions it encompasses, to bear a different meaning than its common law definition. We further hold that while the FCA’s scienter provision is defined via three distinct definitions, a failure to establish the Safeco standard as a threshold matter precludes liability under any of these definitions. Applying this stand- ard to the case at hand, SuperValu did not act with the No. 20-2241 3

requisite knowledge under the FCA. The judgment of the dis- trict court is affirmed. I. Background Underlying this case is a complex regulatory scheme, the details of which inform whether SuperValu has run afoul of the FCA’s prohibition on submitting false claims to the gov- ernment. Before canvassing the case facts, it is necessary to provide a brief overview of both the regulatory schemes un- der Medicare Part D and Medicaid and our FCA precedent involving those statutes. A. Medicare Part D and Medicaid Medicare and Medicaid are government healthcare pro- grams administered by the Department of Health and Human Services through the Centers for Medicare and Medicaid Ser- vices (“CMS”). Medicare Part D is a prescription drug benefit providing insurance coverage to beneficiaries. The govern- ment employs a multi-tier system to provide Medicare pre- scription subsidies. At the outset, CMS awards contracts to private plan sponsors to facilitate the benefits program and pays them directly, based in part on the number of enrolled beneficiaries. 42 U.S.C. § 1395w-115; 42 C.F.R. §§ 423.265, 423.315, 423.329(a), (c). Plan sponsors, in turn, enter agree- ments with pharmacies or with middlemen, known as Phar- macy Benefit Managers (“PBMs”), which deal directly with the pharmacies. The PBMs’ contractual agreements with pharmacies specify the methods of calculating prescription drug rates for reimbursement claims, and the PBMs process claims and oversee reimbursements. See 42 U.S.C. § 1395w- 111(i). 4 No. 20-2241

Medicare Part D limits prescription drug reimbursement rates to the lower of either the “actual charge” or “106 percent of the average sales price,” subject to specific limitations. 42 C.F.R. § 414.904(a). While federal regulations do not define “actual charge,” they do define “actual cost.” 42 C.F.R. § 423.100. The actual cost for a prescription from a “network pharmacy” means the “negotiated price” set by the PBM con- tract with that pharmacy. Id. If an out-of-network pharmacy prescribed the drug, the actual cost is the U&C price. Id. Med- icare regulations define U&C price as the price charged to “a customer who does not have any form of prescription drug coverage.” Id. PBM contracts must comply with the Medicare Part D statute and regulations. Medicaid operates in similar fashion but leverages the co- operative efforts of the states. 42 U.S.C. § 1396 et seq. The fed- eral government and participating states jointly finance Med- icaid, and the states implement the program through “state plans.” To be eligible for federal funding, a state’s plan must comply with the Medicaid statute and federal regulations and obtain approval from CMS. 42 U.S.C. §§ 1396-1, 1396a, 1396b. A state’s plan must describe the state agency’s “payment methodology for prescription drugs,” and the drug reim- bursement methodology must comport with federal require- ments for Medicaid expenditures. 42 C.F.R. § 447.518(a)–(b). Relevant here, federal regulations limit the pharmacy reim- bursement for certain prescription drugs to the lower of either “[Actual acquisition cost] plus a professional dispensing fee” or providers’ “usual and customary charges to the general public.” 1 42 C.F.R. § 447.512(b). Because both Medicare and

1 While the state plans for the four states implicated in this appeal contain definitions of U&C price that have slight variances from the No. 20-2241 5

Medicaid programs involve third-party submission of claims to the government, these reimbursement processes give rise to FCA litigation. B. United States ex rel. Garbe v. Kmart Corporation We confronted one such FCA qui tam suit in United States ex rel. Garbe v. Kmart Corporation. In Garbe, we elaborated on the falsity prong of FCA claims in the context of U&C prices reported by pharmacies. The Garbe relator alleged that Kmart submitted false claims for prescription reimbursements under Medicare and Medicaid by failing to report its discount- program prices as its U&C prices. Garbe, 824 F.3d at 636. Instead, Kmart had reported the higher prices it charged to third-party insurers and non-program cash customers. Id. The district court disposed of the relator’s FCA claim on a motion for partial summary judgment. On interlocutory appeal, we added the question whether the district court correctly held that Kmart’s discount-program prices were U&C prices—the prices “charged to the general public.” Id. at 637. We affirmed that determination. Our decision referenced a variety of sources—dictionary definitions, regulatory definitions, Medicare policy, caselaw, and a CMS manual—to determine the boundaries of “usual and customary price charged to the general public.” We noted that unless state regulations provided a different meaning, the U&C price “is defined as the ‘cash price offered to the gen- eral public.’” Id. at 643. Upon consideration of these sources and the case facts, we determined that Kmart’s program fell

wording in § 447.512(b), the Relators have stipulated that these definitions are substantively equivalent to the federal definition. We consequently an- alyze the federal definition of U&C price for purposes of this appeal. 6 No. 20-2241

within the scope of “U&C price.” Kmart’s generic-drug dis- count program offered set prices and was open to the public— any customer could opt in by paying a $10 fee and providing personal information. Id. at 643. The discount prices were “the lowest prices for which its drugs were widely and consist- ently available”—over 89% of Kmart’s cash customers re- ceived the discount prices. Id. at 635, 645; U.S. ex rel. Garbe v. Kmart Corp., 73 F. Supp. 3d 1002, 1018 n.10 (S.D. Ill. 2014). We also found it significant that Kmart had offered these prices for several benefit years rather than as “a one-time ‘lower cash’ price.” Garbe, 824 F.3d at 644. On those facts, we held that a pharmacy’s discount-program prices could be its U&C prices when the program was offered to the public, even though the discount prices were not the retail prices charged to all customers. Id. at 645. We remanded Garbe without dis- cussing the FCA’s scienter prong. Although the scienter prong is at issue in this appeal, Garbe played a key role in the suit against SuperValu. C. Factual Background SuperValu, through several subsidiaries, operated or con- trolled roughly 2,500 grocery stores with over 800 in-store pharmacies between 2006 and 2016. In 2006, SuperValu’s na- tional headquarters implemented the discount program un- derlying this appeal, which ran until December 2016. The price-match initiative was an attempt to compete with phar- macies such as Wal-Mart, which had launched a discount pro- gram that same year offering hundreds of generic drugs at $4 per 30-day prescription. SuperValu sought to remain compet- itive without adopting Wal-Mart’s program. According to Su- perValu’s Vice President of Prescription Services, implement- ing a $4 generics program would cost SuperValu $40–$50 No. 20-2241 7

million in losses if $4 was the U&C cost passed on to PBMs. Instead, SuperValu employed what it internally characterized as a “‘stealthy’ approach.” Corporate officers framed Super- Valu’s price-match program as an “‘exception’ for customer service reasons” that would not be reported as the U&C price. Under SuperValu’s price-match program, its regional stores could match lower prices on prescription drugs offered by other, local pharmacies within a specific proximity to the regional store. But the discount was not automatic. Customers had to request a price match. Once SuperValu pharmacists verified the competitor’s price, SuperValu automatically ap- plied the discount for that customer on future refills. 2 Any customer could request a price match, including those with insurance or government healthcare plans. When applying a price-match cost for insured customers, the pharmacists over- rode the price in the pharmacy’s automatic system and man- ually entered the price-matched cost. SuperValu instructed pharmacies to process these price-match sales as cash trans- actions rather than third-party payor claims that would go di- rectly to insurers. SuperValu did not report these price-matches when it sub- mitted reimbursement claims to third-party insurers, includ- ing Medicare Part D and Medicaid. Rather, SuperValu listed its retail price—the price for uninsured cash customers—as its U&C price. Many of SuperValu’s PBM contracts contained U&C price clauses, but the contractual definitions of that term varied. Some contracts addressed reporting prices from

2 SuperValu did not implement its automatic price override until 2008.

All SuperValu’s pharmacies had ceased the price-match program by De- cember 2016, a few months after this Court decided Garbe in May 2016. 8 No. 20-2241

discount programs, either including discount programs as a blanket rule or excepting specific types of discounts. Others did not mention discounts at all. None of the contracts ex- pressly included price-matching, although one PBM, Medco, stated in its 2007–2008 manual that it included a “competitor’s matched price” in its definition of U&C price. Between 2006 and 2016, sales under SuperValu’s price- matching policy accounted for 26.6% of SuperValu’s cash drug sales and 1.69% of its total prescription drug sales— roughly 6.3 million sales. In 2012, the majority of the cash sales for 44 of SuperValu’s top 50 prescription drugs were made at a price-match cost rather than SuperValu’s retail price. 3 SuperValu continued its price-match program until December 2016 and did not report its discount prices as its U&C prices to any PBM or state agency during that time. D. Procedural Background In 2011, the Relators filed this suit against SuperValu un- der the FCA on behalf of the federal government and several states. 4 They alleged that SuperValu knowingly caused false payment claims to be submitted to government healthcare programs between 2006 and 2016 by incorrectly reporting their U&C drug prices. The Relators’ theory of the case was

3The dissent cites this statistic without confining it to fiscal year 2012. We note that the Relators have identified no evidence regarding the fre- quency of price-match sales versus retail cash sales for SuperValu’s top 50 drugs during any of the other years between 2006-2016 when its price- match program was active. 4 In the district court, the parties stipulated to a dismissal of all Medi- caid claims on behalf of the states except those on behalf of California, Il- linois, Utah, and Washington. No. 20-2241 9

that SuperValu price-matched to avoid losing customers to competitors with lower drug prices like Wal-Mart and made up the difference by charging government healthcare pro- grams its higher, retail price. In effect, the Relators argued, SuperValu caused the government to subsidize its market competitiveness. The government did not intervene in this case. The district court, relying on Garbe, granted summary judgment to the Relators on the falsity prong. 5 It acknowl- edged that SuperValu’s price-match program required cus- tomers to initiate a discount and found that discount sales comprised a lower portion of SuperValu’s sales—roughly 2% of total transactions and 26.9% of cash sales—compared to Kmart’s discounts in Garbe, which amounted to 89% of its cash sales. Even so, the court held that the fact that SuperValu made its price-match policy available to the general public throughout a benefit year was determinative. In a separate order, the district court sided with SuperValu on the scienter prong. The court first applied Safeco’s standard to the FCA’s scienter prong and held that a failure to establish the objective scienter standard precluded liability under the FCA. Under the Safeco standard, the court held that Super- Valu’s understanding of U&C price, while incorrect, was ob- jectively reasonable at the time. The district court first ob- served that there were multiple district court decisions en- dorsing SuperValu’s view of U&C price or recognizing that the term was open to interpretation. It also took note of the unique circumstance in which Garbe addressed the definition

