Window Covering Manufacturers Association v. CPSC

82 F.4th 1273

• Court: Court of Appeals for the D.C. Circuit
• Decided: September 12, 2023
• Docket: 22-1300
• Status: Published

Opinion

United States Court of Appeals FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued April 24, 2023 Decided September 12, 2023

No. 22-1300

WINDOW COVERING MANUFACTURERS ASSOCIATION, PETITIONER

v.

CONSUMER PRODUCT SAFETY COMMISSION, RESPONDENT

On Petition for Review of a Final Rule of the Consumer Product Safety Commission

Nicole A. Saharsky argued the cause for petitioner. With her on the briefs were Avi M. Kupfer and Erika Z. Jones.

Tyler S. Badgley, Jordan L. Von Bokern, Brett A. Shumate, Anthony J. Dick, Brinton Lucas, Charles E.T. Roberts, and J. Benjamin Aguiñaga were on the brief for amicus curiae Chamber of Commerce of the United States of America in support of petitioner.

John J. Vecchione was on the brief for amicus curiae New Civil Liberties Alliance in support of petitioner. 2 Paul W. Hughes and Andrew A. Lyons-Berg were on the brief for amicus curiae National Association of Manufacturers in support of petitioner.

Steven A. Myers, Attorney, U.S. Department of Justice, argued the cause for respondent. With him on the brief were Brian M. Boynton, Principal Deputy Assistant Attorney General, and Daniel Tenny, Attorney.

Adina H. Rosenbaum and Michael T. Kirkpatrick were on the brief for amici curiae Consumer Federation of America, et al. in support of respondent.

Elizabeth B. Wydra, Brianne J. Gorod, and Brian R. Frazelle were on the brief for amicus curiae Constitutional Accountability Center in support of respondent.

Before: SRINIVASAN, Chief Judge, WILKINS and PAN, Circuit Judges.

Opinion for the Court filed by Circuit Judge PAN.

PAN, Circuit Judge: The Consumer Product Safety Commission (the “Commission”) issues safety standards for potentially dangerous products. In 2022, the Commission promulgated a rule that set stringent safety standards for the operating cords on custom-made window coverings, based on a finding that such cords pose a strangulation risk to young children. The rule sought to eliminate the risk of injury by essentially prohibiting corded window products, and it set an aggressive timeline for industry compliance with the new standards. The Window Covering Manufacturers Association (“WCMA”) filed a petition in this court challenging the rule and its compliance deadline. Because the Commission breached notice-and-comment requirements, erroneously 3 relied on certain data in its cost-benefit analysis, and selected an arbitrary effective date for the rule, we grant the WCMA’s petition for review and vacate the rule.

I.

A.

In 1972, Congress enacted the Consumer Product Safety Act (the “Act”) to “protect the public against unreasonable risks of injury associated with consumer products.” 15 U.S.C. § 2051(b)(1). The Act created the Commission, an independent regulatory agency, and gave it authority to “promulgate consumer product safety standards” and ban “imminently hazardous” products from the market. Id. §§ 2053, 2056(a), 2057, 2061. The Commission employs staff members who perform regulatory analyses, correspond with industry stakeholders, and make recommendations to the Commission. E.g., 16 C.F.R. §§ 1031.4(a)(3), 1031.6(c)(2). The discretion to promulgate safety standards, however, lies with the Commission itself. See 15 U.S.C. § 2056(a).

The Commission has five commissioners, who are appointed by the President with the advice and consent of the Senate. 15 U.S.C. § 2053(a). Each commissioner must have expertise in “areas related to consumer products.” Id. No more than three commissioners may be members of the same political party. Id. § 2053(c). The Act contains a for-cause removal restriction: Commissioners “may be removed by the President for neglect of duty or malfeasance in office but for no other cause.” Id. § 2053(a).

When promulgating a safety standard, the Commission must follow procedures mandated by the Act, as well as those required by the Administrative Procedure Act (“APA”). See 4 15 U.S.C. § 2060(c) (stating that a reviewing court “shall have jurisdiction to review the consumer product safety rule” in accordance with the APA).

Four of the Act’s provisions are relevant in this case:

First, the Act provides that industry organizations, such as the WCMA, may create “voluntary standards” that render regulation unnecessary. The Commission must refrain from issuing a safety standard if: (i) compliance with the voluntary standard “would eliminate or adequately reduce the risk of injury addressed”; and (ii) it is likely that the industry will substantially comply with the voluntary standard. 15 U.S.C. § 2056(b)(1).

Second, the Act requires the Commission to conduct a “final regulatory analysis” — i.e., a cost-benefit analysis — before promulgating a safety standard. The analysis must detail costs, benefits, and alternatives to the proposed standard, and must address any issues raised by commenters. 15 U.S.C. § 2058(f)(2).

Third, the Commission must “make a host of findings” before proposing a safety standard under the Act. Finnbin, LLC v. Consumer Prod. Safety Comm’n, 45 F.4th 127, 131 (D.C. Cir. 2022). Those findings include: (1) “that the rule (including its effective date) is reasonably necessary to eliminate or reduce an unreasonable risk of injury associated with [the] product”; (2) that any voluntary standard is not likely to eliminate or reduce the risk of injury, or that it is “unlikely that there will be substantial compliance” with the voluntary standard; (3) that the rule’s benefits “bear a reasonable relationship to its costs”; and (4) that the rule “imposes the least burdensome requirement” to prevent or reduce the risk of injury. 15 U.S.C. § 2058(f)(3)(A), (D), (E), (F). 5 Fourth, the Commission must set the proposed safety standard’s effective date. The Act imposes a 180-day effective date by default, “unless the Commission [1] finds, for good cause shown, that a later effective date is in the public interest[;] and [2] publishes its reasons for such finding.” 15 U.S.C. § 2058(g)(1). As noted, the Commission must find that the effective date “is reasonably necessary to eliminate or reduce an unreasonable risk of injury associated with [the] product” regulated. Id. § 2058(f)(3)(A).

