District of Columbia v. U.S. Department of Agriculture

UNITED STATES DISTRICT COURT FOR THE DISTRICT OF COLUMBIA DISTRICT OF COLUMBIA, et al., Plaintiffs, Civil Action No. 20-cv-00119 (BAH) v. Chief Judge Beryl A. Howell UNITED STATES DEPARTMENT OF AGRICULTURE, et al., Defendants. BREAD FOR THE CITY, et al., Plaintiffs, v. UNITED STATES DEPARTMENT OF AGRICULTURE, et al., Defendants. MEMORANDUM OPINION Eight months into the COVID-19 pandemic, which has rocked the economy, killed nearly 220,000 Americans, 1 quadrupled the national unemployment rate, 2 and dramatically increased the number of Americans forced to reckon with hunger this year, 3 the United States Department of Agriculture (“USDA”) is pursuing implementation of a Final Rule to “dramatically alter the — 1 — See Cumulative Cases by Days Since 50th Confirmed Case, JOHNS HOPKINS UNIVERSITY (Oct. 18, 2020), https://coronavirus.jhu.edu/data/cumulative-cases (last visited Oct. 18, 2020). — 2 — See BUREAU OF LABOR STATISTICS, THE EMPLOYMENT SITUATION – AUGUST 2020, USDL-20-1838 (Sept. 4, 2020), https://www.bls.gov/news.release/pdf/empsit.pdf (last visited Oct. 18, 2020). — 3 — See ALISHA COLEMAN-JENSEN ET AL., U.S. DEP’T OF AGRIC., ERR-275, HOUSEHOLD FOOD SECURITY IN THE UNITED STATES IN 2019 12 (Sept. 2020), https://www.ers.usda.gov/webdocs/publications/99282/err275.pdf?v=9263.7 (last visited Oct. 18, 2020); The Impact of the Coronavirus on Food Security, FEEDING AMERICA, 3 (Apr. 22, 2020), https://www.feedingamerica.org/sites/default/files/202004/Brief_Impact%20of%20Covid%20on%20Food%20Insecurity%204.22%20%28002%29.pdf (last visited Oct. 18, 2020). — 1 — long-standing operations” of the Supplemental Nutrition Assistance Program (“SNAP”), by stripping States of their current flexibility in providing this food assistance benefit. D.C. v. U.S. Dep’t of Agric. (“D.C. I”), 444 F. Supp. 3d 1 , 6 (D.D.C. 2020); see Final Rule, Supplemental Nutrition Assistance Program: Requirements for Able-Bodied Adults Without Dependents, 84 Fed. Reg. 66782 (Dec. 5, 2019) (to be codified at 7 C.F.R. pt. 273). In March 2020, when this Court largely granted preliminary injunction requests from nineteen States, the District of Columbia and the City of New York, as well as private plaintiffs, USDA estimated the prospective changes to SNAP would affect over one million people, by newly subjecting them to time limits on their eligibility to receive food under this program, and kick almost 700,000 able-bodied adults without dependents (“ABAWDs”) out of the SNAP program altogether. See 84 Fed. Reg. 66782 , 66807, 66809 (touting savings “of about $1.1 billion per year” from reduction in SNAP benefit payments and estimating that 1,087,000 individuals would be newly subject to eligibility time limits and 688,000 individuals, in fiscal year (FY) 2021, will neither meet the new waiver requirement nor be exempt); ABAWD00000431 (Regulatory Impact Analysis), ECF No. 105-1. The agency has been icily silent about how many ABAWDs would have been denied SNAP benefits had the changes sought in the Final Rule been in effect while the pandemic rapidly spread across the country and congressional action had not intervened to suspend any time limits on receipt of those benefits. In the pandemic’s wake, as of May 2020, SNAP rosters have grown by over 17 percent with over 6 million new enrollees. 4 — 4 — Ed Bolen, USDA Rolling Back SNAP Flexibility That States Need in Current Crisis, CENTER ON BUDGET AND POLICY PRIORITIES (Aug. 10, 2020, 11:00 AM), https://www.cbpp.org/blog/usda-rolling-back-snap-flexibilitythat-states-need-in-current-crisis (last visited Oct. 18, 2020). — 2 — More than merely silent, USDA strenuously objects to consideration of estimates, for example, that under the new restrictions in the Final Rule, only 10 percent, rather than the current 97 percent, of U.S. counties would have the flexibility to extend SNAP benefits to ABAWDs. See State Pls.’ Mem. in Supp. of Mot. Summ. J. (“State Pls.’ MSJ”), Ex. 1, Decl. of Edward Bolen, Sr. Policy Analyst, Center on Budget and Policy Priorities (CBPP) (“Bolen Decl.”), ¶ 13, ECF No. 65-1; Defs.’ Cross-Mem. in Supp. of Mot. Summ. J. and in Opp’n to State Pls.’ MSJ (“Defs.’ Opp’n”), at 44 n. 22, ECF No. 92. Despite the agency’s blinkered effort to downplay or disregard the predicted outcomes of the Final Rule, the backdrop of the pandemic has provided, in stark relief, its procedural and substantive flaws. To be sure, States bear the statutory responsibility of aiding ABAWD SNAP recipients to transition into the workforce by providing vocational training and transitional support. To that end, the statutory scheme sets up mechanisms for the federal government to spur States to provide better services to their citizens to enable their self-sufficiency and move them off government assistance. See 84 Fed. Reg. at 66796 (“the Department expects States to support ABAWDs in their efforts to find work and meet the work requirement by expanding access to work programs and other supportive services for ABAWDs.”); id. at 66807 (“the Department expects State agencies to do what they can to increase the employability of ABAWDs, and help them find and gain work.”). At the same time, however, the over-arching goal of SNAP is to enable States to target nutrition benefits to those residents in need, with statutory provisions that protect that critical function. See 7 U.S.C. § 2011 (Congressional findings “that the limited food purchasing power of low-income households contributes to hunger and malnutrition among members of such households” and “[t]o alleviate such hunger and malnutrition, a supplemental nutrition assistance program … will permit low-income households to obtain a more nutritious — 3 — diet through normal channels of trade by increasing food purchasing power for all eligible households who apply for participation.”). The Final Rule at issue in this litigation radically and abruptly alters decades of regulatory practice, leaving States scrambling and exponentially increasing food insecurity for tens of thousands of Americans. Whether USDA could, using a legally proper process, adequately explain how the Final Rule’s changes both comport with the statutory scheme and make sense is a question for another day. For now, the agency has not done so. For the reasons stated below, plaintiffs’ motions for summary judgment are GRANTED while the defendants’ cross-motions for summary judgment are DENIED, and the Final Rule is VACATED. I. BACKGROUND Familiarity with the statutory framework, regulatory background leading up to the December 5, 2019 promulgation of the Final Rule and the procedural history of this case, as comprehensively detailed in this Court’s March 13, 2020 preliminary injunction decision, D.C. I, 444 F. Supp. 3d at 7–15, is assumed and only briefly summarized here. A. Brief Overview of Statutory Requirements and the Regulatory Scheme Changed by Final Rule Pursuant to the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA), which amended the Food Stamp Act of 1977, ABAWDs are generally limited to receiving SNAP benefits for 3 months in a 36-month period, unless certain work requirements are met. See 7 U.S.C. § 2015 (o)(2). 5 This time limit on ABAWD’s SNAP benefits may be — 5 — The time limits applied to ABAWDS are not applicable to individuals who are: “(A) under 18 or over 50 years of age; (B) medically certified as physically or mentally unfit for employment; (C) a parent or other member of a household with responsibility for a dependent child; (D) otherwise exempt under subsection (d)(2); or (E) a pregnant woman.” 7 U.S.C. § 2015 (o)(3). These statutory exceptions to the time limits on eligibility for SNAP benefits are reflected in current regulations. See 7 CFR §273.24 (c). — 4 — temporarily waived or exempted as provided in two separate statutory sections that the Final Rule is intended to implement. See 7 U.S.C. §§ 2015 (o)(4)(A) (waivers), 2015(o)(6) (exemptions). Those statutory provisions and associated long-standing implementing regulations, originally made effective in 2001, see Final Rule, Food Stamp Program: Personal Responsibility Provisions of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996, 66 Fed. Reg. 4438 (Jan. 17, 2001) (codified at 7 C.F.R. pt. 272, 273), are described below. 1. Waiver of Time Limits Applicable to ABAWDs The statute’s waiver provision authorizes the Secretary of Agriculture, upon request of a State agency, with the support of the State’s chief executive officer, to waive the applicability of the time limit and work requirement— to any group of individuals in the State if the Secretary makes a determination that the area in which the individuals reside – (i) has an unemployment rate of over 10 percent; or (ii) does not have a sufficient number of jobs to provide employment for the individuals. 7 U.S.C. § 2015 (o)(4)(A) (emphasis supplied). The “individuals” referenced in this statutory provision are ABAWDs. See State Pls.’ MSJ at 13; Defs.’ Opp’n at 5. The current USDA regulation, in effect since 2001—and continuing in effect during the pendency of this Court’s preliminary injunction, which stayed the waiver process changes in the Final Rule that would otherwise have been effective on April 1, 2020, 84 Fed. Reg. at 6680—permits States to “submit whatever data it deems appropriate” to support a waiver request, with the caveat that “requests based on unemployment rates or labor force data” must use “standard Bureau of Labor Statistics (BLS) data or methods.” 7 C.F.R. § 273.24 (f)(2) (2001). 6 To support waiver requests for an — 6 — 7 C.F.R. §273.24 (f)(2) (2001) largely codified USDA guidance issued in 1996, see ABAWD00000166 (U.S. DEP’T AGRIC., GUIDANCE FOR STATES SEEKING WAIVERS FOR FOOD STAMP LIMITS (1996)), ECF No. 105-1, at the time of PRWORA’s enactment, see State Pls.’ MSJ at 25; Defs.’ Opp’n at 7. — 5 — “area” due to lack of sufficient jobs for ABAWDs, under prong (ii) of § 2015(o)(4)(A), the current regulation provides “[a] non-exhaustive list of the kinds of data a State agency may submit.” Id. 7 As evidence of a “lack of sufficient jobs,” six kinds of data are suggested, including data showing that an area is (1) “designated as a Labor Surplus Area (LSA)” by the Department of Labor (“DOL”); (2) qualifies “for extended unemployment benefits;” (3) has a low and declining employment-to-population ratio; (4) “has a lack of jobs in declining occupations or industries;” (5) “is described in an academic study or other publications as an area where there are lack of jobs;” and (6) “has a 24-month average unemployment rate 20 percent above the national average for the same 24-month period,” id. § 273.24(f)(2)(ii). USDA concedes that the challenged Final Rule upends this long-standing USDA policy on States’ prerogatives by strictly delineating “the area” to which waivers may apply to a metric developed by the U.S. Department of Labor, called a labor market area (“LMA”), and also by eliminating flexibility in probative data sources to restrict the criteria States may use to demonstrate a lack of “a sufficient number of jobs” for ABWADs under 7 U.S.C. § 2015 (o)(4)(A)(ii). See Defs.’ Opp’n at 2, 32. 2. Discretionary Exemptions from Time Limits Applicable to ABAWDs The year following PRWORA’s enactment, the Balanced Budget Act (“BBA”) of 1997 again amended the Food Stamp Act and added the discretionary exemption provision, which — 7 — To support waiver requests under prong (i) of 7 U.S.C. § 2015 (o)(4)(A), the 2001 regulation suggested the following three kinds of data to show an “unemployment rate of over 10 percent”: (1) “a recent 12 month average unemployment rate over 10 percent”; (2) “a recent three month average unemployment rate over 10 percent”; or (3) “an historical seasonal unemployment rate over 10 percent.” 7 C.F.R. § 273.24 (f)(2)(i) (2001). The Final Rule eliminates States’ flexibility in submitting “whatever data it deems appropriate to support its request” under prong (i), id. § 273.24(f)(2), to require requesting States to use only “[d]ata from the Bureau of Labor Statistics (BLS)…that shows an area has a recent 12-month average unemployment rate over 10 percent,” 84 Fed. Reg. at 66811. This elimination of all but the single type of data under prong (i) was criticized during the rulemaking as “inadequate and that other time periods should be allowed to demonstrate an unemployment rate over 10 percent.” 84 Fed. Reg. at 66784. Plaintiffs do not challenge the Final Rule’s change to the implementation of prong (i) of 7 U.S.C. § 2015 (o)(4)(A). — 6 — provided flexibility to States to exempt up to 15 percent of ABAWDs in the State from the time limits on eligibility for SNAP benefits. See BBA, Pub. L. No. 105-33, § 1001 , 111 Stat. Ann. 251, 252 (1997); see also Final Rule, Food Stamp Program: Work Provisions of the [PRWORA] of 1996 and Food Stamp Provisions of the [BBA] of 1997, 67 Fed. Reg. 41589 , 41591 (Jun. 19, 2002) (to be codified at 7 C.F.R. pts. 271, 272, 273, 275 and 277) (“2002 Final Rule”). This percentage was recently reduced to 12 percent “for fiscal year 2020 and each subsequent fiscal year” in the Agriculture Improvement Act of 2018 (“2018 Farm Bill”). 2018 Farm Bill, Pub. L. No. 115-334, § 4005 , 132 Stat. 4490 , 4632 (2018); see also 84 Fed. Reg. at 66802. Each exemption used by a State provides one additional month of SNAP benefits to an ABAWD who would otherwise be ineligible for such benefits due to the time limit. See Proposed Rule, Supplemental Nutrition Assistance Program: Requirements for Able-Bodied Adults Without Dependents, 84 Fed. Reg. 980 , 987 (Feb. 1, 2019); Advance Not. of Proposed Rulemaking (ANPRM), Supplemental Nutrition Assistance Program: Requirements and Services for AbleBodied Adults Without Dependents, 83 Fed. Reg. 8013 , 8014 (Feb. 23, 2018) (“Each 15 percent exemption extends eligibility to one ABAWD for one month.”). To calculate the number of discretionary exemptions each State is entitled to use for the fiscal year, § 2015(o)(6) sets out in six subparagraphs specific directions to USDA’s Secretary, culminating in the final adjustment that: the Secretary shall increase or decrease the number of individuals who may be granted an exemption by a State agency under this paragraph to the extent that the average monthly number of exemptions in effect in the State for the preceding fiscal year under this paragraph is lesser or greater than the average monthly number of exemptions estimated for the State agency for such preceding fiscal year under this paragraph. 7 U.S.C. § 2015 (o)(6)(G) (emphasis supplied). To comply with this mandate, the USDA Secretary must annually adjust the number of exemptions granted to each State based on the — 7 — difference between the “average monthly number of exemptions in effect in the State” and such number “estimated for the State” for the preceding fiscal year under paragraph (6). Id. This statutory direction is implemented in a regulation, 7 C.F.R. § 273.24 (h)(2) (2001), originally promulgated in 1999, see Interim Rule, Food Stamp Program: Food Stamp Provisions of the Balanced Budget Act of 1997, 64 Fed. Reg. 48246 , 48257–58 (Sept. 3, 1999), that remained in effect until October 1, 2020, when the Final Rule changed it. See 84 Fed. Reg. at 66804. The prior regulation in effect for two decades provided that “[i]f the State agency does not use all of its exemptions by the end of the fiscal year, [USDA] shall increase the estimated number of exemptions allocated to the State agency for the subsequent fiscal year by the remaining balance.” 7 C.F.R. § 273.24 (h)(2)(i) (2001); see also Defs.’ Opp’n at 7 (“If a State does not use all of its earned exemptions for a given fiscal year, USDA increases the State’s estimated number of exemptions for the following year by the positive balance.”). Conversely, “[i]f the State agency exceeds its exemptions by the end of the fiscal year, [USDA] shall reduce the estimated number of exemptions allocated to the State agency for the subsequent fiscal year by the corresponding number.” Id. § 273.24(h)(2)(ii). USDA acknowledged in the Final Rule that the prior regulation’s implementation – or “interpretation”—of 7 U.S.C. § 2015 (o)(6)(G) “allows any unused exemptions to carry over and accumulate from one year to the next, indefinitely,” with the result that “States have accumulated extremely high amounts of unused discretionary exemptions that well exceed the number allotted to each State for the fiscal year.” 84 Fed. Reg. at 66783. — 8 — The Final Rule, again, upends this long-standing USDA policy by altering the agency’s interpretation of the statutory mandate set out in 7 U.S.C. § 2015 (o)(6)(G). 8 Specifically, under the Final Rule, “each State agency’s carryover adjustment would be based on the number of exemptions earned in the preceding fiscal year” —rather than “allocated” for use, under the prior regulation—“minus the number of exemptions used in the preceding fiscal year,” with any carryover adjustment “to apply only to the fiscal year in which the adjustment is made.” 84 Fed. Reg. at 66802 (emphasis supplied). The Final Rule’s switch to a “use it or lose it” system for exemptions would significantly reduce the discretionary exemptions available to States by eliminating accumulated exemptions amassed through judicious administration of the SNAP program over the last two decades. See id. at 66803. B. Preliminary Resolution of Plaintiffs’ Challenges to Final Rule Plaintiffs challenge multiple parts of the Final Rule on procedural and substantive grounds. Amended Compl. (“State Pls.’ Am. Compl.”), D.C. et al. v. U.S. Dep’t of Agric. et al., No. 20-cv-00119 (D.D.C. Jan. 29, 2020), ECF No. 19; Compl. (“Priv. Pls.’ Compl.”), Bread for the City, et al. v. U.S. Dep’t of Agric., et al., No. 20-cv-00127, (D.D.C. Jan. 16, 2020), consolidated sub nom, D.C. v. U.S. Dep’t of Agric., No. 20-cv-00119 (D.D.C. Jan. 23, 2020), — 8 — In addition, USDA plans in a future rulemaking to change what these exemptions are called. 