5SuperValu does not contest the district court’s falsity holding in this appeal. 10 No. 20-2241

of U&C price. Because the Seventh Circuit added that ques- tion to the issues certified for interlocutory appeal, the district court suggested that we must have found the matter “suffi- ciently debatable to be addressed.” Based on the available caselaw, the court held that it was unclear that SuperValu’s program fell within the U&C defini- tion. Further, the court held that prior to our 2016 decision in Garbe, there was no authoritative guidance to warn SuperValu away from its interpretation of U&C price. In view of these conclusions, the district court entered summary judgment for SuperValu on all FCA claims, which the Relators now chal- lenge. II. Discussion On appeal, the parties ask us to determine whether Safeco applies to the FCA’s scienter standard and, if so, to what ex- tent. Our answers to those questions will dictate the outcome of the final issue in this appeal—whether the district court properly granted summary judgment for SuperValu on the scienter prong of the FCA claim. We review the district court’s determinations on these legal issues de novo and affirm the grant of summary judgment to SuperValu. Bigger v. Facebook, Inc., 947 F.3d 1043, 1048, 1051 (7th Cir. 2020). A. The False Claims Act The FCA imposes civil liability on any person who “know- ingly presents, or causes to be presented, a false or fraudulent claim for payment or approval.” 31 U.S.C. § 3729(a)(1)(A). FCA civil claims thus require proof of two primary elements: (1) falsity and (2) scienter. The Supreme Court has also inter- preted § 3729(a)(1)(A) to require that knowingly false claims be material to the government’s payment decision for liability No. 20-2241 11

to attach. Univ. Health Servs., Inc. v. U.S. ex rel. Escobar, 136 S. Ct. 1989, 1996 (2016). Although “Congress did not define what makes a claim ‘false’ or ‘fraudulent,’” the Supreme Court has applied the common law meaning of fraud to these terms as they are used in the FCA. Id. at 1999. Under that definition, a claim may be false or fraudulent through either express misrepresentations or “misrepresentations by omission.” Id. Unlike the falsity prong, the FCA’s scienter requirement is statutorily defined. A party who submits a false claim to the government is on the hook for FCA liability only if it acted knowingly. § 3729(a)(1)(A). The FCA defines knowingly to “mean that a person, with respect to information (i) has actual knowledge of the information; (ii) acts in deliberate ignorance of the truth or falsity of the information; or (iii) acts in reckless disregard of the truth or falsity of the information.” § 3729(b)(1)(A). It “require[s] no proof of specific intent to de- fraud.” § 3729(b)(1)(B). The FCA levies significant conse- quences against parties found liable under the Act and bal- ances the severity of its penalties by carefully circumscribing liability, in part through its scienter requirement. See Escobar, 136 S. Ct. at 1995–96 (observing that FCA civil “liability is es- sentially punitive in nature” (internal quotation omitted)). B. Safeco Insurance Company of America v. Burr While the FCA lists the range of scienter levels encom- passed by “knowingly,” it does not further define those terms. SuperValu urges us to look to the Supreme Court’s decision in Safeco for guidance. Safeco involved an interpretation of the FCRA’s common law scienter requirement, under which plaintiffs must show that defendants acted “willfully.” 15 12 No. 20-2241

U.S.C. § 1681n(a). As defined by the Court, the FCRA’s use of that term includes both “knowing” and “reckless disregard.” Safeco, 551 U.S. at 52, 59. In interpreting the FCRA’s scienter prong, the Court first observed “the general rule that a common law term in a stat- ute comes with a common law meaning, absent anything pointing another way.” Id. at 58. Finding none, it employed what amounts to a two-step inquiry for determining reckless disregard. Id. at 69. A defendant who acted under an incorrect interpretation of the relevant statute or regulation did not act with reckless disregard if (1) the interpretation was objec- tively reasonable and (2) no authoritative guidance cautioned defendants against it. Id. at 70. Critically, the Court empha- sized that a defendant’s subjective intent is irrelevant for pur- poses of liability. Id. at 68, 70 n.20. The Court also explained that failure to meet this standard would preclude a finding of knowing violations as well. Id. at 70 n.20. The Court then applied that standard and held that while Safeco may have violated the FCRA, it did not do so with reckless disregard. The FCRA requires that any person who takes an “adverse action” against a consumer based on infor- mation in a consumer report notify that consumer. 15 U.S.C. § 1681m(a). An “adverse action” is statutorily defined as in- cluding “an increase” in the amount charged for “insurance, existing or applied for.” § 1681a(k)(1)(B)(i). The Safeco plain- tiffs argued that Safeco violated the FCRA when it offered new insurance applicants higher rates without notifying them that their credit scores triggered the less favorable policy of- fers. Safeco, 551 U.S. at 55. Safeco thought initial rate offers to new customers fell outside FCRA notice obligations because No. 20-2241 13

it interpreted “increase” to mean rate hikes on existing poli- cies. Id. at 69–70. While Safeco’s interpretation was erroneous, the Court held that it was objectively reasonable. Why? Because Safeco’s “reading ha[d] a foundation” in “the less-than-pellucid statutory text.” Id. Further, there was no court of appeals decision or authoritative guidance from the Federal Trade Commission—the agency charged with enforcing the FCRA—that “might have warned it away from the view it took.” Id. at 70. Under the Court’s two-step inquiry, these facts precluded a finding of reckless disregard. Since this decision, four circuit courts have applied the Safeco standard to the FCA’s scienter prong. SuperValu asks us to do the same today. C. Safeco applies to the FCA To determine what the FCA’s scienter provision requires, we “start, as always, with the statutory text.” Escobar, 136 S. Ct. at 1999. The FCA defines “knowingly” as encompassing three common law standards—actual knowledge, deliberate indifference, and reckless disregard—but is silent as to what those standards mean in the context of this statute. 6 Supreme Court precedent teaches that “a common law term in a statute comes with a common law meaning, absent anything point- ing another way.” Safeco, 551 U.S. at 58. That principle informs our decision today. Here, the Relators have identified no stat- utory indicia that Congress intended the familiar, common

6 The FCA imposes civil liability. We thus reference the civil, not crim-

inal, definitions of these scienter standards throughout our discussion. Safeco, 551 U.S. at 60 (acknowledging distinctions between criminal and civil uses of the same scienter terms and indicating that criminal law usage has no bearing on the definitions of these terms when used in civil laws). 14 No. 20-2241

law terms used in § 3729 to differ from their common law meaning. Indeed, the Supreme Court has confirmed that the FCA does employ the common law meaning for other com- mon law terms—“false” and “fraudulent”—and has limited the common law definition only to the extent that the statute expressly contradicted it. “Congress retained all other ele- ments of common-law fraud that are consistent with the stat- utory text because there are no textual indicia to the contrary.” Escobar, 136 S. Ct. at 1999 & n.2. Given that the common law meaning applies to the FCA’s scienter standard, all that re- mains is to identify that meaning. We need look no further than the Supreme Court’s decision in Safeco. Safeco defined a similar common law term—“willfully,” as used in the FCRA—which the Court interpreted as encom- passing the same common law scienter terms used in the FCA (“knowingly” or “reckless disregard”). Referencing the com- mon law meaning, the Court then announced a standard in- quiry for reckless disregard. While reiterating that “know- ingly” and “reckless disregard” remain distinct terms, the Su- preme Court held that the objective scienter standard it artic- ulated precluded liability under either term. Safeco, 551 U.S. at 60, 70 n.20. There is no reason why the scienter standard established in Safeco (for violations committed knowingly or with reckless disregard) should not apply to the same com- mon law terms used in the FCA. The dissent suggests that Safeco has no bearing simply be- cause it interpreted a different scienter requirement in a dif- ferent statute. We respectfully disagree. Safeco articulated an objective scienter standard for establishing willful violations, which it framed in terms of the scienter floor for that stand- ard—reckless disregard. Likewise, reckless disregard is the No. 20-2241 15

baseline scienter definition encompassed by the FCA’s scien- ter requirement, “knowingly.” United States v. King-Vassal, 728 F.3d 707, 712 (7th Cir. 2013) (observing that reckless disregard “is the most capacious of the three” terms used to define the FCA’s scienter requirement). And Safeco explicitly held that the test for reckless disregard would likewise cover violations committed “knowingly.” Safeco, 551 U.S. at 70 n.20. In view of those parallels, we see no barrier to importing the Safeco standard to the FCA. See Purcell, 807 F.3d at 284, 290. Every other circuit court to discuss the relevance of Safeco’s scienter standard to the FCA has arrived at this conclusion. United States ex rel. Streck v. Allergen, 746 F. App’x 101, 106 (3d Cir. 2018); United States ex rel. McGrath v. Microsemi Corp., 690 F. App’x 551, 552 (9th Cir. 2017); United States ex rel. Donegan v. Anesthesia Assocs. of Kan. City, PC, 833 F.3d 874, 879–80 (8th Cir. 2016); United States ex rel. Purcell v. MWI Corp., 807 F.3d 281, 284 (D.C. Cir. 2015). The dissent claims that the Eleventh Circuit declined to apply Safeco to the FCA in United States ex rel. Phalp v. Lincare Holdings, Inc., 857 F.3d 1148 (11th Cir. 2017). But Phalp did not reject Safeco—it did not even cite Safeco. To support its conclusion, the dissent points to Phalp’s assertion that “scienter is not determined by the ambiguity of a regulation, and can exist even if a defendant’s interpretation is reasonable.” Id. at 1115. That is not inconsistent with Safeco. Under Safeco, an objectively reasonable interpretation of a statute or regulation does not shield a defendant from liability if authoritative guidance warned the defendant away from that interpretation. Regardless of differing views as to whether Phalp is consistent with the Safeco standard, the Elev- enth Circuit did not reject Safeco’s applicability to the FCA. Even though the parties briefed the court on Safeco, that brief- ing does not convert the Eleventh Circuit’s silence into a 16 No. 20-2241