In addition, the Commission’s actions must comport with the APA’s familiar requirements. A reviewing court must set aside agency actions found to be “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706. And when an agency conducts a rulemaking, as here, it must provide the public with notice and an opportunity to comment on the content of the proposed rule. Id. § 553(b)–(c).

B.

The United States window covering market boasts sales of roughly $6.7 billion annually and includes approximately 1,900 manufacturers and retailers. Safety Standard for Operating Cords on Custom Window Coverings, 87 Fed. Reg. 73,144, 73,149 (Nov. 28, 2022) (to be codified at 16 C.F.R. pt. 1260) (“Final Rule”).

The industry divides window coverings into two types: “stock” and “custom.” “Stock” window coverings are “completely or substantially fabricated” before being sold to consumers, including pre-assembled products that are modified or adjusted before sale. See Final Rule, 87 Fed. Reg. at 73,148; see also J.A. 1328. “Custom” window coverings refer to “any window covering that is not classified as a stock window 6 covering.” See Final Rule, 87 Fed. Reg. at 73,148; see also J.A. 1328. As one commenter explained, custom products are based on a consumer’s “desired specifications, including dimensions, fabrics, colors, control mechanisms[,] and mounting techniques.” J.A. 709. Custom products account for 44% of unit sales and a “disproportionate amount of revenue” in the market. Final Rule, 87 Fed. Reg. at 73,149. Their prices “can be as high as $5,000” per covering, and custom prices “are on average higher than similar stock products sold by mass retailers.” Id.

Some window coverings are operated with cords, while others are cordless. Final Rule, 87 Fed. Reg. at 73,147. Corded window coverings are at issue in this case. The industry has developed three varieties of corded window coverings. First, “cord-lock[]” systems “consist[] of two or more cords” that the user pulls to raise the covering and lock it into place. Id. at 73,148. Second, “retractable cord-lift” systems require users to pull a solid, plastic wand to raise the window covering, which retracts into a headrail. See id.; see also Opening Br. 12. Finally, “continuous-loop” systems have a beaded cord or chain strung in a loop. Final Rule, 87 Fed. Reg. at 73,148. A user can pull the loop one way to raise the covering or pull in the opposite direction to lower the covering. Id.

Corded window coverings pose a deadly risk to children. When children become tangled in operating cords, they can strangle themselves or suffer serious injuries. Final Rule, 87 Fed. Reg. at 73,150. To reduce that risk, the industry has developed modifications to corded systems. For example, continuous-loop systems can be installed with a “tension device” — a piece of metal or plastic that attaches to a wall and holds the loop taut. Another safety device is a “rigid cord shroud,” which encases the operating cord in a hard material, 7 thus “allow[ing] the user to use the pull cords while eliminating access to the hazardous cords.” Id. at 73,158.

C.

The WCMA is a trade association that “represents the interests of window covering industry manufacturers, fabricators, and assemblers.” Opening Br. 20. In 1996, the WCMA adopted a voluntary safety standard to mitigate the strangulation risk posed to children by operating cords. The voluntary standard established length requirements for operating cords, required that products include safety warnings and tags, and eliminated loop-ended operating cords. Since then, the WCMA has revised its voluntary standard seven times, most recently in 2018 and 2022. The WCMA publishes its voluntary standards through the American National Standards Institute (“ANSI”), a nonprofit organization that administers such standards in U.S. markets. See J.A. 737–39.

The 2018 voluntary standard “[e]liminated corded stock window coverings” altogether “by requiring that these products be cordless, have only inaccessible cords, or have a short helper cord.” J.A. 739; see also J.A. 1328. That standard, however, did not apply to custom products.

Beginning in late 2021, both the WCMA and the Commission took steps to make custom window coverings safer. The WCMA first presented revisions to its voluntary safety standard in January 2022, proposing to eliminate free- hanging cords, to require rigid cord shrouds to pass tests showing that they kept cords inaccessible, and to impose strict requirements to ensure that tension devices kept cords held taut. J.A. 520. The Commission expressed concerns that the WCMA’s revisions did not adequately mitigate the risk to children of strangulation. The WCMA nevertheless approved 8 its revisions on August 15, 2022, and the ANSI published the voluntary standard on December 23, 2022. The WCMA’s 2022 voluntary standard does not take effect until June 2024.

Meanwhile, on January 7, 2022, the Commission published a notice of proposed rulemaking “to require that operating cords on custom window coverings meet the same requirements as [those for] stock window coverings.” Safety Standard for Operating Cords on Custom Window Coverings, 87 Fed. Reg. 1014, 1014 (Jan. 7, 2022) (“Proposed Rule”). In effect, the Commission proposed to eliminate corded products. See id.

The Commission received 2,060 written comments and heard oral comments from stakeholders at a March 16, 2022, hearing. On November 28, 2022, the Commission unanimously approved the Final Rule and published it in the Federal Register. Final Rule, 87 Fed. Reg. at 73,144. The Final Rule requires custom window coverings to be cordless, to have inaccessible operating cords, or to have operating cords “equal to or shorter than 8 inches.” Id. at 73,145. The Final Rule discusses the 2018 and 2022 versions of the WCMA’s voluntary standard, responds to comments, and makes specific findings about costs and benefits. Over the objections of hundreds of commenters, the Small Business Administration, and the Commission’s own staff, the Commission set a 180- day effective date for the industry to comply with the Final Rule — May 30, 2023. Id. at 73,190.

With the Final Rule’s effective date fast approaching, the WCMA petitioned this court for review and moved for a stay pending appeal. A special panel granted that motion, stayed the promulgation of the Final Rule, and ordered an expedited briefing schedule. 9 The WCMA has representational standing in this matter, as its members consist of companies who are the “object of the action . . . at issue.” Bonacci v. TSA, 909 F.3d 1155, 1159 (D.C. Cir. 2018). We have jurisdiction to review the Final Rule under 15 U.S.C. § 2060(c).

II.

