NATIONAL COMMUNITY REINVESTMENT COALITION et al v. CONSUMER FINANCIAL PROTECTION BUREAU, No. 1:2020cv02074 - Document 26 (D.D.C. 2022)

Court Description: MEMORANDUM OPINION as to plaintiffs' 14 Motion for Summary Judgment and defendant's 18 Cross-Motion for Summary Judgment. Signed by Chief Judge Beryl A. Howell on September 23, 2022. (lcbah4)

Download PDF
NATIONAL COMMUNITY REINVESTMENT COALITION et al v. CONSUMER FINANCIAL PROTECTION BUREAU Doc. 26 UNITED STATES DISTRICT COURT FOR THE DISTRICT OF COLUMBIA NATIONAL COMMUNITY REINVESTMENT COALITION, et al., Civil Action No. 20-2074 (BAH) Plaintiffs, Chief Judge Beryl A. Howell v. CONSUMER FINANCIAL PROTECTION BUREAU, Defendant. MEMORANDUM OPINION For nearly fifty years, the Home Mortgage Disclosure Act (“HMDA”), 12 U.S.C. § 2801, et seq., has provided a window to glean who is being given access to credit to purchase homes across the United States, and who is not. Without prohibiting any particular conduct, HMDA promotes transparency that enables enforcement of other laws and scrutiny of lending practices. HMDA requires covered lenders to collect specified data on mortgages and mortgage applications, with mandated public release of these data to provide communities and public officials “with sufficient information to enable them to determine whether depository institutions are filling their obligations to serve the housing needs of the communities and neighborhoods in which they are located.” Id. § 2801(b). HMDA data also assist public officials in determining how to distribute “public sector investments . . . to improve the private investment environment,” id., and “in identifying possible discriminatory lending practices and enforcing antidiscrimination statutes,” Home Mortgage Disclosure (Regul. C), 12 C.F.R. § 1003.1(b)(1)(iii). 1 Dockets.Justia.com Plaintiffs, which include five nonprofit organizations—the National Community Reinvestment Coalition (“NCRC”), Montana Fair Housing (“MFH”), Texas Low Income Housing Information Service (“TxLIHIS”), Empire Justice Center (“EJC”), and the Association for Neighborhood & Housing Development (“ANHD”)—and the City of Toledo, Ohio, assert, without any dispute from defendant, that HMDA data have been invaluable in “uncovering and addressing redlining, fair lending violations, and other inequitable lending practices” over the decades. Pls.’ Mem. Supp. Mot. Summ. J. (“Pls.’ Mem.”) at 6, ECF No. 14. As examples, plaintiffs cite investigative reporting that garnered a 1989 Pulitzer Prize by documenting the barriers faced by Black communities in Atlanta, Georgia to accessing mortgage loans and the “widening” gap between the rates at which Black and White residents with the same income levels were approved for mortgages by local banks. See Admin. Rec. (“AR”) at 1716–24 (Bill Dedman, The Color of Money, ATLANTA J.-CONST., May 1, 1988, at A1). This series prompted the U.S. Department of Justice to take “enforcement action against a prominent Atlanta lender for violating federal lending discrimination laws.” Pls.’ Mem. at 6. More recently, HMDA data have been used, inter alia, by plaintiff EJC to assess mortgage lending patterns in communities across New York, see Pls.’ Mot., Ex. 3, Decl. of EJC L. Dir. Jonathan Feldman (“Feldman Decl.”) ¶¶ 3–7, ECF No. 14-3 (citing Barbara van Kerkhove, The River Runs Dry II: The Persistent Mortgage Drought in Rochester’s Communities of Color, EJC (July 2015), http://empirejustice.org/wpcontent/uploads/2018/01/river-runs-dry-ii-1.pdf; Barbara van Kerkhove, The Lingering Storm: Mortgage Lending Disparities on Long Island, EJC (Sept. 2015) http://empirejustice.org/wpcontent/uploads/2015/09/the-lingering-storm-mortgage.pdf); and by plaintiff TxLIHIS to “track modern day redlining” by “investigating the number, amount, and providers of home mortgage 2 loans on a census tract level for several major counties in Texas” and “monitor . . . public investments in housing,” id., Ex. 4, Decl. of TxLIHIS Advoc. Co-Dir. Adam Pirtle (“Pirtle Decl.”) ¶¶ 3, 4, ECF No. 14-4. The federal agency statutorily tasked under HMDA with collecting and distributing data on mortgages, mortgage applications, and lines of credit secured by dwellings, and promulgating rules and regulations governing which lenders must collect and report these data, is the Consumer Financial Protection Bureau (“CFPB”). 12 U.S.C. §§ 2802(1); 2804. In implementing HMDA, CFPB rules and regulations distinguish between “closed-end mortgage loans” and “open-end lines of credit.” A “[c]losed-end mortgage loan” is “an extension of credit that is secured by a lien on a dwelling,” 12 C.F.R. § 1003.2(d), like the traditional 30-year mortgage millions of Americans use to purchase their homes. An “[o]pen-end line of credit” is an extension of credit “up to any limit set by the creditor,” upon which the consumer may draw repeatedly throughout the plan’s term as the outstanding balance is repaid, and which is also secured by a lien on a dwelling. Id. § 1003.2(o)(1). 1 In 2015, CFPB overhauled the existing HMDA data collection and reporting requirements. Upon determining that the existing rules for closed-end mortgage loans presented an administrative burden on lending institutions, CFPB promulgated a rule exempting from the disclosure requirements those lending institutions issuing fewer than 25 closed-end mortgage loans. Home Mortgage Disclosure (Regul. C), 80 Fed. Reg. 66128 (Oct. 28, 2015) (codified at 1 One common type of open-end line of credit is a home equity line of credit (“HELOC”). A HELOC is a line of credit allowing homeowners to borrow against their home equity, or “the amount [the] property is currently worth, minus the amount of any mortgages.” My lender offered me a Home equity line of credit (HELOC). What is a Heloc?, CFPB (Sept. 25, 2017), https://www.consumerfinance.gov/ask-cfpb/my-lender-offered-me-a-homeequity-line-of-credit-heloc-what-is-a-heloc-en-246/. During the “draw period” of a HELOC, the homeowner may borrow money repeatedly from the line of credit, while making minimum payments, and then, during the “repayment period” of the HELOC, the homeowner must pay off the balance all at once or over a certain period of time, but risks foreclosure on the property if the HELOC is not repaid. Id. 3 12 C.F.R. pt. 1003) (the “2015 Rule”). Offsetting this contraction of HMDA data collection requirements, CFPB, for the first time, also required lending institutions to collect and submit information on open-end lines of credit secured by dwellings, with an exception for institutions issuing fewer than 100 such loans, id. at 66149, accompanied by a three-year delay in the effective date for lenders to adjust to the new open-end lines of credit requirements, id. at 66162. In the end, the 2015 Rule exempted “22 percent of depository institutions” that had previously been required to report HMDA data, which resulted in a significant loss of data in certain census tracts.” Id. at 66148. Nonetheless, according to CFPB, these thresholds balanced the need for “sufficient data for analyzing mortgage lending at the national, local, and institutional levels” with the need to avoid burdening “lower-volume” lenders with substantial “compliance costs.” Id. at 66146. Following promulgation of the 2015 Rule, CFPB still “heard concerns” that “lowervolume institutions continue[d] to experience significant burden” from the reporting requirements, which “d[id] not justify the small amount of data such institutions would report.” Home Mortgage Disclosure (Regul. C), 85 Fed. Reg. 28364, 28368 (May 12, 2020) (to be codified at 12 C.F.R. pt. 1003) (the “2020 Rule”). Such concerns from lending institutions led CFPB, in 2017, to increase the reporting threshold for open-end lines of credit to 500 loans proactively, before the reporting requirements of the 100-loan threshold took effect. See Home Mortgage Disclosure (Regul. C), 82 Fed. Reg. 43088, 43094 (Sept. 13, 2017) (to be codified at 12 C.F.R. pt. 1003) (the “2017 Rule”). Then, in 2020, “[i]n light of the concerns expressed by industry stakeholders regarding the considerable burden associated with reporting” the data “required by the 2015 HMDA Rule,” CFPB again modified its regulations to reduce further the reporting requirements for HMDA 4 data on closed-end mortgage loans, this time quadrupling the loan-volume threshold triggering the requirement that lending institutions collect and report data on mortgages and mortgage applicants. See 2020 Rule, 85 Fed. Reg. at 28368. As for open-end lines of credit, the 2020 Rule doubled the loan-volume reporting threshold of the 2015 Rule, to which the reporting threshold automatically would have reverted with the expiration of the 2017 Rule’s temporary increase. Id. at 28367. Plaintiffs, all of whom “use HMDA data in their research, education, and advocacy to promote access to credit, and thus to housing opportunities” in minority and rural communities, Pls.’ Mem. at 15, initiated this lawsuit challenging the 2020 Rule as arbitrary and capricious, contrary to law, and in excess of the agency’s statutory authority under the Administrative Procedure Act (“APA”), 5 U.S.C. § 706. See Compl. ¶ 2, Nat’l Cmty. Reinvestment Coal. v. CFPB, No. 20-cv-2074, ECF No. 1 Now pending before the Court are the parties’ cross motions for summary judgment. See Pls.’ Mot. Summ. J. (“Pls.’ Mot.”), ECF No. 14; Def.’s Cross-Mot. Summ. J. (“Def.’s Mot.”), ECF No. 18. For the reasons explained below, plaintiffs are entitled to partial summary judgment as to the portions of the 2020 Rule addressing closed-end mortgage loans, and defendant is entitled to partial summary judgment as to the portions of the same rule addressing open-end lines of credit. I. BACKGROUND A. Statutory and Regulatory Background The statutory framework and regulatory developments leading to promulgation of the 2020 Rule are discussed below, in chronological order. 1. HMDA In 1975, HMDA was enacted to remedy the failure of “some depository institutions . . . to provide adequate home financing to qualified applicants on reasonable terms and conditions,” 5 which, Congress found, had “sometimes contributed to the decline of certain geographic areas.” 12 U.S.C. § 2801(a). Rather than creating a direct enforcement mechanism to ensure financial institutions were serving their communities adequately, Congress opted for a sunshine statute, acting on the belief that public knowledge of local banks’ lending patterns could improve the banks’ affirmative compliance with fair lending and anti-discrimination laws. See id. § 2801(b). One of the bill’s sponsors explained that the new statute “would use the power of market competition—competition for the saver’s dollar—to encourage lenders to do a better job in their own backyards,” because “if depositors are able to learn, through disclosure, which local lenders are treating the community fairly, lenders will become more accountable.” 121 CONG. REC. 25160 (1975) (statement of Sen. Proxmire). In the 1980s, after a “series of investigative reports and studies revealed that discrimination against certain applicants and borrowers was common during the mortgage lending process,” 2015 Rule, 80 Fed. Reg. at 66130, HMDA was amended in 1988 and 1989 to “dramatically improve[] the public’s understanding of how mortgage lending decisions affect[] both communities and individual applicants and borrowers,” id., by expanding both the universe of lending institutions required to report under HMDA and the data points such lenders were required to collect, id.; see also Financial Institutions Report, Recovery and Enforcement Act of 1989, Pub. L. 101-73 § 1211, 103 Stat. 183, 524–26. For more than four decades, HMDA has served “to provide the citizens and public officials of the United States with sufficient information to enable them to determine whether depository institutions are filling their obligations to serve the housing needs of the communities and neighborhoods in which they are located.” 12 U.S.C. § 2801(b). 6 As a so-called “sunshine statute,” 2020 Rule, 85 Fed. Reg. at 28392, HMDA’s principal tools for fulfilling its ends are data collection and disclosure. Under HMDA, financial institutions not otherwise exempted are required to “compile and make available . . . to the public” information including “the number and total dollar amount of mortgage loans” that were either “originated” or “purchased by that institution during each fiscal year.” 12 U.S.C. § 2803(a)(1). A financial institution must disclose this information for each “primary metropolitan statistical area, metropolitan statistical area, or consolidated metropolitan statistical area that is not comprised of designated primary metropolitan statistical areas” (collectively, “MSA”) in which it has a home or branch office. Id. These records must “clearly and conspicuously disclose” the number and total dollar amount of mortgage loans by census tracts and/or by county, id. § 2803(a)(2)(A), as well as additional demographic and geographic information about the mortgagors or mortgage applicants. 2 Institutions with “total assets” below the specific and substantial annually adjusted amount—currently $50,000,000—are exempted from the HMDA collection and disclosure requirements altogether. Id. § 2808. 3 Over time, Congress has expanded the scope of information HMDA requires lenders to report. Initially, depository institutions were only required to disclose the “number and dollar amount” for their loans (1) itemized “by census tracts” or “otherwise by ZIP code,” Act of Dec. 31, 1975, Pub. L. No. 94-200 § 304(a)(2)(A), 89 Stat. 1124, 1124; and (2) indicating which loans were “insured under title II of the National Housing Act,” id. § 304(b)(1). Beginning in 1989, 2 A census tract is defined by the U.S. Census Bureau as a “relatively permanent small-area geographic division[] of a county or statistically equivalent entity” with a population of 1,200 to 8,000 people. See Census Tract Program for the 2010 Decennial Census—Final Criteria, 73 Fed. Reg. 13836, 13836 (Mar. 14, 2008). 3 The current $50,000,000 asset-size exemption threshold means that only lenders holding assets in excess of $50,000,000 are subject to HMDA’s data collection and reporting for the mortgages and loans they issue or for which they receive applications. See Home Mortgage Disclosure (Regul. C) Adjustment to Asset-Size Exemption Threshold, 86 Fed. Reg. 72818, 72818 (Dec. 23, 2021) (to be codified at 12 C.F.R. pt. 1003) (“2022 Adjustment to Asset-Size Exemption”). 7 lenders were also required to disclose “the number and dollar amount of mortgage loans and completed applications . . . grouped according to . . . income level, racial characteristics, and gender.” Financial Institutions Reform, Recovery, and Enforcement Act § 1211(a)(4). 2. The Dodd-Frank Act In 2010, in the wake of the financial crisis attributed in significant part to underregulated, high-risk mortgages, Congress substantially amended HMDA through the Dodd-Frank Wall Street Reform and Consumer Protection Act (“Dodd-Frank Act”), Pub. L. 111-203, 124 Stat. 1376. Among other reforms, the Dodd-Frank Act transferred rulemaking authority under HMDA from the Board of Governors of the Federal Reserve System to CFPB, authorizing the latter agency to prescribe regulations “necessary to carry out the purposes” of the statute, 12 U.S.C. § 2804(a), including to exempt certain financial institutions from HMDA’s requirements, see id. §§ 2805(b) & 2808(a). CFPB now works in tandem with various other federal agencies to ensure the requisite data are collected and published. Although CFPB maintains “principal authority to examine and enforce compliance,” id. § 2804(d), HMDA requirements are otherwise enforced, depending on the type of lending institution, by federal banking agencies, the Administrator of the National Credit Union Administration, and the Secretary of Housing and Urban Development (“HUD”), id. § 2804(b)(1). The Federal Financial Institutions Examination Council (“FFIEC”) synthesizes the reported data to “compile[] . . . aggregate data by census tract for all depository institutions” yearly and “produce tables indicating . . . aggregate lending patterns for various categories of census tracts grouped according to location, age of housing stock, income level, and racial characteristics” for each MSA, which tables are published online. Id. § 2809(a). 4 4 According to its website, the FFIEC “is a formal interagency body empowered to prescribe uniform principles, standards, and report forms for the federal examination of financial institutions by the Board of 8 The Dodd-Frank Act also expanded the scope of information HMDA requires lenders to disclose, including the age of the applicant, and additional loan-specific data points, such as associated fees, interest rates, loan term, value of the collateral, and “such other information as the Bureau may require.” Dodd-Frank Act § 1094(3)(A). 