decision that Safeco does not apply to the FCA. As it stands, no circuit has held Safeco inapplicable to the FCA. The dissent would part ways with the circuits that have applied the Safeco standard to the FCA and look instead to the Restatement (Second) of Torts § 526, which makes subjective intent relevant to the scienter inquiry. Section 526 defines “conditions under which misrepresentation is fraudulent.” It does not define “knowingly” (or any of the common law sci- enter terms listed in § 3729(b)(1)(A)). And it is a different pro- vision than the Restatement provision that the Court refer- enced in Safeco, 551 U.S. at 69 (relying upon § 500, defining “reckless disregard”). We thus disagree that § 526 is relevant to the FCA’s scienter provision. Take out “knowingly,” and perhaps it makes sense to read general, common law fraudu- lent scienter into the Act. But here, Congress has willed a spe- cific scienter requirement—knowingly, not “‘knowing’ of fal- sity,” as the dissent suggests. Unlike § 526, § 500 defines a term that the FCA’s definition of knowingly expressly includes (“reckless disregard”). The dissent insists that because § 500—which defines “reckless disregard of safety”—applies to cases involving physical harm, it is inapplicable to “reckless disregard” as used in the FCA. But the Supreme Court applied this definition outside the physical-harm context in Safeco. Ultimately, the crucial point is that the Court has articulated a standard for acts com- mitted “knowingly” or with “reckless disregard” that ex- cludes subjective intent. In the absence of textual indicia in the FCA supporting that subjective intent matters here, we apply Supreme Court precedent to interpret the same common law terms addressed in Safeco. No. 20-2241 17

While the dissent claims that its countervailing view is tex- tually mandated, nothing in the language of the FCA suggests that a defendant’s subjective intent is relevant. In contrast to § 526, terms such as “believes” or “have [] confidence” are conspicuously absent from the FCA, and the only reference to intent is an express disclaimer that “specific intent to de- fraud” is irrelevant. § 3729(b)(1)(B). We decline to graft as- pects of common law fraudulent scienter into the FCA when Congress chose not to include such requirements. The dissent instead looks to legislative history and out-of- circuit caselaw to support its reading of the FCA. We find nei- ther source persuasive. Legislative history cannot support reading in a subjective-intent requirement that goes beyond the text of the Act’s scienter provision. See Chamber of Com- merce of U.S. v. Whiting, 563 U.S. 582, 599 (2011) (“Congress’s authoritative statement is the statutory text, not the legislative history.” (internal citation and quotation omitted)). And the circuit cases upon which the dissent relies all predate Safeco, as well as subsequent caselaw in each of those circuits apply- ing Safeco to the FCA. Neither Restatement § 526 nor legisla- tive history pose a barrier to applying Safeco. The Relators challenge Safeco’s viability on a separate ba- sis that likewise fails. They contend that subsequent Supreme Court precedent limited Safeco, leaning on a 2016 patent case for this premise—Halo Electronics, Inc. v. Pulse Electronics, Inc., 136 S. Ct. 1923 (2016). Halo Electronics interpreted § 284 of the Patent Act, which provides that courts may award treble dam- ages in infringement cases. Section 284 does not specify a sci- enter standard, and prior to Halo Electronics, the Federal Cir- cuit required plaintiffs to show that an infringer’s conduct was “both objectively baseless and brought in subjective bad 18 No. 20-2241

faith.” Id. at 1932–33. Halo Electronics clarified that § 284 liabil- ity does not depend on objective recklessness. The problem with importing an objective recklessness in- quiry into the patent context was that “such a defense insu- lates the infringer from enhanced damages, even if he did not act on the basis of the defense or was even aware of it.” Id. at 1933. In rejecting that standard, the Court emphasized that the Patent Act targets “consciously wrongful” bad action and held that “[i]n the context of such deliberate wrongdoing, however, it is not clear why an independent showing of ob- jective recklessness … should be a prerequisite to enhanced damages.” Id. at 1932. The defendants, citing Safeco, argued that bad faith is irrel- evant when there is no showing of objective recklessness. Id. at 1933 n.*. While acknowledging the Safeco standard, the Court declined to apply it. It observed that “willfully is a word of many meanings whose construction is often depend- ent on the context in which it appears.” Id. (internal quotation omitted). The Patent Act presented a different context than the FCRA: “[O]ur precedents make clear that ‘bad-faith infringe- ment’ is an independent basis for enhancing patent dam- ages.” Id. The Supreme Court thus did not walk back Safeco or adopt a new standard for objective recklessness. Halo Electronics simply did not apply objective recklessness in the context of a statute focused on defendants’ subjective bad faith. The rea- sons informing that decision do not apply here. Unlike the Pa- tent Act, the FCA expressly includes a scienter standard and limits liability to knowingly false claims. By its own terms, Safeco holds that a failure to establish its objective scienter standard precludes a finding that a defendant acted No. 20-2241 19

knowingly. We thus hold that Safeco’s scienter standard ap- plies to the FCA. D. Failure to meet the Safeco standard precludes liability Beyond the threshold question of Safeco’s applicability to the FCA, the parties also dispute how broadly Safeco reaches. We agree with SuperValu that the Safeco standard reaches all three of the scienter terms that define “knowingly.” The dis- sent takes the Relators’ position that even if it is relevant to the FCA, Safeco defines only “reckless disregard.” Under this view, failure to show that a defendant meets the Safeco stand- ard does not preclude liability under the actual knowledge or deliberate ignorance components of the FCA’s scienter defini- tion. The dissent contends that holding otherwise would col- lapse distinct scienter terms and violate the rule against sur- plusage. We are unconvinced. The Supreme Court has already undermined this line of reasoning. In Safeco, the Court rejected the defendants’ argu- ment that it was conflating scienter terms and reaffirmed that the terms it used to define “willfully” were distinct. Safeco, 551 U.S. at 60. (“[A]ction falling within the knowing subcategory does not simultaneously fall within the reckless alternative.”). It nevertheless held that the standard it articulated in the con- text of “reckless disregard” also functioned as a baseline re- quirement for establishing the more demanding scienter cat- egory of “knowledge.” Id. at 70 n.20 (“Where, as here, the stat- utory text and relevant court and agency guidance allow for more than one reasonable interpretation, it would defy his- tory and current thinking to treat a defendant who merely adopts one such interpretation as a knowing or reckless viola- tor.” (emphasis added)). That holding nullifies the dissent’s contention. 20 No. 20-2241

Even aside from Safeco’s dismissal, the dissent’s argument rests upon a false equivalence. No one disputes that the three scienter terms used to define “knowingly” are distinct and bear different meanings. Both actual knowledge and deliber- ate ignorance indicate higher degrees of culpability and, if im- plicated in a case, might render reckless disregard inapplica- ble. See Purcell, 807 F.3d at 288 (observing that reckless disre- gard is the loosest standard of knowledge under the FCA’s scienter requirement). That does not prevent these terms, however, from sharing a common requirement. Indeed, we do not see how it would be possible for de- fendants to actually know that they submitted a false claim if relators cannot establish the Safeco scienter standard. A de- fendant might suspect, believe, or intend to file a false claim, but it cannot know that its claim is false if the requirements for that claim are unknown. The dissent’s primary concern that the Safeco standard eliminates culpability for deliberately in- different defendants is likewise misplaced. The dissent postu- lates that under the Safeco standard, defendants could escape liability by making a “barely plausible” post-hoc argument about a statute’s meaning, “even though the defendant ig- nored repeated and correct warnings.” That fundamentally misapprehends Safeco. Under Safeco, a defendant will be suc- cessful only if (a) it has an objectively reasonable reading of the statute or regulation and (b) there was no authoritative guid- ance warning against its erroneous view. That test does not shield bad faith defendants that turn a blind eye to guidance indicating that their practices are likely wrong. Nor does Safeco’s standard excuse a company if its executive deci- sionmakers attempted to remain ignorant of the company’s claims processes and internal policies. Safeco covers all three of the scienter standards listed in § 3729. When relators cannot No. 20-2241 21

establish the standard articulated in Safeco, there is no liability under the FCA. E. SuperValu’s interpretation of “usual and customary price” was objectively reasonable under Safeco Although the Safeco Court did not express its standard for reckless disregard in terms of elements, the Court’s objec- tively reasonable inquiry involved two distinct questions— whether the defendant has a permissible interpretation of the relevant provision and whether authoritative guidance nev- ertheless warned it away from that reading.7 1. Permissible Interpretation The objectively reasonable inquiry hinges on the text of the statute or regulation that the defendant allegedly violated and as such is a question of law. Safeco, 551 U.S. at 69; see also Van Straaten v. Shell Oil Prods. Co. LLC, 678 F.3d 486, 489–90 (7th Cir. 2012). If the plain language of the statute precludes the erroneous interpretation, the defendant cannot clear this hur- dle. To decide whether SuperValu had a permissible interpre- tation of U&C price, we must first determine the source of that term and relevant definition.