On appeal, the WCMA challenges the safety standard adopted in the Final Rule, arguing that the Commission: (1) violated notice-and-comment requirements under the APA; (2) acted arbitrarily and capriciously under the APA and violated the Act by giving insufficient weight to the voluntary standard and by performing an erroneous cost-benefit analysis; (3) set an arbitrary 180-day effective date for the Final Rule, in violation of the APA and the Act; and (4) had no authority to promulgate the challenged rule because its members are subject to an unconstitutional for-cause removal provision. For the following reasons, we find several of the WCMA’s arguments under the APA and the Act meritorious. We therefore vacate the Final Rule. Accordingly, we need not reach the WCMA’s constitutional argument.

When an agency seeks to promulgate a rule, the APA requires it to provide “notice” of “either the terms or substance of [a] proposed rule or a description of the subjects and issues involved,” and then “give interested persons an opportunity to participate in the rule making through submission of written data, views, or arguments.” 5 U.S.C. § 553(b)(3), (c). The WCMA argues that the Commission violated notice-and- comment requirements by failing to disclose injury-incident data on which the Proposed Rule and Final Rule relied, by changing its cost-benefit analysis in the Final Rule without 10 industry input, and by ignoring revisions to the industry’s voluntary standard that made the Final Rule unnecessary. We agree that the Commission erred by failing to disclose its injury-incident data but reject the WCMA’s other arguments.

1.

In performing the statutorily required cost-benefit analysis for its proposed safety standard, the Commission evaluated the potential benefits of the Final Rule by estimating how many injuries would be avoided if the Final Rule took effect. See J.A. 580 tbl.15 n.*; see also J.A. 582 tbl.17. The Commission relied on a proprietary database of its own creation to estimate the number of injuries that have been caused by corded window coverings. See Final Rule, 87 Fed. Reg. at 73,152.

“Based on newspaper clippings, consumer complaints, death certificates purchased from states, medical examiners’ reports, reports from hospital emergency department-treated injuries, and in-depth investigation reports,” the Commission found that at least 209 fatal and near-miss strangulations involving operating cords had occurred between January 2009 and December 2021, among children who were eight years old and younger. Final Rule, 87 Fed. Reg. at 73,151–52, 73,152 tbl.2a. The Commission could not determine the type of window covering (i.e., stock or custom) for 123 of the 209 reported incidents. Id. at 73,152 tbl.2b.

The Commission projected that the Final Rule would yield benefits of approximately $23 million if it were enacted. That figure relied on the Commission’s estimate of annual deaths and injuries that would be caused by corded window coverings in the absence of the Final Rule, the “value of a statistical life” of each child who might be injured, and the distribution of expected injuries across different types of window coverings. 11 Final Rule, 87 Fed. Reg. at 73,182. At many steps in this analysis — for example, when estimating the annual deaths and injuries caused by custom window coverings — the Commission relied on its injury-incident database.

According to the WCMA, the Commission violated the APA by failing to disclose internally compiled reports concerning the 209 injury incidents included in the database. Instead, the Commission shared redacted versions of the incident reports, but only after the WCMA filed multiple Freedom of Information Act (“FOIA”) requests. The Commission then produced an undated spreadsheet summarizing the purported injuries, but only after this lawsuit began. The Commission argues that it met the APA’s requirements by providing “summaries of each incident, which were discussed in the notice of proposed rulemaking, and [were] made available to [the WCMA] in connection with this litigation.” Comm’n Br. 49. That approach was proper, in the Commission’s view, “to appropriately protect [the victims’] sensitive personal and medical information.” Id. at 50.

“[W]e have cautioned that the most critical factual material that is used to support the agency’s position on review must have been made public in the proceeding and exposed to refutation.” Air Transp. Ass’n of Am. v. FAA, 169 F.3d 1, 7 (D.C. Cir. 1999) (emphasis omitted); accord Ass’n of Data Processing Serv. Orgs., Inc. v. Bd. of Governors of Fed. Reserve (ADPSO), 745 F.2d 677, 684 (D.C. Cir. 1984). It “is the agency’s duty to identify and make available technical studies and data that it has employed in reaching the decisions to propose particular rules.” Owner-Operator Indep. Drivers v. FMCSA, 494 F.3d 188, 199 (D.C. Cir. 2007) (cleaned up). “Disclosure of staff reports allows the parties to focus on the information relied on by the agency and to point out where that information is erroneous or where the agency may be drawing 12 improper conclusions from it.” Nat’l Ass’n of Regul. Util. Comm’rs v. FCC (NARUC), 737 F.2d 1095, 1121 (D.C. Cir. 1984).

The Commission violated the foregoing principles and acted arbitrarily by failing to disclose the individual incident reports underlying its injury-incident dataset. To make its statutorily mandated cost-benefit assessment, the Commission estimated the number of injuries that would be caused by custom window coverings if the Final Rule were not promulgated, based on historical information from its proprietary dataset. Thus, the injury-incident dataset contained “critical factual material” that supported the rulemaking. Air Transp. Ass’n of Am., 169 F.3d at 7. A “genuine interchange” about the accuracy of the data did not occur during the rulemaking, Conn. Light & Power Co. v. NRC, 673 F.2d 525, 530 (D.C. Cir. 1982), because no usable information was provided to the public at that time. Instead, the agency provided “about one-third of the reports” in “heavily redacted” form to the WCMA in response to certain FOIA requests. Reply Br. 21; see also J.A. 737 n.8. And the agency’s summary of the underlying data was disclosed only to the WCMA, long after the Final Rule was published, during the pendency of this litigation. See Comm’n Br. 49.

We have noted that “requiring agencies to obtain and publicize the data underlying all studies on which they rely would be impractical and unnecessary.” Am. Trucking Ass’ns v. EPA, 283 F.3d 355, 372 (D.C. Cir. 2002) (internal quotation marks omitted). But that concern is inapposite in this circumstance, where the agency “possessed the underlying data but failed to include it in the rulemaking record.” Coalition of Battery Recyclers Ass’n v. EPA, 604 F.3d 613, 623 (D.C. Cir. 2010). Although the Commission later disclosed injury- incident summaries to the WCMA, those summaries should 13 have been made public during the rulemaking itself — not provided to a single party during a post hoc judicial proceeding. See Air Transp. Ass’n of Am., 169 F.3d at 7. In any event, we are skeptical that the minimal information in the summaries satisfies the APA’s disclosure requirements. To provide a meaningful opportunity for commenters to provide substantive feedback, the Commission should have made public the incident reports that it used to support its calculation of benefits.