84 Fed. Reg. at 66802. According to the agency, these exemptions had been “generally referred to as 15 percent exemptions,” id., but then, in the Proposed Rule, were “referred to [] as ‘percentage exemptions’ as a way to avoid confusion as the calculation transitioned from 15 percent to 12 percent,” id. In the Final Rule, USDA “chose[] to refer to these exemptions as ‘discretionary’ exemptions…because States have discretion on whether to use these exemptions,” id., by contrast to other enumerated statutory exemptions, “which are not discretionary” insofar as States must “exempt individuals from the ABAWD time limit if the individual meets at least one of those listed exemptions,” id. — 9 — ECF No. 1. 9 Those challenges, and their preliminary resolution in D.C. I, are summarized next.10 1. Procedural Deficiencies in Promulgation of Final Rule Plaintiffs allege that the Final Rule is unlawful and must be set aside as promulgated “without observance of procedure required by law,” 5 U.S.C. § 706 (2)(D), because USDA failed to provide notice with an ability for meaningful comment for three “key aspects of the Rule.” State Pls.’ Am. Compl. ¶¶ 473, 477; see Priv. Pls.’ Compl. ¶ 99. Despite, in plaintiffs’ view, no suggestion in the Proposed Rule of these changes, the Final Rule: (1) eliminated as a criterion for a waiver request that an area qualified for extended employment benefits, State Pls.’ Am. Compl. ¶ 474 ; see Priv. Pls.’ Compl. ¶ 102; (2) limited States’ discretion “to seek waivers for substate areas like cities or counties,” by making LMAs the exclusive area subject to a waiver, States’ Am. Compl. ¶ 475; see Priv. Pls.’ Compl. ¶ 102; and (3) changed the waiver process in reliance on USDA’s so-called “operational experience,” without defining or explaining the terms and bases to “appraise interested parties fairly such that they had an opportunity to meaningfully comment,” States’ Am. Compl. ¶ 476; see Priv. Pls.’ Compl. ¶ 100. — 9 — The private plaintiffs filed their complaint the same day as the State plaintiffs and “[t]he two suits were consolidated under Federal Rule of Civil Procedure 42 without objection from the parties . . . .” D.C. I, 444 F. Supp. 3d at 14. — 10 — USDA does not renew its prior challenge to private plaintiffs’ standing, see D.C. I, 444 F. Supp. 3d at 40, and the record at summary judgment confirms that all plaintiffs’ have standing—an issue that must be established before a federal court may address the merits, see Hancock v. Urban Outfitters, Inc., 830 F.3d 511, 513 (D.C. Cir. 2016). Standing “must be supported . . . with the manner and degree of evidence required at the successive stages of litigation.” Lujan v. Defs. Of Wildlife, 504 U.S. 555, 561 (1992). “Now, on summary judgment, the plaintiffs must prove injury in fact with ‘specific facts’ in the record.” Humane Soc'y of the United States v. Perdue, 935 F.3d 598, 602 (D.C. Cir. 2019) (quoting Defs. Of Wildlife, 504 U.S. at 561 ). State plaintiffs plainly establish the requisite concrete, particularized, and actual or imminent injury-in-fact that would flow directly from the changes proposed in the Final Rule, including increased administrative burden on state-operated programs, rising health care costs for SNAP participants expelled from the program, and loss of tax revenue. See State Pls.’ MSJ at 41–42. Likewise, the findings, based on sworn affidavits in the record, regarding private plaintiffs’ standing remain intact. D.C. I, 444 F. Supp. 3d at 40-45. — 10 — D.C. I declined to reach these procedural challenges and instead, as USDA suggested, postponed consideration of plaintiffs’ arguments regarding procedural deficiencies in promulgation of the Final Rule, until the full administrative record had been compiled for review. D.C. I, 444 F. Supp. 3d at 21 (“[P]laintiffs’ other APA claims—that the Rule … was promulgated without observing procedural requirements—need not be addressed at this stage of the case.”); id. at 21 n.11 (accepting “suggestion” by USDA “that these procedural challenges may be best evaluated on a full administrative record, which is currently being compiled” (citing Defs.’ Opp’n to State Pls.’ Mem. Supp. Mot. for Prelim. Inj. (“Defs.’ PI Opp’n”) at 4, ECF No. 26). This Court made clear that “[t]he decision to postpone consideration of any arguments [regarding lack of adequate notice] at this stage in no way reflects any assessment that those arguments are less likely to succeed.” Id. 2. Plaintiffs’ Challenge to Waiver Changes D.C. I concluded that “plaintiffs are likely to succeed on their claim that the waiver changes in the Final Rule are arbitrary and capricious,” D.C. I, 444 F. Supp. 3d at 21, and ultimately enjoined these aspects of the Final Rule from going into effect as scheduled on April 1, 2020. Specifically, this Court found that USDA’s decision to eliminate the use of any metric for measuring insufficient jobs, under 7 U.S.C. § 2015 (o)(4)(A)(ii), other than the 24-month general unemployment rate, was inadequately explained and arbitrary and capricious, id. at 22– 27, and that USDA’s new restrictive definition of “the area” subject to a waiver, under 7 U.S.C. § 2015 (o)(4)(A), as limited to only an LMA, was likewise arbitrary and capricious, id. at 27–33. 3. Plaintiffs’ Challenge to Final Rule’s Discretionary Exemption Change Plaintiffs were denied preliminary injunctive relief on their claims that the Final Rule’s changes limiting the carryover of discretionary exemptions to one year and eliminating the States’ stockpile of such exemptions were inconsistent with the statutory scheme and arbitrary — 11 — and capricious, despite the States’ conceded reliance on longstanding agency policy. Id. at 20 . Moreover, plaintiffs were unable to show immediate irreparable harm, since the exemption changes in the Final Rule would not become effective until October 1, 2020, “eight months away.” Id. Plaintiffs argued that the Final Rule’s elimination of discretionary exemption carryovers was contrary to law, under the first step of the well-known test from Chevron U.S.A. Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 842 (1984), which focuses on whether the statutory text at issue is ambiguous. On plaintiffs’ reading, the proposed elimination of these exemption carryovers was foreclosed by the plain meaning of the statutory language referring to “the average monthly number of exemptions estimated for the State agency for the preceding fiscal year,” 7 U.S.C. § 2015 (o)(6)(G) (emphasis added), which, in their view, unambiguously includes all available exemptions, including those carried over from previous years. D.C. I, 444 F. Supp. 3d at 17. As support, plaintiffs cited the 2018 Farm Bill, which reduced the available exemption percentage from 15 percent to 12 percent without touching the exemption carry over provision, and the Conference Report on that legislation, which stated that “States will maintain the ability to . . . accrue exemptions and retain any carryover exemptions from previous years, consistent with current law.” Id. at 17–18 (citing H.R. Rep. No. 115-1072, at 616 (2018) (“2018 Farm Bill Conf. Rep.”)). USDA countered that the statute was ambiguous at best and contrary to plaintiffs’ position at worst. Id. at 17. Finding the readings tendered by both parties to be plausible, the text was found to be ambiguous, warranting assessment of the Final Rule’s exemption change, under Chevron step two and the arbitrary and capricious standard. Id. at 18. Since USDA had both considered numerous comments relaying the importance of the carryover exemptions to enable States to — 12 — respond nimbly to economic downturns and explicitly noted the serious reliance interests of States after two decades of functioning under a regime permitting indefinite carryover, id. at 19– 20, the Court concluded that USDA offered rational reasons for limiting such carryover exemptions, id. at 19. Regarding the “undoubtedly serious” reliance interests of the States, id. at 20, the Court applied the test from Encino Motorcars LLC v. Navarro, 136 S. Ct. 2117 (2016), which permits agencies to make policy changes adverse to affected parties’ reliance interests “‘as long as [agencies] provide a reasoned explanation for the change’ and ground the change in a reasonable interpretation of the governing statute,” D.C. I, 444 F. Supp. 3d at 16 (quoting Encino, 136 S. Ct. at 2125 ), and determined USDA’s actions met that test, id. at 20. Thus, plaintiffs’ bid to enjoin the exemption change failed. Thus, as relief, the Final Rule’s changes only to the regulatory process for time limit waivers were temporarily enjoined. D.C. I at 46. C. Subsequent Procedural History Following entry of the preliminary injunction, USDA appealed this Court’s order to the D.C. Circuit Court, where the appeal remains pending. Notice of Appeal to D.C. Cir. Court, D.C. v. U.S. Dep’t of Agric., No. 20-cv-00119 (D.D.C. May 12, 2020), ECF No. 61; see Notice of Appeal, D.C. v. U.S. Dep’t of Agric., No. 20-5136 (D.C. Cir. May 14, 2020). 11 Meanwhile, the parties filed and briefed cross motions for summary judgment, pursuant to a schedule adopted by the Court from the parties’ joint proposed schedule, see Min. Order (Mar. 23, 2020); Notice of Joint Proposed Schedule, ECF No. 53, and USDA compiled the administrative record, — 11 — A week after entry of the preliminary injunction, Congress passed the Families First Coronavirus Response Act (“FFCRA”), Pub. L. No. 116-127, 134 Stat. 178 (2020), which suspends ABAWD time limits until one month after the public health emergency declaration by the Secretary of Health and Human Services is lifted. FFCRA § 2301. Thus, as soon as the state of emergency is over, the Final Rule will go fully into effect absent the relief plaintiffs request here. — 13 — with a certified notice of the contents filed with the Court, see Notice Of Filing Of Certification Of Corrected Administrative Record, ECF No. 63. 12 Upon filing by the parties of their Joint Appendix of Administrative Record, in four volumes, on September 11, 2020, see Joint App’x, ECF No. 105, the pending motions are ripe for resolution. II. LEGAL STANDARD The Administrative Procedure Act (“APA”) provides for judicial review of any “final agency action for which there is no other adequate remedy in a court,” 5 U.S.C. § 704 , and “instructs a reviewing court to set aside agency action found to be ‘arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.’” Cigar Ass'n of Am. v. United States FDA, 964 F.3d 56, 61 (D.C. Cir. 2020) (quoting 5 U.S.C. § 706 (2)(A)). This standard “‘requires agencies to engage in reasoned decisionmaking,’ and . . . . to reasonably explain to reviewing courts the bases for the actions they take and the conclusions they reach.” Bhd. of Locomotive Eng'rs & Trainmen v. FRA, 972 F.3d 83, 115 (D.C. Cir. 2020) (quoting Dep't of Homeland Sec. v. Regents of the Univ. of Cal., 140 S. Ct. 1891, 1905 (2020)). While judicial review of agency action is limited to “the grounds that the agency invoked when it took the action,” Regents, 140 S. Ct. at 1907 (quoting Michigan v. EPA, 576 U.S. 743, 758 (2015)), the agency, too, “must defend its actions based on the reasons it gave when it acted,” id. at 1909. — 12 — The Final Rule’s administrative record totals 187,464 pages and includes regulations, guidance documents, reports, USDA analyses for the rulemaking, websites referenced in rulemaking, waivers of the ABAWD time limit received by various states from 2007 to 2020, USDA correspondence, analysis of public comments for rulemaking, supplemental state waiver maps, and public comments. Defs.’ Notice of Filing Certification of Corrected Admin. Record, Att. 2 (List of Contents of AR), ECF No. 91-2; see Defs.’ Mem. Pts. & Auth. Supp. Cross Mot. for Summ. J. and Opp’n to Pls.’ Mot. for Summ. J. (Defs.’ Opp’n) at 15, ECF No. 92 (noting “administrative record comprising nearly 180,000 pages of materials”). The parties’ Joint Appendix totals 1,784 pages. See Joint App’x of Administrative Record, ECF No. 105. State plaintiffs also submitted five declarations in support of their Motion for Summary Judgment. State Pls.’ Mot. Supp. Summ. J. (“State Pls.’ MSJ”), Exs. 1–5, ECF Nos. 65-1-5. In addition, the U.S. House of Representatives, counties, cities, think tanks, legal aid providers and various national and statebased non-profit organizations submitted nine amicus curiae briefs in support of plaintiffs. See ECF Nos. 70, 71, 81–86, 90. — 14 — The law is well-settled that “[a] rule is arbitrary and capricious if (1) the agency ‘has relied on factors which Congress has not intended it to consider’; (2) the agency ‘entirely failed to consider an important aspect of the problem’; (3) the agency's explanation ‘runs counter to the evidence before the agency’; or (4) the explanation ‘is so implausible that it could not be ascribed to a difference in view or the product of agency expertise.’” Am. Bankers Ass'n v. NCUA, 934 F.3d 649, 663 (D.C. Cir. 2019) (quoting Motor Vehicle Mfrs. Ass'n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co. (State Farm), 463 U.S. 29, 43 (1983)). Notably, the court “‘may not substitute [its] own judgment for that’ of the agency,” id. (quoting FERC v. Elec. Power Supply Ass'n, 136 S. Ct. 760, 782 (2016)), but nonetheless is “not a ‘rubber stamp’ and [] must ensure that the agency considered all of the relevant factors.” Oceana, Inc. v. Ross, 920 F.3d 855, 863 (D.C. Cir. 2019) (quoting Ethyl Corp. v. EPA, 541 F.2d 1 , 34 (D.C. Cir. 1976) (en banc)). III. DISCUSSION As discussed below, while D.C. I did not address plaintiffs’ procedural challenges to the Final Rule, with the benefit of the administrative record, the substantial flaws in USDA’s noticeand-comment process are apparent, requiring that this rule be set aside on this ground alone. See NRDC v. Wheeler, 955 F.3d 68, 85 (D.C. Cir. 2020) (“Failure to provide the required notice and to invite public comment—in contrast to the agency's failure . . . adequately to explain why it chose one approach rather than another for one aspect of an otherwise permissible rule—is a fundamental flaw that normally requires vacatur of the rule.”) (quoting Heartland Reg'l Med. Ctr. v. Sebelius, 566 F.3d 193, 199 (D.C. Cir. 2009) (internal quotation marks omitted)). Moreover, for the same reasons explained in D.C. I, 444 F. Supp. 3d at 21–34, the Final Rule’s significant changes to the current regulation’s implementation of 7 U.S.C. § 2015 (o)(4)(A)’s — 15 — waiver request process are arbitrary and capricious. Further, although D.C. I declined to find a likelihood of success on the merits of plaintiffs’ challenge to the Final Rule’s change in interpretation of 7 U.S.C. 2015(o)(6)(G) to eliminate discretionary exemption carryovers, D.C. I, 444 F. Supp. 3d at 17–21, a different conclusion is reached here, based on more detailed statutory interpretation arguments presented in the parties’ summary judgment briefing. Finally, USDA’s failure fully to consider both the significant cost impact on States and the disparate impact of the Final Rule bolster the conclusion that this agency is arbitrary and capricious for failure to consider significant critical comments relevant to the over-arching statutory purpose. 13 A. USDA Failed to Provide Sufficient Notice of Changes Adopted In Final Rule Plaintiffs challenge the sufficiency of the notice provided by USDA in the Proposed Rule as to the fundamental changes to the waiver process adopted in the Final Rule, see State Pls.’ MSJ at 11; Private Pls.’ MSJ at 44, which reversed long-standing agency policy that “states should have ‘broad discretion in defining areas that best reflect the labor market prospects of program participants and administrative needs.’” State Pls.’ MSJ at 5 (quoting — 13 — Private plaintiffs reiterate on summary judgment the argument that 7 U.S.C. § 2015 (o)(4)(A) requires the agency to “determine job-insufficiency waivers on a case-by-case basis” and that the Final Rule undermines this adjudicatory process by “predetermin[ing] the outcome of States’ waiver applications through a prospective categorical rule” and “relegat[ing] case-specific waiver adjudications to rare situations involving ‘extraordinary circumstances’ or anomalous geographic areas . . . .” Priv. Pls.’ MSJ at 9, ECF No. 64, 66. D.C. I did not address this argument. D.C. I, 443 F. Supp. 3d at 21, n.11. As an initial matter, the Final Rule simply does not change the requirement that USDA review, on a case-by-case basis, waiver requests when presented by a State. While the Final Rule aims to institute “broader application of the time limit” and “reduce [ABAWD] dependence on government benefits” by restricting how states can qualify for waivers, 84 Fed. Reg. at 66796; see also 84 Fed. Reg. at 981–82 (describing rulemaking to “ensure the waivers are applied on a more limited basis” by restricting how states can qualify for waivers), promulgating such general restrictions by rulemaking is entirely appropriate, see Am. Hosp. Ass’n v. NLRB, 499 U.S. 606, 612 (1991) (recognizing agency’s authority to “rely on rulemaking to resolve certain issues of general applicability”). As USDA explains, the agency is expressly authorized to “issue rules to administer SNAP,” Defs.’ MSJ at 34; see 7 U.S.C. § 2013 (c) (“The Secretary shall issue such regulations consistent with this chapter as the Secretary deems necessary or appropriate . . .”), and, as private plaintiffs concede, agencies have “discretion to proceed by rulemaking or adjudication,” unless Congress has otherwise spoken, Priv. Pls.’ MSJ at 15; Defs.’ MSJ at 34. In short, § 2015(o)(4)(A) does not bar USDA from establishing “general principles to guide the required case-by-case . . . determinations.” Am. Hosp. Ass’n v. NLRB, 499 U.S. at 612 . Consequently, private plaintiffs’ argument carries little weight. — 16 — ABAWD00000166 (U.S. DEP’T AGRIC., GUIDANCE FOR STATES SEEKING WAIVERS FOR FOOD STAMP LIMITS (1996) (“1996 Guidance”)), ECF No. 105-1). The procedural flaws identified by plaintiffs are fatal to the Final Rule. See State Pls.’ MSJ at 12 (“These failures alone require that the Rule be vacated.”). The APA requires agencies to provide the public with notice of a proposed rulemaking, an opportunity to comment, and, “[a]fter consideration of the relevant matter presented,” a “concise general statement” of the rule's basis and purpose. Sherley v. Sebelius, 689 F.3d 776, 784 (2012) (quoting 5 U.S.C. § 553 ); see also Perez v. Mortg. Bankers Ass'n, 575 U.S. 92, 96 , 135 S. Ct. 1199, 1203 (2015) (citing 5 U.S.C. § 553 (c) (describing three-step procedure for “notice-and-comment rulemaking, including to ‘give interested persons an opportunity to participate in the rule making through submission of written data, views, or arguments.’”)); Cigar Ass'n of Am., 964 F.3d at 63–64 (same). The importance of a robust notice-and-comment process is due to the “central purpose” served to “subject agency decisionmaking to public input and to obligate the agency to consider and respond to the material comments and concerns that are voiced.” Make the Rd. N.Y. v. Wolf, 962 F.3d 612, 634 (D.C. Cir. 2020); see also Nat'l Ass'n of Clean Water Agencies v. EPA, 734 F.3d 1115, 1148 (D.C. Cir. 2013) (“A purpose of noticeand-comment provisions under the APA . . . is ‘to ensure that affected parties have an opportunity to participate in and influence agency decision making at an early stage, when the agency is likely to give real consideration to alternative ideas.’” (quoting New Jersey, Department of Environmental Protection v. EPA, 626 F.2d 1038 , 1049 (D.C. Cir. 1980)). In addition, “part of the purpose of notice and comment rulemaking is to ensure the parties develop a record for judicial review.” Wolf, 962 F.3d at 634 (quotation and citation omitted). Put simply, “[rulemaking n]otice requirements are designed . . . to give affected parties an opportunity to — 17 — develop evidence in the record to support their objections to the rule and thereby enhance the quality of judicial review.” Int’l Union, United Mine Workers v. Mine Safety & Health Admin. (“Int’l Union”), 407 F.3d 1250, 1259 (D.C. Cir. 2005). As part of notice-and-comment rulemaking, the notice of proposed rulemaking must contain “either the terms or substance of the proposed rule or a description of the subjects and issues involved,” 5 U.S.C. § 553 (b)(3), a requirement construed “to mean that the final rule the agency adopts must be a logical outgrowth of the rule proposed,” Long Island Care at Home, Ltd. v. Coke (“Long Island Care”), 551 U.S. 158, 174 (2007) (internal quotation and citations omitted). “The object, in short, is one of fair notice.” Id. “A final rule is the ‘logical outgrowth’ of a proposed rule if interested parties should have anticipated that the change was possible, and thus reasonably should have filed their comments on the subject during the noticeand comment period.” Chesapeake Climate Action Network v. EPA, 952 F.3d 310, 319 (D.C. Cir. 2020) (quoting Clean Air Council v. Pruitt, 862 F.3d 1, 10 (D.C. Cir. 2017) (quoting CSX Transp., Inc. v. Surface Transp. Bd., 584 F.3d 1076 , 1080 (D.C. Cir. 2009)). An agency fails the logical outgrowth test if “interested parties would have had to divine the agency’s unspoken thoughts, because the final rule was surprisingly distant from the proposed rule.” Id at 319–20. Set against this standard, USDA failed to provide adequate notice of the Final Rule in three ways, which are discussed seriatim. 1. USDA Provided No Notice of Elimination of Extended Unemployment Benefits Criterion as Basis for Waiver Request No reference to extended unemployment benefits, let alone the potential elimination of this criterion for a State waiver request, was made in the Advance Notice of Proposed Rulemaking for the Final Rule. See generally 83 Fed. Reg. 8013 . The Proposed Rule similarly gave no indication that such elimination was contemplated. See generally 84 Fed. Reg. 980 . To — 18 — the contrary, the Proposed Rule expressly and repeatedly stated that extended unemployment benefits would be retained as a basis for State waiver requests. For example, the Proposed Rule provided explicit assurance that the agency “would continue to approve a State’s waiver request that is based upon the requesting State’s qualification for extended unemployment benefits,” 84 Fed. Reg. at 985, even going so far as to designate extended unemployment benefits as one of three “core standards for [waiver] approval,” id. (“[T]hree provisions . . . the unemployment rate over 10 percent standard, the 20 percent standard, and the qualification for extended unemployment benefits standard” would be the “core standards for [waiver] approval.”). While all other waiver criteria based on substate data would be eliminated, the Proposed Rule promised that extended unemployment benefits would be retained as a basis for statewide waivers. Id.; see also id. at 987 (“. . . as stated previously, the proposed rule would no longer provide for statewide waivers except for those waivers approved based upon a state’s qualification for extended unemployment benefits.”). Consistent with these assurances, the proposed amended language of the applicable regulation, 7 CFR § 273.24 , set out extended unemployment benefits as one criterion on the exhaustive list of waiver request qualifications. Id. at 992. The Final Rule, however, did exactly the opposite of what was promised and proposed and instead eliminated all bases, other than the prescribed unemployment rate, for a State’s waiver request, including a State’s qualification for extended unemployment benefits. 84 Fed. Reg. at 66789 (conceding that USDA proposed to “continue to approve any waiver request that is supported by the requesting State’s qualification for extended unemployment benefits . . . [and] to prohibit statewide waivers when substate data is available, except for those States qualifying under the extended unemployment benefits standard.”). State plaintiffs correctly — 19 — point out that such a “surprise switcheroo” is impermissible under the APA. State Pls.’ MSJ at 11 (citing Envt’l. Integrity Project v. EPA, 425 F.3d 992 , 996 (D.C. Cir. 2005)). USDA defends its about-face, arguing that “an agency’s refusal to adopt its proposal is always a logical outgrowth of the proposal.” Defs.’ MSJ at 67 (citing New York v. EPA, 413 F.3d 3, 44 (D.C. Cir. 2005) (per curiam) and Ariz. Pub. Serv. Co. v. EPA, 211 F.3d 1280 , 1299– 1300 (D.C. Cir. 2000)). To be sure, when an agency ultimately decides not to adopt a proposed change but instead to retain the status quo ante, the logical outgrowth test is satisfied because exercising the option of making no change is always, at least implicitly, on the table. See Long Island Care, 551 U.S. at 174–75 (rejecting procedural challenge, explaining that “[s]ince the proposed rule was simply a proposal, its presence meant that the Department was considering the matter; after that consideration the Department might choose to adopt the proposal or to withdraw it” and the “possibility” of such withdrawal was “reasonably foreseeable”) (emphasis in original); Idaho Conservation League v. Wheeler, 930 F.3d 494 , 508–09 (D.C. Cir. 2019) (“an agency's decision to withdraw a proposed rule is a logical outgrowth of the proposal” since a proposed rule “adequately put interested parties on notice that the [agency] was planning regulatory action that might, as happened, not materialize”); New York v. EPA, 413 F.3d at 44 (rejecting procedural challenge where EPA proposed “‘menu of alternatives’ approach” but instead adopted mandatory approach of “the status quo ante,” observing that EPA had “received extensive comments on all aspects of the rule, including whether to integrate the” menu elements into a set of minimum requirements, showing no “mystification here,” and opining that “[o]ne logical outgrowth of a proposal is surely . . . to refrain from taking the proposed step” (quoting Am. Iron & Steel Inst. v. EPA, 886 F.2d 390, 400 (D.C. Cir. 1989))). Likewise, when an agency adopts a sub-part of a proposed rule change, rather than the proposal in its entirety, — 20 — this, too, may be deemed reasonably noticed under the logical outgrowth test. See Arizona Pub. Serv. Co., 211 F.3d at 1299–1300 (finding sufficient notice where agency first proposed that Indian tribes be required to meet the “same requirements” as States with respect to judicial review of Clean Air Act, but then adopted a final rule that exempted tribes from some, though not all, such requirements). Neither of these scenarios occurred here: first, the Final Rule eliminates, rather than retains, the status quo ante authorizing waivers based on a State’s qualification for extended unemployment benefits, and, second, such elimination was never suggested as part of the proposed changes in the Proposed Rule, which instead assured retention of this criterion. USDA’s reliance on New York v. EPA, Defs.’ Opp’n at 67, and Arizona Pub. Serv. Co., Defs.’ Reply at 33, to excuse the agency’s lack of notice is therefore wholly misplaced. Essentially conceding, as it must, that no specific notice of the elimination of extended unemployment benefits was provided in the Proposed Rule, USDA falls back to posit that the “full context of the Proposed Rule” put interested parties “on notice” that the extended unemployment benefits qualification could be eliminated to avoid “obvious tension with USDA’s proposal to drastically curtail the availability of statewide waivers.” Defs.’ Opp’n at 68. This is not so. While “[r]estricting [s]tatewide [w]aivers” was an express goal of the rule changes, the two specific mechanisms outlined to achieve that goal in the Proposed Rule, 84 Fed. Reg. at 985-86 (describing elimination of statewide waiver approvals when substate data is available and “prohibit[ing] States from grouping areas”), made no mention of eliminating extended unemployment benefits. In any event, USDA’s excuse of a “wholesale revision to its waiver criteria,” Defs.’ Opp’n at 68, does not overcome this procedural flaw since a generally announced goal does not — 21 — excuse adoption of a policy altering an aspect of the status quo ante that the agency promised to retain. Indeed, the D.C. Circuit has not hesitated to reject as procedurally deficient reliance on a generally outlined scheme of proposed agency action, explaining that “general notice that a new standard will be adopted affords the parties scant opportunity for comment,” and thus the “agency’s obligation is more demanding--it must ‘describe the range of alternatives being considered with reasonable specificity. Otherwise, interested parties will not know what to comment on, and notice will not lead to better-informed agency decision-making.’” Horsehead Res. Dev. Co. v. Browner, 16 F.3d 1246 , 1267-69 (1994) (quoting Small Refiner Lead PhaseDown Task Force v. EPA, 705 F.2d 506 , 549 (D.C. Cir. 1983)). Given the Proposed Rule’s assurances that extended unemployment benefits would be retained, no reasonable interested party would have “anticipated” elimination of this criterion in the Final Rule and thus adoption of that position in the Final Rule was not reasonably on the table for comment. Cf. Long Island Care, 551 U.S. at 174-75 ; Idaho Conservation League v. Wheeler, 930 F.3d at 508-09 ; New York v. EPA, 413 F.3d at 44 ; see also CSX Transp. Inc. v. Surface Transp. Bd., 584 F.3d at 1079–80 (noting that final rule is only a logical outgrowth if “interested parties ‘should have anticipated’ that the change was possible, and . . . reasonably should have filed their comments on the subject during the notice-and-comment period.” (citing Ne. Md. Waste Disposal Auth. V. EPA, 358 F.3d 936, 952 (D.C. Cir. 2004)). Not surprisingly, USDA “did not receive many comments with regard to retaining the extended unemployment benefits standard,” 84 Fed. Reg. at 66789, because the agency failed to give the public any inkling that the extended unemployment benefits standard might be eliminated. Cf. New York v. EPA, 413 F.3d at 44 (noting “extensive comments” received). — 22 — The distance between the Proposed Rule’s assurances of retention of extended unemployment benefits to qualify for a waiver and the Final Rule’s elimination of that criterion altogether could not be greater. See Int’l Union, 407 F.3d at 1259–60 (finding notice insufficient where final rule is “‘surprisingly distant’ from the proposed rule” (quoting Ariz. Pub. Serv. Co. v. EPA, 211 F.3d at 1299)). A “blunder [in] a proposed rule” may be remedied by a “retreat to the status quo ante,” but eliminating the latter without so much as a hint that the agency might do so is impermissible under the APA. Am. Iron & Steel Inst. v. EPA, 886 F.2d at 400 . 2. USDA Provided No Notice of Limiting Waiver Area to LMA The rulemaking process leading up to promulgation of the Final Rule made no mention that “the area” subject to waiver, under 7 U.S.C. § 2015 (o)(4)(A), would be restricted only to LMAs. In its 2018 Advance Notice of Proposed Rulemaking, USDA recognized that “States currently have discretion to define the area they are requesting to waive,” 83 Fed. Reg. at 8015; id. (“The decision to request and implement an ABAWD time limit waiver rests with the States.”), and posed the questions, “Should States maintain this flexibility? Should an ‘economic area’ be limited in geographic scope, such as to a single county, metropolitan area, or labor market area?” id. To the extent this query obliquely alluded to restricting “the area” to an LMA, the next step of the rulemaking process, with issuance of the Proposed Rule, seemingly dropped this proposal. The Proposed Rule repeatedly acknowledged the fact that current regulations and agency guidance “provide States with the discretion to define the areas to be covered by waivers,” 84 Fed. Reg. at 985, and mentioned limiting, not eliminating, that flexibility in only two circumstances. First, USDA proposed to “eliminate statewide waiver approvals when substate data is available . . . except for waivers based upon the State’s qualification for extended unemployment benefits,” id., again providing false assurance that such extended unemployment — 23 — benefits would remain a waiver qualification, see supra Part III.A.1. USDA explained this proposal would allow States to “target[] those particular areas in which unemployment rates are high,” and “that a more targeted approach would ensure that waivers exist only in areas that do not have a sufficient number of jobs to provide employment for the individuals living in that specific area.” 84 Fed. Reg. at 985. This first proposed limitation on statewide waivers made no mention of restricting waiver areas to LMAs, but simply explained the proposal was intended to have States utilize substate data when such data was “available” for substate areas for “a more targeted approach.” Id. The second circumstance in which USDA proposed limiting State flexibility in defining “the area” subject to waiver, was in the grouping of areas, stating that “[t]he proposed rule would prohibit States from grouping areas, except for areas that are designated a Labor Market Area (LMA)…to ensure that only areas that are economically tied are grouped together, [such that] States would be unable to omit certain areas within the LMA in the State for the purpose of achieving a qualifying unemployment rate for part of an LMA.” 84 Fed. Reg. at 986. As an alternative to limiting grouped areas to LMAs, the Proposed Rule asked whether USDA “should prohibit grouping entirely,” id., noting that “[i]f grouping were prohibited entirely, waived areas would be limited to individually qualifying jurisdictions with corresponding data (for example, counties and their equivalents, cities, and towns),” id. This focus on allowing grouping of areas only in LMAs, versus disallowing grouping altogether for substate areas, did not raise the prospect of eliminating altogether the long-standing policy of State flexibility in defining “the area” subject to waiver by restricting all waiver areas to an LMA, no matter the geographic size, scope, number of individual counties, cities or towns encompassed by an LMA. — 24 — Nevertheless, in the Final Rule, USDA shifted gears to “only allow for waivers covering LMAs; not individual jurisdictions within LMAs, such as counties or county equivalents, and not for any State-defined groupings of substate areas,” in direct contradiction to the proposals set out in the Proposed Rule. 84 Fed. Reg. at 66796. Plaintiffs understandably complain that such an about-face in eliminating all State discretion in defining “the area” subject to waiver “deprived [p]laintiffs and the public of the chance to comment on key provisions of the Rule.” State Pls.’ MSJ at 12. The Court agrees. Merely mentioning in the Proposed Rule the potential usefulness of LMAs as a measure of waiver areas in a specific context of substate area grouping did not provide adequate notice that the Final Rule would adopt LMAs as the sole measure of any waiver area. See Horsehead Res. Dev. Co. v. Browner, 16 F.3d at 1267-68 (finding procedural fault with agency rulemaking that failed to indicate contemplation of the possibility of combining parts of proposals, explaining “[t]his omission is critical because notice of individual parts of a proposed rule is not necessarily notice of the whole. One purpose of notice is to promote informed decisionmaking, and comments addressed to one specific component part of the standard do not necessarily bear on the viability … as a whole.”). This is yet another fatal procedural flaw in the Final Rule. 3. USDA’s Opaque Reference to “Operational Experience” Provided Inadequate Notice of Elimination of State Flexibility in Both Defining Waiver Area and Using More Probative Data Finally, State plaintiffs posit that USDA denied the opportunity for meaningful comment by relying, vaguely, on the agency’s “operational experience” as the basis for its proposed changes. State Pls.’ MSJ at 12. That term certainly made no appearance in the agency’s Advance Notice of Proposed Rulemaking, when indicating that “the ABAWD time limit waivers represent an area of concern for the Department.” 