3. Regulation C CFPB has implemented HMDA through Regulation C. Under Regulation C, covered financial institutions are required “to disclose certain data to the public, about covered loans for which the financial institution receives applications, or that it originates or purchases, and that are secured by a dwelling located in a State of the United States of America, the District of Columbia, or the Commonwealth of Puerto Rico.” 12 C.F.R. § 1003.1(c). In addition to differentiating between closed-end mortgage loans and open-end lines of credit, HMDA and its implementing regulations distinguish between two types of financial institutions, both of which are subject to HMDA disclosure requirements. See id. §§ 1003.2(g); 1003.5(a)(4). “Depository financial institution[s]” include banks, savings associations, and credit unions that, inter alia, (1) have assets above a certain annually adjusted threshold, id. § 1003.2(g)(1)(i); and (2) are federally insured or regulated, or issue loans “insured, guaranteed, or supplemented by a Federal agency,” or where intended to be sold to the Federal National Mortgage Association (“Fannie Mae”) or the Federal Home Loan Mortgage Corporation (“Freddie Mac”), id. § 1003.2(g)(1)(iv). “Nondepository financial institution[s],” by contrast, are “for-profit mortgage-lending institution[s]” other than banks, savings associations, or credit Governors of the Federal Reserve System (FRB), the Federal Deposit Insurance Corporation (FDIC), the National Credit Union Administration (NCUA), the Office of the Comptroller of the Currency (OCC), and the [CFPB]” in order to “promote uniformity in the supervision of financial institutions.” About the FFIEC, FFIEC (Apr. 15, 2020), https://www.ffiec.gov/about.htm. HMDA data is publicly available on the FFIEC’s website. See HMDA Data Publ’n, FFIEC (last accessed September 20, 2022), https://ffiec.cfpb.gov/data-publication/. 9 unions. Id. § 1003.2(g)(2). Under Regulation C, both depository and nondepository financial institutions are required “to submit data to the appropriate Federal agency.” Id. § 1003.1(c). 5 Through Regulation C’s data disclosure requirements, CFPB discharges HMDA’s goals of “provid[ing] the public with loan data” that can be used to (1) “help determine whether financial institutions are serving the housing needs of their communities;” (2) “assist public officials in distributing public-sector investment so as to attract private investment to areas where it is needed;” and (3) “assist in identifying possible discriminatory lending patterns and enforcing antidiscrimination statutes.” Id. § 1003.1(b)(1). Specifically, Regulation C’s § 1003.4 sets out the data that financial institutions must collect and report for each loan application they receive and each covered loan they originate or purchase within the calendar year, and has remained largely unchanged since its last modification in 2018. These data include, inter alia, for each loan or application, three general types of information— (a) Information about the loan itself, such as: • the purpose for the loan (i.e., home purchase, home improvement, refinancing), id. § 1003.4(a)(3); • “[t]he amount of the covered loan,” meaning, for a closed-end mortgage loan, “the amount to be repaid as disclosed on the legal obligation” or, for an open-end line of credit, “the amount of credit available to the borrower under the terms of the plan,” id. § 1003.4(a)(7); • “[t]he interest rate applicable to the approved application, or to the covered loan at closing or account opening,” id. § 1003.4(a)(21); • “the difference between the covered loan’s annual percentage rate and the average prime offer rate for a comparable transaction as of the date the interest rate is set,” id. § 1003.4(a)(12)(i); • data about “the property securing the covered loan,” including its value, id. § 1003.4(a)(28), and address, state, county, and census tract, id. § 1003.4(a)(9), as well as whether the dwelling on that property “is site-built or a manufactured home,” id. § 1003.4(a)(5); 5 Depository institutions report data variously, as designated, to the Office of the Comptroller of the Currency, the Federal Deposit Insurance Corporation, the Board of Governors of the Federal Reserve System, or the National Credit Union Administration Board. See 12 C.F.R. § 1003.5(a)(4); 12 U.S.C. §§ 1813(q); 2803(h)(2). Other financial institutions report data to the HUD Secretary. 12 U.S.C. § 2803(h)(2)(D). 10 • • whether the terms of the contract include or would have included balloon payments or interest-only payments, id. § 1003.4(a)(27), and whether the covered loan is a high-cost mortgage, id. § 1003.4(a)(13); and a universal loan identifier “that can be used to identify and retrieve the covered loan or application file,” id. § 1003.4(a)(1). (b) Information about actions taken on the loan, such as: • the “action taken by the financial institution,” meaning, “[w]hether [the] covered loan was originated or purchased,” or, if an application did not result in origination, whether it was “not accepted, denied, withdrawn by the applicant, or closed for incompleteness,” id. § 1003.4(a)(8); and • for denied applications, “[t]he principal reason or reasons the financial institution denied” it, id. § 1003.4(a)(16). (c) Information about the loan applicant, such as: • the ethnicity, race, age, and sex of the applicant “and whether this information was collected on the basis of visual observation or surname,” as well as “the gross annual income relied on in making the credit decision” or “processing the application,” id. § 1003.4(a)(10); • “the credit score or scores” on which the financial institution relied “in making the credit decision,” as well as “the name and version of the scoring model used to generate each credit score,” id. § 1003.4(a)(15); and • the “ratio of the applicant’s or borrower’s total monthly debt to the total monthly income relied on in making the credit decision,” id. § 1003.4(a)(23), and ratio of “total amount of debt secured by the property to the value of the property relied on in making the credit decision,” id. § 1003.4(a)(24). 4. The 2015 Rule In 2015, in light of the Dodd-Frank Act’s changes to HMDA, CFPB promulgated a new rule revising Regulation C by expanding its scope in three important ways, while also creating a new category of exempt institutions. In terms of expansion, the 2015 Rule more than doubled the number of data points financial institutions were required to collect and report under Regulation C, see id. § 1003.4(a), and extended the HMDA reporting requirements for depository institutions to nondepository institutions, which previously did not have to collect and report these data, see 2015 Rule, 80 Fed. Reg. at 66146. It also expanded the types of transactions reported under HMDA, for the first time requiring lending institutions to submit information about “dwelling-secured, consumer-purpose open-end lines of credit,” based on 11 CFPB’s recognition that, “[h]ad open-end line of credit data been reported in HMDA,” in the years leading up to the 2008 financial crisis, “the public and public officials could have had a much earlier warning and a better understanding of potential risks, and public and private mortgage relief programs could have better assisted distressed borrowers in the aftermath of the crisis.” Id. at 66149. At the same time, the 2015 Rule, for the first time, established loanvolume thresholds that exempted from the definition of “financial institution” any lending institution—depository or nondepository—originating or purchasing fewer than 25 closed-end mortgage loans, or fewer than 100 open-end lines of credit, in each of the two preceding calendar years. Id. at 66128. CFPB justified these loan-volume thresholds as properly “balanc[ing] the burden on financial institutions with the value of the data reported,” which was crucial to “HMDA’s ability to achieve its purposes.” Id. at 66147. Many “industry commenters” had advocated for “[h]igher closed-end mortgage loanvolume thresholds” than the 25-mortgage number adopted by the 2015 Rule, but CFPB rejected the proposed higher thresholds. Id. The agency explained that while higher volume exemption thresholds “might not significantly impact the value of HMDA data for analysis at the national level,” they “would have a material negative impact on the availability of data about patterns and trends at the local level,” which data was “essential to achieve HMDA’s [] purposes.” Id. Citing multiple examples of local and state officials using HMDA data to “identify and target relief to localities impacted by high-cost lending or discrimination,” see id. at 66147–148, CFPB further explained that “the loss of data in communities at closed-end mortgage loan-volume thresholds higher than 25 would substantially impede” the ability of the public and public officials in these locales and others “to understand access to credit in their communities.” Id. at 66148. CFPB quantified the impact if the mortgage loan-volume threshold were set at 100 loans, estimating 12 that the number of census tracts that would lose at least 20% of reported data under the 2015 Rule would be eight times higher “than the number with a threshold set at 25,” and six times higher for the number of lower-middle income census tracts affected by the 2015 Rule. Id. The new thresholds took effect January 1, 2017. Id. As for the open-end line of credit exemption threshold, CFPB concluded that setting the threshold at 100 loans would “improve the availability of data concerning open-end dwellingsecured lending,” id. at 66150, the expansion of which had “contributed to the foreclosure crises that many communities experienced in the late 2000s,” id. at 66149. Setting the threshold at 100 loans would also ensure data was collected “from a sufficient array of institutions and about a sufficient array of transactions”—with “nearly 90 percent of all open-end line of credit originations” being reported under the 2015 Rule—to “improve the public and public officials’ ability to monitor and understand all sources of dwelling-secured lending and the risks posed to consumers and communities by those loans.” Id. at 66150. To give lenders time to adjust to the new reporting requirements for open-end lines of credit, the effective date for the open-end reporting threshold was January 1, 2018. Id. at 66162. 5. The 2017 Rule Before the reporting requirements for open-end lines of credit took effect, however, in 2017, CFPB responded to “concerns that the open-end threshold at 100 transactions [was] too low” by issuing a rule temporarily increasing, for 2018 and 2019, the loan-volume threshold for open-end lines of credit, exempting financial institutions that originated or purchased fewer than 500 open-end lines of credit in each of the two preceding calendar years from the data collection and disclosure requirements. See 2017 Rule, 82 Fed. Reg. at 43088. CFPB explained the openend lines of credit reporting threshold was revised so soon after promulgating the 2015 Rule because the agency had underestimated the start-up costs of reporting to lending institutions, due 13 to its mistaken assumption that many institutions would collect and report their HMDA data on open-end lines of credit using the same systems and processes already in place to report closedend mortgage loans. Id. at 43094. CFPB also noted the growth, over several years, in the number of dwelling-secured open-end lines of credit originated across the United States, reasoning that the increased prominence of open-end lines of credit in the mortgage market would mean “that more institutions would have reporting responsibilities under the 100-loan open-end threshold than estimated in the 2015 HMDA Final Rule.” Id. Originally, the agency had decided on the 100-loan threshold because this “would require reporting by only 749 financial institutions, all but 24 of which would also report data on their closed-end mortgage lending,” and “avoid imposing the burden of establishing open-end reporting on approximately 3,000” other institutions with “lower volumes of open-end lending.” Id. at 43091. Now, CFPB explained, “the total number of open-end reporters exceeding the transactional coverage threshold could be estimated at 980.” Id. at 43094. In light of these belated insights into the costs of the 2015 Rule, CFPB promulgated the 2017 Rule to temporarily increase the open-end reporting threshold to 500 loans, in order to “provide time for [CFPB] to consider whether to initiate another rulemaking to address the appropriate level for the open-end threshold for data collected beginning January 1, 2020.” Id. at 43088. 6. EGRRCPA In 2018, the availability of HMDA information was contracted somewhat by enactment of the Economic Growth, Regulatory Relief, and Consumer Protection Act (“EGRRCPA”), which codified partial exemptions for certain smaller-volume lenders from the Dodd-Frank Act’s expanded disclosure requirements. Pub. L. 115-174 § 104(a)(2), 132 Stat. 1296, 1300–01. EGRRCPA exempted “insured depository institution[s] or insured credit union[s]” that “originate[] fewer than 500” closed-end or open-end loans “in each of the 2 preceding calendar 14 years” from the expanded Dodd-Frank mortgage-data reporting requirements, while still requiring these low-volume lenders to collect and report the HMDA data required before enactment of the Dodd-Frank Act. 12 U.S.C. §§ 2803(i)(1)–(2). Sensitive to the value of HMDA data for communities and local and federal government agencies to enforce other laws against discriminatory lending, EGRRCPA expressly made ineligible for the new exemption any lending institution with a bad record “in meeting community credit needs.” Id. § 2803(i)(3). 6 7. The 2018 Rule After passage of EGRRCPA, CFPB issued an interpretive rule “to implement and clarify the requirements of section 104(a) of [EGRRCPA],” which had amended HMDA by “adding partial exemptions from HMDA’s requirements for certain insured depository institutions and insured credit unions.” Partial Exemptions from the Requirements of the Home Mortgage Disclosure Act Under the Economic Growth, Regulatory Relief, and Consumer Protection Act (Regul. C), 83 Fed. Reg. 45325, 45325 (Sept. 7, 2018) (to be codified at 12 C.F.R. pt. 1003) (the “2018 Rule”). In relevant part, the 2018 Rule clarified “which of the data points in Regulation C are covered by the partial exemptions.” Id. Specifically, the 2018 Rule stated that institutions partially exempted under EGRRCPA would no longer need to collect and report more than half of the data “currently set forth in Regulation C” (26 out of 48 data points), which the agency interpreted as falling within EGRRCPA’s carveout, and clarified that these partially exempt institutions would still be required to report 22 data points “currently specified in Regulation C,” 6 Specifically, EGRRCPA exemptions do not apply to insured depository institutions that have “received a rating of ‘needs to improve record of meeting community credit needs’ during each of [their] 2 most recent examinations or a rating of ‘substantial noncompliance in meeting community credit needs’ on [their] most recent examination” under the Community Reinvestment Act, such that those institutions must continue to comply with the expanded Dodd-Frank Act reporting requirements. 12 U.S.C. §§ 2803(i)(3); 2906(b)(2). 15 id. at 45329, including basic information about the loan and loan applicant. See 12 C.F.R. § 1003.4(a). 8. The 2019 Notice of Proposed Rulemaking In May 2019, based on the information gathered after temporarily increasing the openend line of credit threshold in 2017 and in light of EGRRCPA, CFPB decided to proceed with changes to the 2015 Rule, publishing a notice of proposed rulemaking to consider amendments to the loan-volume reporting thresholds. See Home Mortgage Disclosure (Regul. C), 84 Fed. Reg. 20972, 20972 (proposed May 13, 2019) (the “2019 NPRM”). The 2019 NPRM proposed two alternate increased thresholds for closed-end mortgage loans higher, at either 50 or 100 closed-end loans, an increase from the then-current 25. Id. For open-end lines of credit, the 2019 NPRM proposed extending the temporary 500 open-end line of credit threshold imposed under the 2017 Rule for several more years, until January 1, 2022, after which the reporting threshold would decrease permanently to 200 open-end lines of credit. Id. During the first comment period, which closed on June 12, 2019, id., CFPB received over 300 comments, and after the comment period was reopened until October 15, 2019, see Home Mortgage Disclosure (Regul. C); Reopening of Comment Period, 84 Fed. Reg. 37804, 37805 (Aug. 2, 2019), CFPB received roughly 400 additional comments, see 2020 Rule, 85 Fed. Reg. at 28367. 9. The Challenged 2020 Rule On May 12, 2020, CFPB promulgated the challenged 2020 Rule, amending Regulation C by, inter alia, expanding the class of lending institutions exempt from HMDA’s reporting requirements beyond both the original HMDA exemption for any lending institution with less than $50,000,000 in assets and the 2018 EGRRCPA exemption from expanded reporting for low-volume mortgage lenders. Today, under the 2020 Rule, depository and nondepository institutions only qualify as “financial institutions” subject to HMDA’s data collection and 16 disclosure requirements if they have originated at least 100 closed-end mortgage loans or at least 200 open-end lines of credit not otherwise excluded under HMDA in each of the two preceding calendar years. 2020 Rule, 85 Fed. Reg. at 28364; see also 12 C.F.R. §§ 1003.2(g)(1)(v); 1003.2(g)(2)(ii). Summarized below are salient parts of the agency’s explanation for the 2020 Rule. a. Reasons for the 2020 Rule CFPB cited “a few developments” since the issuance of the 2015 Rule as influencing the agency’s decision to revisit the reporting thresholds. 