The Commission asserts that disclosing the injury-incident reports would have been improper because that would have revealed protected “personal, medical, or manufacturer- specific information.” Comm’n Br. 50. That argument misses the mark. The deaths and injuries enumerated in the dataset no doubt raise personal-privacy concerns. But the Commission could have redacted sensitive information from the reports before releasing them. The Commission has not justified its decision to withhold critical information altogether because of generalized personal-privacy concerns.1

1 Although the Commission cites 15 U.S.C. § 2055(a) for the proposition that “certain information obtained by [the] Commission [is] confidential,” Comm’n Br. 50, it overstates that statute’s scope. Section 2055(a)(1) restates a general rule that information “otherwise protected by law from disclosure to the public” shall remain confidential. 15 U.S.C. § 2055(a)(1). The remainder of that subsection designates trade secrets and information relating to a “manufacturer” or “private labeler” of a consumer product — i.e., those who manufacture, import, or own a consumer product’s brand or trademark — as confidential. Id. § 2055(a)(2)–(3). Section 2055(a) does not, as the Commission suggests, impose a categorical confidentiality requirement on all personal information obtained during an informal rulemaking. Nor does it suggest that the Commission may solve any privacy concerns by withholding the 14 2.

Notice-and-comment violations are subject to “the rule of prejudicial error.” 5 U.S.C. § 706. We “will not set aside a rule absent a showing by the petitioner[] ‘that [it] suffered prejudice from the agency’s failure to provide an opportunity for public comment.’” Am. Radio Relay League v. FCC, 524 F.3d 227, 237 (D.C. Cir. 2008) (quoting Gerber v. Norton, 294 F.3d 173, 182 (D.C. Cir. 2002)). A petitioner may demonstrate prejudice “by showing that it ‘ha[s] something useful to say’” about undisclosed information “that may allow it to ‘mount a credible challenge’ if given the opportunity to comment.” Id. at 237–38 (first quoting Chamber of Com. v. SEC, 443 F.3d 890, 905 (D.C. Cir. 2006); then quoting Gerber, 294 F.3d at 184). “A petitioner is not required to show that its comments would have persuaded the agency to reach a different outcome in order to establish prejudice.” 1 KRISTIN E. HICKMAN & RICHARD J. PIERCE, JR., ADMINISTRATIVE LAW TREATISE § 5.3.2 (6th ed. 2019).

The WCMA has made the requisite showing of prejudice from the Commission’s nondisclosure of its injury-incident dataset and underlying incident reports. In the WCMA’s comment on the Proposed Rule, the WCMA expressed several concerns about the Commission’s data that might have been further developed or amplified if the WCMA had been given access to the incident reports. The WCMA noted: (1) a “spike” in injuries for 2009 and 2010 that may have stemmed from a specific type of window covering, “Roman shade[s],” which the WCMA “swiftly and decisively” addressed by amending

data altogether. Cf. United States v. Am. Honda Motor Co., Inc., 143 F.R.D. 1, 8–9 (D.D.C. 1992) (reviewing Commission procedures for creating “a hot line and a central library depository” to disclose information about all-terrain vehicle injuries while allowing manufacturers to raise § 2055(a) objections). 15 the voluntary standard; (2) the use of “anecdotal” data, which the WCMA claimed did not “support inferences based on the percentage of incidents attributed to different product types”; and (3) the fact that the injury dataset “[did] not identify the age of the custom window coverings involved in the incidents,” making it impossible to draw conclusions about the effectiveness of the WCMA’s voluntary standard “in reducing the risk of incidents and injuries.” J.A. 745–47, 749–50. The Commission ignored those comments when promulgating the Final Rule. The arguments made by the WCMA during the rulemaking demonstrate that it “had something useful to say” about the underlying incident reports, and that it was prejudiced by its inability to evaluate critical information relied upon by the agency. Chamber of Com., 443 F.3d at 905.

3.

We reject the WCMA’s remaining notice-and-comment objections.

First, the WCMA claims error in the Commission’s failure to discuss the 2022 revision to the voluntary standard “until the [F]inal Rule itself.” Opening Br. 43–44. In other words, the WCMA believes that the Proposed Rule should have addressed the 2022 revision. But the Commission published its notice of proposed rulemaking on January 7, 2022, long before the 2022 revision to the voluntary standard was approved by the WCMA on August 15, 2022. Nothing in the Act or the APA required the Commission “to evaluate an unfinished draft when promulgating its rule.” Comm’n Br. 44. The Final Rule was then published on November 28, 2022; and the 2022 revision to the voluntary standard was published on December 23, 2022. See J.A. 103–04. The Final Rule discussed the 2022 revision as an alternative to the adopted safety standard. Final Rule, 87 Fed. Reg. at 73,165–67. Under the circumstances, the 16 Commission was not required to address the 2022 revision any earlier in the rulemaking process.

Second, the WCMA argues that the Commission “made substantial changes to its cost-benefit analysis” in the Final Rule without industry input. Opening Br. 45–46. That criticism lacks merit because commenters “should have anticipated” that such changes were possible, and they therefore had an opportunity to comment. See Int’l Union, United Mine Workers of Am. v. MSHA, 626 F.3d 84, 94–95 (D.C. Cir. 2010). In the Proposed Rule, the Commission explained the “[m]any sources of uncertainty” in its cost- benefit analysis — such as cost estimates, the value of a statistical life, and the number of window coverings in use — and suggested that it might increase the compliance-cost estimate. Proposed Rule, 87 Fed. Reg. at 1044, 1047. Those statements by the Commission gave commenters meaningful notice that the cost-benefit analysis might change, and thus provided an opportunity for them to comment. AFL-CIO v. Donovan, 757 F.2d 330, 338 (D.C. Cir. 1985). The WCMA took advantage of that opportunity by extensively critiquing the proposed cost-benefit analysis in a comment. See J.A. 767–70. We therefore conclude that the cost-benefit analysis in the Final Rule was a “logical outgrowth” of the one included in the Proposed Rule. See Air Transp. Ass’n of Am. v. Civil Aeronautics Bd., 732 F.2d 219, 224 (D.C. Cir. 1984).