83 Fed. Reg. at 8015. Indeed, rather than accuse States of waiver manipulation or abusive practices, the agency acknowledged that “not all — 25 — States that are eligible for ABAWD time limit waivers request one,” id., and, further, that “as economic conditions [have] improved, there has been a decline in the use of these waivers.” id. (citing that, in FY 2013, 45 States, D.C., Guam and the Virgin Island used waivers, with 42 covering the entire State or jurisdiction, compared, in FY 2017, to 33 States, D.C., Guam and the Virgin Islands using waivers, with only 9 covering the entire State or jurisdiction). Despite the overall decline in the number of waivers, marked by a deep decline in the number of Statewide waivers, the agency expressed concern “that the number of areas waived has not decreased as much as would be expected during the continued decline in unemployment rates over this time period,” id., though no information was provided about what that “expected” number might be. Consistent with the Advance Notice, the Proposed Rule issued the following year, documented a continuing decline in Statewide waivers, stating that as of “April 2018, 8 waivers applied to an entire State, and 28 covered specific areas within a State.” 84 Fed. Reg. at 982. Thus, between July 2013 and April 2018, the number of overall waivers dropped from 45 to 36 and the number of state-wide waivers dropped from 42 to 8, while the number of waivers covering only substate areas, reflecting a more targeted approach, increased from 6 to 28. Id. Nevertheless, the Proposed Rule viewed as problematic that “[d]espite the national unemployment rate’s decline, . . . a significant number of States continue to qualify for and use ABAWD waivers under the current waiver standards.” Id. at 981. While the number of waivers used over a five-year period show declining numbers of waivers and deep reductions in use of Statewide waivers in favor of more targeted substate waivers, USDA cited its “operational experience,” without offering any supporting definition, timeframe, or analytical framework for the term, as the basis for wholesale changes to the waiver process. The Proposed Rule identified “key concerns,” without specific identification of States — 26 — or jurisdictions for accurate rebuttal or commenters’ analysis, that “some States have maximized the number of areas or people covered by waivers by combining data from areas with high unemployment with areas with low unemployment [such that] not all individual sub-areas would have qualified on their own,” id.; and “several States that have historically requested 12-month waivers on a fiscal year basis . . . have shifted their waiver request and implementation dates to later in the fiscal year . . . based on the 20 percent standard . . . to capitalize on older data and qualify for waivers of the ABAWD time limit for additional time,” id. at 986. Plaintiffs take issue with USDA’s attempt to claim states have manipulated the waiver criteria to “reciev[e] waivers in areas that do not clearly demonstrate a lack of sufficient jobs,” id. at 981, by relying on “operational experience,” without specific backup, as the agency’s sole evidence for such conclusion. State Pls.’ MSJ at 12. The caselaw supports plaintiffs’ position. USDA’s vague invocations of “operational experience” reference no “technical studies [or] data that [the agency] employed in reaching the decision[] to propose” the rule at issue here. Conn. Light & Power v. Nuclear Regulatory Comm’n, 673 F.2d 525, 530 (D.C. Cir. 1982). Such evidence is necessary “to allow for useful criticism,” id., because “in order to have a ‘meaningful’ opportunity to comment, [the public] must be aware of the information the agency finally decides to rely on in taking agency action.” Nat’l Asphalt Pavement Ass’n v. Train, 539 F.2d 775 , 779 n.2 (D.C. Cir. 1976) (citing Portland Cement Ass’n v. Ruckelshaus, 486 F.2d 375 , 393 n.67 (D.C. Cir. 1973)). Instead, USDA offers only assertions “that cannot be meaningfully tested or addressed.” State Pls.’ MSJ at 12. USDA responds that the agency “cited its own experience with the ABAWD time limit to identify perceived weaknesses in the 2001 Regulation,” asserting that it “clearly explained why it thought these were weaknesses and identified examples of exactly these sorts of requests.” — 27 — Defs.’ Opp’n at 69. The agency’s explanation was far from “clear[]” in at least three respects, however. First, the examples USDA offers do not identify any specific instances of alleged State abuse of waiver criteria to obtain a time limit waiver in an area that actually had a sufficient number of jobs for ABAWDs. Rather, the agency speaks only generally about state grouping practices that sometimes include low-unemployment areas for the general population, not the ABAWD population. Second, the agency justified effectively eliminating statewide waivers in order to minimize inclusion of low-unemployment areas and “ensure that waivers of the ABAWD time limit are more appropriately targeted to those particular areas that have unemployment rates of over 10 percent or lack sufficient jobs, as required by the Act.” 84 Fed. Reg. at 66790; see also id. (finding “that waivers should be targeted to economically-tied areas with a lack of sufficient jobs, rather than entire states that contain distinct economic regions.”). Yet, the cited problem of statewide waivers is simply not borne out by the data. The agency had already documented a deep decline over five years in use of statewide waivers and more “appropriately targeted” use of substate waivers, undercutting the agency’s claimed “operational experience” showing manipulative practices by States. Third, and most notably, USDA’s complaint, based on purported “operational experience” of “waiver misuse and abuse,” id. at 66806, due to States inappropriately obtaining waivers in areas with sufficient jobs for ABAWDs, is contradicted by the agency’s description of its process of “carefully review[ing] all State waiver requests,” id. at 66800, “includ[ing] independently obtaining and validating the data and evidence presented by the State in support of all requested areas to determine if the areas meet the standards for approval,” id. As part of that process, the agency noted that “discrepancies or inaccuracies in the data presented by the waiver — 28 — requesting State” have been identified and, “[i]n some cases,” result in denial or only partial approval of the “waiver request because not all areas meet the standards for approval.” Id. 14 In other words, the approved waiver requests had been fully vetted by the agency as meeting the requirements for ABAWD time limits waivers. For the agency then to cite these vetted and agency-approved waiver requests as “questionable as to whether the statutory conditions for approval…are present,” id. at 66783, and as evidence of States’ “manipulation,” id., not only makes “questionable” the “operational experience” justification for the waiver process changes and amplifies the need for clearer notice of the factual basis for the agency’s claim of States’ abuse of the waiver process. D.C. I determined, as a preliminary matter, that USDA provided “‘no evidence’ that such state manipulation is ‘a real problem’,” D.C. I, 443 F.Supp.3d at 27 (citing Nat’l Fuel Gas Supply Corp. v. FERC, 468 F.3d 831, 841 (D.C. Cir. 2006)), explaining that, in order to rely on “experience,” an agency must “‘adequately record[] and explain[] that experience on the record.’” Id. (citing Nat’l Tour Brokers Ass’n v. ICC, 671 F.2d 528 (D.C. Cir. 1982)). The record on summary judgment does nothing to alter the Court’s prior conclusion that USDA failed to provide adequate factual support for the agency’s critical and predicate finding based on its experience that the States were “receiving waivers in areas that do not clearly demonstrate a — 14 — At the same time USDA touts its rigorous vetting process for waiver requests, the agency doubles down on its “years of experience” for the belief “that the 2001 Regulation included a weakness, and therefore that the regulatory framework needed strengthening,” Defs.’ Opp’n at 54, because the “very fact that, under the 2001 Regulation, waiver requests . . .... satisfied the standard for a lack of sufficient jobs was the impetus for the redefinition of the waiver area,” id. n. 25, since “under that framework, USDA simply could not reject waivers that, for example, grouped counties separated by hundreds of miles into a single area or conspicuously omitted nearby jurisdictions whose inclusion would mean the request no longer met the 20% standard,” id. at 55; see also 84 Fed. Reg. at 982 (“current regulations give States an opportunity to qualify for waivers and avoid the ABAWD time limit when economic conditions do not justify such relief”). In faulting both the States’ purported abuse of the waiver process and the 2001 Regulation for permitting “strategically grouped waivers,” Defs.’ Opp’n at 54 n.23, the agency still fails to explain precisely why and identify which such groupings fall short of meeting the statutory requirement for a waiver, as plaintiffs correctly point out. State Pls.’ Reply at 30. — 29 — lack of sufficient jobs” for ABAWDS, 84 Fed. Reg. at 981, depriving plaintiffs and other interested parties of the opportunity to test the agency’s proffered conclusions and analysis, by providing correction, context or alternative interpretations of the data on which the agency relied. Consequently, the agency failed to provide a meaningfully opportunity to participate in the notice-and-comment process. *** The procedural flaws in USDA’s notice-and-comment process require vacatur of the Final Rule. As discussed further, procedural blunders by USDA are not the only shortcoming in the Final Rule. B. The Final Rule’s Waiver Changes are Arbitrary and Capricious and Not in Accordance with Law The Final Rule’s elimination of States’ historic flexibility in delineating “the area” for a requested waiver by limiting such an “area” to an LMA, is both contrary to law and inadequately justified. D.C. I, 444 F. Supp. 3d at 27–33. Compounding this error, the Final Rule’s elimination of all measures for assessing employment opportunities for ABAWDs, under 7 U.S.C. § 2015 (o)(4)(A)(ii), other than a measure based on a 24-month average standard unemployment rate, arbitrarily writes a statutory “distinction out of the Rule,” D.C. I, 444 F. Supp. 3d at 16, and is otherwise arbitrary and capricious, id. at 22-27. 1. Final Rule’s Restriction of Waiver “Area” to U.S. Department of LaborDefined LMA is Both Contrary to Law and Arbitrary and Capricious The Final Rule profoundly re-interprets 7 U.S.C. § 2015 (o)(4)(A) by eliminating the States’ long-standing role in delineating “the area” for which a State may make a waiver request. The Final Rule accomplishes this by “[e]stablishing a [s]trict [d]efinition of [w]aiver ‘[a]rea,’” 84 Fed. Reg. at 66793, using LMAs, a broad measure developed by the U.S. Department of Labor, without regard to political, geographic or State boundaries, to designate “an economically — 30 — integrated area within which individuals can reside and find employment within a reasonable distance or can readily change jobs without changing their place of residence,” id. USDA acknowledges that “it was changing its position” on “the appropriate level of State flexibility,” Defs.’ Opp’n at 50, but boldly asserts that “[t]he States’ role was a matter of grace, not of right,” id. at 32. USDA has overplayed its hand, however, in taking this aggressive stance vis a vis the States in altering the role of the Federal government in this program designed to operate as a partnership between the Federal and State governments. Plaintiffs are correct that this new policy is both contrary to law and arbitrary and capricious. 15 a. Restricting Waiver “Area” to LMA Is Contrary to Law The parties agree that § 2015(o)(4)(A) instructs USDA to determine the answer to a simple question: whether there are “insufficient jobs to provide employment for a group of ABAWDs in the area where they reside?” State Pls.’ MSJ at 13 (citing 7 U.S.C. 2015(o)(4)(A)); Defs.’ Opp’n at 19. In plaintiffs’ view, this question requires use of a definition of “the area” for purposes of a waiver request that corresponds to where “group[s] of individuals” live within a State. State Pls.’ MSJ at 15. By contrast, the LMA definition comprising a “statistical area covering multiple states and tens of counties” is, according to plaintiffs, inappropriate. Id. Although reference to “any group of individuals,” 7 U.S.C. § 2015 (o)(4)(A), may have “a wide reach,” State Pls.’ MSJ at 15 (quoting Babb v. Wilkie, 140 S. Ct. 1168 , 1173 & n.3 (2020)), the Court agrees that the statutory text firmly tethers these individual ABAWDs for whom a State may request a time limit waiver to a specific State. The statutory language of § 2015(o)(4)(A) referring to “any group of individuals in the State,” clearly contemplates that the “individuals” for whom a State may request a waiver are within the state, see Priv. Pls. MSJ at 41; otherwise, — 15 — D.C. I did not address whether USDA’s redefinition of waiver “area” was contrary to law at the preliminary injunction stage of these proceedings. See D.C. I, 444 F. Supp. 3d at 21. — 31 — the text would have referred to only “any group of individuals,” without the qualifier “in the State.” Moreover, the statute grants each State the authority to request a waiver only for the benefit of “individuals in the State,” 7 U.S.C. § 2015 (o)(4)(A) (“On the request of a State agency . . .”), with the concomitant “responsibility for designing the request, including selecting for what ‘group of individuals’ to request a waiver,” State Pls.’ MSJ at 16 (citing Miller v. Casey, 730 F.2d 773 , 776–77 (D.C. Cir. 1984)). USDA concedes that waivers are only issued at the request of a State agency, but counters that this language does not mean the scope of “area . . . must be defined by the States.” Defs.’ Opp’n at 30. That may be so. The statute gives the agency ultimate authority to grant or deny a waiver request, see 7 U.S.C. § 2015 (o)(4)(A) (“ . . . the Secretary may waive the applicability of paragraph (2) . . . if the Secretary makes a determination that the area in which the individuals reside” fits the waiver criteria), and, according to USDA, this “broad delegation of authority to USDA implies that the agency retains the authority to define the scope of the area subject to a waiver approval because information about the labor market characteristics of the ‘area’ where the selected group resides is relevant to the statutory determination about job availability,” Defs.’ Opp’n at 30. 16 Even if USDA retains the authority to define the term “area,” however, this means only that the agency must exercise this responsibility by adopting a sensible definition that comports with the statute, including State-specific statutory directives. The Final Rule, however, adopts a definition of “area” using LMAs, notwithstanding the fact that — 16 — USDA further argues that, “if States controlled the area subject to a waiver request,” it would undercut the Secretary’s “authority to ‘make[] a determination’” that an area qualifies for waiver and “would effectively [] delegate[]” the determination “to States.” Defs.’ Opp’n at 31 (citing 7 U.S.C. § 2015 (o)(4)(A)(i)). This argument is logically flawed. Merely because States’ waiver requests may identify an “area” with insufficient job opportunities for ABAWDs does nothing to curtail the Secretary’s authority to grant or deny the application based on the agency’s own assessment of the requests and the data on which the requests are predicated. — 32 — some LMAs extend beyond state boundaries. This definition therefore contorts, rather than comports with, the State-specific statutory directives. This misfit between the new LMA definition of “the area” and the statutory directives in § 2015(o)(4)(A) has at least two consequences that persuade this Court that the Final Rule’s definition is contrary to law. First, restricting waiver “areas” to LMAs is simultaneously too narrow and overbroad for the operation of the statute. An LMA may define too narrow an “area,” because the key phrase “any group of individuals in the State” is “not limited to only all ABAWDs living in the expansive boundaries of a particular statistical designation,” State Pls.’ MSJ at 15, such as an LMA, but allows States to seek relief for any defined group of ABAWDs residing in towns, cities or counties in the State, whether in a DOL-defined LMA or not. An LMA may also define an overbroad “area” because LMAs can “cover[] multiple states and tens of counties,” id., while “the area” subject to waiver must be for “individuals in the State” where “the individuals reside,” 7 U.S.C. § 2015 (o)(4)(A). Second – and related to the overbroad aspect of the LMA definition – restricting the term “area” to an LMA “renders the phrase ‘any group of individuals’ largely meaningless.” State Pls.’ MSJ at 16. If the statute were intended to allow waivers only in federally-defined statistical areas that cross numerous state and county boundaries “to the exclusion of all other metrics,” id. at 16–17, then the law would have no need to require States to identify “any group of individuals in the State,” for whom employment opportunities were insufficient. Instead, however, neighboring statutory subsections “require state-designed programs responsive to ‘State or local workforce needs.’” Id. at 16 n.4 (citing 7 U.S.C. § 2015 (d)(3)). Statutory construction requires statutes to be construed so as to “‘give effect . . . to every clause and word of a statute.’” Duncan v. Walker, 533 U.S. 167, 174 (2001) (citing U.S. v. Menasche, 348 U.S. 528 , 538–39 (1955)), — 33 — and the Final Rule’s definition of waiver “area” runs afoul of this requirement by shifting focus far from the “group of individuals,” for whom a waiver request must be tailored in a State and where they reside. See Nat’l Credit Union Admin v. First Nat. Bank & Trust Co., 522 U.S. 479, 501 (1998) (finding administrative interpretation, which rendered a statutory phrase surplusage, out of step with what the statute provides). As the next part of this discussion illustrates, the misfit of restricting “the area” under § 2015(o)(4)(A) to LMAs is so great that this agency choice is also arbitrary and capricious. b. Restriction of Waiver “Area” to LMA was Arbitrary and Capricious D.C. I granted plaintiffs’ request for temporary injunctive relief as to the Final Rule’s adoption of LMA to define a waiver “area” because this change was likely arbitrary and capricious in two respects: first, USDA’s state abuse justification was “wholly unsupported,” D.C. I, 444 F. Supp. 3d at 27, and, second, the agency failed adequately to respond to the “considerable evidence suggesting that the new definition of area was inappropriate,” id. Nothing in the record at summary judgement warrants altering these findings. See id. at 27-33. 