7 Some courts have divided the Safeco inquiry into three steps, adding as a preliminary question whether the relevant text is ambiguous. Done- gan, 833 F.3d at 878; Purcell, 807 F.3d at 288. We elect to condense the in- quiry into the two issues expressly discussed in Safeco—permissible inter- pretation and authoritative guidance. The Safeco Court did not require a separate determination of ambiguity, and we think that the issue of textual ambiguity is subsumed within the permissible-interpretation inquiry. A defendant’s erroneous interpretation cannot be reasonable if the meaning of the text is unambiguous. 22 No. 20-2241

Medicaid regulations define U&C price without much elaboration as the price that a pharmacy “charges to the gen- eral public.” 8 42 C.F.R. § 447.512(b); see also Garbe, 824 F.3d at 643–44. Federal regulations do not elaborate beyond that cur- sory definition or guide pharmacies on identifying the “gen- eral public” when they charge customers various prices for the same prescription.9 “Usual and customary” might mean the price that is “charged” most frequently for a drug, but it could also indicate the retail rather than discount price. See GAO, Report to Congress on Trends in Usual and Customary Prices for Drugs Frequently Used by Medicare and Non-

8 The Relators argue that for Medicare, pharmacies that have con- tracted with a plan sponsor or PBM report the “negotiated price” deter- mined by the contract. On appeal, the Relators consequently look to the various formulations of U&C price in SuperValu’s PBM contracts. In the district court, however, they took the opposite position: “Relators dispute that the contracts between PBMs and pharmacies ‘govern the terms’ by which Defendants are required to submit claims to the PBMs and in turn, whether and how much the PBMs should pay Defendants for dispensing drugs to their beneficiaries.” Relators’ Resp. to Defs.’ Mot. for Partial Summ. J. at 9 [191-1]. As a result, they have waived any argument on ap- peal that the contractual definitions of U&C price are distinct from the Medicaid regulatory definition. We thus examine only § 447.512(b)’s defi- nition of U&C price and treat the PBM contract definitions of U&C price as consistent with it. 9 The Relators also identified four states’ regulations defining U&C price, as Medicaid is implemented through the states. The regulations that were concurrent with SuperValu’s price-match program used substan- tially the same definition of U&C price as 42 C.F.R. § 447.512(b). The Rela- tors also claim that they are “consistent with the controlling federal defi- nition and the U&C framework analyzed in Garbe.” We consequently treat our analysis of the federal definition of U&C price as extending to these states and the FCA claims related to Medicaid. No. 20-2241 23

Medicare Enrollees at 1 (Oct. 6, 2004) (“The usual and custom- ary price is the undiscounted price individuals without drug coverage would pay.“). “General public” may mean that dis- count prices qualify only if applied to all consumers or, alter- natively, if they constitute the price most frequently charged to consumers. But it just as easily might encompass any dis- count program offered to the public, regardless of whether all consumers take advantage of it. Garbe, 824 F.3d at 643. As is, the U&C price definition is open to multiple interpretations. Here, SuperValu interpreted its set, retail price for a pre- scription drug as the “price it charges to the general public.” Unlike its retail price, the discount prices under SuperValu’s price-match program depended upon the prices charged by local competitors and initially applied only upon customer re- quest. In short, while its program was available to any cus- tomer requesting a valid price match, SuperValu would not necessarily charge all or most of its customers lower, price- matched costs. SuperValu thus did not view its competitor price-matching as the price that it “charged to the general public.” That interpretation is not inconsistent with the text of the U&C price definition. See Garbe, 824 F.3d at 644 (citing § 447.512(b)). The Relators spend little time discussing the compatibility of SuperValu’s interpretation of U&C price with the regulatory text. Instead, they contend that Garbe forecloses any argument on objective reasonableness. Garbe characterized the federal regulations at issue here as having a “clear” purpose—ensuring that the government receives the benefit of the “prevailing retail market price” that pharmacies provide to consumers. Garbe, 824 F.3d at 644. From this, the Relators claim that we have already held that the meaning of 24 No. 20-2241

U&C price is unambiguous. The flaws in this argument are two-fold. As an initial matter, it overextends our holding in Garbe. Garbe held that the correct interpretation of U&C price in- cluded certain discount program prices—it did not hold that this was the only objectively reasonable interpretation of the term. In fact, Garbe did not discuss Safeco at all. We had no reason to do so because we explicitly did not address the FCA’s scienter prong. The decision that we did reach in Garbe—interpreting “U&C price”—does not influence the ob- jectively reasonable inquiry here, either. Safeco’s scienter standard has bite only if a defendant’s interpretation may be objectively reasonable even if it is erroneous. That Super- Valu’s interpretation of U&C price is incorrect under Garbe does not de facto render its interpretation unreasonable. The Relators also err by calibrating objective reasonable- ness against the clarity of a statute or regulation’s policy ob- jective. Their Garbe argument rests on the assumption that any regulation with a clear purpose cannot be ambiguous. But Safeco tethered the objectively reasonable inquiry to the legal text, not its underlying policy. Safeco, 551 U.S. at 69–70 (hold- ing that Safeco’s erroneous interpretation was reasonable be- cause it had a foundation in the “less-than-pellucid” statutory text). The Relators’ failure to engage with the regulatory text is fatal to their objections. They have not shown that Super- Valu’s erroneous interpretation of U&C price was unreasona- ble. Apart from the Relators’ arguments based on Garbe, the dissent suggests a more fundamental concern with Super- Valu’s interpretation of U&C price. It argues that for an erro- neous interpretation to be objectively reasonable, the No. 20-2241 25

defendant must have held that view at the time that it submit- ted its false claim. 10 Otherwise, the dissent insists, defendants can avoid liability by concocting “post-hoc arguments” to jus- tify their conduct under an objectively reasonable reading of the applicable regulation—even if they acted in bad faith. The dissent essentially argues that SuperValu believed it was vio- lating the requirement to report its U&C price and arrived at its “interpretation” of U&C price after the fact. Even if the Relators can raise an issue of fact on this point, it is irrelevant. The FCA establishes liability only for knowingly false claims—it is not enough that a defendant suspect or be- lieve that its claim was false. See Purcell, 807 F.3d at 288 (hold- ing that defendants did not violate the FCA because they “could reasonably have concluded” that their conduct com- plied with the law, even though they believed—and testified that they “knew”—it did not). Indeed, Safeco emphasized that a defendant’s subjective intent does not matter for its scienter analysis—the inquiry is an objective one. This standard re- flects the limits of FCA liability. See Escobar, 136 S. Ct. at 2003 (“The False Claims Act is not an all-purpose antifraud statute or a vehicle for punishing garden-variety breaches of contract or regulatory violations.” (cleaned up)). We apply the stand- ard as we find it and hold that SuperValu has offered an ob- jectively reasonable interpretation of U&C price.

10 The dissent does not—and cannot—rely on Safeco for this assertion. Safeco made no mention of a temporal requirement when it articulated the objectively reasonable inquiry. The Relators cited Halo Electronics when they raised this same argument on appeal. But as explained previously, we reject the applicability of that case to the FCA. 26 No. 20-2241

2. Authoritative Guidance This moves SuperValu but halfway across the scienter line. Safeco makes clear that a permissible interpretation is no de- fense if there existed authoritative guidance that should have warned defendants away from their erroneous interpreta- tion. 11 “Authoritative guidance,” as the moniker implies, must come from a source with authority to interpret the relevant text. Safeco also suggests that the guidance must be suffi- ciently specific to the defendant’s incorrect interpretation. The Supreme Court did not flesh out the boundaries of au- thoritative guidance, but at minimum, Safeco supports that it must come from a governmental source—either circuit court precedent or guidance from the relevant agency. 12 Safeco, 551 U.S. at 70. We are not alone in this view. Other circuit courts likewise have limited authoritative guidance to these two sources. See Purcell, 807 F.3d at 289 (considering only circuit court caselaw and guidance from the controlling agency); Streck, 746 F. App’x at 106, 108 (same). Our reading of Safeco automatically excludes one of the three sources of guidance proposed by the Relators—the PBM contract definitions of U&C price. The Relators also identify federal and state regu- lations defining U&C price, but we have considered the rele- vant regulatory definition above and determined that it does

11 The authoritative-guidance inquiry is a question of law in this case, as it entails only the interpretation of regulatory guidance. 12 The parties agree that Garbe is no help to the Relators on this front, despite its status as circuit court precedent that would otherwise consti- tute authoritative guidance. Recall that we decided that case in May 2016, the same year that SuperValu shelved its discount program. The Supreme Court did not deny the Garbe certiorari petition until 2017. No. 20-2241 27

not preclude SuperValu’s interpretation. As a result, those definitions cannot constitute warnings that SuperValu’s inter- pretation was erroneous. The remaining source of guidance identified by the Relators is the CMS Medicare Prescription Drug Benefit Manual (“CMS manual” or “manual”). The Relators contend that the manual constitutes authoritative guidance which should have warned SuperValu that its discount prices amounted to U&C prices. SuperValu responds that it did not, for two reasons. First, SuperValu suggests that the manual is not “authoritative” guidance as defined by Safeco. It reads Safeco to require that authoritative agency guidance not only originate from the agency charged with implementing the relevant statute but that it be binding on the agency, such as notice-and-comment rulemaking or agency adjudication. The circuits that have addressed Safeco’s applicability to the FCA appear split on this question. But we need not—and do not— decide this matter today because we agree with SuperValu’s second argument: the CMS manual was not sufficiently specific to warn SuperValu that its program likely would fall within the definition of U&C price. Safeco suggests that authoritative guidance must have a high level of specificity to control an issue. In Safeco, the agency guidance at issue was an FTC letter to Safeco explain- ing that an adverse action “occurs when ‘the applicant will have to pay more for insurance at the inception of the policy than he or she would have been charged if the consumer re- port had been more favorable.’” Safeco, 551 U.S. at 70 n.19 (in- ternal citation omitted). That guidance certainly related to the question on appeal—whether an “increase” in insurance rates based on a consumer report could “be understood without 28 No. 20-2241

reference to prior dealing (allowing a first-time applicant to sue).” Id. at 64–65. Nevertheless, the Supreme Court rejected the FTC letter in part because the Court thought that it “did not canvass the issue.” 13 Id. at 70 n.19. Upon review of the CMS manual, we conclude that it is similarly flawed. Footnote one of the manual is most salient and reads in relevant part as follows: We note that in cases where a pharmacy offers a lower price to its customers throughout a bene- fit year, this would not constitute a “lower cash price” situation that is the subject of this guid- ance. For example, Wal-Mart recently intro- duced a program offering a reduced price for certain generics to its customers. The low Wal- Mart price on these specific generic drugs is considered Wal-Mart’s “usual and customary” price, and is not considered a one-time “lower cash” price. Part D sponsors consider this lower amount to be “usual and customary” and will reimburse Wal-Mart on the basis of this price. CENTERS FOR MEDICARE & MEDICAID SERVICES, Chapter 14— Coordination of Benefits, in MEDICARE PRESCRIPTION DRUG BENEFIT MANUAL 19 n.1 (2006), https://perma.cc/MW6AH4P6. The footnote clarifies that a pharmacy’s consistent, lower- price offers are included within U&C prices. But it says noth- ing about price-match programs like that employed by

13 The Court’s other reason for considering the letter unauthoritative was that the FTC had expressly stated that it was not binding on the agency. Safeco, 551 U.S. at 70 n.19. No. 20-2241 29