Next, the WCMA challenges the Commission’s factual findings. It argues that the Commission gave insufficient weight to the efficacy of the 2018 and 2022 versions of the voluntary standard; made errors in performing its cost-benefit analysis; and arbitrarily imposed a 180-day effective date. 17 We may not uphold a safety standard unless the Commission’s findings and conclusions are “supported by substantial evidence on the record taken as a whole.” 15 U.S.C. § 2060(c). “Substantial evidence is not a high bar.” DHSC, LLC v. NLRB, 944 F.3d 934, 937 (D.C. Cir. 2019). “It means — and means only — such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Biestek v. Berryhill, 139 S. Ct. 1148, 1154 (2019) (internal quotation marks omitted).

“When applied to rulemaking proceedings, the substantial evidence test ‘is identical to the familiar arbitrary and capricious standard.’” S. Carolina Pub. Serv. Auth. v. FERC, 762 F.3d 41, 54 (D.C. Cir. 2014) (quoting Wis. Gas Co. v. FERC, 770 F.2d 1144, 1156 (D.C. Cir. 1985)). Under the arbitrary and capricious standard, an agency must “articulate a satisfactory explanation for its action including a ‘rational connection between the facts found and the choice made.’” Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983) (quoting Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 168 (1962)). The agency’s explanation must “enable us to conclude that the [agency’s action] was the product of reasoned decisionmaking.” Id. at 52.2

2 The WCMA claims that the Act’s reference to “substantial evidence” imposes a heightened standard of review. See Opening Br. 21–22. Not so. We once alluded to (but did not decide) the “interesting question” that the Act’s substantial evidence standard might impose “more strict” procedural requirements than an arbitrary and capricious standard of review. Forester v. Consumer Prod. Safety Comm’n, 559 F.2d 774, 789 n.22 (D.C. Cir. 1977). The Fifth Circuit adopted that view shortly thereafter. Aqua Slide ‘N’ Dive Corp. v. Consumer Prod. Safety Comm’n, 569 F.2d 831, 837 (5th Cir. 1978). But we later rejected such an interpretation. In an 18

We conclude that the Commission properly weighed the voluntary standard, but erred in performing its cost-benefit analysis and setting the Final Rule’s effective date. We take each issue in turn.

When a regulated industry has imposed a voluntary standard, the Commission may not promulgate a safety rule unless it finds that: (i) compliance with the voluntary standard “is not likely to result in the elimination or adequate reduction of [the] risk of injury”; or (ii) “it is unlikely that there will be substantial compliance” with the voluntary standard. 15 U.S.C. § 2058(f)(3)(D); see also id. § 2058(b)(2). Here, the Commission fulfilled its obligation to examine the 2018 voluntary standard, which was the standard in effect at the time of the rulemaking; and the Commission properly issued the required findings based on the 2018 standard when it promulgated the Final Rule. See Final Rule, 87 Fed. Reg. at 73,192–94. Additionally, the Commission assessed whether the 2022 proposed revisions to the voluntary standard would sufficiently mitigate the risk of injury, even though the 2022 revision had not yet been published at the time of the rulemaking. See id. at 73,183–84.

The WCMA essentially claims that the Commission’s required findings about the efficacy of the voluntary standard were not supported by substantial evidence. We disagree. The

opinion by then-Judge Scalia, we emphasized “that the distinction between the substantial evidence and the arbitrary or capricious test is largely semantic” and that the two tests “are one and the same.” ADPSO, 745 F.2d at 683–684 (cleaned up). The WCMA’s contrary view is “no longer viable.” Zen Magnets, LLC v. Consumer Prod. Safety Comm’n, 841 F.3d 1141, 1148 n.8 (10th Cir. 2016). 19 Commission found that the voluntary standard failed to adequately address the risk of strangulation associated with custom products. Final Rule, 87 Fed. Reg. at 73,194. That finding was clearly explained and well supported by evidence. Although the WCMA insists that the voluntary standard effectively reduced the risk of injury from corded window products, the Commission apparently took the permissible position that such risk should be eliminated. We highlight three flaws that the Commission found with the 2018 and 2022 versions of the voluntary standard.

First, the 2018 revision allowed “hazardous operating cords” that were either “long enough for a child to wrap around their neck” or capable of becoming “tangled and creat[ing] a loop large enough for a child to insert [its] head.” Final Rule, 87 Fed. Reg. at 73,157. Although that revision required window coverings to be operated with wands instead of operating cords, purchasers of custom products could override the wand requirement. Id. at 73,160 (“Firms typically allow consumers to easily change the default options during the custom order process.”). The 2018 revision also mandated that operating cords could not exceed 40% of a window covering’s height, but such operating cords could still be dangerous to children when extended. See id. at 73,157; see also J.A. 1329.

Second, the 2018 revision permitted manufacturers to sell continuous-loop systems, provided that the products used cord tension devices. J.A. 1329. The Commission rejected this potential solution. “[C]onsumers must use or install [a tension device] separately” from the window covering itself, which allows them to avoid utilizing the tension device. Final Rule, 87 Fed. Reg. at 73,161. And “depending on how taut the cord loop is, it can still allow a child’s head to enter the opening” when a tension device is used. Id. at 73,161, 73,192. The Commission noted that several reported injuries involved 20 continuous-loop systems that had tension devices installed. Id. at 73,161. Because the 2022 revision continued to rely on tension devices, the Commission deemed it inadequate. The Commission reasoned that consumers could still “set up the window covering in an unsafe manner . . . by removing the tension device from the loop” or “by leaving the tension device on the loop, but not attaching it on the wall.” Id. at 73,165–66. Moreover, the Commission reiterated that a tension device, even when properly installed, “still allows an accessible hazardous loop.” Id. at 73,166–67.