2020 Rule, 85 Fed. Reg. at 28370. First, CFPB referenced “concerns” from “industry stakeholders” that “lower-volume institutions continue[d] to experience significant burden” from the reporting requirements. Id. at 28368. Second, CFPB now had “access to HMDA data from 2018, . . . the first year that financial institutions collected data under the 2015 HMDA Rule,” which put it in “a better position to assess both the benefits and burdens of the reporting required under the 2015 HMDA Rule.” Id. at 28371. Third, specific to the threshold for open-end lines of credit, CFPB had discovered that cost estimates on which it had relied in promulgating the 2015 Rule “may understate the burden that open-end reporting would impose on smaller institutions” and had “significantly” underestimated the “total number of institutions exceeding the threshold of 100 open-end lines of credit in 2018” to be 749 institutions, when in fact the total would be 1,014 institutions. Id. at 28378. Finally, CFPB cited the passage of EGRRCPA as “chang[ing] the costs and benefits associated with different coverage thresholds” because the “partial exemptions [now] available to the vast majority of the depository financial institutions” originating fewer than 500 open-end lines of credit or closed-end mortgage loans annually, id., would relieve them “of the obligation to report many of the data points generally required by Regulation C,” id. at 28371. Put simply, CFPB explained that it now understood the burdens of HMDA data collection and reporting to 17 be different than when it had promulgated the 2015 Rule, leading the agency to conclude it was necessary to revisit the reporting requirements. CFPB acknowledged the conclusion reached in the 2015 Rule that the “loss of data at the local level would substantially impede the public’s and public officials’ ability to understand access to credit in their communities” if the closed-end threshold were set higher than 25 loans, id. at 28368, but nonetheless determined that, “in light of these developments,” raising the reporting thresholds would “provide[] sufficient information” on both types of lending “to serve HMDA’s purposes, while appropriately reducing ongoing costs that smaller institutions [were] incurring under the current threshold[s],” id. at 28371. b. Legal Authority As for its legal authority to amend the thresholds, CFPB explained it was acting “[p]ursuant to its authority under . . . HMDA section 305(a),” id. at 28367, and its judgment “that the final rule’s amendments to the thresholds . . . are necessary and proper to effectuate the purposes of HMDA and facilitate compliance with HMDA by reducing burden and establishing a consistent loan-volume test, while still providing significant market coverage,” id. at 28368. c. Comments Supporting and Critiquing the 2020 Rule CFPB documented numerous comments received in response to the 2019 NPRM in support of the amended thresholds, including statements by “most” industry commenters “that the cost of complying with regulations ha[d] affected their ability to serve their communities,” with “[a] number of small financial institutions” further explaining that their employees had to “spend a considerable amount of time on HMDA compliance” and a “national trade association” stating that “certain institutions manage[d] their mortgage lending to stay below the threshold for HMDA reporting, . . . ultimately leav[ing] customers with fewer lending options.” Id. at 28369. CFPB did not cite to any specific comments in the record making these points. 18 Notably absent from these comments was any identification of precisely which institutions were managed with a focus on staying below the HMDA thresholds, and no commenting institution admitted to doing this themselves. Despite this notable lack of specificity, this commenter “suggested that an increase in the . . . threshold[s] could increase the flow of credit by small banks into their communities.” Id. A commenter also “stated that, in areas where government authorities do consider HMDA data in making public investment decisions, HMDA data from lower-volume institutions ma[de] up [such] a small percentage of the overall lending data within the area” that they did not “impact such investment decisions.” Id. Again, this broad characterization was lacking in any specificity about which areas were referenced to enable any due diligence or double-checking on the representation. Regarding open-end lines of credit, commenters argued that under the revised threshold, “the costs associated with reporting such lines of credit would make them unprofitable, leading banks to either discontinue offering such loans or to pass on cost increases to consumers.” Id. at 28377. Again, no data or other empirical or direct support for this argument was presented beyond the speculative warning. CFPB recognized receipt of many comments opposed to increasing the reporting thresholds, including from “many community organizations, consumer advocates, research organizations and individuals,” id. at 28369, who asserted that the increased thresholds “would imperil HMDA’s purpose of assessing whether financial institutions are meeting the housing needs of their communities,” id. at 28369–70. Noting that “the public visibility of HMDA data [had] motivated financial institutions to increase lending to traditionally underserved borrowers and communities,” these critics of the 2019 NPRM voiced “concern that the smaller institutions that would no longer be required to report closed-end data at the proposed higher thresholds disproportionately len[t] in underserved neighborhoods,” making use of HMDA data more 19 difficult “to uncover and address redlining and other fair lending and fair housing violations,” especially since “the proposed threshold increase would result in a more notable decrease in closed-end data for distressed urban areas, rural areas, tribal areas, communities of color, and neighborhoods that have a high number of immigrants.” Id. at 28370. Both federal and state officials echoed concerns of these community and research organizations and advocates. A letter submitted by a group of “19 U.S. Senators . . . pointed out that the 2015 HMDA Rule exempted 22 percent of depository institutions that had previously been required to report HMDA data, which resulted in a significant loss of data in certain census tracts” and raised concerns that “even if the loss of data from smaller-volume institutions would be limited to the overall market, the loss of that data would have a real and meaningful impact for residents of affected communities.” Id. Comments of a State attorney general were also cited for the point “that the proposed closed-end threshold increase would all but eliminate its ability to enforce lending laws in ‘hyper-localized’ markets in rural areas” where “small, local lenders [were] disproportionately represented.” Id. “Many commenters” opposing the proposed increased exemption thresholds “also stated that the cost savings that would result from excluding lenders from HMDA reporting would be modest,” noting that “most of the lenders that would be excluded . . . [were] already exempt from reporting many of the new HMDA data points because they qualif[ied] for partial exemptions under the EGRRCPA and would therefore be reporting data that lenders have been reporting for decades,” and because “much of the data reportable under HMDA must be collected for other rules,” including the Truth in Lending Act, “and ordinary underwriting standards.” Id. Commenters took further issue with the impact that an increase in the thresholds “would have on visibility into specific loan products, such as loans for multifamily housing and manufactured 20 housing.” Id. For open-end lines of credit specifically, they expressed fears that “too many lenders and open-end lines of credit might escape public scrutiny . . . and thus make it more likely that events similar to those that led to the 2008 financial crisis would occur again.” Id. at 28377. They noted the 2018 HMDA data “suggested that the most vulnerable borrowers were obtaining the open-end lines of credit with the highest interest rates” and demonstrated “that open-end lines of credit have a high incidence of features that can be risky for borrowers, particularly when layered on top of one another.” Id. CFPB reported these critical concerns without disputing or rebutting the underlying data and the adverse impact on enforcement of anti-discriminatory lending. d. Cost, Benefits, and Impacts of the 2020 Rule In Part VII of the 2020 Rule, CFPB “considered the potential benefits, costs, and impacts of the final rule,” as required under Dodd-Frank Act § 1022(b)(2)(A), id. at 28388, which instructs the agency to consider “the potential benefits and costs of a regulation” to both “consumers and covered persons, including the potential reduction of access by consumers to consumer financial products or services; the impact on depository institutions and credit unions with $10 billion or less in total assets”; and “the impact on consumers in rural areas,” id. at 28388 n.161. Plaintiffs’ challenge to the 2020 Rule is highly critical of CFPB’s cost-benefit analysis, see Pls.’ Mem. at 20–39, and, consequently, this portion of the 2020 Rule is discussed in detail. i. Benefits and Costs to Covered Persons The 2020 Rule identified the principal benefits of the increased volume-threshold exemptions to “covered persons,” i.e., lending institutions, as “the reduction of costs . . . relative to the compliance costs the covered persons would have to incur” if the 2015 Rule thresholds 21 were maintained (or, in the case of the open-end line of credit reporting threshold, reinstituted). 2020 Rule, 85 Fed. Reg. at 28389. To quantify the burdens of HMDA’s data collection and reporting requirements on financial institutions, CFPB relied on the same “basic framework” as used in the 2015 Rule, id., with “some updates, mainly to reflect the inflation rate,” and with efforts to account for the impact of EGRRCPA and “operational improvements” in reporting, id. at 28390. The agency first identified “18 discrete compliance ‘tasks’” and “seven key dimensions of compliance operations that were significant drivers of compliance costs, including the reporting system used, the degree of system integration, the degree of system automation, the compliance program, and the tools for geocoding, performing completeness checks, and editing.” Id. at 28389. These tasks and “compliance operations” were then used essentially to categorize covered financial institutions into three tiers. Id. (explaining definition of “three broadly representative financial institutions according to the overall level of complexity of their compliance operations,” with the institutions’ compliance operations decreasing in complexity from “Tier 1” down to “Tier 3” institutions). Based on this framework, CFPB developed “market-level estimates” for the cost of HMDA compliance, id. at 28391, estimating that 401 financial institutions that otherwise would have been required to report their data on open-end lines of credit would now be exempt, but because 378 of those institutions were already partially exempt from reporting under EGRRCPA, id. at 28399, the total estimated “savings in the operational costs associated with open-end lines of credit [would be] about $3.7 million per year,” compared to what the costs would be if the open-end threshold reverted to 100 loans, id. at 28400. Regarding the impact of the new closedend reporting thresholds, CFPB concluded that 1,700 financial institutions previously required to 22 collect and report data on their closed-end mortgage loans under HMDA would now be entirely exempt from these requirements, despite the fact that 1,630 of these financial institutions were already eligible for partial exemptions under EGRRCPA. Id. at 28396. As for closed-end savings, CFPB cited conflicting estimates for the annual costs savings from the heightened closed-end threshold. For example, in Part V of the 2020 Rule, titled “Section-by-Section Analysis,” CFPB “estimate[d] that with a threshold of 100 closed-end mortgage loans, . . . institutions that originate between 25 and 99 closed-end mortgage loans [would] save approximately $11.2 million per year, relative to the current threshold of 25.” Id. at 28374 (emphasis added); see also id. at 28383 (citing same estimated cost savings for nondepository institutions). Part VII of the 2020 Rule, however, discusses the Dodd-Frank Act costbenefit analysis, and cites the annual savings to covered institutions from increasing the closedend mortgage loan threshold as $6.4 million, id. at 28392; 28396, without any explanation for the discrepancy, and with other asserted estimated savings from the new rule to be nearly double that amount, at $11.2 million.7 In its discussion of the cost savings from the 2020 Rule, CFPB also referenced comments responsive to the 2019 NPRM that questioned the extent any burden suffered by lending institutions would actually be relieved by exempting more institutions from HMDA’s requirements, because (1) “smaller-volume lenders already benefit from the EGRRCPA’s partial exemptions;” (2) “almost all of the data that such institutions must report under HMDA would already need to be collected to comply with other statutes like the Truth in Lending Act, to sell loans to Fannie Mae or Freddie Mac, or to acquire Federal Housing Administration insurance for 7 As discussed in Part I.B. infra, after plaintiffs filed the instant suit, CFPB eventually issued a Notice of Correction clarifying that the annual costs savings from the closed-end threshold were in fact $6.4 million, and calling the $11.2 million estimation a “clerical error.” See 2020 Notice of Correction, 85 Fed. Reg. at 69119. 23 loans;” and (3) it would be “hard to imagine that a bank would not keep an electronic record of its lending, even if it were not subject to HMDA reporting.” Id. at 28390. CFPB summarily stated that it had “considered these comments and conclude[d] . . . that they do not undermine the Bureau’s approach,” without further explanation. Id. Regarding the costs of the 2020 Rule to “covered persons,” CFPB acknowledged the possibility that the new thresholds might create onetime costs for financial institutions “related to training and system changes,” but stated that no substantial, ongoing costs to covered persons imposed by the increased reporting thresholds had been identified. Id. at 28391. ii. Benefits and Costs to Consumers Turning to the benefits to consumers under the 2020 Rule, CFPB predicted that some of the costs saved by the financial institutions would be passed onto consumers, id., although the estimated savings per loan would only amount to between $6 and $42 per loan (where the median total loan costs per closed-end mortgage in 2018 were $6,056), id. at 28397. At the same time, given the acknowledged imperfect competitive structure of the mortgage marketplace, the agency conceded that consumers would only expect to receive a portion of these already relatively small dollar figure savings. Id. at 28397; 28402. CFPB speculated that the 2020 Rule might increase the availability of mortgages for consumers, by allowing “smaller banks . . . to enter the mortgage market at more profitable levels,” given that their newfound ability to issue up to 99 closed-end mortgage loans per year without taking on the burden of collecting and reporting HMDA data, and the assurance that they would be able to issue up to 199 open-end lines of credit without having to comply with these requirements. Id. at 28391. Regarding the costs to consumers, CFPB conceded that “[a]s a sunshine statute . . . , most of the benefits of HMDA are realized indirectly” and that the reduction in “data required to be collected and reported under HMDA” would mean “the HMDA data available to serve HMDA’s 24 statutory purposes w[ould] decline.” Id. at 28392. Under the revised closed-end threshold, CFPB anticipated that coverage of “all closed-end mortgage loan originations in the entire mortgage market” would decrease only slightly from 88% to 86%, id. at 28393, and coverage of all open-end line of credit originations across the market would diminish from 89% to 84%, id. at 28398. As for the coverage of lenders across the market, however, the new thresholds reduced the portion of lenders required to report their data from 43% to 27% for closed-end mortgage loans, id. at 28393, and from 15% to 9% for lenders issuing open-end lines of credit, see id. at 28398. 8 Consequently, CFPB further conceded that “[t]he decreased data about excluded institutions may lead to adverse outcomes for some consumers.” Id. at 28397. The agency acknowledged, without disputing or otherwise trying to counter, comments suggesting that increasing the reporting thresholds would (1) “lead to another round of abusive and discriminatory lending similar to abuses that occurred in the years before the financial crisis,” id. at 28392; (2) make it more difficult to “identify possible discriminatory lending patterns,” id. at 28397; and (3) leave “the general public, researchers, . . . Federal agencies,” and States with “an incomplete picture of lending trends in thousands of census tracts and neighborhoods if affected institutions no longer report[ed] HMDA data,” id. at 28392. Notwithstanding these costs and potential adverse outcomes to consumers identified by commenters and the agency itself, CFPB repeatedly emphasized the “substantial challenges” of “quantify[ing] the reduction of . . . benefits to consumers,” lamenting that “none of [the] 8 The 2020 Rule’s estimations are largely based on 2018 data. For that year, CFPB estimated “the total number of institutions that were engaged in closed-end mortgage lending . . . regardless of whether they met all HMDA reporting criteria, was about 11,600, and the total number of closed-end mortgage originations in 2018 was about 7.2 million.” 2020 Rule, 85 Fed. Reg. at 28393. Using these same market-size estimations, the 3,160 financial institutions that continue to report their data after the 2020 Rule took effect (i.e., those remaining after the 1,700 lower-volume reporters were exempted) constitute 27% of the total population of lenders issuing closed-end mortgage loans. As for open-end reporters, CFPB estimates the total universe of institutions issuing open-end lines of credit in 2018 to be 6,615 lenders. Id. at 28399. Setting the reporting threshold at 200 lines of credit means that 613 institutions will continue to report these data, representing 9.3% of the total universe of lenders. 