SuperValu. Further, the majority of the footnote discusses a specific example—Wal-Mart’s $4 generics program—which differed in significant respects from SuperValu’s price-match guarantee. Wal-Mart’s program employed a set lower price ($4 for 30-day generic prescriptions) automatically applied to any customer. By contrast, SuperValu’s discount prices could vacillate. Its discounts depended upon the pricing of local competitors, which could vary between SuperValu’s regional stores. SuperValu’s discounts also were customer-initiated in the first instance. The manual did not put SuperValu on notice that this type of discount program fell within the definition of U&C price—at least, not with the specificity required to be authoritative guidance. We hold that no authoritative guid- ance warned SuperValu away from its permissible interpreta- tion of U&C price. The district court correctly granted sum- mary judgment to SuperValu on the question of scienter. III. Conclusion Our resolution of this case is controlled by Safeco. Today, we hold that Safeco’s standard both applies to the FCA’s sci- enter requirement and precludes liability under it, regardless of whether relators premise their case on reckless disregard or the other scienter terms. Because SuperValu had an objec- tively reasonable understanding of the regulatory definition of U&C price and no authoritative guidance placed it on no- tice of its error, the Relators have not shown that SuperValu acted knowingly. The district court’s judgment is AFFIRMED. 30 No. 20-2241

HAMILTON, Circuit Judge, dissenting. We should reverse summary judgment for defendant SuperValu. The relators have come forward with evidence that SuperValu knowingly misled the government’s agents about its “usual and custom- ary” prices for a significant number and volume of prescrip- tion drug sales. For forty-four of the fifty top-selling drugs, SuperValu was charging the government prices eight to fifteen times higher than the prices it was actually charging a majority of the relevant customers. Binding circuit precedent holds that those price claims were false. SuperValu’s defense is that it did not “know” its “usual and customary” price claims were false. When the False Claims Act is properly understood, however, genuine factual disputes over SuperValu’s conduct and state of mind should preclude summary judgment. This appeal presents a broad and important issue for the False Claims Act. The issue is whether the Act can reach busi- nesses that submit false claims for government payment but claim there is some legal ambiguity that kept them from “knowing” for certain that their claims were false. Under the text and history of the Act, the answer should be yes. The majority answers no. It thus creates a safe harbor for deliberate or reckless fraudsters whose lawyers can concoct a post hoc legal rationale that can pass a laugh test. The major- ity’s new safe harbor even makes subjective bad faith “irrele- vant” in fraud cases. Ante at 25. That undermines the 1986 amendments to the False Claims Act and turns the Act up- side-down, losing touch with the statutory text and its history and links to the common law of fraud. I respectfully dissent. Part I of this opinion explains the relators’ claims and sup- porting evidence. Part II explains the better understanding of the False Claims Act’s “knowledge” standard based on the No. 20-2241 31

statutory text, the common law of fraud, and statutory his- tory. Part III explains the majority’s two fundamental errors in reading the statute. First, rather than focusing on the statu- tory text, history, and purpose of the False Claims Act itself, the majority reads far too much into Safeco Insurance Co. v. Burr, 551 U.S. 47 (2007), where the Supreme Court interpreted a different term under a different statute. Second, the majority turns into surplusage two-thirds of the False Claims Act’s def- inition of “knowing” added in 1986. I. The Relators’ Claims A. The Relators’ Evidence The majority explains helpfully the important role of “usual and customary” drug prices in Medicare and Medi- caid. Congress has not allowed the government to do what private insurance companies do: use bargaining power to ne- gotiate for lower drug prices. Instead, the government tries to take advantage of private competition in so-called “cash” sales of prescription drugs. See United States ex rel. Garbe v. Kmart Corp., 824 F.3d 632, 644 (7th Cir. 2016). Those are sales to customers whose drug purchases are not covered by insur- ance. Under the statutes and regulations, SuperValu’s “usual and customary” drug prices for those cash sales were caps on what the government would pay SuperValu for drugs pro- vided to Medicare and Medicaid patients. Starting in 2006, Walmart began offering cash sales of ge- neric drug prescriptions for four dollars for a one-month sup- ply and ten dollars for a three-month supply. SuperValu re- sponded to Walmart’s move with an aggressive, widely-ad- vertised price-matching program. The result, giving relators the benefit of reasonable inferences from the evidence, was 32 No. 20-2241

dramatic reductions in the prices SuperValu charged most “cash” customers for many drugs for over a decade. The applicable regulation describes the price cap as “Pro- viders’ usual and customary charges to the general public.” 42 C.F.R. § 447.512(b)(2). Regulations also include this defini- tion: “Usual and customary (U&C) price means the price that an out-of-network pharmacy or a physician’s office charges a customer who does not have any form of prescription drug coverage for a covered Part D drug.” 42 C.F.R. § 423.100. In this appeal, SuperValu does not dispute that under now-binding circuit precedent, a discounted price can be the “usual and customary” price. See Garbe, 824 F.3d at 644–45. SuperValu also does not dispute that it pushed its price match as a matter of company policy and that it usually charged the four-dollar price for many drugs. Relators offered evidence that SuperValu told the federal government for years that its “usual and customary” prices were much higher than those that it actually charged most cash customers for many drugs. The question here is whether plaintiffs have come forward with evidence to support a find- ing that SuperValu made these many false claims “know- ingly.” There is room for reasonable disagreement about exactly how to interpret “usual and customary” prices when a seller matches a competitor’s prices to keep a customer. That room for argument, says the majority, entitles SuperValu to sum- mary judgment. As applied to these facts, though, it should be easy to find that SuperValu’s claims were false and that Su- perValu knew they were false. No. 20-2241 33

At one end of a spectrum, imagine a local mom-and-pop pharmacy that occasionally grants a few customers’ informal requests for lower prices after some comparison shopping. At the other end, imagine a nationwide chain with a nationwide program advertising that the seller will match any competi- tor’s lower prices. Then imagine that the seller tells its phar- macists and cashiers to offer the discounted prices to all cus- tomers paying cash for drugs (i.e., without insurance or gov- ernment coverage). And then imagine that the seller makes a majority of its cash drug sales at the discounted rates, not at the much higher prices that it officially tells the government are “usual and customary.” Relators’ evidence here fits this end of the spectrum—SuperValu’s price matches were availa- ble to any members of the general public, who were encour- aged to ask for them. Then consider relators’ evidence about the results of this nationwide, decade-long program. Focus on SuperValu’s sales of the fifty highest-volume drugs, where most of the rel- evant money is. For forty-four of those top fifty drugs, Super- Valu was making a majority of its cash sales for less than its claimed “usual and customary” prices. For thirty of those drugs, SuperValu was making more than eighty percent of its cash sales for less than its claimed “usual and customary” prices. Then consider that SuperValu was claiming that its “usual and customary” prices for those drugs were as much as eight to fifteen times the discounted prices it was actually charging most of the time. See Dew Rebuttal Report at 7–8. Given those facts, a reasonable jury could easily find both that SuperValu’s claims for reimbursement based on its “usual and customary” 34 No. 20-2241

prices were false under any reasonable interpretation of the term and that SuperValu knew its claims were false. B. Ambiguity and Knowing Fraud Smart lawyers and judges can debate exactly how to de- fine “usual and customary” under the infinite variety of situ- ations we might hypothesize. But with respect, I do not see room for reasonable disagreement about whether claimed prices eight to fifteen times the actual cash prices that Super- Valu charged most of the time were in any sense “usual and customary.” Without even reaching the direct evidence of knowledge, discussed below, a reasonable jury could infer that SuperValu’s decade-long practice of claiming higher re- imbursement levels by disregarding the much higher prices it actually charged a majority of the time was a “knowing” fraud on the government. See 31 U.S.C. § 3729(a)(1). The dis- connect between its representations (the higher prices were usual and customary) and reality (much lower prices were charged most of the time) is great enough that a jury could infer knowledge, as the term is defined in the False Claims Act, on that basis alone. Then we come to the more direct evidence that SuperValu knew that what it was doing was fraudulent. The huge gaps between actual sale prices and claimed “usual and custom- ary” prices did not escape notice by executives. Documents show that they paid close attention to the results of the price- matching program. That’s no surprise. The program re- sponded to a major disruption in retail drug markets, with a big financial impact on cash sales. The executives also knew it had huge implications for the even higher volume of Medi- care and Medicaid sales of those drugs. The executives esti- mated that the correct application of “usual and customary” No. 20-2241 35

prices could cost SuperValu tens of millions of dollars per year. Executives recognized that widespread price-matching could undermine what they euphemistically called the “in- tegrity” of SuperValu’s “usual and customary” price claims for government reimbursement as price-matching became more than an “‘exception’ for customer services reasons.” And in the face of that concern, they chose what one called in an email the “stealthy” approach (scare-quotes in the Super- Valu original) to ensure that word about this “exception” did not reach too many customers. The problem, of course, is that SuperValu’s price-matching was not only widespread but also advertised in all its stores. It knew that these practices were undermining the “integrity” of its certifications to the govern- ment, yet went forward anyway. The False Claims Act requires proof that a defendant knowingly submitted false claims. It defines “knowing” of fal- sity to include acting “in deliberate ignorance of the truth or falsity” of the information submitted to the government. 31 U.S.C. § 3729(b)(1). A jury could reasonably find that Super- Valu’s widespread adoption of price-matching on a scale far beyond an “exception” was at least a deliberate choice to re- main ignorant about whether its ongoing claims based on supposedly “usual and customary” prices were false. A jury could also reasonably infer actual knowledge from the obvi- ous and known effects of the gap between most actual sale prices under the nationwide price-matching and the claimed “usual and customary” prices. See Farmer v. Brennan, 511 U.S. 825, 836 (1994) (obvious risk of harm justifies inference of knowledge), cited in Safeco Insurance, 551 U.S. at 68. 36 No. 20-2241