Third, the Commission disagreed with the 2018 revision’s finding that retractable cord-lift systems were an acceptable safety feature. The Commission noted that those systems still posed a strangulation risk because the retractable cords, when pulled, had a “low retraction force” and exposed enough cord such that “a child [could] manipulate the cord and wrap the cord around their neck.” Final Rule, 87 Fed. Reg. at 73,157– 58. Although the 2022 revision mandated that retractable cord- lift systems use wands instead of cords, it still allowed a length of exposed cord, or “stroke length,” of thirty-six inches when pulled, which poses a strangulation risk. Id. at 73,165.

Ultimately, the Commission made the common-sense observation that “[h]aving no operating cords effectively eliminates the strangulation hazard . . . because there is no cord to cause strangulation.” Final Rule, 87 Fed. Reg. at 73,162. The WCMA’s argument that the Commission relied on “old data, hypotheticals, and speculation,” Reply Br. 6–7, to make that finding lacks merit. See Nasdaq Stock Mkt. LLC v. SEC, 38 F.4th 1126, 1142–43 (D.C. Cir. 2022) (rejecting a criticism of an agency’s “purely theoretical” concern because the agency relied on industry comments and its own experience when promulgating a rule); cf. Associated Gas Distribs. v. FERC, 824 F.2d 981, 1008 (D.C. Cir. 1987) (“Agencies do not need to 21 conduct experiments in order to rely on the prediction that an unsupported stone will fall.”). The Commission reasonably relied on data, industry comments, and its own expertise when determining that the voluntary standards would not sufficiently address the risk of strangulation.3

2.

The WCMA asserts that the Commission erred by finding that the rule’s expected benefits bore a “reasonable relationship” to expected costs. Opening Br. 29 (citing 15 U.S.C. § 2058(f)(3)(E)). Furthermore, the WCMA claims that the Commission’s analysis is “not accurate” because it underestimates costs and overstates benefits. Id. at 30. We agree with the WCMA that the Commission’s estimate of cost increases from implementation of the Final Rule is flawed because it relies on price data for stock products, not custom products. Id. at 30–31. That error undermines its finding about the relationship between expected benefits and costs.

We review cost-benefit analyses deferentially. Nat’l Ass’n of Home Builders v. EPA, 682 F.3d 1032, 1040 (D.C. Cir. 2012). “Such cost-benefit analyses epitomize the types of decisions that are most appropriately entrusted to the expertise of an agency.” Off. of Commc’n of United Church of Christ v. FCC, 707 F.2d 1413, 1440 (D.C. Cir. 1983). But “a serious

3 Referring to the Commission’s database of injury incidents, the WCMA implies that the Commission “relied on injury data that did not capture subsequent regulatory actions” reducing the injury rate, suggesting that conclusions drawn from that injury data are unreliable. See Opening Br. 26–27; Reply Br. 6–7. But the Commission also referenced studies that independently and substantially supported its finding that the voluntary standards would not sufficiently reduce the risk of strangulation. Final Rule, 87 Fed. Reg. at 73,153–54. 22 flaw” in a cost-benefit analysis “can render the [resulting] rule unreasonable” and warrant vacatur on arbitrary and capricious grounds. Nat’l Ass’n of Home Builders, 682 F.3d at 1040; see also Thompson v. Clark, 741 F.2d 401, 405 (D.C. Cir. 1984) (reasoning that “an unreasonable assessment of social costs and benefits” can render a rule arbitrary and capricious). An agency contravenes the APA when it “fails to examine the relevant data,” which could reveal “that the figures being used are erroneous.” Dist. Hosp. Partners, L.P. v. Burwell, 786 F.3d 46, 57 (D.C. Cir. 2015).

Here, the Commission crafted its cost estimate by: (i) making low-end and high-end estimates of the percentage increase in manufacturing costs resulting from the Final Rule; then (ii) multiplying those estimates by the mean price of different types of window covering units. J.A. 572–73. To find the mean prices, the Commission appears to have relied on a third-party study that averaged prices for products listed in the online catalogs of two major retailers. See id. at 572 n.20, 1289–90 & n.1. Virtually all of the products surveyed in this study were stock products. Id. at 1289. According to the WCMA, these selectively sourced prices created an “artificially low” cost estimate. Opening Br. 31. One commenter complained that this data was based on “mass- produced, high volume, low end window coverings,” and provided its own (much higher) estimates of mean unit prices for custom products. See J.A. 720–21. Moreover, both the Commission and its underlying study acknowledged that custom products are more expensive on average than stock products. See Final Rule, 87 Fed. Reg. at 73,193; J.A. 1290.

Nothing in the Rule or the final regulatory analysis demonstrates that the Commission reckoned with the discrepancy between the prices for stock and custom window coverings, an issue that was repeatedly brought to its attention. 23 The Commission erroneously argues that it accounted for “data relating to stock products” by producing low-end and high-end cost estimates, implying that the high-end estimate adequately incorporated costs relating to custom products. See Comm’n Br. 37. But both the low-end and high-end cost estimates were derived by applying expected percentage increases in cost to the inaccurate mean unit prices. See J.A. 572 tbl.9, 573 tbl.10. If the mean unit prices were too low, as the WCMA contends, then both the low-end and high-end estimates would increase as a result of any corrections. It is not just that “the low-end estimate was too low,” Comm’n Br. 37, but that the entire range of cost estimates would shift upward if the mean unit prices were corrected to account for custom products.4

While we could speculate about the Commission’s justifications for relying on stock window covering data, it is not our role to “supply a reasoned basis for the agency’s action that the agency itself has not given.” Bowman Transp., Inc. v. Ark.-Best Freight Sys., Inc., 419 U.S. 281, 285–86 (1974) (citing SEC v. Chenery Corp., 332 U.S. 194, 196 (1947)). On this administrative record, we cannot say that substantial evidence supports the Commission’s finding that the Final Rule’s expected benefits bore a “reasonable relationship” to its