25 commenters” expressing concern about the reduction in data “provided specific quantifiable estimates of the loss of benefits from decreased information.” Id. Without “the data to qualify all HMDA benefits” and an inability “to assess completely” how the 2020 Rule would “reduce those benefits,” the agency stated it would “generally provide[] a qualitative (not quantitative) consideration of the costs.” Id. Without elaborating on the qualitative analysis, however, the agency instead simply reiterated throughout the analysis that it “d[id] not have the data to quantify those costs.” Id.; see also id. at 28402 (stating that it had “no quantitative data that c[ould] sufficiently measure the magnitude of any [adverse] impact of setting the permanent open-end threshold at 200” on consumers); id. at 28398 (recognizing the costs to consumers under the 100-loan threshold were greater than under the 50-loan alterative proposal, but stating that it “currently lacks sufficient data to quantify these costs” other than by comparing “the estimated numbered of covered loans and covered institutions” under each threshold). The 2020 Rule’s new thresholds took effect on July 1, 2020, for closed-end mortgage loans, and on January 1, 2022, for open-end lines of credit. Id. at 28364. B. Procedural Background Two months after issuance of the final 2020 Rule, plaintiffs initiated this lawsuit challenging the 2020 Rule as arbitrary and capricious, contrary to law, and exceeding CFPB’s statutory authority under the APA, 5 U.S.C. §§ 706(2)(A), (C). See Compl. ¶¶ 29–52. Four months after plaintiffs filed the Complaint, highlighting “the presence of inconsistencies and errors that belie the exercise of reasoned decision-making” in the 2020 Rule, id. ¶ 48, CFPB issued a Notice of Correction addressing “several clerical errors regarding the estimated cost savings in annual ongoing costs” for financial institutions that would be newly exempt from reporting requirements under the 2020 Rule, and clarifying that the yearly cost savings from the new rule for closed-end mortgage loans were in fact $6.4 million, not $11.2 million, Home 26 Mortgage Disclosure (Regul. C); Correction of Supplementary Information, 85 Fed. Reg. 69119, 69119 (Nov. 2, 2020) (to be codified at 12 C.F.R. pt. 1003) (“2020 Notice of Correction”). The briefing schedule for cross-motions for summary judgment proposed by the parties was adopted, see Min. Order (Sept. 30, 2020), which schedule was extended twice at the parties’ request, see Min. Order (Oct. 15, 2020) (granting plaintiffs’ request); Min. Order (Jan. 25, 2021) (granting defendant’s request). CFPB thereafter compiled the administrative record, with a certified notice of the contents filed with the Court. See Not. Filing of Certified List of Contents of Admin. Rec., ECF No. 13. 9 On February 5, 2021, after plaintiffs had filed their motion for summary judgment, see Pls.’ Mot., the parties moved to stay the proceedings to afford CFPB’s “Acting Director time to consider and review” the litigation “[i]n light of the possibility the Bureau may decide to undertake further rulemaking activity potentially affecting the challenged rule,” Jt. Mot. Stay Procs. at 1–2, ECF No. 16, which request was granted and this litigation stayed until March 22, 2021, Min. Order (Feb. 5, 2021). Following entry of another briefing schedule proposed by the parties, see Min. Order (Mar. 22, 2021), and the completion of briefing on cross motions for summary judgment, this matter is now ripe for resolution. II. LEGAL STANDARD The 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 9 The 2020 Rule’s administrative record totals just over 3,865 pages and includes regulations; guidance documents; reports; analyses for the rulemaking; statements from the congressional record; websites and news articles referenced in rulemaking; CFPB decision memoranda and correspondence; and public comments. See Not. Filing of Certified List of Contents of Admin. Rec. The parties’ Joint Appendix totals 622 pages, and includes CFPB regulations and rulemaking documents; public comments; and a series of news articles. See Not. Filing of Admin. Rec. App’x, ECF No. 24. 27 accordance with law,’” Cigar Ass’n of Am. v. 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. (“Regents”), 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. 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. Nat’l Credit Union Admin., 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, 577 U.S. 260, 292 (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)). In APA cases such as this one, involving cross-motions for summary judgment, “the district judge sits as an appellate tribunal,” Rempfer v. Sharfstein, 583 F.3d 860, 865 (D.C. Cir. 28 2009) (quoting Am. Bioscience, Inc. v. Thompson, 269 F.3d 1077, 1083 (D.C. Cir. 2001)), since the “entire case on review is a question of law,’ and the ‘complaint, properly read, actually presents no factual allegations, but rather only arguments about the legal conclusion to be drawn about the agency action,’” id. (quoting Marshall Cnty. Health Care Auth. v. Shalala, 988 F.2d 1221, 1226 (D.C. Cir. 1993)). III. DISCUSSION Plaintiffs raise two challenges to the 2020 Rule. 10 First, they argue that the 2020 Rule exceeded CFPB’s statutory authority under HMDA, Pls.’ Mem. at 39–45, which authorizes the agency to make “adjustments and exceptions for any class of transactions, as in the judgment of the Bureau are necessary and proper to effectuate the purposes of this chapter, and prevent 10 CFPB does not contest, see generally Def.’s Opp’n, and the record sufficiently establishes, that plaintiffs have standing to challenge the 2020 Rule because they suffer an injury-in-fact due to the deprivation of “data to which they are entitled by statute and that they would obtain and use” to further “their organizational objectives to promote fairness in and access to lending, to combat housing discrimination, and to shape decisions about community investment,” Pls.’ Mem. at 20; see Decl. of MFH Exec. Dir. Pam Bean (“Bean Dec.”) ¶¶ 2, 4, ECF No. 14-1 (stating plaintiff MFH “investigates alleged violations of fair housing and anti-discrimination laws and files complaints of discrimination in administrative and legal proceedings on behalf of its constituents, in the public interest, or as an organization” and has used and intends to use “HMDA data to identify discriminatory lending practices and areas in need of investment”); Decl. of City of Toledo Chief of Staff Catherine Crosby (“Crosby Decl.”) ¶ 3, ECF No. 14-2 (stating the City of Toledo uses “HMDA data to assess which areas of Toledo are most in need of public investment to ensure adequate access to credit”); Decl. of NCRC Chief Exec. Officer Jesse Van Tol (“Van Tol Decl.”) ¶¶ 5, 7, ECF No. 14-5 (plaintiff “NCRC uses HMDA data and makes it available to the public to promote fairness in lending and combat housing discrimination” and “intended to use HMDA data . . . that would have been submitted under the 2015 HMDA Rule but will not be submitted under the 2020 HMDA Rule to conduct research on issues related to fair lending, access to credit across different geographies and demographic groups, and the determination of which areas are in greatest need of public investment to ensure sufficient home lending.”); Decl. of ANHD Exec. Dir. Barika X. Williams (“Williams Decl.”) ¶¶ 4, 6, ECF No. 14-6 (“HMDA data frequently forms the basis of research released by” plaintiff ANHD to analyze “affordable housing trends by neighborhood in New York City, and the records of individual lenders.”); Feldman Decl. ¶¶ 6, 8 (Plaintiff EJC has “released over a dozen reports analyzing mortgage lending using HMDA data during the past 27 years” and “intended to use HMDA data . . . that would have been submitted under the 2015 HMDA Rule but will not be submitted under the 2020 HMDA Rule to further its fair housing work.”); Pirtle Decl. ¶¶ 4–5 (Plaintiff “TxLIHIS uses HMDA data to track modern day redlining” and “intended to use HMDA data on closed-end mortgage loans that would have been submitted under the 2015 HMDA Rule but will not be submitted under the 2020 HMDA Rule to promote fair housing in Texas.”). See also Fed. Election Comm’n v. Akins, 524 U.S. 11, 21 (1998) (“[A] plaintiff suffers an ‘injury in fact’ when the plaintiff fails to obtain information which must be publicly disclosed pursuant to a statute.”); Waterkeeper All. v. EPA, 853 F.3d 527, 533 (D.C. Cir. 2017) (to show injury in fact, a plaintiff must “assert ‘a view of the law under which the defendant . . . is obligated to disclose certain information that the plaintiff has a right to obtain,’” and when an agency’s exemption “reduces the information that must be publicly disclosed,” thereby denying plaintiff “a statutory right to access it,” such denial is “injury enough” (quoting Am. Soc’y for Prevention of Cruelty to Animals v. Feld Ent., Inc., 659 F.3d 13, 22–23 (D.C. Cir. 2011))). 29 circumvention or evasion thereof, or to facilitate compliance therewith,” 12 U.S.C. § 2804(a). CFPB defends the 2020 Rule as “within the scope of its authority under HMDA,” arguing the statute “does not unambiguously foreclose the Bureau’s interpretation,” such that its interpretation must be given deference under Chevron U.S.A., Inc. v. Nat. Res. Def. Council, Inc. (“Chevron”), 467 U.S. 837, 842–43 (1984). Def.’s Mem. Supp. Cross-Mot. Summ. J. & Opp’n Pls.’ Mot. Summ. J. (“Def.’s Opp’n”) at 43–55, ECF No. 18 Second, plaintiffs contend that the cost-benefit analysis underlying the 2020 Rule is flawed because CFPB exaggerated the “benefits” of increasing the loan-volume reporting thresholds by failing adequately to account for comments suggesting that the savings would in fact be much smaller than estimated, and by relying on overinflated estimates of the cost savings to newly-exempted lending institutions with smaller loan volumes. See Pls.’ Mem. at 20–28. Relatedly, plaintiffs also argue that CFPB miscalculated the “costs” of amending the 2015 Rule by failing to consider the nonquantifiable harms of raising the reporting thresholds, and the disproportionate impacts of those harms. See id. at 28–38. CFPB counters that its cost-benefit analysis was “reasonable” and accounted for all “the relevant factors in assessing the benefits and costs to covered persons” and to consumers “that would result from increasing the HMDA reporting thresholds.” See Def.’s Opp’n at 21–43. For the reasons set forth below, promulgation of the 2020 Rule did not exceed CFPB’s statutory authority, since HMDA grants broad discretion “in the judgment of the” agency to create “exceptions” to the statutory reporting requirements, 12 U.S.C. § 2804(a), but CFPB failed adequately to explain or support its rationales for adoption of the closed-end reporting thresholds under the 2020 Rule, rendering this aspect of the rule arbitrary and capricious. The parties’ 30 dispute over the statutory authority for promulgation of the 2020 Rule is discussed first, followed by analysis of plaintiffs’ challenge to the 2020 Rule as arbitrary and capricious. A. CFPB’s Statutory Authority for Promulgation of the 2020 Rule Plaintiffs’ challenge to CFPB’s authority under HMDA to promulgate the challenged rule is predicated on two arguments. First, plaintiffs take issue with CFPB’s interpretation of HMDA § 305(a), see Pls.’ Mem. at 39, which the agency relied upon as the source of authority for the 2020 Rule. This statutory provision authorizes CFPB to “provide for such adjustments and exceptions for any class of transactions, as in the judgment of the Bureau are necessary and proper to effectuate the purposes of [HMDA] . . . or to facilitate compliance therewith.” 12 U.S.C. § 2804(a). Plaintiffs posit that a rule relieving roughly forty percent of institutions otherwise subject to reporting requirements, from these obligations is too broad to be considered an “exception,” and further that HMDA only authorizes CFPB to create exceptions for classes of transactions, not for classes of institutions. See Pls.’ Mem. at 40–44. 11 Second, plaintiffs contend that promulgation of the 2020 Rule is so arbitrary and capricious that it “fails to effectuate the purposes of HMDA” and thereby falls outside the scope of HMDA’s § 305(a), which cabins the exercise of the agency’s authority “to effectuate the purposes of . . . or to facilitate compliance with” the statute. See id. at 39–40 (quoting 12 U.S.C. § 2804(a)). CFPB counters that it acted within the scope of its statutory authority when promulgating the 2020 Rule, because the text of HMDA “does not unambiguously foreclose [CFPB’s] interpretation” of the statute, Def.’s Opp’n at 46, and CFPB’s interpretation—authorizing it “to 11 Plaintiffs further argue that CFPB’s general rulemaking authority under the Dodd-Frank Act, see 12 U.S.C. § 5512(b)(1), is insufficient to justify the broad exemptions of the 2020 Rule. See Pls.’ Mem. at 44–45. Although the 2020 Rule states that it was issued “pursuant to [CFPB’s] authority under the Dodd-Frank Act and HMDA,” 2020 Rule, 85 Fed. Reg. at 28367, the agency has clarified that, “in amending Regulation C’s reporting thresholds,” CFPB “relied solely on its authority under HMDA,” Def.’s Opp’n at 45 n.17. Accordingly, plaintiffs’ arguments regarding CFPB’s general rulemaking authority under the Dodd-Frank Act are not addressed further. 31 provide an exception for the classes of transactions made by institutions whose prior loan activity fell below certain thresholds”—is reasonable, id. at 45. Despite the force of plaintiffs’ arguments that the 2020 Rule tests the limits of CFPB’s statutory authority to promulgate rules to effectuate HMDA’s purposes, “prevent circumvention or evasion” of those purposes, or “facilitate compliance” with HMDA, 12 U.S.C. § 2804(a), this Court concludes that the challenged rule falls within the broad grant of rulemaking authority assigned to CFPB under the statute. To understand the scope of an agency’s authority to promulgate rules, “we consider the statute’s text, structure, and context,” Truck Trailer Mfrs. Ass’n, Inc. v. EPA, 17 F.4th 1198, 1201 (D.C. Cir. 2021), starting by “apply[ing] ordinary tools of statutory construction to determine ‘whether Congress has directly spoken to the precise question at issue,’” Air Transp. Ass’n of Am., Inc. v. Dep’t of Agric., 37 F.4th 667, 672 (D.C. Cir. 2022) (quoting Merck & Co., Inc. v. Dep’t Health & Hum. Servs., 962 F.3d 531, 535 (D.C. Cir. 2020)). “At step one, we ask ‘whether the agency-administered statute is ambiguous on the precise question at issue,’” Nasdaq Stock Mkt. LLC v. SEC, 38 F.4th 1126, 1135 (D.C. Cir. 2022) (quoting Eagle Pharms., Inc. v. Azar, 952 F.3d 323, 330 (D.C. Cir. 2020)), and “[i]f the statute unambiguously resolves the question, that is the end of our inquiry,” Sault Ste. Marie Tribe of Chippewa Indians v. Haaland, 25 F.4th 12, 17 (D.C. Cir. 2022); see also Cal. Cmtys. Against Toxics v. EPA, 928 F.3d 1041, 1053 (D.C. Cir. 2019). “If not, we assume at step two that the ‘Congress has empowered the agency to resolve the ambiguity’ and accordingly defer to the agency’s interpretation so long as it is a reasonable construction of the statute.” Nasdaq Stock Mkt. LLC, 38 F.4th at 1135 (quoting Util. Air Regul. Grp. v. EPA, 573 U.S. 302, 315 (2014)). The “precise question at issue” in this case is whether HMDA § 305(a) permits CFPB to create an “exception” to 32 HMDA’s collection and reporting requirements for a “class” of up to “40% of otherwise covered financial institutions.” Pls.’ Mem. at 40; Def.’s Opp’n at 45. Based on the unambiguous meaning of the relevant terms, including “exception” and “class of transactions,” and the operational terms in this provision, HMDA § 305(a) authorizes CFPB to promulgate rules along the lines of the 2020 Rule. Even dispensing with the Chevron framework, which the parties invoke, and “[u]sing a statutory interpretation lens,” CFPB has “offered the best construction of the statute,” and thus the Court adopts the agency’s understanding. Guedes v. Bureau of Alcohol, Tobacco, Firearms & Explosives, 45 F.4th 306, 313 (D.C. Cir. 2022); cf. Am. Hosp. Ass’n v. Becerra, 142 S. Ct. 1896, 1906 (2022) (rejecting agency’s interpretation of governing statute “after employing the traditional tools of statutory interpretation”). “In addressing a question of statutory interpretation, [courts] begin with the text.” City of Clarksville, Tenn. v. FERC, 888 F.3d 477, 482 (D.C. Cir. 2018); see also Eagle Pharms., Inc., 952 F.3d at 330 (“Of the tools of statutory interpretation, ‘[t]he most traditional tool, of course, is to read the text.’” (quoting Engine Mfrs. Ass’n v. EPA, 88 F.3d 1075, 1088 (D.C. Cir. 1996), alteration in original)). Generally, when a statutory term is undefined, courts look first to “that term’s ‘ordinary, contemporary, common meaning.’” Food Mktg. Inst. v. Argus Leader Media, 139 S. Ct. 2356, 2362 (2019) (quoting Perrin v. United States, 444 U.S. 37, 42 (1979)); see also Board of Cnty. Comm’rs v. Fed. Hous. Fin. Agency, 754 F.3d 1025, 1028–29 (D.C. Cir. 2014) (“[W]here a statute’s terms are undefined, our interpretation is guided by the terms’ ‘regular usage.’”) (quoting Lopez v. Gonzales, 549 U.S. 47, 53 (2006))). Here, Congress did not explicitly define “exception” or “class of transactions” as used in HMDA’s § 305(a), but the “traditional tools of statutory interpretation,” Eagle Pharms., Inc., 952 F.3d at 330, provide the plain meaning of both terms. 