17 The Proposed Rule justified the rulemaking to address States’ abuse of the waiver process by “receiving waivers in areas that do not clearly demonstrate a lack of sufficient jobs,” 84 Fed. Reg. at 981, leading to the Final Rule’s adoption of a restrictive definition of waiver “area” that was predicated on “operational experience” showing States had “too much flexibility” and were “grouping areas in such a way to maximize waived areas,” 84 Fed. Reg. at — 17 — Private plaintiffs also argue that the waiver “area” definition change is merely an effort by the agency to remove people from the SNAP roster, as evidenced by internal inconsistencies in USDA’s reasoning. Priv. Pls. MSJ at 35–36. Even if the agency’s underlying purpose for the Final Rule is to reduce ABAWDs’ reliance on SNAP, this purpose is within the agency’s discretion. As the Final Rule explains, the purpose of PRWORA “was to ‘[promote] work over welfare and self-reliance over dependency, thereby showing true compassion for those in America who need a helping hand, not a handout.’” 84 Fed. Reg. at 66782 (quoting H.R. REP. NO. 104-725, at 261 (1996) (Conf. Rep.)). As such, Congress’ intent for the agency, in enacting time limits and work requirements for ABAWDs, was to encourage “those who can work . . . [to] work,” id. at 66783, and the agency cannot be faulted for pursuing that end, except insofar as it does so in violation of the law, including the APA. — 34 — 66794. D.C. I determined that the agency “‘provided no evidence’ that such state manipulation [was] ‘a real problem,’” D.C. I, 444 F. Supp. 3d at 27 (citing Nat’l Fuel Gas Supply Corp. v. FERC, 468 F.3d 831, 841 (D.C. Cir. 2006)), and that states’ practice of “grouping contiguous counties with relatively high unemployment while omitting contiguous counties with relatively low unemployment” was “entirely consistent with good faith efforts by states to target waiver requests to areas that lack sufficient jobs” for ABAWDs, id. at 28–29. The agency’s own documentation, showing a five-year decline in States’ use of waivers generally and deep decline in statewide waivers, from 45 in 2013 to only 8 in 2018, in favor of more targeted use of substate waivers, already reflects a trend in line with the agency’s view that statewide waivers should be restricted to “more appropriately target[]” areas of need “rather than the larger areas of entire states.” 84 Fed. Reg. at 66790. USDA permitted states to exercise “broad discretion in defining areas that best reflect the labor market prospects of program participants” for nearly twenty-five years. ABAWD00000166–68 (1996 Guidance), ECF No. 105-1; see also ABAWD00000089 (Proposed Rule, Food Stamp Program: Personal Responsibility Provisions of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996, 64 Fed. Reg. 70920 , 70945 (Dec. 17, 1999) (same)), ECF No. 105-1; ABAWD00000316 (U.S. DEP’T AGRIC., SUPPORTING REQUESTS TO WAIVE THE TIME LIMIT FOR ABLE-BODIED ADULTS WITHOUT DEPENDENTS (2016) (same)), ECF No. 105-1; ABAWD00000124 (66 Fed. Reg. at 4463 (maintaining that “State agencies have complete discretion to define the geographic areas covered by waivers”)), ECF No. 105-1; ABAWD00000212 (U.S. DEP’T AGRIC., GUIDANCE ON REQUESTING ABAWD WAIVERS (2006) (“2006 Guidance”) (confirming that the “State is responsible for clearly saying which areas are to be waived and under what criteria.”)), ECF No. 105-1. An agency proposing to make a — 35 — significant change to an existing policy based “upon factual findings that contradict those which underlay its prior policy,” is required to support its “good reasons” for the new policy with a “more detailed justification than what would suffice for a new policy created on a blank slate.” FCC v. Fox Television Stations, Inc., 556 U.S. 502, 515 (2009). Such “good reason” requires more than “nodding to [commenters’] concerns . . . only to dismiss them in a conclusory manner.” Gresham v. Azar, 950 F.3d 93, 103 (D.C. Cir. 2020) (citing Am. Wild Horse Pres. Campaign v. Perdue, 873 F.3d 914, 932 (D.C. Cir. 2017) and then citing Getty v. Fed. Savs. & Loan Ins. Corp., 805 F.2d 1050, 1055 (D.C. Cir. 1986)). Further, when the agency is rescinding a prior policy, “its reasoned analysis must consider the ‘alternative[s]’ that are ‘within the ambit of the existing [policy].’” Regents, 140 S. Ct. at 1913 (quoting State Farm, 463 U.S. at 51 ). Failing to provide such an explanation can render the agency action arbitrary and capricious. See Wisconsin Valley Improvement v. FERC, 236 F.3d 738, 748 (D.C. Cir. 2001) (“[A]n agency acts arbitrarily and capriciously when it abruptly departs from a position it previously held without satisfactorily explaining its reason for doing so.”). USDA’s reliance on States’ purported abuse of the waiver process to restrict waiver areas to LMAs fails to meet this standard for the same reason that the agency failed to provide adequate notice, with underlying and specific data, for meaningful comment. See supra Part III.A.3. Specifically, the agency neither grappled with nor sufficiently analyzed reasons provided in the administrative record for States’ grouping practices. For example, the Final Rule asserts that “flexibility allows States to strategically group substate areas to maximize the geographic coverage of waived areas rather than to demonstrate high unemployment or a lack of sufficient jobs for ABAWDs,” 84 Fed. Reg. at 66794, and cited this flexibility as a “problem” and “one of the primary reasons why about half of the ABAWDs participating in SNAP live in — 36 — waived areas, despite current low unemployment levels across the majority of the country,” id. Citing national standard unemployment rates, however, does not address the sufficiency of jobs for ABAWDs in the waived areas and therefore does not show that States’ grouping practices produced any outcomes inconsistent with the statute. See State Pls. MSJ at 32. Indeed, commenters provided ample alternative explanations for why states would group towns other than by LMA, including “cost of living, lack of access to or availability of transportation, or lack of employers with a certain job field.” ABAWD00008242 (U.S. DEP’T AGRIC., FOOD & NUTRITION SERV., FNS-2018-0004, SUPPLEMENTAL NUTRITION ASSISTANCE PROGRAM: REQUIREMENTS FOR ABLE-BODIED ADULTS WITHOUT DEPENDENTS, SUMMARY OF PUBLIC COMMENTS (2019) (“Summary Comment”)), ECF No. 105-2. Without engaging these comments, the Final Rule merely “disagree[d]” with comments supporting States’ flexibility to group substate areas. 84 Fed. Reg. at 66794. USDA takes issue with several points made on this issue in D.C.I, but none of the agency’s arguments holds up. First, D.C. I found the agency’s assertion that “half” of ABAWDs live in waived areas to be “unsurprising,” since the waived areas are supposed to be targeted to areas where ABAWDs actually live and cannot find sufficient jobs. D.C. I, 444 F. Supp. 3d at 28. Although the agency stated that “half” was a “suspiciously high share of ABAWDs . . . living in waived areas,” the insinuation that this exemplified an abuse of the waiver process was simply unsupported. Id. On this point, USDA points to evidence “confirm[ing] that strategic grouping was a real problem,” Defs.’ Opp’n at 51, citing (1) the agency’s 2016 OIG Report findings that States had “specifically requested ABAWD time limit waivers in as many parts of the State as possible to minimize the areas where they needed to track the ABAWD time limits,” and that waivers covered even “parts of the States where unemployment rates [were] as low as 0 — 37 — percent,” id. (citing ABAWD00000289 (U.S. DEP’T AGRIC., OFFICE OF INSPECTOR GEN., Audit Report 27601-0002-31, FNS CONTROLS OVER SNAP BENEFITS FOR ABLE-BODIED ADULTS WITHOUT DEPENDENTS, (2016) (“OIG Report”)), ECF No. 105-1); and (2) individual waiver requests from States that grouped the majority of counties in the State to reach the 120% national unemployment rate threshold, id. 52–54 (citing ABAWD00002477–78 (Georgia FS ABAWD Waiver Request for FY2016), ECF No 105-2; ABAWD00003536 (California FS ABAWD Waiver Request for FY2018), ECF No. 105-2). As noted in D.C. I, the OIG Report findings were not recounted or even referenced as justification for the Final Rule, and thus are not properly considered. 18 D.C. I, 444 F. Supp. 3d at 28 n.16. Further, examples of “gerrymander[ed]” waiver areas neither explain why States’ efforts to “maximize” waivers are impermissible, nor whether or how the State grouping practices have provided benefits to ABAWDs whom the statute is not meant to cover. See State Pls. MSJ at 32; see also D.C. I, 444 F. Supp. 3d at 28 n.16 (finding that the OIG report was “consistent with the idea that the state grouping behavior USDA deemed manipulative is a feature of the statutory design rather than a product of state abuse.”). Indeed, given USDA’s avowed “careful[] review[]” of “all State waiver requests” by “independently obtaining and validating the data and evidence…to — 18 — USDA vigorously argues that the OIG Report should be considered as an “amplified explanation for the reasons [the agency] offered at the time of the action,” rather than presenting a “new basis for its action.” Defs.’ MSJ at 51 n.23 (citing Nat’l Oilseed Processors Ass’n v. Browner, 924 F. Supp. 1193, 1204 (D.D.C. 1996)). If that were true, the omission of any reference to this OIG Report in the Proposed and Final Rules is even more puzzling. In any event, the agency may only present these findings as a “clearer or more detailed explanation” of the justification offered in the Final Rule if it offered any explanation to speak of. Nat’l Oilseed Processors Ass’n, 924 F. Supp. at 1204 . Here, the agency offered only conclusory justifications based on vague claims of “operational experience” and “problems” of state abuse, while failing altogether to refute the various, good faith explanations for state waiver grouping. An agency explanation not described in any meaningful way in a Final Rule cannot be amplified, after the fact, for the purposes of litigation. See D.C. I, 444 F. Supp. 3d at 28 n.16 (citing SEC v. Chenery Corp., 318 U.S. 80 , 94–95 (1943)). — 38 — determine if the areas meet the standards for approval,” 84 Fed. Reg. at 66800, the agency’s current critique of agency-vetted and approved waivers as somehow improper is strained. Second, D.C. I determined that the agency’s new policy eliminating States’ flexibility in determining waiver areas was “a solution in search of a problem,” taking note that the agency could simply take the “sensible course” of denying or requesting review of inappropriate waiver applications. D.C. I, 444 F. Supp. 3d at 31. 19 Plaintiffs observe that no evidence has been presented that USDA took steps to address any “state abuse,” either “by notifying the state in question, seeking a modified waiver application, or denying an application in whole or in part.” State Pls. MSJ at 32. The Final Rule itself touts the thoroughness of agency review of waiver requests, with in-depth analysis of the data presented, identification of “discrepancies or inaccuracies,” and denial in whole or in part of “the waiver request because not all areas meet the standards for approval.” 84 Fed. Reg. at 66800. USDA now tries to walk away from the thoroughness of its waiver review process, saying that the agency “simply could not reject waivers that . . . grouped counties separated by hundreds of miles into a single area or conspicuously omitted nearby jurisdictions whose inclusion would mean the request no longer met the 20% standard,” when such grouping was authorized under the 2001 regulation. Defs.’ Opp’n at 55. This is a punt. As USDA concedes, the Secretary has broad discretion over the determination of whether the waiver request meets the statutory qualification under 7 U.S.C. § 2015 (o)(4)(A)(i) or (ii), see Defs.’ MSJ at 18–19, and if, as USDA alleges, States are seeking — 19 — The Final Rule mentioned the agency’s effort in 2016 “to clarify its intention that areas be economically tied through policy guidance,” D.C. I, 444 F. Supp. 3d at 31 n.18 (citing 84 Fed. Reg. at 66794 & n.8), but did not address why that guidance failed to deter any purportedly improper grouping to maximize waiver areas and the agency “did not then begin denying or questioning waiver areas that it viewed as inconsistent with this guidance,” id. — 39 — waivers for areas with sufficient jobs for ABAWDs, the Secretary may deny the request, a power the Final Rule says has been exercised, 84 Fed. Reg. at 66800. Further, USDA’s expressed concern about State manipulation of large waiver areas, Defs.’ MSJ at 51–54, is difficult to reconcile with the agency’s new commitment to LMAs, which multiple commenters claim are too large to be appropriate for a measure of ABAWDs job opportunities, State Pls.’ MSJ at 34; see ABAWD00008242 (Summary Comment (“[L]ocal economic conditions may not be captured by BLS data”), ECF No. 105-2). This concern also ignores the fact that the number of statewide waivers has already sharply dropped in the five years between 2013 and 2018, with States opting for a more targeted substate approach, suggesting that rather than use statewide unemployment averages, which, in the agency’s view, “may mask ‘slack’ job markets (insufficient jobs) in some substate areas,” and also “tight labor markets in some substate areas,” 84 Fed. Reg. at 66790, States were relying on available data sources to identify those areas of the State with insufficient jobs for ABAWDS. In sum, the Final Rule provided insufficient evidence and inadequate explanation that States were abusing their authority by requesting and obtaining time limit waivers in areas where sufficient jobs actually existed for ABAWDs to support the severe restriction on State flexibility in delineating waiver areas imposed by limiting such areas to LMAs. Third, D.C. I was unconvinced by USDA’s alternative justification to State abuse for adopting LMA as a “strict definition of waiver ‘area.’” Id. at 66795. Specifically, the Final Rule adopted LMAs because they are “economically integrated area[s] within which individuals can reside and find employment within a reasonable distance or can readily change jobs without changing their place of residence,” id. at 66793, citing the usefulness of LMAs in delineating “commuting patterns of the general workforce,” id. (emphasis supplied). D.C. I found a — 40 — fundamental “mismatch between available employment for ABAWDs and available employment across an LMA,” D.C. I, 444 F. Supp. at 31, since LMAs do not capture “available employment for ‘low-income, low-skilled ABAWDs who lack affordable transportation options,’” id. (citing 84 Fed. Reg. at 66793), rendering LMAs an unsuitable—and therefore unreasonable—measure of sufficient jobs where ABAWDs reside. One example of the insufficiency of using LMAs for the express statutory purpose is in large urban areas, such as New York City and the District of Columbia, where LMAs include multiple adjacent counties that extend well into areas of bordering states that are inaccessible to low-income ABAWDs for work. Id. at 31–32 (observing that the new LMA definition of “area” would subject ABAWDs residing in the District to “the work requirement if the unemployment rate is low in the suburbs and exurbs of the District as far out as West Virginia, although these out-of-state counties are inaccessible by public transportation from the District and although unemployment rates in parts of the District are as high as 11.6%”). Despite “considerable evidence before [the agency]” on this point, the agency “ignored critical factors” and merely “nodd[ed] to concerns raised by commenters only to dismiss them in a conclusory manner,” which is “not a hallmark of reasoned decisionmaking.” Id. at 31–32 (quoting Gresham, 950 F.3d at 103 ). The full administrative record supports the conclusion in D.C. I that the agency failed to contend with the evidence presented during the rulemaking that the LMA definition is too attenuated from ABAWDs’ residential, transportation, and job patterns to inform effectively the statutorily-mandated inquiry into whether there are sufficient jobs for ABAWDs in “the area[s] in which [they] reside.” 7 U.S.C § 2015(o)(4)(A). Numerous comments before the agency explained and provided research that the use of LMAs for waiver areas would fail to incorporate “local nuances” in work opportunities, see ABAWD00008242 (Summary Comment), ECF No. — 41 — 105-2, and “hamper states’ ability to target waivers to specific areas,” State Pls.’ MSJ at 34 (citing ABAWD00008239-40 (Summary Comment), ECF No. 105-2), remarking on the significant issues that arise when using LMAs to assess waivers in both rural and urban parts of a State, see ABAWD00008239, -8243 (Summary Comment), ECF No. 105-2. The Final Rule summarizes comments expressing these facts. See, e.g., 84 Fed. Reg. at 66793 (noting “examples in which some LMAs are too big to properly define commuting patterns for ABAWDs because it could take more than two hours without traffic to commute one way from one end of an LMA to another by car and [additional] examples where it is impossible to access most of communities within an LMA using public transportation.”); id. (acknowledging that “in some counties, workers may have to travel in all directions and often beyond a contiguous county for their job, and, therefore LMAs are too small in some cases” and citing “research…[that] poor, minority residents [are] seeing the biggest decline in jobs within a reasonable commuting distance.”). Yet, USDA made no effort to rebut these comments nor the underlying research cited in the comments. Instead, the agency offered only the blunt conclusory statement that, “[t]he Department is not compelled by the commenters’ suggestions . . . [and] as there are no Federally-designated areas that specifically assess commuting patterns and other related economic factors for ABAWDs, . . . the Department maintains that [LMAs] are the best available and most appropriate area delineation at this time.” Id. USDA concedes, as it must in the face of the many comments and research cited in the Final Rule, that LMAs are not “perfectly aligned with job markets solely for ABAWDs.” Defs.’ Opp’n at 33; see also id. at 57 (recognizing that LMA “delineation is not a perfect fit for ABAWDs”). Absent “anything closer to an administrable definition of an ABAWD-labor market area,” id. at 33, however, the agency contends that use of LMAs is “particularly — 42 — reasonable” compared to the alternative of “a civil jurisdiction . . . based on a political boundary,” id.; see also id. at 56 (asserting that “LMA definition was motivated by the need to align waiver areas with job markets…unlike individual jurisdictions, which turn on invisible boundaries . . . .”). In this “line-drawing” exercise, id. at 57, USDA argues that the agency has been tasked with broad discretionary authority to determine “how best to strike the balance inherent in defining a waiver area,” id. at 58; see also 84 Fed. Reg. at 66794 (asserting that USDA is “within its authority to revise its regulations [since] the statute does not define what constitutes an ‘area’”), concluding that “‘in light of the alternatives, the agency construction was reasonable,’” Defs.’ MSJ at 58 (quoting Bernhart v. Thomas, 540 U.S. 20, 21 (2003)). This convoluted response argument needs unpacking, because the inferential leaps tying it together are flawed. At the outset, broad discretion to define terms does not give an agency carte blanche to adopt new standards “unsupported by the record before the agency and by the agency’s explanation.” D.C. I, 444 F. Supp. 3d at 32. The agency’s choice to use only LMAs meant disregarding political boundaries, including State geographic limits, even though State agencies are tasked statutorily with making waiver requests for ABAWDs only within the State, as well as providing those ABAWDs with opportunities to transition to jobs, creating an inherent administrative tension. See, e.g., ABAWD00168348 (Food Research & Action Center, Comment Letter on Proposed Rule: SNAP – Requirements for Able-Bodied Adults without Dependents RIN 0584-AE57 (2019) (noting “the proposed rule would make it more difficult for states and counties to coordinate effective employment and training efforts, including for areas covered by Workforce Development Boards that are inconsistent with the BLS labor market area definition”)), ECF No. 105-4. This tension is only exacerbated by the new restriction on data sources for waiver approvals imposed by the Final Rule. See infra Part III.B.2. Ultimately, — 43 — USDA’s argument boils down to LMAs being the best measure of job markets, even if not precisely measuring the job market for ABAWDs. To reach that conclusion, however, the Final Rule had to examine the criticisms about, and alternatives to, LMAs, and this is exactly where the Final Rule falls far short. As to whether LMAs are actually the best measure of job markets for ABAWDs, many commenters said otherwise and pointed to States as being in the best position to assess such job opportunities. The Final Rule, again, summarizes the many commenters pointing out that “each State has the contextual knowledge and experience to identify the most appropriate grouping areas for waiver,” 84 Fed. Reg. 66793 -94; that the agency’s prior findings “that county unemployment rates were the most available measure of the vitality of local labor markets” and “that States were best-equipped to determine whether high unemployment in some areas adversely affect employment prospects in others,” id. at 66794; and that over the past two decades, the agency “has never expressed that commuting patterns [should] be the primary or only basis for whether or not substate areas could be grouped together,” id. Again, USDA did not rebut, refute or even debate these points about States’ expertise in identifying waiver areas. Rather, the agency merely stated “the Department disagrees,” id., citing “its extensive operational experience,” to conclude that strategic grouping of substate areas was a problem of State flexibility addressed by the LMA definition, id. The agency highlighted as examples of such problematic substate grouping “to maximize [the] waived areas in the State,” id.—without specific details that could be subjected to challenge or rebuttal—that “some States have grouped nearly all contiguous counties in the State together while omitting a few counties with relatively low unemployment” and others “have grouped certain towns together that share the same economic region while omitting others with relatively low unemployment from the group,” id. — 44 — These examples gloss over, without addressing, the “numerous reasons” commenters suggested for such grouping, “such as cost of living, lack of access to or availability of transportation, lack of employers with a certain job field, or other demographic considerations,” id.; the fact, previously accepted by USDA, that “patterns of employment and mobility for the low-skilled employment market can be quite different from those for the overall employment market,” id.; and “economic ties” in grouped areas, “such as employer recruiting practices, regional workforce development strategies, regional economic development and investment patterns, service delivery models and migration patterns,” id. at 66795. In other words, the agency rejected its long-standing policy of States’ flexibility in delineating the areas for which a State believes a waiver request is warranted with only lip service acknowledgement of the problems with the LMA alternative adopted, and without any consideration of the complex factors underlying the substate groupings the agency seized on as problems. USDA’s treatment of even other problems commenters pointed out with restricting waiver areas to LMAs, including reliance on “outdated data,” infrequent updating “to capture recent labor market trends,” cautions from OMB about use of LMA delineations “to develop and implement Federal, State, and local non-statistical programs and policies,” id., was similarly dismissive. In response to these additional problems with LMAs, USDA stated that “[t]he Department is resolute” that “LMAs represent the best available and most practical solution,” to “the problem of States’ manipulative usage of grouping substate areas to maximize waived areas.” Id. at 66795. Thus, the agency’s reasoning may be summed up as follows: LMAs do not measure what the statute requires in terms of job opportunities for ABAWDs, but this metric at least addresses an issue of substate grouping that the agency holds up as a problem without addressing any of the cited complex factors showing this may be no problem at all. — 45 — As determined in D.C. I, in defending the reasonableness of the Final Rule’s adoption of LMAs, USDA did little more than state that the agency was rejecting critical comments about problems with this choice as a definition for “areas.” See D.C. I, 444 F. Supp. 3d at 32. As such, the agency’s failure to contend meaningfully with the valid concerns and criticisms raised by commenters renders the waiver “area” redefinition to LMA arbitrary. 2. Final Rule’s Adoption of Standard Unemployment Rate Criteria as Sole Basis for Time Limit Waiver Is Arbitrary and Capricious Recall, as described, supra, in Part I.A.1, that waivers of the time limits on ABAWDs’ eligibility for SNAP benefits are statutorily authorized in the “area in which” ABAWDs reside if the area either “(i) has an unemployment rate of over 10 percent; or (ii) does not have a sufficient number of jobs to provide employment for” ABAWDs. 7 U.S.C. § 2015 (o)(4)(A). The Final Rule eliminates the flexibility States have been granted for over two decades to support waiver requests with “whatever data [they] deem appropriate,” 7 C.F.R. § 273.24 (f)(2) (2001), and restricts States to using only the standard U-3 unemployment rate generated by BLS for both prongs (i) and (ii), see supra note 7. The U-3 unemployment rate is one of six measures of labor underutilization published by BLS and represents the total number of unemployed persons available to take a job and actively seeking a job in the past four weeks, and, thus, unlike other BLS unemployment rates, does not include so-called “discouraged workers,” who looked for a job in the past twelve months, or involuntary part-time workers, who would like to work full time. 84 Fed. Reg. at 66788 & n.5. The Final Rule would implement prong (ii) by requiring States requesting waivers under this provision to show the LMA at issue has a recent 24-month average unemployment rate 20 percent of more above the national rate but no less than 6 percent. Id. at 66811. This Rule, had it been in effect during the COVID-19 pandemic, “would have prevented almost every area throughout the country from qualifying. As a result, despite record — 46 — unemployment, the final rule would offer almost no relief to unemployed workers and overwhelmed state benefit systems and local food banks.” Bolen Decl. at ¶ 8. In making this significant policy change to rely solely on the U-3 standard unemployment rate for a 24-month period, USDA concedes that this new required measure fails to gauge what prong (ii) of the statute calls for, namely: whether an area has “a sufficient number of jobs to provide employment for” ABAWDs. As USDA points out, no available BLS statistical data provides a measure tailored to this statutory requirement. See 84 Fed. Reg. at 66787 (“recogniz[ing] that there is no measure available for determining the number of available jobs specifically for ABADS participating in SNAP in any given area.”). Furthermore, despite adopting this standard unemployment rate as the measure of the availability of jobs in an area where ABAWDs reside, USDA also concedes that ABAWDs face difficulties in finding jobs different than the general population. Id. (“recogniz[ing] that ABAWDs may face barriers to employment and have more limited employment prospects than the general public due to low educational attainment or other factors . . . .”). Notwithstanding the conceded misfit between the standard unemployment rate and the statutory waiver provision in prong (ii), however, the Final Rule eliminates States’ flexibility in garnering relevant evidence to meet this statutorily authorized basis for waiver by making the sole data source for such waiver requests the 24-month average U-3 standard unemployment rate. Id. at 66785, 66790, 66811. Plaintiffs challenge this change in implementation of prong (ii) as arbitrary and capricious. State Pls.’ MSJ at 24–28; Private Pls.’ MSJ at 24–32. The Court agrees the Final Rule’s adoption of standard unemployment rates as the measure for prong (ii) is flawed because, first, this change conflates and obstructs the purpose of this statutory provision — 47 — without adequate justification and, second, fails adequately to address the significant critiques and alternatives to the option adopted. First, to the extent general unemployment data is relevant to time limit waivers, that kind of evidence is expressly cited in prong (i), which provides that waiver requests may be based on “an unemployment rate over 10 percent.” 7 U.S.C. § 2015 (o)(4)(A). By contrast, prong (ii) requires more nuanced and targeted focus on job opportunities for ABAWDs in the area where ABAWDs reside. If Congress intended for the agency to use only general unemployment rates to determine whether to grant waivers, either prong (ii) would have been unnecessary or would have mirrored the language from prong (i). See Dep’t of Homeland Sec. v. MacLean, 574 U.S. 383 , 391 (2015) (“Congress generally acts intentionally when it uses particular language in one section of a statute but omits it in another.”); Office of Consumers’ Counsel v. FERC, 783 F.2d 206 , 222 (D.C. Cir. 1986) (explaining that requiring a term in the statute to encompass a definition that is “flatly inconsistent with Congress’ clear message on [a] precise issue” is impermissible). The Final Rule’s reliance on general unemployment data for both prongs of § 2015(o)(4)(A) largely conflates the two statutory provisions, forces States to rely on the same kind of data, subject to slightly different parameters, for waiver requests under both prongs and functionally forecloses the specific focus on ABAWD job opportunities that is called for under prong (ii). 20 The Final Rule never adequately explains why the changes made in prong (ii) — 20 — State plaintiffs press their argument that the Final Rule’s singular reliance on unemployment rates flunks the Chevron step one test, State Pls. MSJ at 13–15, because eliding the distinction in the statute’s two-pronged structure disqualifies this agency action. See New York v. EPA, 443 F.3d at 887 (setting aside an agency action where the action rendered the words in the statute “insignificant” or “superfluous”); Shays v. FEC, 414 F.3d 76, 105 (D.C. Cir. 2005) (vacating agency rules at Chevron step one because the agency’s “narrow interpretation of” a statutory term was “implausible.”). As explained in D.C. I, 444 F. Supp. 3d at 24 n.13, however, the broad discretion granted to the USDA Secretary to make the requisite determination under § 2015(o)(4)(A)(ii), compounded by the broad scope of what data may—or may not—be covered in prong (ii), means both that this language is “not obviously unambiguous at Chevron Step one,” id., and that the agency’s interpretive overlay relying solely on general unemployment rates as the measure in prong (ii) is not barred at Chevron Step two. — 48 — furthers this statutory directive in a manner different than prong (i), nor even appears to acknowledge this problem. Further, as explained in D.C. I, by adopting “unreasonably narrow criteria for implementing the statutory requirements for evaluating waivers,” D.C. I, 444 F. Supp. 3d at 25, the Final Rule frustrates the statute’s tacit instruction to the agency to calculate “sufficient jobs” with “some attention . . . to ABAWD-related measures,” by making the sole criteria one that is incapable of measuring what the statute requires, id. No more than a passing relationship exists between the 24-month unemployment rate and ABAWD job availability and USDA does not claim otherwise, asserting only that “the Department finds the 20 percent [general unemployment rate] standard . . . to be one of the most objective and defensible ways of determining a lack of sufficient jobs,” without claiming this measures a lack of sufficient jobs for ABAWDs. 84 Fed. Reg. at 66787. Indeed, the statutory phrase “any group of individuals in the State” for which waivers may be requested signals that such requests may be granted when the area in the State where ABAWDs reside lacks jobs “for the individuals,” 7 U.S.C. § 2015 (o)(4)(A)(ii), meaning ABAWDs. While U-3 standard unemployment data may show the percentage of jobless persons looking for work the past month, this measure misses the requisite focus on ABAWDs, who face labor market disadvantages that the general public does not, including low levels of education, part-time work, irregular hours, seasonal work, underemployment, high turnover, homelessness, former incarceration, racial and ethnic discrimination, and therefore have “significantly higher” unemployment rates than that reflected in the general unemployment rate. 84 Fed. Reg. at 66787. Consequently, plaintiffs’ assert, this Nonetheless, as discussed in the text, USDA’s rationales for adoption of this policy must still be adequately explained and in furtherance of the statute, and the agency fails to do this under an arbitrary and capricious analysis. — 49 — “concededly imperfect measure” would “prevent states from securing waivers for areas that lack sufficient jobs for ABAWDs,” contrary to the intent of the statute. State Pls.’ MSJ at 25. Despite this mismatch between the standard unemployment rate and measuring job opportunities for ABAWDs, USDA’s explanation for eliminating the multi-criteria waiver basis set out in 7 C.F.R. § 273.24 (f)(2)(ii) (2001), repeatedly cites the preferred reliability and consistency of “standard unemployment data in demonstrating a lack of sufficient jobs.” 84 Fed. Reg. at 66791. This explanation simply avoids grappling with the fact that this data source is not more reliable and consistent for showing the lack of sufficient jobs for ABAWDs, as the statute directs. Rather than address the multiple comments that said just this, see e.g., id. (“commenters assert, as previously noted, that unemployment rates do not precisely capture job availability for ABAWDs”); id. (commenters noted “that not including these criteria . . . would undercut a more nuanced understanding of local job markets”), USDA simply rejects these comments as “not sufficiently compelling to justify making changes to the proposed rule.” Id. In doing so, the agency expresses the suspicion that “areas that do not clearly lack sufficient jobs [are] qualify[ing] for waivers,” id. at 66787, noting that “a state could . . . qualify for a waiver in areas with an unemployment rate as low as 4.7 percent,” id. Thus, the agency’s over-arching intent in dramatically restricting waiver criteria was to address the suspected misuse of waivers, not to identify areas with insufficient jobs for ABAWDs, which at least partially explains why the new rule adopts a measure inherently ill-suited for the statutory purpose and the agency made little effort to explain how the general unemployment rate was supposed to serve the specific statutory directive of measuring job opportunities for a particular group of ABAWDs. Second, and relatedly, as explained in D.C. I, USDA’s jettisoning of two decades of reliance on a multiple criteria to measure job opportunities for ABAWDS, in favor of singular — 50 — reliance on the standard unemployment rate in both prongs of § 2015(o)(4)(A), was arbitrary because the agency failed adequately to respond to commenter concerns and data showing that the general unemployment rate was so general as to be an “inappropriate measure of lack of sufficient jobs for ABAWDs.” D.C. I, 444 F. Supp. 3d at 23; see also State Pls. MSJ at 24–25 (contending that rule change restricting waiver criteria to 24-month average standard unemployment rate, combined with an unemployment rate floor, is “contradicted by both the agency’s prior factual findings and evidence provided by commenters that unemployment levels alone are not an adequate measure of sufficient jobs for ABAWDs.”). USDA has long deemed the general unemployment rate alone as inadequate for measuring ABAWD employment opportunities, stating twenty-five years ago that “[t]he statute recognizes that the unemployment rate alone is an imperfect measure of the employment prospects of individuals with little work history and diminished opportunities.” ABAWD00000166 (1996 Guidance), ECF No. 105-1. The agency has also consistently maintained, and reiterated, in 1999, 2001, 2006, and 2016 the position that there are no singular “standard data or methods to make the determination of the sufficiency of jobs” for ABAWDS. ABAWD00000090 (64 Fed. Reg. at 70946 (conceding that there are “no standard data or methods to make the determination of the sufficiency of jobs”)), ECF No. 105-1; ABAWD00000212 (2006 Guidance (maintaining position)), ECF No. 105-1; ABAWD00000330–31 (U.S. DEP’T AGRIC., SUPPORTING REQUESTS TO WAIVE THE TIME LIMIT FOR ABLE-BODIED ADULTS WITHOUT DEPENDENTS (2016) (maintaining position)), ECF No. 1051. Given the conceded lack of available BLS or other precise measures of job opportunities for ABAWDs in the areas where they reside, USDA policy and regulations have long permitted — 51 — States to use different kinds of evidence to inform this query with a focus, as the statutory directive requires, on the availability of jobs for ABAWDs. USDA concedes that these different kinds of evidence authorized in the 2001 regulation, such as “low and declining employment-to-population ratio, a lack of jobs in declining occupations or industries, or an academic study or other publication(s)[,] can enhance the understanding of the job market.” 