SuperValu of course has arguments and evidence pointing toward its honesty and innocence. But we are reviewing a grant of summary judgment. The account set forth above is a reasonable view of the evidence in the light most favorable to the non-moving relators. A reasonable jury could find that Su- perValu either actually knew or deliberately chose to keep it- self in ignorance that it was submitting false, hugely inflated claims for reimbursement. SuperValu does not dispute that it was selling forty-four of the top fifty drugs most of the time for much less than it claimed to the government were its “usual and customary” prices. Nor does it dispute that it was selling thirty of those drugs more than eighty percent of the time for much less than its claimed “usual and customary” prices. Instead, SuperValu points out that relators’ case is not limited to those high-vol- ume drugs (“cherry-picked examples,” says SuperValu). Per- haps, but even if the relators tried to reach too far with other drugs, that would not mean their claims based on the “cherry- picked” drugs lack merit. The evidence of SuperValu’s actions and state of mind regarding those “cherry-picked” drugs can also shed light on others. After all, the price-match program covered lots of drugs over the decade it was in place. Relators’ case here is factually complex because of time, geography, and the number of drugs involved. Their claims span a decade, during which SuperValu’s price-matching practices changed in arguably important ways. Their claims also span drug sales across a host of local and regional retail markets with different competitors and matched prices. And their claims cover hundreds of different drugs. That complex- ity should not distract us from the sound theory at the core of relators’ case. Where the price-matching program produced a No. 20-2241 37

majority of actual sales at prices below the claimed “usual and customary” prices, the claimed prices could no longer be hon- estly deemed “usual and customary.” II. Knowledge Under the False Claims Act SuperValu and the majority do not dispute this evidence or even the inferences that relators seek to draw from it. In- stead, SuperValu and the majority say the evidence of Super- Valu’s actual knowledge and intentions is “irrelevant.” Ante at 25. If that’s correct, this case creates a safe harbor for fraud- sters who claim taxpayer funds in bad faith, but whose barely- straight-faced lawyers offer an innocent explanation for their conduct. The majority even says it is irrelevant whether Su- perValu actually believed and/or relied upon the post hoc jus- tifications offered in litigation. “[I]t is not enough that a de- fendant suspect or believe that its claim was false.” Id., citing United States ex rel. Purcell v. MWI Corp., 807 F.3d 281, 288 (D.C. Cir. 2015); ante at 20 (“A defendant might suspect, be- lieve, or intend to file a false claim, but it cannot know that its claim is false if the requirements for that claim are un- known.”). From 40,000 feet, that interpretation of the False Claims Act, or any cause of action for fraud, is extraordinary. It is a standard of knowledge we do not accept in any other areas of law, including criminal law. How many chief financial offic- ers could say they did not “know”—not really—that the earn- ings reports were inflated, even if they suspected or believed they were? How many drug couriers could assert they did not really “know” that they were carrying drugs? Federal law- suits and prosecutions are not seminars in such radical epis- temological doubt. Federal courts routinely give “ostrich” in- structions in response to such defenses, even in criminal cases: 38 No. 20-2241

“You may find that the defendant acted knowingly if you find beyond a reasonable doubt that he believed it was highly probable that [state fact as to which knowledge is in question, e.g., ‘drugs were in the suitcase,’ ‘the financial statement was false,’] and that he took deliberate action to avoid learning that fact.” Seventh Circuit Pattern Criminal Jury Instructions 4.10 (2020). We would never accept a defense theory based on such Cartesian doubt, and certainly not as a matter of law, in any other case requiring proof of knowledge of the key facts. A. Statutory Text and the Common Law Looking at the analysis more closely, the majority’s inter- pretation conflicts with the statutory text of the False Claims Act, its common-law foundations, and its history and pur- poses. Let’s start with the text of the Act. The key language imposes liability on a person who “knowingly presents, or causes to be presented, a false or fraudulent claim for pay- ment or approval” or who “knowingly makes, uses, or causes to be made or used, a false record or statement material to a false or fraudulent claim.” 31 U.S.C. § 3729(a)(1). The scienter standard is “knowingly,” and the Act then de- fines the term: (b) Definitions. For purposes of this section– (1) the terms “knowing” and “knowingly”– (A) mean that a person, with respect to information– (i) has actual knowledge of the information; (ii) acts in deliberate ignorance of the truth or falsity of the information; or (iii) acts in reckless disregard of the truth or falsity of the information; and No. 20-2241 39

(B) require no proof of specific intent to defraud…. 31 U.S.C. § 3729(b). The three prongs of the statutory definition closely track the most authoritative summary of the common law’s treat- ment of fraudulent scienter, used by the Supreme Court to in- terpret the False Claims Act. The Restatement (Second) of Torts § 526 (1977) also offers three prongs: A misrepresentation is fraudulent if the maker (a) knows or believes that the matter is not as he represents it to be, (b) does not have the confidence in the accuracy of his representation that he states or implies, or (c) knows that he does not have the basis for his representation that he states or implies. The majority itself emphasizes that the Supreme Court has interpreted the False Claims Act consistently with the com- mon law of fraud. Ante at 14, quoting Universal Health Ser- vices, Inc. v. United States ex rel. Escobar, 136 S. Ct. 1989, 1999 & n.2 (2016). That’s certainly correct. Escobar relied on the Re- statement (Second) of Torts, as did Safeco in interpreting the Fair Credit Reporting Act. 551 U.S. at 69. 1

1The majority thinks § 526 is irrelevant in interpreting the False Claims Act’s scienter standard, and that § 500 is a better guide because that’s what Safeco cited for “reckless disregard.” Ante at 16, citing Safeco, 551 U.S. at 69. That reasoning is circular. Section 500 addresses reckless disregard for the safety of another person. In other words, the majority is relying on the common law of reckless driving, not the common law of fraud. Safeco seems to have cited § 500 for lack of anything more pertinent to violations of the technical notice requirements of the Fair Credit 40 No. 20-2241

As Restatement § 526 shows, the common law definition of fraud makes subjective bad faith central to fraudulent sci- enter. Yet the majority concludes that bad faith is irrelevant … in a fraud case! I would follow the Restatement, as echoed in the text of the False Claims Act itself. A reasonable jury could infer that SuperValu “knew” or “believed” that its higher prices were not its usual and customary prices, or, at the very least, did “not have the confidence in the accuracy” of its rep- resentations to the United States government that its certifica- tions stated or implied. But see ante at 20 (“A defendant might suspect, believe, or intend to file a false claim, but it cannot know that its claim is false if the requirements for that claim are unknown.”).

Reporting Act. But § 526 appears in the Restatement Division on Misrep- resentation, the Chapter on Misrepresentation and Nondisclosure Caus- ing Pecuniary Loss, the Topic on Fraudulent Misrepresentation (Deceit), and Title A, Fraudulent Character of Misrepresentation. Section 526 is ti- tled “Conditions Under Which Misrepresentation is Fraudulent (Scien- ter).,” Each of its three prongs is phrased in terms of what the maker of the misrepresentation “knows.” Thus, for the common-law understanding of the False Claims Act’s definition of “knowing,” § 526 is right on target. (In Escobar, the Supreme Court relied on § 529, from the same topic on fraudulent misrepresentations. 136 S. Ct. at 1999.) And I confess to being baffled by the majority’s assertion: “We decline to graft aspects of com- mon law fraudulent scienter into the FCA when Congress chose not to include such requirements.” Ante at 17. With respect, given the majority’s stated adherence to common-law understandings, what the maker of the false claims believes or suspects fits squarely into both the second and third prongs of 31 U.S.C. § 3729(b) and Restatement (Second) of Torts § 526. The common law of reckless driving (§ 500) does not provide the relevant scienter standard for a fraud case or a fraud statute. No. 20-2241 41

B. Origins of the Statutory Definition The majority fails to appreciate the importance of the False Claims Act’s textual definitions of “knowingly” and their common-law roots. The majority instead focuses on the Su- preme Court’s interpretation of a different term in a different statute. That’s a mistake. The False Claims Act’s three-part definition of knowingly, with the disclaimer that specific in- tent to defraud is not required, did not come from nowhere. It was a clear instruction from Congress to courts to relax their restrictive interpretations of “knowing” under the Act. Before 1986, the False Claims Act used the terms “know- ing” and “knowingly” without elaboration. When Congress added the definitions in 1986, it acted in response to court de- cisions that were making it difficult to bring claims against dishonest claimants absent clear evidence of actual knowledge of the falsity of the claim. We should not ignore this history. The statutory text and history show Congress’s clear intent to allow False Claims Act lawsuits to proceed against businesses that fail to do basic due diligence in re- sponse to warning signs that their government payments are ill-gotten. 2

2 The majority asserts it is an error to rely on statutory history to go “beyond the text” of the statute. Ante at 17. If the statutory text were clear as applied to this case, I might agree, but the majority obviously does not believe the statutory text of § 3729 is clear. Otherwise the majority would not need to rely on Safeco, addressing a different statute and different sci- enter standard. Since the text is not self-explanatory, it makes good sense to use reliable evidence to figure out what problem Congress was trying to solve. See also Safeco itself, where the Supreme Court said it was decid- ing as it did because there was “no indication that Congress had some- thing different in mind.” 551 U.S. at 69. The Court’s comment invites 42 No. 20-2241

The amended three-pronged definition of “knowledge” in the False Claims Act was added in 1986 as part of a broader revision to the Act. As sponsor Senator Grassley explained, the government needed “lots of help” from Congress to iden- tify fraudsters and bring them to justice. 132 Cong. Rec. S11243 (Aug. 11, 1986). Expanding the statute’s definition of “knowledge” to reach broader degrees of culpability was an important tool to reach that goal. Id. The problem, as explained in the Senate Committee report, was that courts had applied too narrow a definition of “knowledge,” often requiring actual literal knowledge of a claim’s falsity or even specific intent to defraud to find liabil- ity under the Act. S. Rep. 99-345 at 7, citing United States v. Aerodex, 469 F.2d 1003 (5th Cir. 1972) (collecting cases). Given the “remedial” goals of the False Claims Act, the Committee sought to prevent courts from allowing unscrupulous claim- ants, acting in bad faith, to evade liability through legal tech- nicalities about the definition of “knowledge.” See S. Rep. 99- 345 at 7, 21. The result of these earlier court decisions had been pre- dictable: unscrupulous claimants could structure claim-pro- cessing procedures so that false claims could be filed without the relevant decision-makers truly “knowing” of the fraud. Id. at 7. Even if hints of possible wrongdoing surfaced, decision- makers could insulate themselves from liability by ignoring problems that even a cursory investigation would have un- covered. Id.