33 First, the ordinary meaning of the phrase “class of transactions” as used in § 305(a) is neither elusive nor ambiguous. The ordinary meaning of “class” is “[a] set or category of things having some related properties or attributes in common, grouped together, and differentiated from others under a general name or description; a kind, a sort.” OXFORD ENG. DICTIONARY, https://www.oed.com/view/Entry/33874?rskey=JZctov&result=1&isAdvanced=false#eid (last visited Sept. 22, 2022). When that definition of “class” is combined with “transactions,” as used in § 305(a), the meaning is that the transactions encompassed within a class must share some identifiable common attributes. CFPB’s interpretation of the phrase “class of transactions” comports with this understanding: here, the “related propert[y]” or common “attribute[]” of the “category of things” is that all of these loans were made (or loan applications were denied) by a small-volume lending institution—specifically, a lending institution that was responsible for fewer than 100 closed-end mortgage loans or 200 open-end lines of credit in each of the two preceding years. See Def.’s Opp’n at 46–48; Def.’s Reply Supp. Def.’s Cross-Mot. Summ. J. (“Def.’s Reply”) at 22–24, ECF No. 23. A “class of transactions” may plainly be imbued with, and defined by, the features of the institution that issues them. Plaintiffs’ attempts to place artificial limits on this statutory phrase, by arguing “the 2020 Rule is not an ‘exception for [a] class of transactions,’” but instead “an exemption for institutions,” Pls.’ Reply Supp. Mot. Summ. J. & Opp’n Def.’s Cross-Mot. Summ. J. (“Pls.’ Opp’n”) at 25, ECF No. 21 (quotations omitted), are unavailing. See NYC C.L.A.S.H., Inc. v. Fudge, No. 20-5126, 2022 WL 3694872, at *2 (D.C. Cir. Aug. 26, 2022) (denying challenge to HUD’s authority to promulgate the challenged rule where “the plain language of the statute encompasses the Rule,” which therefore “lies within the statute’s grant of authority” to HUD). 34 Likewise, “exception” is defined as “[s]omething that is excepted; a particular case which comes within the terms of a rule, but to which the rule is not applicable; a person or thing that does not conform to the general rule affecting other individuals of the same class.” OXFORD ENG. DICTIONARY, https://www.oed.com/view/Entry/65724?rskey=v5rOd6&result=1&isAdvanced=false#eid (last visited Sept. 22, 2022); see also Exception, BLACK’S LAW DICTIONARY (11th ed. 2019) (“Something that is excluded from a rule’s operation.”); Exception, MERRIAM-WEBSTER’S UNABRIDGED DICTIONARY, https://unabridged.merriam-webster.com/unabridged/exception (last visited Sept. 22, 2022) (an “exclusion or restriction (as of a class, statement, or rule) by taking out something that would otherwise be included”). These definitions provide no express limit on how “big” or “broad” the exclusion can be to qualify as an “exception,” but the implication is that an exception applies less broadly than a general rule, by “taking out” what would “otherwise be included” by default. Contrary to plaintiffs’ assertions, see Pls.’ Mem. at 40–44, even a regulation relieving roughly forty percent of institutions from data collection and reporting requirements is an exception to the “rule” of disclosure, which continues to apply to the majority of institutions. As used in § 305(a), then, CFPB’s authority to exercise its judgment to create “adjustments or exceptions” allows the agency to exempt broad classes of covered entities, although not so broadly as to gut the statutory definition of “covered persons” and put those obliged to comply with the HMDA reporting requirements in the minority. See Def.’s Reply at 21 (acknowledging that the agency’s “authority is not limitless,” as CFPB must “ensur[e] that there is sufficient publicly available HMDA data to effectuate HMDA’s purposes”). The 2020 Rule does not convert the general reporting rule to a minority rule applicable only to a small percentage of “covered persons,” but instead “preserve[s] HMDA’s reporting requirements, as 35 compared to the 2015 Rule, for most institutions, the vast majority of loans, and the vast majority of communities.” Id. at 25. Thus, plaintiffs’ challenge to the breath of coverage of the reporting-threshold exemption fails. 12 Plaintiffs next argue that the 2020 Rule exceeded CFPB’s statutory authority because it “fails to effectuate the purposes of HMDA,” and thereby falls outside the scope of HMDA § 305, Pls.’ Mem. at 39–40, given that “[e]xempting institutions from compliance with HMDA does not ‘facilitate compliance’ with HMDA,” Pls.’ Opp’n at 25. CFPB asserts that the 2020 Rule is consistent with HMDA’s purposes by “maintaining such a large volume of data” on both types of loans, Def.’s Opp’n at 51, and furthers the statute’s goals “by potentially enabling smaller institutions to focus on lending activities and serving their communities,” Def.’s Reply at 20 (alterations omitted), and by “reducing compliance burdens for only the smallest financial institutions, who find reporting the most onerous, and, in fact, a deterrent to lending in their communities,” id. at 21. CFPB’s efforts to explain how the goals of a sunshine statute like HMDA are furthered by a rule newly exempting forty percent of institutions—all of which hold at least $50,000,000 in assets—from any HMDA public disclosure requirements, beyond even what Congress had so recently deemed appropriate in EGRRCPA, are on shaky ground. Yet, the authority to enact rules to further the availability of data about mortgage lending necessarily implies some ability to calibrate those rules to maximize the benefits to all. If HMDA only permitted CFPB to ratchet up data collection and reporting requirements, without any ability to relieve some burden on 12 As for plaintiffs’ argument that the 2020 Rule exceeded CFPB’s statutory authority because HMDA distinguishes between “exceptions (applying to classes of transactions) and exemptions (applying to classes of institutions),” see Pls.’ Opp’n at 29–30 (emphasis in original), the Court agrees with the agency that “HMDA’s plain text shows that Congress has not ascribed the rigid definitions that Plaintiffs claim,” Def.’s Reply at 23, given the overlap of usage in 12 U.S.C. § 2810 (referring to “mortgage loans exempted under section 2803(g) of this title” (emphasis added)), and because plaintiffs have not rebutted the general presumption against “giv[ing] great weight to statutory headings,” Holland v. Williams Mountain Coal Co., 256 F.3d 819, 822 (D.C. Cir. 2001). 36 lending institutions when the agency’s well-reasoned judgment deems such relief necessary—for example, to prevent certain lending institutions from leaving the market en masse—the effectiveness of this tool for achieving the statutory goals would be undermined. More importantly, the raw effects of the new thresholds on overall compliance with HMDA are minimal, since data on 98% of the closed-end transactions that were reported and 95% of the open-end transactions that would have been reported under the 2015 Rule continue to be reported under the 2020 Rule, Def.’s Opp’n at 1, and the new rule still provides coverage of 86% of the entire market for closed-end loans and 84% of the total market for open-end lines of credit, see 2020 Rule, 85 Fed. Reg. at 28393–94, 28398. The fact that the 2020 Rule affects such a comparatively small percentage of transactions means it can hardly be said to have created a nefarious loophole so large as to swallow the rule of reporting, or otherwise undermine the statutory scheme. Instead, in this numbers comparison, CFPB’s interpretation of HMDA § 305(a) “to make exceptions for ‘any class of transactions’ that [CFPB] judges to be necessary and proper” as permitting it to “authoriz[e] exceptions for the small percentage of transactions that are made by institutions that originated relatively few loans in each of the two preceding years,” Def.’s Opp’n at 50 (emphasis in original), does not exceed the agency’s statutory authorization. 13 Nonetheless, the statutory authority to create exemptions along the lines of those in the 2020 Rule does not fully vindicate the agency’s decisionmaking reflected therein. Plaintiffs’ questions regarding the extent to which the 2020 Rule was appropriately attuned to HMDA’s 13 Regarding plaintiffs’ argument that CFPB’s interpretation of its “exception” authority unreasonably conflicts with the “careful statutory balance” struck by EGRRCPA, see Pls.’ Mem. at 42–43, this argument fails because plaintiffs are unable to point to any move by Congress, in amending certain provision of HMDA in 2018, to amend § 305 or otherwise limit CFPB’s authority to create exceptions for classes of transactions. “[R]epeals by implication are not favored.” Rodriguez v. United States, 480 U.S. 522, 524 (1987). The extent to which CFPB adequately accounted for EGRRCPA in promulgating the 2020 Rule, is a different matter, one best addressed via an arbitrary-and-capricious challenge to the rulemaking. See Section III.B. infra. 37 goals and CFPB’s rationales for adoption of the new thresholds are more appropriately addressed under an arbitrary and capricious analysis. As discussed next, CFPB’s explanations and justifications for the 2020 Rule fare less well under this framework, at least with regard to the revised threshold for closed-end mortgage loans. B. Plaintiffs’ Challenge to the 2020 Rule as Arbitrary and Capricious and Contrary to Law Turning from CFPB’s authority to issue the 2020 Rule to the agency’s promulgation process, plaintiffs challenge the changes in loan-volume reporting thresholds as arbitrary and capricious, specifically raising several concerns regarding the cost-benefit analysis underlying this rule. An agency acts arbitrarily and capriciously when it “entirely fail[s] to consider an important aspect of the problem,” offers an explanation for its decision that “runs counter to the evidence before the agency,” or “is so implausible that it could not be ascribed to a difference in view or the product of agency expertise.” Am. Bankers Ass’n, 934 F.3d at 663 (quoting State Farm, 463 U.S. at 43); see also Nasdaq Stock Mkt., LLC, 38 F.4th at 1135 (“Besides adhering to statutory and regulatory requirements, the [agency] is required to ‘examine the relevant data and articulate a satisfactory explanation for its action including a rational connection between the facts found and the choice made.’” (quoting State Farm, 463 U.S. at 43)). While this “is not a high bar, . . . it is an unwavering one.” Judulang v. Holder, 565 U.S. 42, 45 (2011). Set against these applicable standards, plaintiffs’ challenges, as arbitrary and capricious, to the 2020 Rule’s revised reporting threshold for open-end lines of credit and increased reporting threshold for closed-end mortgage loans are discussed next, with differing conclusions reached as to the different types of lending involved. The 2020 Rule put in place the 200-loan volume threshold for open-end lines of credit and effectively lowered the loan-volume threshold from the temporary 500-loan mark in effect from 2017 through 2022, thereby actually expanding 38 the universe of lenders required to report this HMDA data. As such, comparing the rates of coverage and data loss under the 2020 Rule to the open-end reporting threshold under the 2015 Rule, which never actually took effect, is of limited significance. Plaintiffs have not met their burden of showing that CFPB, in revising the open-end lines of credit threshold, “failed to consider an important aspect of the problem,” or otherwise offered an explanation for its decision that “runs counter to the evidence before the agency,” or is “implausible,” Am. Bankers Ass’n, 934 F.3d at 663, or otherwise demonstrated that the 2020 Rule was arbitrary and capricious on this aspect of the regulatory change. As to the 2020 Rule’s increased reporting threshold for closed-end mortgage loans, however, the Court agrees with plaintiffs that the cost-benefit analysis was arbitrary and capricious. CFPB exaggerated the savings to “covered persons” under the new rule, and did not engage appropriately with the nonquantifiable “harms” of the 2020 Rule, and the disparate impact of those harms on the traditionally underserved populations HMDA is intended to protect, even as it conceded the revised threshold would certainly result in some harm to consumers. Given that the benefits of raising the closed-end reporting threshold were inflated, and its costs inadequately addressed, CFPB’s conclusion that the benefits of doing so outweighed the harms was arbitrary and capricious and grounds for vacatur of this aspect of the 2020 Rule. 14 14 As discussed supra in Part I.A.9, the 2020 Rule variously cites the estimated cost savings to “covered persons” under the 2020 Rule as $6.4 million and $11.2 million, a fact plaintiffs seize upon to claim that this rule relied on “erroneous projected cost savings” to an extent that “requires vacatur and remand.” Pls.’ Mem. at 21; see also 2020 Rule, 85 Fed. Reg. at 28374 (citing savings for depository institutions under revised 100-loan threshold for closed-end mortgage loans as $11.2 million per year); id. at 28383 (citing same estimated cost savings for nondepository institutions); id. at 28392 (citing savings for financial institutions under revised 100-loan threshold for closed-end mortgage loans as $6.4 million per year). While the use of contradictory estimates in the rule is emblematic of the sloppiness that pervades CFPB’s reasoning in the 2020 Rule, these “clerical errors,” Def.’s Opp’n at 17, are not, themselves, grounds for vacatur. First, the 2020 Rule “repeatedly emphasized” that the Dodd-Frank Act cost-benefit analysis in Part VII, which relied on the $6.4 million figure, “contained the most detailed and comprehensive analysis of the Bureau’s cost estimates,” id. at 17–18, and actually demonstrated how the agency calculated the $6.4 million figure, see 2020 Rule, 85 Fed. Reg. at 28396, lending support to the conclusion that CFPB in fact relied on the $6.4 million figure in preparing its cost-benefit analysis, rather than the $11.2 million figure, for which the agency offered no explanation as to how this number was calculated. Moreover, unlike in the 39 The 2020 Rule’s effect on open-end lines of credit is discussed first, followed by this rule’s treatment of closed-end mortgage loans, after a preliminary discussion of the parties’ use of numbers. 1. Parties’ Use of Numbers At the outset, a brief review of the parties’ use of numbers is helpful in framing their dispute. As previously noted, HMDA implementing regulations distinguish between two types of loans: closed-end mortgages and open-end lines of credit. According to CFPB’s own estimates, of the 4,860 financial institutions required to collect and report data on closed-end mortgage loans under the 2015 Rule, 1,700 of these institutions are now completely exempted under the 2020 Rule, without any public access to data on any of the approximately 112,000 closed-end mortgage loans and loan applications these institutions process each year. 2020 Rule, 85 Fed. Reg. at 28393. While this complete exemption is a boon to the financial institutions, the vast majority of these lenders—1,630 out of the 1,700 institutions exempted by the 2020 Rule— already qualify for partial, statutory exemptions under EGRRCPA, id., and thus would only be required to report basic demographic information on the mortgage loans and applicants in the absence of the 2020 Rule. cases on which plaintiffs rely, the correct figures were not totally absent from the 2020 Rule, making less clear that CFPB “relied on” incorrect data, as in N.C. Wildlife Fed’n v. N.C. Dep’t of Transp., 677 F.3d 596, 602–04 (4th Cir. 2012), nor did these errors create any of the same reliance concerns at issue in Am. Wild Horse Pres. Campaign v. Perdue, 873 F.3d 914, 924 (D.C. Cir. 2017), where all concerned parties had relied on what the U.S. Forest Service termed “an administrative error” in the official map of a wild horse territory since the 1980s. In contrast to the “error” in that case, the erroneous savings estimations in the 2020 Rule were corrected less than a year later by CFPB’s Notice of Correction, and have not been “reconfirmed repeatedly,” let alone “by two decades of agency practice and official pronouncements.” Here, CFPB “would have made the same decision absent its errors,” Hermes Consol., LLC v. EPA, 787 F.3d 568, 579 (D.C. Cir. 2015), given that in the more comprehensive analysis of Part VII of the 2020 Rule, the agency relied on the same figures now claimed to be accurate, and the logic underlying the agency’s analysis in Part V, in which the incorrect figure was cited, still applies when the correct numbers are substituted. While the mismatched savings estimates in the 2020 Rule undermine confidence that the rule was the result of a careful and comprehensive decision process, this discrepancy, standing alone, does not rise to the level of “a serious flaw undermining [the] analysis,” Nat’l Ass’n of Home Builders v. EPA, 682 F.3d 1032, 1040 (D.C. Cir. 2012), to such an extent as to require vacatur. 40 As for open-end lines of credit, of the 1,014 institutions that would have been subject to the reporting requirements under the 2015 threshold—which, as discussed above, was raised temporarily to 500 loans before the rule took effect—401 institutions are completely exempted from the disclosure requirements under the 2020 Rule, and data on their 69,000 lines of credit will not be made public. Id. at 28399. Of these 401 institutions, 378 of them already qualify for the same partial, EGRRCPA exemptions. Id. The parties take different approaches to these numbers: plaintiffs object to the 2020 Rule’s higher reporting thresholds for exempting 35% to 40% of lending institutions that otherwise would have been required to collect and report data on their lending practices. Pls.’ Opp’n at 1. 