84 Fed. Reg. at 66791. In fact, the Final Rule continues to include these otherwise eliminated criteria as “appropriate alternative measures when standard unemployment data is not available for an area,” id., such as a reservation or U.S. Territory, id. at 66811. Not surprisingly, given the reliance among States and other stakeholders on this longstanding policy, many of the “over 100,000” comments before the agency, id. at 66787, opposed eliminating the multiple criteria basis to implement § 2015(o)(4)(A)(ii), and supported retention as data sources employment-to-population ratios, see, e.g., ABAWD00008237 (Summary Comment), ECF No. 105-1, information about declining industries or occupations, see, e.g., ABAWD00008238 (Summary Comment), ECF No. 105-2, and academic studies and LSAs, see 84 Fed. Reg. at 66791. The Final Rule summarily dismissed such data sources as “less reliable and consistent than standard unemployment data” based on the agency’s “operational experience,” id., or “inconsistent with the final rule’s definition of an area,” id. at 66800; see id. at 66791 (adopting “a set of consistent criteria for [waiver] approval based on reliable and robust available evidence for evaluating labor market conditions”). While BLS’ U-3 standard unemployment rate may be reliable and consistent for some purposes, extensive comments and research data before the agency demonstrated the distinct demographic factors unique to ABAWDs that make a general unemployment measure — 52 — inadequate for identifying sufficient jobs under the statute, as the agency’s long-standing position had previously acknowledged. ABAWD unemployment is significantly higher than that of the general population, particularly during economically depressed periods because the ABAWD population intersects with a variety of socio-economic factors that make finding employment especially difficult. Priv. Pls. MSJ at 26–30. For example, 75 percent of SNAP beneficiaries have only a high school education or less, making them four to six percent more likely to be unemployed than those with bachelor’s degrees, see ABAWD00110132-33 (CBPP, Comment Letter on Proposed Rule: SNAP – Requirements for Able-Bodied Adults without Dependents RIN 0584-AE57 (Apr. 1, 2019) (“CBPP Comment”)), ECF No. 105-3; 40 percent of SNAP participants are African American or Latino, which are populations that had unemployment rates averaging between two and three percent above the overall unemployment rate for white workers in 2018, ABAWD00110136 (CBPP Comment), ECF No. 105-3; see, e.g., Priv. Pls.’ MSJ at 28 (explaining that the unemployment rate for African Americans in the District of Columbia is nearly 8.5 times higher than that of white residents); and 20 percent of ABAWDs reported a “health problem or disability that prevents them from working or limits the type of work they can do,” ABAWD00110142 (CBPP Comment), ECF No. 105-3. In addition, some of the most commonly reported occupations for ABAWDs are in sales or service, which are sectors that report unemployment rates between 23 percent and 56 percent higher than the general rate. ABAWD00110139 (CBPP Comment), ECF No. 105-3. The inadequacies of the general unemployment rate as a tool for assessing sufficient jobs for ABAWDs were highlighted by comments comparing other data sources and criteria, which incorporated factors pertinent to ABAWDs: BLS’s U-6 unemployment rate, low and declining employment-to-population ratio, and the State’s qualification for extended unemployment — 53 — benefits. Priv. Pls.’ MSJ at 31 – 32; State Pls.’ Reply at 15. For example, commenters urged consideration of the U-6 unemployment rate, which more broadly than the U-3 standard unemployment rate, includes the unemployed, those “marginally attached to the labor force,” who have been looking for work for longer than four weeks, and involuntary part-time workers. ABAWD00110183-85 (CBPP Comment), ECF No. 105-3; see ABAWD00034696 (Harry J. Holzer, Comment Letter on Trump Administration Proposal to Limit SNAP Waivers for the ABAWD Population (Mar. 2019)), ECF No. 105-3; Priv. Pls. MSJ at 31. Commenters explained that the U-6 rate would “more accurately capture[] the condition of the labor market for ABAWDs,” 84 Fed. Reg. at 66789, and that the “the employment-to-population ratio, along with BLS’ U-6 measurement, captures valuable information about adults who are not actively looking for work,” ABAWD00008237 – 38 (Summary Comment), ECF No. 105-2. The Final Rule rejected both measures solely “because [they are] not available at the substate level,” 84 Fed. Reg. at 66789–90, the same reason the USDA abruptly eliminated, without appropriate notice, see supra Part III.A.1, a State’s qualification for extended unemployment benefits to support a waiver, 84 Fed. Reg. at 66789–90 (eliminating the criterion because “qualification for extended unemployment benefits is designated only at the state level, not at the LMA level”). Elimination of the extended unemployment benefits criterion is particularly troubling since USDA specifically noted that this qualification “has been a clear indicator of lack of sufficient jobs and an especially responsive indicator of sudden economic downturns, such as the Great Recession,” 84 Fed. Reg. at 985, and an “appropriate indicator that a State lacks sufficient jobs,” 84 Fed. Reg. at 66789. These comments put into stark relief the fact that more suitable measures are available for the ABAWD population, who, due to their distinguishing attributes, are likely excluded from the general unemployment rate applicable to the conventional civilian labor force. — 54 — The agency’s rejection of these more appropriate measures due solely to the simultaneous adoption in the Final Rule of LMAs as the only definition of waiver area, and concomitant elimination of Statewide waivers, only bolsters the conclusion that this agency action is arbitrary and capricious. Such radical changes in long-standing policies, on which States, their agencies and others have long relied for such a critical purpose as necessary nutritional assistance, requires that the agency adequately consider “the ‘alternative[s]’ that are ‘within the ambit of the existing [policy].’” Regents, 140 S. Ct. at 1913 (quoting State Farm, 463 U.S. at 51 ). USDA’s explanation that more appropriate measures of ABAWD job opportunities than the measure adopted were rejected due to their mismatch with the new LMA waiver area definition, does not provide a sufficiently “reasoned explanation,” Fox, 556 U.S. at 515 , why carve-outs or exceptions to the LMA waiver area were impossible or otherwise unacceptable. In failing to meaningfully contend with submitted comments, the agency did not “consider an important aspect of the problem,” producing a decision that “runs counter to the evidence” the agency received, which renders it arbitrary. State Farm, 463 U.S. at 43 . C. The Final Rule’s Discretionary Exemption Change is Contrary to Law State agencies may use discretionary exemptions, authorized under 7 U.S.C. § 2015 (o)(6), to provide SNAP benefits to ABAWDs, who do not reside in a waiver area and are otherwise subject to time limits. Recall that paragraph (6) of § 2015(o) sets out directions to USDA’s Secretary on how to calculate the number of exemptions each State is awarded annually. See supra in Part I.A.2. As a general matter, the number of discretionary exemptions available for use by each State is a percentage—currently 12 percent, under subparagraph (E)— of the number of “covered individuals in the State,” subject to various adjustments by the Secretary. 7 U.S.C. § 2015 (o)(6)(E) (making the “number of exemptions” that “a State agency may provide” “[s]ubject to subparagraphs (F) through (H)”). — 55 — As such, the “12 percent of the covered individuals in the State” provides the baseline number of exemptions available, which number is then adjusted for the total number of discretionary exemptions allocated or available for the State’s use. Id. One such adjustment by the Secretary is “to reflect changes in the State’s caseload,” id., and is guided by subparagraph (F), titled “Caseload adjustments.” The adjustment at issue here is set out in subparagraph (G), titled “Exemption adjustments,” and directs that in “each” fiscal year— the Secretary shall increase or decrease the number of individuals who may be granted an exemption by a State agency under this paragraph to the extent that the average monthly number of exemptions in effect in the State for the preceding fiscal year under this paragraph is lesser or greater than the average monthly number of exemptions estimated for the State agency for such preceding fiscal year under this paragraph. 7 U.S.C. § 2015 (o)(6)(G) (emphasis supplied). As described, supra in Part I.A.2, the Final Rule changed USDA’s long-standing policy that States were entitled, under subparagraph (G), to carryover from year to year, the “number of exemptions that had been estimated for the State agency,” meaning allocated by operation of the other subparagraphs, but unused, and adopted a “use it or lose it” policy for such exemptions. 84 Fed. Reg. at 66802–03. Specifically, the Final Rule now allows carryover of only “a portion of the unused exemptions not to exceed 12 percent of the covered individuals in the State estimated by [USDA] for the preceding fiscal year.” 84 Fed. Reg. at 66811. Plaintiffs argue that USDA’s elimination of discretionary exemptions accumulated over the last twenty years and the new limitation on carryover exemptions beyond one year contravene the statutory text and Congress’ “clearly articulated intent.” State Pls.’ MSJ at 18; State Pls.’ Reply at 18. D.C. I declined to temporarily enjoin the effectiveness of this part of the Final Rule at the preliminary injunction stage of proceedings, 444 F. Supp. 3d at 16-19, but, based on plaintiffs’ new interrogation of the — 56 — dense statutory language and additional briefing at summary judgment, the contrary to law challenge to this aspect of the Final Rule now persuades. Plaintiffs read the distinct steps for computing discretionary exemptions under § 2015(o)(6)(G) as requiring, first, a comparison of “the average monthly number of exemptions in effect,” meaning used, during the preceding fiscal year with such “number of exemptions estimated for the State agency for such preceding fiscal year under this paragraph.” State Pls. MSJ at 18; see also State Pls.’ Reply at 19 (“Nobody disputes that Y [the number of exemptions in effect in the State for the preceding fiscal year] is the number of exemptions the state actually used last year.”). Second, if the number “in effect” (or used) is “less[]… than” the number “estimated,” subparagraph (G) directs that the Secretary “‘shall increase’ the state’s exemptions for the present fiscal year by the difference.” Id. (quoting 7 U.S.C. § 2015 (o)(6)(G)). Plaintiffs posit that “the only question is whether the number ‘estimated…under this paragraph’ includes the number of carryover exemptions that also were available in the preceding fiscal year,” id. (emphasis in original), and contend “[t]he statute’s plain language demonstrates that it does,” id. The last phrase in subparagraph (G) expressly states that the discretionary exemption adjustment is increased (or decreased) by the “exemptions estimated for the State agency for such preceding fiscal year under this paragraph,” thereby “includ[ing] exemptions carried into the preceding fiscal year from the prior year.” State Pls.’ Reply at 19. As plaintiffs point out, agency guidance and rules over the last two decades have, until the challenged Final Rule, reflected this reading of § 2015(o)(6)(G). USDA’s 2002 Rule explained the operation of paragraph (6)(G), stating “[i]f the State agency does not use all of its exemptions by the end of the fiscal year, [USDA] will increase the estimated number of exemptions allocated to the State agency for that subsequent fiscal year by the remaining — 57 — balance.” 2002 Final Rule, 67 Fed. Reg. 41619 . Conversely, “[i]f more exemptions are used than authorized in a fiscal year, the State’s allocation for the next year will be reduced.” Id. Notably, the agency made clear that “[t]he Department has no discretion in implementing this provision.” Id. at 41602. Given its conceded lack of discretionary authority to modify the statutorily required calculation of discretionary exemptions, the Final Rule simply reinterprets the text of subparagraph (G). 21 Plaintiffs’ argument is persuasive. When read as a whole, subparagraph (G)’s explicit inclusion of all unused exemptions from the preceding fiscal year in the calculation of the available exemptions for the current fiscal year, necessarily includes those carried over. While subparagraph (G) may be “ambiguous in isolation,” in context and “clarified by the remainder of the statutory scheme . . . ,” United Sav. Assn. of Tex. v. Timbers of Inwood Forest Associates, Ltd., 484 U.S. 365, 371 (1988), the “exemption adjustments” required under subparagraph (G) are a distinct step in the exemption calculation rubric. By design, the exemption adjustment for any given year is necessarily informed by the exemption estimate of the previous year, which is, in turn, informed by the exemption estimate of the year before that, and so on in a procedure that creates a carryover effect. USDA disputes this interpretation with a repeat of its earlier arguments during preliminary injunction proceedings, contending that plaintiffs “point to no text in the statute that — 21 — Plaintiffs point to the 2018 Farm Bill’s rejection of a House proposal to eliminate the carryover of discretionary exemptions as suggesting Congressional support for their interpretation of the § 2015(o)(6)(G). State Pls.’ MSJ at 20–21. D.C. I concluded that the “Conference Report’s statement was made in the specific context of explaining the Conference Committee’s decision to reject the House’s total elimination of carryover exemptions” and that the report “does not purport to bar agency action in the future.” D.C. I, 443 F. Supp. 3d at 18. To be sure, the Conference Report for the 2018 Farm Bill confirms Congressional support for the prevailing and long-standing agency regime permitting indefinite carryover of discretionary exemptions, but that support alone simply does not mean the statutory text is either clear or ambiguous. Given the finding that subparagraph (G), in the context of the full paragraph (6), is clear in requiring such carryover, no more need be said about the 2018 Farm Bill and its Conference Report. — 58 — entitles States to indefinitely stockpile unused exemptions for use at any point in the future.” Defs.’ Opp’n at 36. Certainly, the text of § 2015(o)(6)(G) does not use the terms “indefinite” or “stockpile,” but the structure of this whole paragraph, paired with close examination of how the calculation works in practice, demonstrates a clear meaning that aligns only with plaintiffs’ construction. See State Pls.’ MSJ at 18–19; State Pls.’ Reply at 19–20. The “Secretary shall” carryover from the preceding FY unused discretionary exemptions that were “estimated for the State agency” for that year by operation of the other subparagraphs. The mandate to the Secretary to calculate the carryover adjustment as directed simply leaves little wiggle room for the limitations USDA now seeks to impose. USDA next argues that, even if subparagraph (G) is construed to refer to the number exemptions computed under all of paragraph (6), this “does nothing to alter the statute’s focus on the ‘preceding fiscal year,’” Defs.’ Opp’n at 37 (quoting 7 U.S.C. § 2015 (o)(6)(G)), such that the one-year time limit on carryover exemptions adopted in the Final Rule can also make sense. Not so. The exemption estimate for the current fiscal year is informed by the number of exemptions estimated for the previous fiscal year. 7 U.S.C. § 2015 (o)(6)(G). This means that for each subsequent year, one of the distinct steps in the exemption calculation process will pull the difference between the estimated exemptions and the used exemptions from the preceding year to inform the estimated exemptions for the current year, thus “incentiviz[ing S]tates to use fewer than the available number of exemptions in a given year in better economic times . . . .” State Pls.’ Reply at 20. Defendants’ focus on the “preceding fiscal year” fails to take into account that each “preceding fiscal year,” under the design of the statute, relies on the year preceding the “preceding fiscal year.” USDA’s reading would require a re-write of subparagraph (G). Instead of requiring carryover, if a State has unused exemptions at the end of a fiscal year, of the — 59 — “greater…number of exemptions estimated for the State agency for such preceding fiscal year under this paragraph,” 7 U.S.C. § 2015 (o)(6)(G), the Final Rule would implement the statute as if it were written to allow carryover of the “greater…number of exemptions estimated for the State agency in only the immediately preceding fiscal year but not to exceed the pre-adjustment number of exemptions under subparagraph (E) for that year.” Thus, USDA’s new interpretation in the Final Rule of the discretionary exemption adjustment is predicated on wishful thinking about what the statute says, not its plain text.22 The text, structure and complex operation of § 2015(o)(6) shows that the Final Rule’s elimination of all accumulated discretionary exemptions and imposition of a limit on carryover to only one year’s worth of exemptions from the prior year, is contrary to law, requiring vacatur of this part of the Final Rule. 23 — 22 — State plaintiffs revive an argument made briefly during preliminary injunction proceedings that the Final Rule’s extinguishing of carryover discretionary exemptions is also contrary to law for violating the APA’s prohibition on retroactive rules. See State Pls’s MSJ at 23 (contending Final Rule “summarily extinguished” accumulated exemptions “usable within an agency-run regulatory scheme under clear prior regulation with the agency’s repeated assurance and recognition of those balances,” representing “millions of dollars in vital food assistance to state residents”). In D.C. I, “at least as [then] presented,” the argument was rejected since “the legal status quo ante did not create inalienable rights to maintain unused exemptions beyond the next fiscal year.” 444 F. Supp. 3d at 18 n.10 (internal quotations omitted). This conclusion must be adjusted in light of the Court’s determination that subparagraph (G) requires carryover of unused exemptions from year to year. The law is wellsettled that “an agency may not promulgate retroactive rules without express congressional authorization,” Arkema Inc. v. EPA, 618 F.3d 1, 7 (D.C. Cir. 2010) (citing Bowen v. Georgetown Univ. Hosp., 488 U.S. 204, 208 (1988)), and “[a] rule operates retroactively if it takes away or impairs vested rights,” id. (quoting Nat'l Mining Ass'n v. United States Dep't of Interior, 177 F.3d 1, 8 (D.C. Cir. 1999) (quoting Ass'n of Accredited Cosmetology Sch. v. Alexander, 979 F.2d 859, 864 (D.C. Cir. 1992)). USDA does not claim to have congressional authorization for a retroactive rule, basing the Final Rule’s elimination of carryover exemptions on the agency’s change in interpretation of § 2015(o)(6)(G). In so doing, USDA has done more than merely “alter[] the future effect” of the regulation and “upset[] expectations based on prior law,” Mobile Relay Assocs. v. FCC, 457 F.3d 1, 11 (D.C. Cir. 2006), but in fact “altered the past legal consequences of past actions,” id., making this part of the Final Rule an improper retroactive rule requiring vacatur. See Arkema, 618 F.3d at 7 (where agency “changed its interpretation of” statutory requirement, “contradict[ing]s its past practice,” with the effect of “undo[ing] what [the agency] had, in practice, approved,” thereby attaching new legal consequences to past actions, the challenged rule “is impermissibly retroactive”); Ne. Hosp. Corp. v. Sebelius, 657 F.3d 1, 13-15 (D.C. Cir. 2011) (concluding that “Secretary's decision to apply her present interpretation of the DSH statute to fiscal years 1999–2002 violates the rule against retroactive rulemaking.”). — 23 — Given this conclusion that the Final Rule’s changes to the discretionary exemptions are contrary to law, the parties’ arguments regarding whether these changes are also arbitrary and capricious need not be considered. — 60 — D. Final Rule’s Consideration of Cost and Disparate Impact Are Arbitrary and Capricious Plaintiffs contend that USDA failed adequately to consider the cost and disparate impact of the Final Rule and that these shortcomings render the Final Rule arbitrary and capricious for “ignor[ing] evidence of the Rule’s significant impacts on state governments and protected groups.” State Pls.’ MSJ at 37. Despite the abundance of comments citing these adverse impacts, the agency gave them short shrift and concluded that the Final Rule’s changes will result in only a “small increase ($1.4 million)” in certain administrative costs to States, 84 Fed. Reg. at 66807, and that the changes will “affect potential SNAP program participants in all groups who are unable to meet the ABAWD work requirements,” id. at 66808. The Court agrees with plaintiffs that the agency’s treatment of these two issues of cost and disparate impact bolster the finding that the Final Rule is based on flawed and inadequate reasoning. 1. Cost Impact of Final Rule Throughout the Final Rule, USDA notes that commenters repeatedly raised alarms about the downstream administrative costs to the States likely to result from its significant changes to the waiver process. For example, commenters advised that restricting waiver areas to LMAs “would reduce States’ ability to allocate and coordinate E&T resources effectively… [and] make State planning more difficult,” id. at 66795, but despite “appreciat[ing]” the States’ “administrative needs,” id., the agency was “not compelled by [these] arguments” and chose to move forward without any analysis of the associated costs. Similarly, commenters complained that the Final Rule’s restriction of statewide waivers “would limit State flexibility and . . . increase administrative complexities and burdens,” id. at 66798, but, again, the agency was “not compelled by commenters’ suggestions that the elimination of statewide waivers is arbitrary,” id., citing its “operational experience,” id., without any analysis of the associated costs. — 61 — Commenters also explained that “eliminating the LSA designation criterion would increase administrative burden on States and the Department,” id. at 66799, since “the LSA designation criterion is one of the least burdensome ways for States to submit a request and for the Department to evaluate a request,” id. Again, USDA expressed “appreciat[ion]” for comments but, without any analysis of the associated costs, moved ahead with the proposal, because “LMAs and LSAs are often geographically inconsistent,” id. at 66800. USDA was plainly cognizant that the changes made by the Final Rule would generate many administratively burdensome tasks for the States. To comply with the changes in the Final Rule, States would have to, inter alia, obtain “clarifying guidance” from the agency, id. at 66802; send the agency “request[s] to amend their current waivers” and have those requests reviewed, id. at 66801; “devote resources to quickly analyz[e] data for new requests and implement[] the time limit in new areas,” id.; “coordinate with counties and provide adequate notice so that individuals properly understand the ABAWD time limit,” id. at 66801-02; “plan sufficiently and provide appropriate oversight and training for counties,” id. at 66802; provide guidance and coordination “for E&T providers and community based organizations to prepare for the impacts of the waiver changes,” id.; “train individuals who develop waiver requests,” id. at 66809; “issue Notice of Adverse Action (NOAAs) to those 688,000 ABAWDs who do not meet the work requirement” under the new rule, id.; and “verify work hours and exemptions for 399,000 ABAWDs” newly subject to the work requirement, id. In fact, USDA responded to complaints about administrative burden arising from these myriad implementation tasks by slightly shifting the effective date for the waiver process changes from October 1, 2019 to April 1, 2020. Id. at 66802. — 62 — Despite this recognition of the breadth of the tasks confronting States in implementing the Final Rule’s wholesale changes in the waiver process, USDA estimated in its Regulatory Impact Analysis (“RIA”) that the increase in costs to States would be a paltry $1.4 million “related to [the] administrative burden for verifying work hours and exemptions and sending notices,” id. at 66807, without calculating any other associated costs of compliance. 24 On that point, the agency bluntly states, “this rule does not impose substantial or direct compliance costs on State and local governments . . . .” Id. at 66808. This conclusion, given the many comments about the increased administrative burdens and associated costs, “is so implausible that it could not be ascribed to a difference in view or the product of agency expertise.” State Farm, 463 U.S. at 43 . The inadequacies of the agency’s regulatory impact analysis are illuminative of its overall failure to consider fully or provide any substantive analysis of the costs to States of complying with the policy and process changes reflected in the Final Rule, as well as other “foreseeable costs of the Rule,” detailed in critical comments, including costs “to amplify current state programs to accommodate the increased number of ABAWDs subject to the time limit,” “[i]ncreased poverty and hunger,” “health care costs,” and the “impact on economic activity of cutting SNAP benefits.” State Pls.’ MSJ at 39 & n.18 (citing ABAWD00008257-60 (Summary — 24 — USDA argues that judicial review of the cost analysis in the RIA is impermissible because the analysis was conducted pursuant to Executive Orders 12,866, 13,563, and 13,771 instead of a statutory mandate and that “‘Executive Orders cannot give rise to a cause of action’ under the APA.” Defs.’ Opp’n at 59 (citing Fla. Bankers Ass’n v. U.S. Dep’t of Treasury, 19 F. Supp. 3d 111 , 118 n.1 (D.D.C. 2014), vacated on other grounds, 799 F.3d 1065 (D.C. Cir. 2015)). This argument is not responsive, however, to plaintiffs’ point. Plaintiffs do not challenge the adequacy of the agency’s performance of the RIA, but instead challenge the sufficiency of the agency’s consideration of significant evidence regarding States’ costs. See State Pls.’ MSJ at 37 (“The Rule also is arbitrary and capricious because it ignores evidence of the Rule’s significant impacts on state governments and protected groups.”). The agency’s choice to “brush[] aside critical facts” about States’ costs presented to the agency by commenters and the agency’s subsequent failure to give those facts any meaningful consideration is appropriately reviewed under the APA. See Am. Wild Horse Preservation Campaign v. Perdue, 873 F.3d 914, 932 (D.C. Cir. 2017). — 63 — Comment), -110316–17; -168812–14; -78135–36; -79151–52; -168358–59; -173986–89), ECF No. 105-2; State Pls.’ MSJ, Ex. 3, Decl. of Alexis Carmen Fernández, Chief of CalFresh & Nutrition Branch, California Department of Social Services (“Fernández Decl.”) ¶ 16, ECF No. 65-3 (“. . . any reduction in CalFresh eligibility has a negative impact on an already hard-hit economy . . . [and] the Final Rule would increase state and local administrative costs during a time when public entities are experiencing significant budget cuts.”); id., Ex. 5, Decl. of Laura Zeilinger, Director of the District of Columbia Department of Human Services (“Zeilinger Decl.”) ¶ 11, ECF No. 65-5 (“The economic impact of the public health crisis is likely to last long after the emergency declarations are lifted, leaving many without jobs in the coming months who will still need to rely on . . . providers of nutrition assistance. These providers cannot realistically absorb the demand that would be created if individuals subject to ABAWD requirements in the District cannot receive SNAP benefits.”); D.C. I, 444 F. Supp. 3d at 39 (finding likelihood of immediate, irreparable harm from Final Rule due to “significant regulatory and administrative burden[]” on the States); id. at 41 (same for private plaintiff Bread for the City, which would have to expend “additional resources on non-educational services, including food assistance and ‘medical and social work.’” (citing Jones Decl. ¶¶ 9, 11, 13, ECF No. 1-6)). In sum, the comments were replete with concerns about administrative burdens on and associated costs for States, but USDA did little more than express appreciation for their receipt, essentially dismissing the concerns as unpersuasive for the agency’s goal of “establish[ing] clear limitations under which waivers can be approved.” 84 Fed. Reg. at 66810. This goal, standing alone, may be unassailable. Yet, when commenters complain that the mechanisms used in the Final Rule to achieve this goal significantly increase States’ administrative burdens and costs, with concomitant adverse second order impacts on health and local economies, to such an extent — 64 — that the rule risks undermining the over-arching policy of the statutory scheme to “safeguard the health and well-being of the Nation’s population by raising levels of nutrition among lowincome households,” 7 U.S.C. § 2011 , the agency’s utter failure to address the issue renders the agency action arbitrary and capricious. See Reytblatt v. U.S. Nuclear Regulatory Comm’n, 105 F.3d 715 , 722 (D.C. Cir. 1997) (“An agency need not address every comment, but it must respond in a reasoned manner to those that raise significant problems.” (citing Action on Smoking and Health v. Civil Aeronautics Bd., 699 F.2d 1209 , 1216 (D.C. Cir. 1983))). 2. Disparate Impact of Final Rule Hundreds of comments submitted to USDA provided “quantified feedback as to the impacts of the proposed rule on protected populations.” ABAWD00008226 (Summary Comment), ECF No. 105-2. Indeed, “[m]any” of the 430 commenters who submitted such comments “argued that several vulnerable populations, including racial and ethnic minorities, LGBT populations, veterans and those with disabilities . . . would be disproportionately harmed by the proposed rule.” ABAWD00008263 (Summary Comment), ECF No. 105-2. Commenters explained that the Final Rule would likely have a disproportionate impact on racial minorities, who “in addition to facing higher unemployment rates, [] are more likely to be marginally attached to the workforce, and thus ignored by the U-3 unemployment rates,” id. at 66789, and “provided research indicating that the change in proximity to jobs in recent years varies by socioeconomic characteristics, with poor, minority residents seeing the biggest decline in jobs within a reasonable commuting distance,” id. at 66793. USDA corroborated these critiques, conceding in a self-described “comprehensive Civil Rights Impact Analysis” (“CRIA”), 84 Fed. Reg. at 66808, that “implementation of the final rule may impact African Americans and Hispanic groups at a higher rate due to factors more strongly associated with potential program users in these minority groups,” ABAWD00000358 (CRIA), ECF No. 105-1, and has “the — 65 — potential for impacting certain protected groups due to factors affecting rates of employment of members of these groups,” 84 Fed. Reg. at 66808. See also id. ABAWD00000351 (CRIA), ECF No. 105-1 (“The Department acknowledges the changes outlined in the final rule will likely impact members of the populations outlined above,” i.e., “vulnerable populations, including racial and ethnic minorities, LGBTQ populations, veterans, homeless populations, and those with disabilities, given the disproportionate rate of food insecurity and unemployment in those communities.”). Notwithstanding the comments alerting the agency to the adverse disparate impact of the Final Rule on minorities, women and persons with disabilities, USDA essentially kicked this problem bucket down the proverbial road, citing a lack of “[s]pecific race, ethnicity, and gender data regarding the ABAWDs that will be impacted” by the Final Rule. ABAWD00000358 (CRIA), ECF No. 105-1; see Defs.’ Opp’n at 63. That excuse rings hollow given the many comments supported by “empirical data.” See, e.g., ABAWD00008264 (Summary Comment), ECF No. 105-2; id. (“50 percent of the individuals that would be impacted by the proposed rule come from communities of color”). Despite asserting “that the implementation of mitigation strategies and monitoring by the [agency’s] Civil Rights Division and [] SNAP may lessen these impacts,” id. at 66808, the agency remains vague as to what precisely those “mitigation strategies” may be, retorting only that plaintiffs “cite nothing for the proposition that USDA was required to prospectively set forth how it would respond to hypothetical circumstances,” Defs.’ Opp’n at 63. As with USDA’s dismissive treatment of the increased costs and administrative burdens to States’ resulting from the Final Rule, the agency’s recognition of the disparate impact on protected groups, without any meaningful discussion of the issue in the context of alternatives to — 66 — the rule’s policy choices, points to the agency’s failure to “consider an important aspect” of the effects of the Rule. State Farm, 463 U.S. at 43 ; see also Am. Bankers Ass'n v. NCUA, 934 F.3d 649, 656 (2019) (remanding, as arbitrary and capricious under 5 U.S.C.S. § 706 (2)(A), part of agency rule intended to promote credit union growth among suburban residents, because agency did not adequately respond to comments objecting that low-income and minority residents in urban core areas would be disparately impacted and excluded). IV. CONCLUSION For the foregoing reasons, state plaintiffs and private plaintiffs are entitled to summary judgment and the Final Rule must be vacated. Accordingly, the plaintiffs’ motions for summary judgment, ECF Nos. 64, 65 and 66, are GRANTED, and defendants’ Cross-Motion for Summary Judgment, ECF No. 92, is DENIED. An Order consistent with this Memorandum Opinion will be filed contemporaneously. Date: October 18, 2020 __________________________ BERYL A. HOWELL Chief Judge — 67 —