reliance on statutory history to answer these questions where the text is not entirely clear. No. 20-2241 43

In explaining the statutory text, the House Judiciary Com- mittee noted the problems from the lack of a definition of “knowledge” and reported: By adopting this [three-pronged] definition of knowledge, the committee intends not only to cover those individuals who file a claim with ac- tual knowledge that the information is false, but also to confer liability upon those individuals who deliberately ignore or act in reckless disre- gard of the falsity of the information contained in the claim. It is intended that persons who ignore “red flags” that the information may not be accurate or those persons who deliberately choose to remain ignorant of the process through which their company handles a claim should be held liable under the Act. This definition, therefore, enables the Govern- ment not only to effectively prosecute those per- sons who have actual knowledge, but also those who play “ostrich.” H. Rep. 99-660 at 21 (emphasis added). The Senate Committee also focused on proverbial “os- triches” who stick their heads in the sand instead of verifying that they are not cheating taxpayers. S. Rep. 99-345 at 7, 15, 21. These ostriches need not have “conscious culpability” of wrongdoing: people who submit claims that they have “rea- son to know” are potentially false run the risk of violating the Act if they “fail[] to inquire” as to the falsity of the claims. 132 Cong. Rec. S11243–44 (Aug. 11, 1986; statement of Senator Grassley). 44 No. 20-2241

The Committee reports explained that the added defini- tion was aimed at claimants who acted in bad faith by failing to investigate potential problems: “those doing business with the Government have an obligation to make a limited inquiry to ensure the claims they submit are accurate.” S. Rep. 99-345 at 7. Congress chose statutory language that could have been custom-tailored for SuperValu’s approach in this case. Super- Valu knew that the “integrity” of its “usual and customary” prices would be suspect if price-matching were not the “ex- ception” but the rule, yet it kept submitting those claims through a nationwide price-matching campaign anyway, net- ting tens of millions of dollars of public funds annually. This is the same standard that the Eleventh Circuit adopted in United States ex rel. Phalp v. Lincare Holdings, Inc., 857 F.3d 1148 (11th Cir. 2017), another case involving arguable regulatory ambiguity. After considering the statutory text and legislative history, the court concluded that “scienter is not determined by the ambiguity of a regulation, and can exist even if a defendant’s interpretation is reasonable.” Id. at 1155, citing the Senate Committee Report indicating Congressional intent to require claimants to engage in “limited inquiry.” Phalp also squarely rejected the majority’s position here: “The district court’s conclusion that a finding of scienter can be pre- cluded by a defendant’s identification of a reasonable inter- pretation of an ambiguous regulation that would have per- mitted its conduct is erroneous.” Id. (Phalp’s treatment of this issue refutes the majority’s attempt to explain it away. See ante at 15.) The Phalp court’s interpretations of the Act’s sci- enter definition should be obviously correct. In fact, before the Safeco progeny cited by the majority, our colleagues in other circuits followed the amended text of the No. 20-2241 45

False Claims Act and common sense: a claimant could be lia- ble under the Act notwithstanding a purported regulatory ambiguity if the defendant deliberately ignored the falsity of the claim or otherwise acted in bad faith. United States v. Sci- ence Applications International Corp., 626 F.3d 1257, 1272–73 (D.C. Cir. 2010) (affirming verdict for United States; jury could infer that defendant knew its claims were false notwithstand- ing “regulatory divide” in how to interpret a regulation); United States ex rel. Oliver v. Parsons Co., 195 F.3d 457, 464 (9th Cir. 1999) (reversing summary judgment: “In short, [defend- ant’s] petition arguing that the sky will fall upon government contractors if they are precluded from relying on a ‘reasona- ble interpretation’ is not only unsupported by case law, it is also ungrounded in reality.”); see also Minnesota Ass’n of Nurse Anesthetists v. Allina Health System Corp., 276 F.3d 1032, 1053 (8th Cir. 2002) (citing Parsons for the proposition that “any possible ambiguity of the regulations is water under the bridge” where contractor’s misinterpretation is “knowing”). 3 In this case, the relators’ evidence shows that SuperValu knew it was claiming high “usual and customary” prices that it was charging less than half the time, often less than one fifth

3 We took a similar approach in United States ex rel. Sheet Metal Workers

Int'l Ass’n, Local Union 20 v. Horning Investments, LLC, 828 F.3d 587 (7th Cir. 2016). The defendant argued that it relied on advice of professional experts in determining that its claims were not false. We rejected that ar- gument on grounds inconsistent with the majority approach here. Rather than treat a professional’s ability to find ambiguity as a defense in itself, we applied a much more demanding five-part test that required proof of timely, good-faith, and full disclosure to competent experts. Id. at 594–95. We ultimately affirmed summary judgment for the defendants, but on a different ground, that the relator simply did not have evidence that de- fendants were on notice that their claims were false. Id. at 595. 46 No. 20-2241

of the time. SuperValu knew that its practices raised questions about the “integrity” of its “usual and customary” prices but nonetheless ignored those concerns. The False Claims Act’s statutory definition of “knowing” reaches those who know their claims are false or who act in deliberate ignorance of whether their claims were true or false. We should reverse summary judgment for SuperValu. III. The Majority’s Safeco Tangent Rather than focusing on the language of the False Claims Act itself, and its origins in the common law of fraud and re- sponses to crabbed judicial interpretations, the majority opin- ion takes a very different approach. It borrows the Supreme Court’s treatment of a different term, “willfully,” under a dif- ferent statute, the Fair Credit Reporting Act, in Safeco Insur- ance Co. v. Burr, 551 U.S. 47 (2007). The majority adopts Safeco’s treatment of reckless disregard for law as a branch of “willful” misconduct. The majority then goes even further and concludes that relators must meet that standard for reck- less disregard for any False Claims Act case, even if they rely on the actual-knowledge or deliberate-ignorance prongs of the Act’s definition of knowing. The majority makes two fundamental mistakes. First, the reliance on Safeco to understand “reckless disregard” is nei- ther necessary nor fitting for the False Claims Act. The Act draws on a different branch of the common law (of fraud, not reckless driving), and the history of the statutory amend- ments shows that Congress thought it was enacting a stand- ard quite different from the majority’s. Second, by saying re- lators must satisfy the Safeco reckless-disregard standard in any case, the majority effectively nullifies two-thirds of the statutory definition of “knowing.” To explain: No. 20-2241 47

The question in Safeco was whether an insurer’s decision about an initial premium rate for an insured could qualify as an “adverse action” based on a credit report that could require notice to the consumer in question. The Supreme Court ulti- mately held that it could but also held that Safeco had not “willfully” violated that Act because the statute and regula- tion were not clear as applied to initial premium decisions, so that Safeco had not acted willfully. The Fair Credit Reporting Act does not define “willfully,” which the Court described as a “word of many meanings whose construction is often dependent on the context in which it appears.” 551 U.S. at 57, quoting Bryan v. United States, 524 U.S. 184, 191 (1998). Without more specific guid- ance for interpreting the term in that act, the Safeco Court had little choice but to construct a working definition from multi- ple sources. The Court focused on civil law, noting that in sev- eral civil contexts, willful violations of statutes could be shown by recklessness, which was consistent with common- law use of the term. Id. Then, because there was “no indication that Congress had something different in mind,” and because the term “recklessness” is not “self-defining,” the Court an- nounced an application of that scienter standard to the Fair Credit Reporting Act. Id. at 57–58, 68–69. The Court then drew on common-law definitions of “reck- lessness” that apply to actions putting others in physical dan- ger. The Court described recklessness as action entailing “an unjustifiably high risk of harm that is either known or so ob- vious that it should be known,” id. at 68, quoting Farmer v. Brennan, 511 U.S. 825, 836 (1994), and conduct involving “un- reasonable risk of physical harm … substantially greater than that which is necessary to make his conduct negligent. Id. at 48 No. 20-2241

69, quoting Restatement (Second) of Torts § 500 (also regard- ing putting another person in physical danger). The Court summarized its view for the Fair Credit Reporting Act: There being no indication that Congress had some- thing different in mind, we have no reason to de- viate from the common law understanding in applying the statute. Thus, a company subject to FCRA does not act in reckless disregard of it un- less the action is not only a violation under a reasonable reading of the statute’s terms, but shows that the company ran a risk of violating the law substantially greater than the risk asso- ciated with a reading that was merely careless. 551 U.S. at 69 (emphasis added; citation omitted). Along the way, the Court added footnote 20, saying that evidence of subjective bad faith would not be relevant to the definition of willfulness in 15 U.S.C. § 1681n(a) where a company followed an objectively reasonable interpretation of the Fair Credit Re- porting Act. The majority here and four other circuits have borrowed this reasoning from Safeco and grafted it onto the False Claims Act. Two of those circuits did so in non-precedential deci- sions. The two precedential decisions are United States ex rel. Purcell v. MWI Corp., 807 F.3d 281 (D.C. Cir. 2015), and United States ex rel. Donegan v. Anesthesia Associates of Kansas City, 833 F.3d 874 (8th Cir. 2016). The Eleventh Circuit reached a differ- ent conclusion in Phalp, 857 F.3d 1148, discussed above. (The Phalp opinion did not discuss Safeco or Purcell, but both cases were briefed extensively, including by the United States in an amicus brief arguing that Safeco provided no meaningful guidance for False Claims Act cases. There is no doubt that No. 20-2241 49

the Eleventh Circuit rejected Purcell’s borrowing of Safeco. It did not cite Safeco because, for reasons explained here, Safeco simply is not needed to interpret the scienter requirement of the False Claims Act.) In the absence of better guidance for the False Claims Act and common law, reliance on Safeco might be understandable, if a bit of a stretch. The majority here errs, however, by over- looking Safeco’s directive: first check to see if “Congress had something different in mind.” 551 U.S. at 57, 69. With the False Claims Act, we do have meaningful guidance from the statu- tory text, the common law, and legislative history, as dis- cussed above. If the majority limited its reliance on Safeco to the reckless- disregard prong of the False Claims Act’s definition of know- ing, its mistake would be more understandable. It’s the ma- jority opinion’s next move that is more extraordinary and much more damaging. The majority concludes that a relator under the False Claims Act must satisfy the Safeco definition of reckless disregard—show that no reasonable understand- ing of law could justify the defendant’s action, or show that the defendant disregarded “authoritative guidance”—in every case, even those relying on the actual-knowledge and deliber- ate-ignorance prongs of the definition of “knowingly.” As a result, the majority holds in effect that those two-thirds of the statutory definition add zero meaning to the statute. The majority’s major premise is that “reckless disregard” is the broadest of the three prongs. Its minor premise is that any case of “actual knowledge” or “deliberate ignorance” would always fall within “reckless disregard,” as that term was defined in Safeco. That too-simple heuristic may be useful in 50 No. 20-2241