15 CFPB defends the 2020 Rule’s thresholds on the grounds that 95% to 98% of the data collected under the previous thresholds will continue to be reported, since the newly fully exempted lenders occupy such a small portion of the overall market. Def.’s Reply at 1. The main justification for the 2020 Rule (as well as for the 2015 Rule) proffered by CFPB is to decrease the burdens of HMDA reporting on “small” institutions. Yet, since HMDA’s original passage, the data collection and reporting requirements have only applied to the portion of the total market consisting of larger lending institutions. Truly “small” institutions, with total assets below an annually-adjusted amount, currently set at $50,000,000, are already entirely exempted by statute from collecting and reporting HMDA data. See 12 U.S.C. § 2808(a). Both the 2015 and 2020 Rules give scant attention to this built-in statutory 15 In their opening brief, plaintiffs state that raising the closed-end mortgage loan threshold will exempt “47% of financial institutions that would otherwise be required to report on mortgage loans” while setting the open-end threshold at 200 loans will exempt “39.5% of institutions that would be required to report on lines of credit.” Pls.’ Mem. at 1. As CFPB correctly notes, the reduction in the number of covered institutions under the increased reporting threshold from 4,860 closed-end reporters to 3,160 reporters results in “a decrease of 35%, not 47%.” Def.’s Opp’n at 46 n.18 (citing 2020 Rule, 85 Fed. Reg. at 28394). In reply, plaintiffs adopt the correct percentages. See Pls.’ Opp’n at 1 (describing the 2020 Rule as “exempting approximately 35% of financial institutions that make mortgage (closed-end) loans and 39.5% of institutions that issue open-end lines of credit” from HMDA reporting requirements). 41 exemption that Congress already provided for small lenders. See 2015 Rule, 80 Fed. Reg. at 66137 n.85 (citing 18 U.S.C. § 2808(a) only in reference to ability to regulate “for-profit mortgage-lending institutions”); id. at 66150 n.157 (same, in reference to nondepository institutions); 2020 Rule, 85 Fed. Reg. at 28380 n.124 (same). The focus in this case is thus on the next tier of financial institutions, which are large enough to hold more than $50,000,000 in assets, with a sufficiently robust mortgage lending practice to issue more than 25 closed-end mortgages or 100 open-end lines of credit each year, but nonetheless for which the collection and reporting requirements for HMDA data amounted to such a noticeable burden that the 2020 Rule was necessary for amelioration. As discussed more fully below, rather than look to gross transaction numbers as CFPB does, plaintiffs’ focus on the financial institutions newly exempt is more probative of the impact of the 2020 Rule on the ability of regulatory agencies, policymakers and HMDA users to identify problematic home lending practices at a community level, in furtherance of the statute’s express goals. 2. Open-End Lines of Credit Reporting Threshold is Not Arbitrary and Capricious Although the parties treat the 2020 Rule as raising the loan-volume reporting thresholds for both types of loans covered under Regulation C, see, e.g., Pls.’ Mem. at 20; Def.’s Opp’n at 1, the 2020 Rule in fact represents the lowest loan-volume threshold ever applied to open-end lines of credit. As such, plaintiffs’ comparison of the 2020 Rule to the 2015 Rule, which was amended before taking effect, fails to render the 2020 Rule arbitrary and capricious. Given the positive effects of the 2020 Rule on the availability of HMDA data on open-end lines of credit, plaintiffs’ challenge to the Rule on these grounds is unpersuasive. As discussed in Part I.A.4. supra, the requirement that lenders collect and submit HMDA data on their open-end lines of credit began with the 2015 Rule, which exempted institutions 42 issuing fewer than 100 open-end lines of credit each year, and gave lenders until January 1, 2018, to prepare for the new reporting requirements. 2015 Rule, 80 Fed. Reg. at 66128. The 100-loan reporting threshold set by the 2015 Rule never actually took effect. In 2017, CFPB temporarily increased the loan-volume threshold to 500, citing “concerns that the open-end threshold at 100 transactions [was] too low,” 2017 Rule, 82 Fed. Reg. at 43088, as well as the agency’s realization that, due to the growth in the number of dwelling-secured, open-end lines of credit in the preceding years, the 100-loan open-end threshold would affect significantly more institutions than estimated in the 2015 Rule, id. at 43094. The 2017 Rule’s temporary threshold of 500 loans extended until January 1, 2022. See Home Mortgage Disclosure (Regul. C), 84 Fed. Reg. 57946, 57946 (Oct. 29, 2019) (to be codified at 12 C.F.R. pt. 1003) (extending temporary 500-loan threshold from 2017 Rule). CFPB’s rulemaking on open-end lines of credit reporting between the 2015 and 2020 Rules thus means that the 2020 Rule’s revised open-end reporting threshold effectively lowered the reporting threshold for open-end lines of credit, for the first time, to 200 loans. To be sure, absent the 2020 Rule, the loan-volume threshold for this type of loan would have “reverted” to 100 loans in 2020, see 2017 Rule, 82 Fed. Reg. at 43088, but this “reversion” would in fact break new ground. The suggestion, therefore, that the 2020 Rule changed the status quo by creating “exempt[ions] [for] more institutions from the disclosures mandated by statute,” Pls.’ Mem. at 20, or leading to a “loss of data” on open-end lines of credit, id. at 21, is not reflective of reality. Plaintiffs’ argument that the 2020 Rule’s departure from a standard that never actually took effect was arbitrary and capricious, or in fact, harmful, is simply not persuasive. When the 2020 Rule’s loan-volume threshold took effect at the beginning of this year, well after this lawsuit was filed, more financial institutions—i.e., all lenders issuing between 200 43 and 499 open-end lines of credit in each of the two preceding years who were otherwise eligible—became newly subject to HMDA’s data collection and reporting requirements. 16 Information on how those burdens have affected the lenders, and the value of the new data reported under the 2020 Rule to plaintiffs and the public is not before the Court, nor could it have had any effect on CFPB’s decisionmaking in promulgating the 2020 Rule. The 2020 Rule broadened the applicability of the reporting requirements for open-end lines of credit beyond the temporary thresholds set in 2017. Consequently, any suggestion that the 2020 Rule actually diminished the availability of HMDA data on open-end lines of credit defies logic. Plaintiffs’ frustration with the revised threshold for not matching a threshold that never actually took effect, does not mean that that revised threshold was arbitrarily and capriciously established. In response to CFPB’s asserted justifications for setting the permanent threshold for open-end lines of credit at 200 loans, plaintiffs contend that CFPB unreasonably failed to respond to comments suggesting that “an increased number of lenders engaged in a higher volume of a kind of lending activity that the agency previously acknowledged was particularly high-risk and prone to be abusive suggested that it was more important to capture data from these lenders, not less.” Pls.’ Opp’n at 15–16 (citing 2015 Rule, 80 Fed. Reg. at 66148–50; 66157–62, emphasis in original). In deciding to reduce the open-end reporting threshold from the temporary 500-loan limit down to the permanent threshold of 200 lines of credit, however, CFPB cited the need for “monitoring of the potential risks, as noted by many commenters, that could be associated with such loans.” 2020 Rule, 85 Fed. Reg. at 28379. CFPB incorporated this risk into its decisionmaking in opting for a permanent loan-volume threshold that required more reporting 16 This is true despite the passage of EGRRCPA, and the impact of its partial exemptions from HMDA reporting requirements on the availability of HMDA data. Plaintiffs’ argument that the agency failed to account adequately for “the impact of the EGRRCPA,” Pls.’ Mem. at 28, is therefore also unpersuasive. 44 than under the regime in effect from 2017 to 2022, even if the permanent loan-volume threshold demanded less reporting than that contemplated in the 2015 Rule, which never actually took effect and for which there are no data. Based on the record before the Court, plaintiffs have not demonstrated that CFPB, in revising the open-end reporting threshold, “relied on factors which Congress ha[d] not intended it to consider,” “entirely failed to consider an important aspect of the problem,” or that its proffered explanations “run[] counter to the evidence before the agency” or are “so implausible that [they] could not be ascribed to a difference in view.” Am. Bankers Ass’n, 934 F.3d at 663 (quoting State Farm, 463 U.S. at 43). As such, the revised loan-volume threshold for open-end lines of credit under the 2020 Rule cannot be considered arbitrary and capricious. 3. Closed-End Mortgage Loan Reporting Threshold is Arbitrary and Capricious The Dodd-Frank Act requires CFPB, when exercising its rulemaking authority, to “consider the potential benefits and costs” of a regulation to both consumers and “covered persons”—here, the lending institutions—and, in conducting this analysis, to account for (1) “the potential reduction of access by consumers to consumer financial products or services resulting from such rule”; (2) the impact on “covered persons”; and (3) “the impact on consumers in rural areas.” 12 U.S.C. § 5512(b)(2)(A). Plaintiffs complain that CFPB did not adequately address or account for these statutorily required considerations. According to plaintiffs, in considering the 2020 Rule’s “benefits” to “covered persons,” CFPB overstated the benefits of increasing the closed-end loan volume threshold by failing meaningfully to consider several relevant factors highlighted in comments, and relying on erroneous projected cost savings. Pls.’ Mem. at 21. In assessing the 2020 Rule’s “costs” to consumers, the agency “barely discussed the loss of data that would result from the Rule” and “failed to respond to the detailed discussion of resulting 45 harms identified by commenters,” in violation of the statutory requirement that the agency “consider the impacts of its actions on consumer access to credit, and the specific impact on consumers in rural communities.” Id. CFPB’s compliance with the statutory directions in conducting the requisite analysis of the “benefits” of increasing the closed-end loan volume reporting threshold and the concomitant “costs” falls so short as to render this aspect of the 2020 Rule arbitrary and capricious. a. The 2020 Rule’s Benefits to Covered Persons Analysis The benefit of the 2020 Rule overwhelmingly driving the agency’s decision to increase the reporting threshold was “the reduction of the costs to covered persons relative to the compliance costs [they] would have to incur” under the already-increased threshold of the 2015 Rule. 2020 Rule, 85 Fed. Reg. at 28389. By quadrupling the closed-end reporting threshold to 100 loans, CFPB calculated that the annual savings to newly exempted financial institutions would be $6.4 million for closed-end mortgage lenders, id. at 28392, or between $6 and $42 per loan (when the median total loan costs per closed-end mortgage in 2018 were $6,056), id. at 28397. Plaintiffs challenge this relatively small per-institution estimated cost savings as exaggerated, asserting that CFPB failed to respond adequately to comments in the record calling “into question how meaningful the savings would be.” Pls.’ Mem. at 24. Moreover, plaintiffs argue these asserted “savings” to covered institutions ring hollow in light of other legal obligations requiring lenders to collect the same HMDA data for other purposes, and where, just two years earlier, Congress directly altered, but purposely left intact, pre-Dodd Frank Act HMDA data reporting requirements for lower-volume lenders. Id. at 24–26. The Court agrees that the “benefits” analysis underlying the 2020 Rule exaggerated the savings from the increased 46 reporting threshold, and thus the benefits of raising them, and that the agency’s failure to engage with comments raising these concerns was arbitrary and capricious. 17 i. Failure to Account for Data Collection for Non-HMDA Purposes Plaintiffs’ question about “the true savings to such institutions” newly excluded from HMDA’s reporting requirements under the 2020 Rule, Pls.’ Opp’n at 11, is well-placed because CFPB nowhere disputes that these “institutions will still be required to maintain most HMDA data to comply with other statutory and regulatory requirements,” such as the Truth in Lending Act and the Real Estate Settlement Procedures Act, to acquire loan insurance from the Federal Housing Administration, and for underwriting purposes, Pls.’ Mem. at 24–25. See, e.g., 12 C.F.R. §§ 1026.19(e); 1026.37 (listing data points that creditors are required to disclose for mortgage transactions subject to these other statutes, approximately a dozen of which points overlap with the HMDA data reporting requirements under 12 C.F.R. § 1003.4(a), as summarized in Section I.A.3. supra). Other comments in the record make the more obvious 17 Although plaintiffs do not make this argument, CFPB’s analysis of the benefits to consumers is also flawed. Based on comments submitted to the 2019 NPRM, the agency posited that the 2020 Rule could be a boon to consumers because smaller financial institutions, relieved of the burden of collecting and reporting these data, would pass the savings on to consumers and/or increase their presence in the mortgage market. See 2020 Rule, 85 Fed. Reg. at 28391; see also Def.’s Opp’n at 34 (repeating this reasoning); Def.’s Reply at 1, 20 (same). While the 2020 Rule cites no specific comments for this proposition, two comments in the Joint Appendix make this argument, but without in any way offering any definitive statement that such smaller financial institutions would actually increase their presence in the market in response to the increased threshold, or attesting that the only impediments to increasing their home mortgage lending are the data collection and reporting requirements. See AR at 2332 (comment from community bank stating “[e]xcessive regulation and the onerous costs of HMDA data collection prevent small banks from doing what they do best – serving existing or new customers who otherwise might not have access to banking, much less a first mortgage”); id. at 2670 (comment from “multi-bank financial holding company” stating that “[i]ncreasing the threshold will also spur more mortgage-loan lending by community banks by freeing those institutions that have opted not to fully engage their local home-lending demand in light of the costs and compliance requirements that they would face if they exceeded the low current threshold of only 25 loans”). The benefits to consumers cited by the agency, therefore, appear to be speculative. To the extent CFPB relied on this “benefit” to consumers in deciding to increase the reporting thresholds—without any empirical data or even broader survey data—such reliance was arbitrary and capricious. See Sorenson Commc’s Inc. v. FCC, 755 F.3d 702, 708–09 (D.C. Cir. 2014) (“Though an agency’s predictive judgments about the likely economic effects of a rule are entitled to deference, . . . deference to such . . . judgment[s] must be based on some logic and evidence, not sheer speculation” and an agency’s failure “to articulate a satisfactory explanation for its action” renders “promulgation of the . . . Rule arbitrary and capricious.” (quoting State Farm, 463 U.S. at 43, alterations in original)). 47 point that data on the identity of the mortgage recipient and the terms of the mortgage itself “should be collected as a matter of sound banking practices,” including “Know Your Customer” (“KYC”) anti-money laundering requirements, and that, “given modern technological advancements,” submitting this information “is unlikely to require substantial resources,” 2020 Rule, 85 Fed. Reg. at 28370, let alone oblige the lender to seek out additional information on mortgages and mortgagors beyond that data already in its possession. CFPB’s only counter is that plaintiffs have “focus[ed] too narrowly on one sentence” in the Dodd-Frank Act cost-benefit analysis—namely, the statement that CFPB “recognized that much of the information required for HMDA reporting is information that financial institutions would need to collect, retain, and secure as part of their lending process, even if they were not subject to HMDA reporting,” Pls.’ Mem. at 25 (quoting 2020 Rule, 85 Fed. Reg. at 28391)— “without considering the entire administrative record, including the description of the Bureau’s methodology” as detailed “more fully” in the 2015 Rule and “incorporated by reference” into the 2020 Rule, Def.’s Opp’n at 26. While “an agency’s obligation to address specific comments is not particularly demanding,” Ass’n of Priv. Sector Colls. & Univs. v. Duncan, 681 F.3d 427, 449 (D.C. Cir. 2012) (quotation omitted), “significant comments” must be addressed “in a reasoned manner,” FBME Bank Ltd. v. Mnuchin, 249 F. Supp. 3d 215, 222 (D.D.C. 2017) (quoting Reytblatt v. Nuclear Regul. Comm’n, 105 F.3d 715, 722 (D.C. Cir. 1997)), to allow the court “to see what major issues of policy were ventilated and why the agency reacted to them as it did,” Del. Dep’t of Nat. Res. & Env’t Control v. EPA, 785 F.3d 1, 15 (D.C. Cir. 2015) (internal alterations and citation omitted). An agency’s “failure to address . . . comments, or at best its attempt to address them in a conclusory manner, is fatal to its defense,” Ass’n of Priv. Sector Colls. & Univs., 681 F.3d at 449 (citation omitted), as the agency’s robust participation in the 48 notice-and-comment process is crucial to fulfilling the APA’s goals of “ensur[ing] that affected parties have an opportunity to participate in and influence agency decision making at an early stage,” Nat’l Ass’n of Clean Water Agencies v. EPA, 734 F.3d 1115, 1148 (D.C. Cir. 2013) (quoting N.J., Dep’t of Evn’t Prot. v. EPA, 626 F.2d 1038, 1049 (D.C. Cir. 1980)). CFPB’s response to these critical comments within the 2020 Rule itself does not satisfy this standard. Citing comments “that much of the data reportable under HMDA must be collected for other rules . . . and ordinary underwriting standards,” 2020 Rule, 85 Fed. Reg. at 28370, CFPB responded summarily that the comments did “not undermine the Bureau’s approach” to the cost-benefit analysis because, when the agency had “estimat[ed] compliance costs associated with HMDA reporting” in the 2019 NPRM, it had done so recognizing “that much of the information required for HMDA reporting is information that financial institutions would need to collect, retain, and secure as part of their lending process, even if they were not subject to HMDA reporting,” id. at 28390–91. With this circular reasoning in place, CFPB did “not believe that the comments received” in response to the 2019 NPRM “provide[d] a basis for departing from the approach for analyzing costs and benefits.” Id. at 28391. Yet, nowhere in the 2019 NPRM does CFPB mention reporting obligations under the Truth in Lending Act or Real Estate Settlement Procedures Act, or otherwise explain how it dissociated the costs of fulfilling HMDA obligations from the costs of complying with other, overlapping regulations. While fathomable that HMDA reporting might create specific burdens not imposed by other reporting requirements, CFPB’s failure to articulate or measure those unique burdens, beyond opaque references to the 2019 NPRM’s “framework,” and the “burdensome” tasks of “scrubbing data, training personnel, and preparing for HMDA-related examinations,” id. at 28390, does not 49 satisfy its duty to “address significant comments” in a “reasoned manner,” FBME Bank, Ltd., 249 F. Supp. 3d at 222. Moreover, CFPB made no effort to rebut the more general comment that the vast majority of the HMDA data institutions are required to report is so fundamental to a swathe of business operations that a lender would collect this information for its own purposes. See 2020 Rule, 85 Fed. Reg. at 28390. The agency’s lack of explanation for how its savings calculations accounted for the continued need to collect data for other purposes leaves the Court without confidence that the closed-end loan-volume threshold in the 2020 Rule was developed “based on a consideration of” the full spectrum of “relevant factors.” Inv. Co. Inst. v. CFTC, 720 F.3d 370, 377 (D.C. Cir. 2013) (quoting Ctr. for Auto Safety v. Peck, 751 F.2d 1336, 1342 (D.C. Cir. 1985))). Further attempts by CFPB to explain its response to these comments in this litigation create no greater certainty. The agency claims that, even if its direct response within the 2020 Rule to the comments questioning the savings is insufficient, “the entire administrative record, including the description of the Bureau’s methodology, which was more fully detailed in the 2015 Rule and incorporated by reference” into the 2020 Rule, demonstrates that the agency adequately accounted for the potential for overlapping requirements between HMDA and other statutes and regulations. Def.’s Opp’n at 26 (citing New Lifecare Hosps. of Chester Cnty. LLC v. Azar (“New Lifecare Hosps.”), 417 F. Supp. 3d 31, 44 (D.D.C. 2019)). Specifically, the agency points to a section of the 2015 Rule where, in analyzing the Rule’s impact, the agency discussed the “alignment of data fields to existing regulations or industry data standards,” id. (quoting 2015 Rule, 80 Fed. Reg. at 66287, emphasis omitted), accounting for “whether analogous data existed in the Uniform Loan Delivery Dataset” (“ULDD”) or the Mortgage Industry Standards 50 Maintenance Organization (“MISMO”), 2015 Rule, 80 Fed. Reg. at 66287. 18 Where a datapoint required under HMDA was “align[ed] with ULDD or MISMO” data, the agency assigned it a lesser reporting cost to the financial institution in the 2015 Rule analysis. Id. at 66289. The 2020 Rule “relied on the same framework for calculating its compliance cost estimates as it did in the 2015 Rule,” with “some updates,” and thus, CFPB argues, the “cost savings estimates in the 2020 Rule reflect how some data points are required by other existing statutes and regulations.” Def.’s Opp’n at 26–27 (citing 2020 Rule, 85 Fed. Reg. at 28390). CFPB’s reliance on the 2015 Rule to justify the 2020 Rule is, at a minimum, ironic, given that the agency concluded in the earlier rule that setting the closed-end mortgage loan reporting threshold any higher than 25 loans “would have a material negative impact on the availability of data about patterns and trends at the local level,” 2015 Rule, 80 Fed. Reg. at 66147, and “substantially impede” the ability of the public and public officials in affected communities “to understand access to credit” at a local level, id. at 66148. Even with “some updates” to the data from 2015, Def.’s Opp’n at 27, the agency fails to explain how its reliance “on the same framework for calculating its compliance cost estimates as [CFPB] did in the 2015 Rule,” id. at 26, reasonably led to such a different conclusion, namely, that a threshold loan number quadruple the previous figure would not “substantially impede” the goals of HMDA or otherwise harm consumers. That CFPB looks to the 2015 Rule to defend the reasoning of the 18 ULDD is a “common set of data elements required by Fannie Mae and Freddie Mac for single-family loan deliveries.” Uniform Loan Delivery Dataset, FANNIE MAE (Dec. 14, 2021), https://singlefamily.fanniemae.com/delivering/uniform-mortgage-data-program/uniform-loan-delivery-dataset. MISMO, a “subsidiary of the Mortgage Bankers Association,” is the “standards development body” of the mortgage industry and maintains “a common language for exchanging information for the mortgage finance industry.” MISMO FAQs, MISMO (last accessed Sept. 22, 2022), https://www.mismo.org/about-MISMO/mission-andvision/mismo-faqs. 51 2020 Rule, therefore, only lends further support for the conclusion that the 2020 Rule was capriciously promulgated. 19 Putting aside the irony of CFPB relying on the cost analysis in the 2015 Rule, which explicitly rejected the result reached by the agency in the 2020 Rule, the record gives no indication that either Rule reasonably accounted for the specific regulations cited repeatedly by commenters on the 2019 NPRM as placing overlapping collection obligations on lenders, even accounting for some analogous data collection needs, including data collected in the ULDD or MISMO. See supra n.18 and accompanying text. As for the agency’s assertion that the 2020 Rule “incorporated by reference” the 2015 Rule’s accounting for costs, Def.’s Opp’n at 26, such argument is spurious where a central premise the agency gave for the new rulemaking was that, “[w]ith the benefit of . . . new data to supplement the Bureau’s analyses, the Bureau [was] now in a better position to assess both the benefits and burdens of the reporting required under the 2015 19 At the same time, the Court is not convinced by plaintiffs’ argument that the 2020 Rule is arbitrary and capricious simply because the rule reflects “a 180-degree reversal of the agency’s conclusion[s] in the 2015 Rule,” Pls.’ Mem. at 20, citing, in particular, that the 2015 Rule concluded that “a loss of 20% of reportable data in 385 census tracts would substantially impede the public’s and public officials’ ability to understand access to credit in their communities,” Pls.’ Opp’n at 21 (quotations omitted), compared to the 2020 Rule’s conclusion “that a loss of 20% of reportable data . . . in 1,200 census tracts would” not, id. at 21; see also Pls.’ Mem. at 20 (relatedly, 2020 Rule’s general conclusion “that exempting more institutions from the disclosures mandated by statute would reduce the burden on those institutions, and that this reduction in burden outweighs the harms caused by depriving the public of the information that would otherwise be disclosed,” contrasted with contrary conclusion in 2015 Rule). Agencies may change course and do so without needing to demonstrate “to a court’s satisfaction that the reasons for the new policy are better than the reasons for the old one.” FCC v. Fox Television Studios, Inc. (“Fox”), 556 U.S. 502, 515 (2009) (emphasis in original). Moreover, the burden of justification is no greater when an agency is changing course than when it is exploring new waters. See Am. Farm Bureau Fed. v. EPA, 559 F.3d 512, 521–22 (D.C. Cir. 2009) (“[I]f the relevant facts have changed or the [agency] has reasonably made a different policy judgment, then it need only explain itself and we will defer.”); Nasdaq Stock Mkt., LLC, 38 F.4th at 1141 (describing an agency’s duty to “acknowledge and explain its departure from past policy” as “not [an] especially high bar” to clear (quoting Sw. Airlines Co. v. FERC, 926 F.3d 851, 856 (D.C. Cir. 2019))). Nevertheless, when the “new policy rests upon factual findings that contradict those which underlay its prior policy,” the agency must provide “a reasoned explanation . . . for disregarding facts and circumstances that underlay or were engendered by the prior policy.” Fox, 556 U.S. at 515–16. Here, although CFPB pointedly acknowledged that the 2020 Rule reached different conclusions than those reflected in the 2015 Rule about the consumer harms and benefits from a 100 closed-end loan threshold, the agency offered no new factual findings for the new policy, and instead relied on the previous analysis in the 2015 Rule even though the two rules support dramatically different conclusions. CFPB’s changed policy poses no per se problem under the APA, but rather the agency’s insufficient factual and analytical support for the changed policy falls short of what is necessary on this record. 52 HMDA Rule,” 2020 Rule, 85 Fed. Reg. at 28371. Unlike in New Lifecare Hosps., the single case on which the agency relies, CFPB may not rely on its incorporation of past reasoning in lieu of a direct response to comments, because here “nearly identical comments” were not ventilated and addressed in that prior reasoning. See 417 F. Supp. 3d at 44. CFPB’s response in the 2020 Rule to commenters’ skepticism—essentially, “don’t worry, we thought of that five years ago”— is not well “reasoned” when a different result was reached five years ago and instead suggests that the concerns were simply brushed past. 20 Remand to CFPB is therefore necessary for a more reasoned explanation as to whether and how the cost-benefit analysis accounted for the ongoing need to collect data on home mortgages pursuant to other statutory requirements and underwriting purposes, and why, when a lender must collect and report multiple data points for each mortgage and loan application, the marginal cost of collecting the additional, HMDAspecific data points is so significant that the increased reporting threshold of the 2020 Rule renders unique cost savings. 20 Plaintiffs also contend that CFPB’s savings analysis for both types of loans relied too heavily on “aggregate figures, completely divorced from context,” obscuring that the costs saved per institution were insignificant. Pls.’ Mem. at 26. Since HMDA only covers depository institutions with a minimum of $50 million in assets, see supra n.3 (citing 2022 Adjustment to Asset-Size Exemption), plaintiffs argue that the institutions affected by the rule change are all sufficiently large that “a cost savings of $2,000 per year is less ‘meaningful,’” even if the institution “issues only a small number of closed-end loans,” Pls.’ Mem. at 26. CFPB acknowledged, however, that “costs vary by institution due to many factors, such as size, operational structure, and product complexity,” and made efforts to account for per-institution costs by developing “three representative tiers of financial institutions” and producing “a series of compliance cost estimates for each representative tier.” Def.’s Opp’n at 30 (quoting 2020 Rule, 85 Fed. Reg. at 28389–90); see also id. at 23–25. Although no financial institution covered under HMDA today can reasonably be thought of as “tiny,” given the requirement that covered institutions hold at least $50 million in assets, CFPB rightly notes that “total asset size does not necessarily relate to the size of that institution’s mortgage origination business or compliance operations.” Def.’s Opp’n at 32. Instead, “regardless of their asset size,” lenders classified as low complexity, Tier 3 institutions under CFPB’s hierarchy “generally have few HMDA-compliance staff, do not have automated systems, and are making fewer than one mortgage loan a day each year,” id., giving greater comparative effect to a savings of even a few thousand dollars per year. CFPB’s decision to focus on “the complexity of institutions’ HMDA-compliance operations in assessing the estimate cost savings and overall impact of the Rule,” id., was a permissible approach. The fact that plaintiffs might have adopted a different approach to the cost-benefit analysis focused on the overall size of the lending institution does not render CFPB’s analysis invalid. See Nicopure Labs, LLC v. FDA, 266 F. Supp. 3d 360, 407 (D.D.C. 2017) (“While plaintiffs would surely have assessed the various costs and benefits in a different manner, the Court does not have the power to take up the agency’s analysis de novo.”). 53 ii. Failure to Account for Effects of EGRRCPA The 2020 Rule was promulgated just two years into “a post-EGRRCPA world,” 2020 Rule, 85 Fed. Reg. at 28389, a new reality with which, plaintiffs allege, CFPB failed to reckon fully. Highlighting comments that opposed the new rule on the grounds that the objective of “reducing lender costs[] has already been largely achieved by . . . EGRRCPA,” AR at 2970 (first comment from plaintiff NCRC), and that in the absence of the 2020 Rule, “the vast majority of lenders that make fewer than 500 closed-end loans or 500 open-end loans . . . [would] report essentially the same data they have collected and reported for decades,” id. at 3765 (second comment from plaintiff NCRC), plaintiffs argue that the agency “gave no indication how or that it took” the impact of EGRRCPA “into account when analyzing the benefits of the Rule,” Pls.’ Mem. at 26. CFPB, in turn, argues “the administrative record shows that [it] adequately accounted for the impact of the EGRRCPA in estimating the compliance costs and resulting savings to institutions that would be exempted by an increase in HMDA’s reporting thresholds.” Def.’s Opp’n at 29. The agency’s accounting was far from “adequate,” however, and its failure to reckon with the huge shifts in the reporting landscape following the enactment of EGRRCPA— and with the contradictions between the 2020 Rule’s threshold and the balance struck by Congress between burdens of reporting and the value of disclosure—constitute a fundamental “fail[ure] to consider an important aspect of the problem.” State Farm, 463 U.S. at 43. The agency’s decision to revisit the congressional balance so recently after Congress had made a recalibration, using largely data collected prior to congressional action, is not well-reasoned, and an arbitrary and capricious exercise of its rulemaking authority. As discussed in Part I.A.6. supra, enactment of EGRRCPA to “promote economic growth, provide tailored regulatory relief, and enhance consumer protections,” EGRRCPA § 1, 54 significantly altered the HMDA statutory scheme. This legislative action was intended to strike the right balance between promoting “fair lending” and “meaningfully reduc[ing] substantial costs imposed on small lenders from HMDA data collection,” Examining the Federal Reserve’s Semiannual Report to Cong. on Monetary Pol’y and the State of the Economy: Hearing Before the S. Comm. on Banking, Hous., and Urb. Affs., 115th Cong. 35 (Mar. 1, 2018) (statement of Sen. Heitkamp, Member, S. Comm. on Banking, Hous., and Urb. Affs.), despite concern that the partial exemptions for “85 percent of banks” went too far, and would lead to “entire communities in America where there will be no data whatsoever, which means there will be no ability to monitor whether people are getting cheated,” 164 CONG. REC. S1363 (daily ed. Mar. 6, 2018) (statement of Sen. Warren). The EGRRCPA amendments to HMDA substantially lightened the reporting burden on a large swathe of smaller loan-volume financial institutions, reducing by more than half the number of data points a depository institution was required to collect and disclose when the institution had issued fewer than 500 mortgage loans in each of the past two years. EGRRCPA § 104(a). Post-EGRRCPA, qualifying institutions were required to report data largely limited to information about the loan’s “number and dollar amount,” as well as the mortgagor’s “census tract, income level, racial characteristics, age, and gender.” See 12 U.S.C. § 2803(b)(4). At the same time, EGRRCPA reflects Congress’s continued belief in the importance of the “sunshine” cast by HMDA: Congress chose not to relieve these lower-volume lenders from reporting requirements entirely, but instead decided to continue eliciting basic data on the loan applications they were receiving, approving, and rejecting. Id.; see also Part. I.A.6. supra. Furthermore, when a lender’s behavior had been identified as problematic in the recent past—through receipt of a rating of “needs to improve record of meeting community credit needs” on each of its two 55 most recent Community Reinvestment Act examinations, or a rating of “substantial noncompliance in meeting community credit needs” on its most recent examination—Congress created an exception to the EGRRCPA exemption, requiring full reporting until the lender improved its Community Reinvestment Act rating. See 12 U.S.C. § 2803(i)(3). This exception to EGRRCPA effectively serves twin goals of allowing the public and public officials to keep a closer eye on riskier lenders, and incentivizing good behavior on the part of financial institutions, rewarding those who affirmatively comply with fair lending requirements with a lesser degree of scrutiny. In the wake of this congressional pronouncement as to the appropriate balance of the burdens and benefits of HMDA data disclosure, however, CFPB decided to take reduction in HMDA reporting far further than Congress allowed, without any well-reasoned explanation of how blanket exceptions for lending institutions issuing fewer than 100 closed-end mortgage loans furthered the goals of HMDA, as so recently amended by EGRRCPA. CFPB’s purported justification is unhelpful: addressing comments on this issue, the agency merely “recognize[d] that the estimated ongoing costs savings” were “less than they would have been absent the relief provided by the EGRRCPA,” but asserted, without more, that the “ongoing cost savings” to smaller institutions still would “provide meaningful burden reduction.” 