4 Additionally, there appear to be many discrepancies in the Commission’s calculations when quantifying costs and benefits. Consider one example: When estimating annual window covering shipments for interior shutters, the Commission suggested that 54.4% of 134,867 shipments is equal to zero. See J.A. 574 tbl.11; see also, e.g., J.A. 572 tbl.9 (wrongly stating for cellular shades that 4% of 94.51 is 5.67, rather than 3.78). While most of the discrepancies are minor — perhaps resulting from rounding or transcription errors — the number of arithmetic mistakes undermines the Commission’s analysis and suggests that greater care is warranted on remand. 24 expected costs because the cost calculation failed to properly examine the relevant data.5

Under the Act, consumer product safety standards have a presumptive 180-day effective date unless the Commission finds, for “good cause shown,” that “a later effective date is in the public interest.” 15 U.S.C. § 2058(g)(1). A safety standard also must include a finding “that the rule (including its effective date) is reasonably necessary to eliminate or reduce an unreasonable risk of injury associated with [the regulated] product.” Id. § 2058(f)(3)(A) (emphasis added). Here, the Commission declined to find “good cause shown” that a later date was “in the public interest”; and it affirmatively found that the 180-day deadline was “reasonably necessary” to eliminate or reduce the risk of injury posed by corded custom window coverings. Final Rule, 87 Fed. Reg. at 73,194.

At the threshold, the Commission suggests that the Act’s references to “good cause shown” and “public interest” commit

5 The WCMA further argues that the cost-benefit analysis “completely ignore[d] the commercial side of the custom market,” underestimated industrywide compliance costs, and improperly assumed that custom window products were involved in over 40% of accidents despite representing 20% of window coverings in use. Opening Br. 31–34. We reject those challenges. An administrative agency must “exercise its expertise to make tough choices” about costs and benefits, which often means “hazard[ing] a guess” in “conditions of serious uncertainty.” Pub. Citizen v. Fed. Motor Carrier Safety Admin., 374 F.3d 1209, 1221 (D.C. Cir. 2004). While commenters might disagree with the agency’s assumptions and reasoning, such judgments fall within the agency’s purview when conducting a cost-benefit analysis. 25 the effective-date decision to agency discretion, requiring “highly deferential[]” review, if any. See Comm’n Br. 51–52. We disagree. The APA “codifies the traditional exception that agency action is unreviewable” when “committed to agency discretion by law.” Sierra Club v. EPA, 47 F.4th 738, 745 (D.C. Cir. 2022). That exception applies only when “courts have no legal norms pursuant to which to evaluate the challenged action, and thus no concrete limitations to impose on the agency’s exercise of discretion.” Physicians for Soc. Resp. v. Wheeler, 956 F.3d 634, 643 (D.C. Cir. 2020) (cleaned up). To be sure, an effective-date extension requires “good cause shown,” 15 U.S.C. § 2058(g)(1), which is an “unrestricted and undefined” term, Clifford v. Peña, 77 F.3d 1414, 1417 (D.C. Cir. 1996). But we have reviewed such a determination despite the Commission’s “broad discretion,” observing that “the requirement of reasoned decisionmaking has vitality as to such deferral measures.” ASG Indus., Inc. v. Consumer Prod. Safety Comm’n, 593 F.2d 1323, 1335 (D.C. Cir. 1979). Moreover, the Commission must separately find that the effective date is “reasonably necessary” to mitigate injury risks. 15 U.S.C. § 2058(f)(3)(A). Regardless of the standard the Commission relied on, its justifications for a 180- day effective date do not reflect the “reasoned decisionmaking” required by the APA. Allentown Mack Sales & Serv., Inc. v. NLRB, 522 U.S. 359, 374 (1998) (quoting State Farm, 463 U.S. at 52).

At least 401 industry commenters informed the Commission that a 180-day compliance deadline was not feasible. Final Rule, 87 Fed. Reg. at 73,177; see also, e.g., J.A. 699 (“There are countless changes by numerous parties throughout the supply chain that would be required to be made in a short amount of time.”); id. at 717 (“Even if it were possible to just substitute all lost corded product volume with a cordless version of the same product (which is not possible), 26 we still would not be able to meet the 180-day timeline.”). The Small Business Administration also “expressed concerns” about the compliance deadline. Final Rule, 87 Fed. Reg. at 73,179. Even the Commission’s own staff emphasized that a short-term effective date “ha[d] the potential to be very disruptive for producers and consumers” and that postponing the effective date “would reduce the benefits of the rule by only a very small amount as most noncompliant window coverings will take years to cycle out of use.” J.A. 590. For these and other reasons, the staff recommended an effective date of either one or two years, depending on the product. Id. Nevertheless, the Commission declined to extend the deadline, relying instead on the comment of a single manufacturer, Safe T Shade, LLC, which asserted that “a 180-day lead time is more than sufficient for a painless [i]ndustry implementation.” See Final Rule, 87 Fed. Reg. at 73,177. Under the circumstances, that decision was arbitrary. The Commission did not explain why it chose to credit the opinion of Safe T Shade’s company president, see id. at 73,177, 73,189, over the contrary feedback that it received from 401 other commenters, the Small Business Association, and its own staff.

Moreover, the Commission’s stated reasons for declining to find “good cause” to extend the compliance deadline were flawed. See Final Rule, 87 Fed. Reg. at 73,177.

First, the Commission highlighted that its economic analysis “[did] not conclude that a longer effective date creates a material reduction in the estimated costs of the rule, and commenters [did] not show that this would be the case.” Final Rule, 87 Fed. Reg. at 73,177. That statement makes little sense. The “estimated costs of the rule” refers to the total cost to the window covering industry that would result from the Final Rule. See J.A. 581 tbl.16. That figure appears to bear little relationship to determining whether an extended 27 compliance deadline is in the public interest. Indeed, when assessing good cause for an extension, the agency looked to a different cost: “the possibility that some styles of custom window coverings may be less available during a transition period.” Final Rule, 87 Fed. Reg. at 73,194.