some easy cases, but its application here is inconsistent with how courts should read statutes. The key logical error lies in the minor premise, that any case of actual knowledge or deliberate ignorance would nec- essarily also be covered by Safeco-reckless disregard. There is no basis for that assumption, which leads away from the com- mon law of fraud, where subjective bad faith is central. Consider a hypothetical close to this case. A government contractor submits claims believing, subjectively, that the claims are probably false. The agency has not yet provided what Safeco would call “authoritative guidance,” but the con- tractor reads the controlling regulation (correctly) to preclude its claim. Still, it decides to stay quiet, hoping it will not get caught, or at least not too quickly. In that situation, judges and jurors can say that claims were fraudulent and the contractor knew it, even if a creative lawyer can later make a non-frivo- lous legal argument for its innocence. Likewise, the contractor acted with fraudulent intent because it “believed” the claims were false and submitted claims in which it did not have the “confidence” it claimed. See 31 U.S.C. § 3729(b); Restatement (Second) of Torts § 526. This bad-faith “catch us if you can” approach to public funds is exactly what Congress thought it was outlawing when it decided in 1986 that it needed to define “knowledge” more specifically for the False Claims Act, including to reach deliberate ignorance of falsity. Recall also that under the ma- jority’s approach, there is no need for a defendant to show that it actually “followed” any “objectively reasonable” inter- pretation of the law that would supposedly save the claims from being false. See ante at 25. No. 20-2241 51

The majority’s logic thus takes the False Claims Act in a direction 180 degrees away from common-law fraud. It makes subjective bad faith, including deliberate ignorance, “irrele- vant.” Id. That’s contrary to both the actual-knowledge and deliberate-ignorance prongs of the Act’s textual definition. It loses sight of the fact that the Act applies to “fraudulent” con- duct. And it’s also contrary to the common-law scienter stand- ard in the Restatement (Second) of Torts, which is satisfied if the defendant “knows or believes that the matter is not as he represents it to be,” or if he “does not have the confidence in the accuracy of his representation that he states or implies….” § 526 (emphasis added). The majority rests heavily on Safeco’s footnote 20 to sup- ports its new safe harbor where subjective state of mind is ir- relevant. See ante at 19. With respect, the majority reads far too much into that footnote, which by its own terms is limited to “determining whether a company acted knowingly or reck- lessly for purposes of § 1681n(a).” By the majority’s reading, that footnote in an opinion on credit reporting requirements, which borrowed from the common law of reckless driving, upended the common law of fraud, one of the paradigmatic intentional torts, where state of mind is critical. The Safeco Court gave no sign that its footnote intended to reach beyond § 1681n(a) or that it was creating a new element for fraud claims—the absence of any plausible reading that would ren- der the false statement true. The majority’s too-broad reading leads it to depart from the text of the False Claims Act and loses sight of Congress’s clear intent. In fact, the Supreme Court itself has warned against read- ing Safeco’s footnote 20 so broadly. It did so in Halo Electronics, Inc. v. Pulse Electronics, Inc., 136 S. Ct. 1923, 1933 n.* (2016). The 52 No. 20-2241

Court declined to extend the Safeco definition of “willfully” to treble-damage awards for patent infringement under 35 U.S.C. § 284. Subjectively bad-faith infringement, focused on the defendant’s state of mind when it acted, had long been an independent basis for enhanced patent damages. 136 S. Ct. at 1933 & n.*. As the majority points out here, ante at 17–18, in Halo Electronics, differences in the two statutes produced dif- ferent scienter standards. Exactly the same reasoning should apply here. The differences between the texts and histories of the Fair Credit Reporting Act and False Claims Act should lead us to decline to extend the Safeco standard and its foot- note 20 to the False Claims Act. Returning to False Claims Act cases, consider, for exam- ple, United States ex rel. Prather v. Brookdale Senior Living Com- munities, Inc., 892 F.3d 822, 837–38 (6th Cir. 2018), where the Sixth Circuit reversed dismissal of a relator’s complaint. The complaint alleged that the relator and other nurses had “con- cerns about the defendants’ compliance with Medicare regu- lations, but were told to ignore any problems.” When relator raised issues about regulatory compliance, executives told her on multiple occasions that “‘[w]e can just argue in our favor if we get audited’ as a solution to any compliance issues.” The Sixth Circuit reasoned that the allegations about notice of compliance problems imposed an obligation on the defend- ants to inquire whether they were actually in compliance with regulations. The Sixth Circuit concluded the allegations sup- ported “knowledge” under both the deliberate-ignorance and reckless-disregard prongs of the definition. Id. at 838. Yet un- der the majority’s approach here, that case would have been dismissed so long as an attorney could later offer a barely- plausible theory of innocence, even though the defendant ig- nored repeated and correct warnings that it was violating the No. 20-2241 53

regulations. Worse yet, the majority here would have dis- missed the case even if the supervisors had admitted that they knew their submissions were non-compliant. The majority’s bottom line—that only objectively reckless disregard matters, and subjective bad faith does not—also vi- olates one of the most common tools of statutory interpreta- tion. It renders the actual-knowledge and deliberate-igno- rance prongs of the statutory definition utterly superfluous. See City of Chicago v. Fulton, 141 S. Ct. 585, 591 (2021) (“The canon against surplusage is strongest when an interpretation would render superfluous another part of the same statutory scheme.”), quoting Yates v. United States, 574 U.S. 528, 543 (2015) (plurality); National Ass’n of Mfrs. v. Dep’t of Defense, 138 S. Ct. 617, 632 (2018), quoting Reiter v. Sonotone Corp., 442 U.S. 330, 339 (1979); In re Southwest Airlines Voucher Litig., 799 F.3d 701, 710 (7th Cir. 2015); Antonin Scalia & Bryan Garner, Reading Law: The Interpretation of Legal Texts 174 (2012) (“The surplusage canon holds that it is no more the court’s function to revise by subtraction than by addition.”). The canon against surplusage is not absolute, of course. Sometimes drafters of legal documents may “intentionally err on the side of redundancy to ‘capture the universe.’” Sterling Nat’l Bank v. Block, 984 F.3d 1210, 1218 (7th Cir. 2021), quoting Abbe R. Gluck & Lisa Schultz Bressman, Statutory Interpreta- tion from the Inside—an Empirical Study of Congressional Draft- ing, Delegation, and the Canons: Part I, 65 Stan. L. Rev. 901, 934 (2013); accord, e.g., Rimini Street, Inc. v. Oracle USA, Inc., 139 S. Ct. 873, 881 (2019); White v. United Airlines, Inc., 987 F.3d 616, 622 (7th Cir. 2021). The False Claims Act definition of “knowingly” is about as strong a case for the canon against surplusage as one is 54 No. 20-2241

likely to find. The three prongs mirror three distinct common- law prongs for fraudulent scienter. Congress adopted them to give courts clearer guidance because Congress was disap- pointed with courts’ interpretations of the undefined “know- ing.” Congressional leaders on the subject, such as Senator Grassley and Representative Berman, were concerned that courts would continue to misinterpret the statute. They ex- plained exactly how the definition of “knowing” should be applied, as did the respective committees. The three prongs may overlap in many cases, but the adoption of the three dis- tinct prongs in the same paragraph of the statutory text was unmistakably an effort to be both thorough and broad. Con- gress said as clearly as it could that the False Claims Act should reach just this kind of case. I close with two final observations about the majority’s misguided holding. First, even under the Safeco standard, a reasonable jury could find that SuperValu’s more extreme conduct here was not reasonable. There is simply no reasona- ble definition of “usual and customary” that means “some- thing we do less than half the time and that we instruct our employees not to do.” Defining “usual and customary” to mean the opposite of what those two words actually mean is simply not reasonable. Second, the majority’s approach actually leaves the False Claims Act definition of knowledge narrower than when the 1986 amendment was passed. Consider, for example, United States v. Mead, 426 F.2d 118, 122–23 (9th Cir. 1970), which Rep- resentative Fish singled out as applying a too-narrow defini- tion of knowledge. 132 Cong. Rec. H6480 (Sept. 9, 1986); see also Aerodex, 469 F.2d at 1007, cited negatively in S. Rep. 99- 345 (collecting Mead as an example of then-operative No. 20-2241 55

knowledge standard). In Mead, the court explained that where regulatory language is uncertain and even the district court misinterpreted the regulations, scienter is still a question of fact. If the government had shown that Mead knew his regu- latory interpretation was wrong or had fraudulent intent, he would still be liable under the Act. See also United States v. Ueber, 299 F.2d 310, 314 (6th Cir. 1962), cited negatively in S. Rep. 99-345 (where contract distinguished between “direct” and “indirect” labor costs, falsity of claims for “direct” labor costs and defendant’s knowledge of their falsity are questions of fact for trial; remanding for further fact-finding). Whatever “reckless disregard” means, we should not use it to narrow the definition of knowledge that Congress thought it was expand- ing. To sum up, relators have come forward with substantial evidence of knowing fraud, as SuperValu claimed reimburse- ment at supposedly “usual and customary” prices for drugs that were as much as eight to fifteen times higher than the prices it was actually charging the general public a majority of the time. The evidence supports a reasonable inference of actual knowledge or at least deliberate ignorance or reckless disregard for whether its reimbursement claims were false. We should reverse summary judgment and remand for trial on relators’ claims. With respect, I believe that both Congress and the Supreme Court will be surprised by this decision and the others extending Safeco to the False Claims Act. If the False Claims Act is to remain effective in discouraging and remedy- ing fraudulent raids on taxpayer dollars, Congress or the Su- preme Court or both will need to respond to this line of cases.