Second, the Commission noted that stock window coverings have been subject to a voluntary standard since 2018 that essentially prohibits corded products, and that the same compliance methods “can be used for, or at a minimum can be adapted to, custom window coverings.” Final Rule, 87 Fed. Reg. at 73,177. That assertion “failed to consider an important aspect of the problem.” State Farm, 463 U.S. at 43. Industry commenters emphasized that the safety features used in stock products — e.g., cordless technologies — are not yet feasible for custom window coverings. See, e.g., J.A. 740 (“Size, weight, and location requirements for custom products prevent the use of many of the cordless lift systems offered for stock products, despite [the Commission’s] claims to the contrary.”); id. at 849–50 (explaining that, for one product, changing from a continuous-loop system to a cordless system would “reduce the maximum available product size” and “add a $105 surcharge” to each unit). Moreover, the Commission failed to address whether the proposed adaptations, even if feasible, could be implemented during the 180-day grace period.

Third, the Commission reasoned that a short timeline would not affect compliance because manufacturers have been on notice of Canada’s regulations for window coverings, which contain restrictions on operating cords and apply equally to stock and custom products. Final Rule, 87 Fed. Reg. at 73,177. That conclusion, however, relies on a false assumption that the Final Rule and the referenced Canadian regulations impose the same safety requirements. To the contrary, as the WCMA points out, the two rules use different tests for “cord 28 accessibility” and “rigid shrouds and restraining devices.” Opening Br. 39–40. The Canadian regulations assess cord length only when the cord is not in use, but the Final Rule imposes length requirements on operating cords even when pulled. Id. at 39; compare Stay Add. 32–33, with Final Rule, 87 Fed. Reg. at 73,190, and J.A. 1334, 1336. Thus, it is not fair to assume that industry compliance with the Canadian regulations would also satisfy the requirements of the Final Rule, thereby justifying a shorter effective date for the Final Rule.6

Fourth, the Commission stated that “manufacturers have been aware of [the Commission’s] proposed rule for at least one year already,” since the Commission issued its notice of proposed rulemaking. Final Rule, 87 Fed. Reg. at 73,177. That is, the Commission suggested that window covering manufacturers should have begun complying with the new safety standard as soon as they learned of the Proposed Rule. We reject that argument and its implications. Agency action is not “final” until a conclusion is reached that “mark[s] the

6 The Commission appears to suggest that the WCMA has forfeited any argument about how “the Canadian rule impacts compliance” with the Final Rule because industry commenters assertedly did not address that issue in the administrative proceedings. Comm’n Br. 54 (citing Fla. Power & Light Co. v. Lorion, 470 U.S. 729, 743 (1985)) (indicating that a reviewing court must assess “the administrative record already in existence, not some new record made initially in the reviewing court” (cleaned up)); see also Final Rule, 87 Fed. Reg. at 73,177. But the Proposed Rule acknowledged differences between the Canadian regulations and the WCMA’s 2018 voluntary standard, including those referenced by the WCMA on appeal. Proposed Rule, 87 Fed. Reg. at 1033. The WCMA’s argument is therefore firmly grounded in the administrative record. 29 consummation of the agency’s decisionmaking process,” from which “rights or obligations have been determined” or “legal consequences will flow.” Bennett v. Spear, 520 U.S. 154, 177– 78 (1997) (cleaned up). Here, the culmination of the agency’s yearlong revision process was the Final Rule, and manufacturers of window coverings had no new compliance obligations until the Final Rule was promulgated. We will not credit an agency explanation that requires regulated entities to tailor their operations to adhere to an agency’s proposed rules. That would make the subsequent notice-and-comment proceedings superfluous and undermine the entire rulemaking process.

Finally, the Commission also erred in finding that a 180- day effective date was “reasonably necessary to address the unreasonable risk of strangulation from operating cords on custom window coverings.” Final Rule, 87 Fed. Reg. at 73,194. That finding was not supported by substantial evidence. Although the Commission extensively discussed the risk of strangulation posed by corded window coverings, see supra Section II.B.1, neither the Final Rule nor the administrative record examined how the timing of the effective date would affect the risk of injury. The Commission therefore had no apparent basis to conclude that the risk of injury necessitated a 180-day effective date, as opposed to a one-year or two-year deadline. Indeed, a comment from Commission staff supported a contrary conclusion, suggesting that near- term implementation of the Final Rule was not necessary. See J.A. 590 (postponing effective date “would reduce the benefits of the rule by only a very small amount”).

The Commission argues that interpreting § 2058(f)(3)(A) to require a separate reasonable-necessity finding would read § 2058(g)(1)’s good-cause provision “out of the statute” and render it “a nullity.” Comm’n Br. 52–53. But it is a “cardinal 30 principle of interpretation that courts must give effect, if possible, to every clause and word of a statute.” Liu v. SEC, 140 S. Ct. 1936, 1948 (2020) (quoting Parker Drilling Mgmt. Servs., Ltd. v. Newton, 139 S. Ct. 1881, 1890 (2019)). The Commission’s interpretation asks us to violate that principle by ignoring § 2058(f)(3)(A)’s mandate for the Commission to find that “the rule (including its effective date) is reasonably necessary to eliminate or reduce an unreasonable risk of injury associated with [the regulated] product.” 15 U.S.C. § 2058(f)(3)(A) (emphasis added). The two provisions at issue are reconcilable because they impose separate obligations: If the Commission wishes to extend a safety standard’s effective date, it must find good cause to do so; and regardless of such an extension, the Commission must find that the effective date is reasonably necessary to reduce or eliminate a risk of injury. See id. § 2058(g)(1), (f)(3)(A).

The WCMA claims that the Commission’s decision should be vacated because its commissioners are “unconstitutionally insulated from Presidential oversight” by their for-cause removal restriction. Opening Br. 49; see also Seila Law LLC v. CFPB, 140 S. Ct. 2183, 2201 (2020). Because other flaws in the Commission’s rulemaking lead us to vacate and remand the Final Rule, we need not reach that argument. See Bond v. United States, 572 U.S. 844, 855 (2014) (recognizing the “well-established principle” that a federal court “will not decide a constitutional question if there is some other ground upon which to dispose of the case” (quoting Escambia Cnty. v. McMillian, 466 U.S. 48, 51 (1984) (per curiam))). 31 * * *

For the foregoing reasons, we grant the petition for review, vacate the Final Rule, and remand this matter to the Commission for further proceedings.

So ordered.