2020 Rule, 85 Fed. Reg. at 28374. The agency had nothing to say about the Rule’s interference with Congress’s desire to maintain some data on lower-volume institutions, while substantially relieving their burden. CFPB rejected calls to increase the reporting threshold even higher to 500 loans “to harmonize the thresholds with the EGRRCPA provisions,” on the grounds that “[d]oing so would provide a complete exclusion from reporting all . . . data for institutions below the threshold of 500, even though Congress opted to provide only a partial exemption at the threshold of 500,” and “would 56 extend that complete exclusion to institutions that Congress did not include in even the partial exemption.” Id. at 28372. This same logic holds for the institutions exempted under the 100 closed-end threshold adopted under the 2020 Rule, none of which Congress thought should receive a complete exclusion from HMDA’s requirements, especially those most basic reporting requirements intended to ensure lenders were on the right track and to disincentivize redlining and other anti-consumer behavior. The agency’s haphazard reckoning with fact that 1,630 of the 1,700 lenders that would be newly exempt from reporting closed-end mortgage loan data, see id. at 28396, had just two years before had their reporting burdens significantly lightened through a careful congressional rebalancing, constitutes a serious “fail[ure] to consider an important aspect of the problem,” and is “so implausible” that it is difficult to “ascribe[] to a difference in view or . . . agency expertise,” Am. Bankers Ass’n, 934 F.3d at 663 (quoting State Farm, 463 U.S. at 43). While the increased threshold might have “effectuate[d] the purposes” of HMDA, 12 U.S.C. § 2804(a), before Congress’s latest pronouncement on the issue, the logic is elusive to say, as the agency did, that the 2020 Rule effectuated the purposes of HMDA as amended by EGRRCPA in 2018. Absent deeper engagement with the effects of EGRRCPA, the 2020 Rule’s increased closed-end volume threshold and resultant broadening of exempt institutions from any HMDA data reporting comes so close to running afoul of the agency’s rulemaking authority as to be arbitrary and capricious. See Farmers Union Cent. Exch., Inc. v. FERC, 734 F.2d 1486, 1501 (D.C. Cir. 1984) (While an agency “enjoys substantial discretion” in exercising its rulemaking authority, “this discretion must be bridled in accordance with the statutory mandate.”). CFPB’s response to comments that the increased loan volume threshold and concomitant expansion of fully exempt institutions would conflict with EGRRCPA, was that the costs savings 57 estimations accounted for EGRRCPA exemptions and thereby adequately accounted for EGRRCPA’s effects on the statutory scheme. See 2020 Rule, 85 Fed. Reg. at 28374 n.68. This wholly fails to engage in any significant way with the fundamental critique that the 2020 Rule undid the balance struck by Congress in EGRRCPA, in violation of the agency’s duty under the APA when engaging in notice-and-comment rulemaking. Where a statute is designed to provide transparency in business practices to enable enforcement of laws designed to bar discriminatory and risky lending practices, and where Congress has so recently carefully crafted a framework to maximize the universe of lending institutions required to report some data, while minimizing the extent of the burden of reporting on those institutions most likely to feel the brunt of it, CFPB’s decision to essentially undo Congress’s carefully selected balance with blanket exceptions for this share of the lending market without explanation is arbitrary and capricious. b. The 2020 Rule’s Costs to Consumers Analysis Turning to the costs to consumers of the revised closed-end reporting threshold, CFPB conceded, in the 2020 Rule itself, that the increased threshold would mean “the HMDA data available to serve HMDA’s statutory purposes w[ould] decline,” id. at 28392, therefore leading “to adverse outcomes for some consumers,” id. at 28397. Given all of the evidence suggesting the benefits of the 2020 Rule—in the form of savings to covered persons—are less than an estimated $6.4 million in annual savings, CFPB’s treatment of the 2020 Rule’s known harms to consumers is arbitrary and capricious. Plaintiffs persuasively identify two principle problems with the 2020 Rule’s costs analysis: first, that CFPB gave short shrift to the “nonquantifiable harms” resulting from the loss of coverage under the 2020 Rule; and second, that CFPB failed to consider and address the disparate impact of the “costs” of the 2020 Rule on rural and lower-income communities, despite 58 the statutory direction to do so. Both problems with the 2020 Rule’s cost-benefit analysis are addressed seriatim. i. Nonquantifiable Harms Plaintiffs argue that, in preparing the cost-benefit analysis for the 2020 Rule, CFPB arbitrarily and capriciously failed to account for certain harms to consumers posed by the 2020 Rule’s increased closed-end reporting threshold on the grounds that these “costs” were “nonquantifiable.” See Pls.’ Mem. at 30. The harms stemming from the increased loan-volume reporting threshold, all of which were brought to CFPB’s attention in comments on the 2019 NPRM, include (1) a decrease in HMDA’s “deterrent effect on predatory and discriminatory practices” and local authorities’ “enforcement activity,” leading to an “increase in predatory lending”; (2) the compromise of public and private actors’ “ability to assess whether credit and housing needs are being met”; (3) diminished accuracy and rigor to “Community Reinvestment Act and fair lending enforcement monitoring”; and (4) an impaired ability of organizations “to do representative and accurate data analyses.” Id. at 31. According to plaintiffs, the agency’s “refusal to address these harms solely because it lacked quantifiable data is arbitrary and capricious” and “contrary to the CFPB’s obligations” under the Dodd-Frank Act. Id. CFPB takes the different view that the 2020 Rule “discussed—qualitatively—the impact that increasing the reporting threshold[] would have on consumers.” Def.’s Opp’n at 36 (citing 2020 Rule, 85 Fed. Reg. at 28392, 28397, 28402). Indeed, on each occasion in the 2020 Rule that CFPB “discussed” the costs to consumers, the agency conceded that the 2020 Rule would lead to a decline in “the HMDA data available to serve HMDA’s statutory purposes,” and acknowledged a concomitant increase in the potential for “abusive and discriminatory lending.” 2020 Rule, 85 Fed. Reg. at 28392. Further consideration of these acknowledged harms was deflected, however, with the excuse that it “lack[ed] sufficient data to quantify th[ose] costs” or 59 otherwise measure the 2020 Rule’s impact. Id. at 28398; see also id. at 28392 (stating agency was “unable to readily quantify the loss of some of the HMDA benefits to consumers with precision, both because the Bureau does not have the data to quantify all HMDA benefits and because the Bureau is not able to assess completely how this final rule will reduce those benefits”). Put another way, CFPB conceded the new rule would cause identifiable harms to the public, but effectively threw up its proverbial hands, citing an inability to incorporate these harms into its analysis as quantifiable “costs,” and moved on to the next topic of discussion. CFPB’s treatment of the nonquantifiable harms stands in sharp contrast to the methodology used in the 2015 Rule. There, when confronted with the same sorts of harms, which did not lend themselves to “reliable quantitative estimates,” 2015 Rule, 80 Fed. Reg. at 66262, CFPB supplemented the discussion of the potential harm to consumers without HMDA data reporting by smaller-volume institutions with examples of how these data had been used in the past “to analyze access to credit at the neighborhood level and to target programs to assist underserved communities and consumers” from Lawrence, Massachusetts, to Albuquerque, New Mexico, id. at 66280. While balancing numerical evidence with anecdotal evidence is, to a certain extent, an exercise in comparing apples to oranges, CFPB’s inclusion of past uses of these data for the public good helped flesh out the treatment of the harms, and gave substance to its statutory duty to consider the impact of these data on consumers. See Nicopure Labs, LLC v. FDA, 266 F. Supp. 3d 360, 406 (D.D.C. 2017) (affirming the agency’s treatment of a rule’s benefits that were “difficult to quantify” because in addition to explaining “the reasons why quantification was not possible,” the agency “provided substantial detail on the benefits of the rule”). 21 21 CFPB’s reliance on Am. Council of the Blind v. Mnuchin in support of the proposition that “agencies are not required to consider costs quantitatively,” see Def.’s Opp’n at 36 (citing Am. Council of the Blind v. Mnuchin, 60 CFPB insists that the agency’s treatment of nonquantifiable costs complied with its statutory duties, as the Dodd-Frank Act “requires only that the Bureau consider benefits and costs,” Def.’s Opp’n at 36 (emphasis in original), as if mere lip service mentioning the matter suffices. Taking “consider” to mean “[t]o view or contemplate attentively, to survey, examine, inspect, scrutinize,” OXFORD ENG. DICTIONARY, https://www.oed.com/view/Entry/39593?redirectedFrom=consider#eid (last visited Sept. 22, 2022), CFPB’s superficial attention to the costs to consumers of the decrease in HMDA data hardly fulfills the obligation for full consideration. Mere reference to an identified cost does not qualify as an effort to “consider,” when that effort lacks discussion or analysis of possible avenues to develop assessment tools for measuring or surveying the cost, which tools might include survey data of users of the data, review of academic research or even press reporting. Contrary to this simplified view of what an agency actually must do to consider an issue, an agency is not “excuse[d] . . . from its statutory obligation to determine as best it can the economic implications of the rule it has proposed,” even where it is “without a reliable basis for determining” the precise costs of its action. Chamber of Com. v. SEC, 412 F.3d 133, 143 (D.C. Cir. 2005) (citation omitted). 22 977 F.3d 1, 8 (D.C. Cir. 2020)), is misplaced. There, plaintiffs argued that the Secretary of the Treasury’s estimation of a delay in providing court-ordered relief was not fully explained because it lacked quantitative support, Am. Council of the Blind, 977 F.3d at 7, but the D.C. Circuit rejected this argument, finding the Secretary’s estimate was sufficiently “concrete” that additional “quantitative support” was unnecessary, id. at 8. Here, plaintiffs do not allege that the 2020 Rule is contrary to law because of insufficient reliance on quantitative data, but because—and the Court agrees—the agency’s superficial engagement with qualitative data in the absence of convincing quantitative data was insufficient. 22 Defendant relies on Lindeen v. SEC, 825 F.3d 646 (D.C. Cir. 2016), for the proposition that an agency conclusion that a rule may have some effect is “sufficient for [an agency] to have considered the benefits of the rule.” Def.’s Reply at 13. While a “single paragraph” explaining how the agency considered “a factor [it] must consider under its organic statute” may suffice to comply with its statutory obligations, and understanding that an agency need not show that a new rule “will actually mitigate” a potential harm, Lindeen, 825 F.3d at 657–58 (emphasis in original), CFPB’s failure to engage in the 2020 Rule with the qualitative harms identified by commenters and conceded by the agency falls short of complying with the most basic standard of “reasoned analysis.” 61 The minimal weight given to these consumer harms in CFPB’s cost-benefit analysis is more troubling in the context of HMDA’s express purposes to (1) “help determine whether financial institutions are serving the housing needs of their communities;” (2) “assist public officials in distributing public-sector investment so as to attract private investment to areas where it is needed;” and (3) “assist in identifying possible discriminatory lending patterns and enforcing antidiscrimination statutes.” 12 C.F.R. § 1003.1(b)(1). These goals do not lend themselves to numerical analysis, and yet the agency is still explicitly instructed to exercise its rulemaking authority in a way that furthers these goals. See id. When the impact of a cost or benefit is “both evident and massively significant,” that cost or benefit “cannot be dismissed without further inquiry,” even if its value is “unquantified.” Md. People’s Couns. v. FERC, 761 F.2d 768, 776 (D.C. Cir. 1985). As plaintiffs attest in their individual declarations, the loss of HMDA data “that would have been submitted under the 2015 HMDA Rule but will not be submitted under the 2020 HMDA Rule,” Van Tol Decl. ¶ 7, is both already “evident and massively significant,” Md. People’s Couns., 761 F.2d at 776; see also Bean Decl. ¶ 5 (stating data lost under 2020 Rule would have been “of particular interest to [plaintiff] MFH because many lenders in rural areas of Montana are likely to be below the reporting thresholds in the 2020 Rule”); Crosby Decl. ¶ 3 (stating plaintiff City of Toledo, Ohio was intending to use these data “to assess which areas of Toledo are most in need of public investment to ensure adequate access to credit”); Williams Decl. ¶ 7 (stating intent of plaintiff ANHD to use these lost data “to conduct research into unfair lending practices and access to credit throughout New York City”). The 2020 Rule’s failure to “further consider[]” the ongoing value of these data, Md. People’s Couns., 761 F.2d at 777, even while repeatedly acknowledging that this data loss came at a harmful “cost” to consumers, was arbitrary and capricious and 62 contrary to law. HMDA is clear that CFPB must exercise its authority with careful consideration of the effects on consumers; CFPB gave short shrift to that requirement here, with its failure to engage with the nonquantifiable “costs” of the 2020 Rule to the same extent it dealt with the benefits and other, quantifiable costs. ii. Disparate Impacts The 2020 Rule’s costs analysis was further deficient, plaintiffs explain, “because the agency addressed only how much data was lost at the aggregate level,” while ignoring the question of “which data would be lost,” which is “necessary to fully see the loss of data from the 2020 Rule.” Pls.’ Mem. at 34 (emphasis in original). In this regard, plaintiffs point to the disproportionate impacts of the data losses on rural and low-to-moderate income census tracts, which is “particularly problematic since the impacted rural and lower-income areas are already underserved, underrepresented and underreported,” id. at 37 (quotation omitted), and on the “multi-family housing market,” id., which “has disproportionate effects, including in particular geographic areas and on particular racial and ethnic groups,” id. at 38. Moreover, credit unions themselves acknowledged, in their comments, that the increased closed-end threshold would exempt 750 credit unions, a significant portion of an industry that “disproportionately serve[s] low- and moderate- income homeowners.” Id. at 34; see also AR at 3697 (comment from association of credit unions). Plaintiffs argue that CFPB’s refusal to consider the disproportionate impact of data losses in these “very communities that HMDA and the DoddFrank Act were designed to protect,” is arbitrary and capricious. Pls.’ Mem. at 35. CFPB, in response, dismisses these arguments, asserting that the 2020 Rule “considered the amount of data los[t] at both the national and census tract level,” and reasonably concluded that the new threshold balanced the need for data from small institutions, and institutions serving low-to-moderate income and rural communities, with the desire to reduce the burdens of 63 compliance costs on newly exempted institutions. Def.’s Opp’n at 38. CFPB’s “consideration” of these discrepancies, however, in the face of a statute explicitly designed to combat invidious discrimination in the housing market, falls far short of its responsibilities under both HMDA and the APA. First, comparing the rates of data that would be lost under the new closed-end reporting threshold compared to that of the 2015 Rule, CFPB conceded that “rural and low-to-moderate income census tracts will lose proportionately more data as the threshold increases than other areas”—specifically, 3% of low-to-moderate income census tracts would lose more than 20% of their data and 5% of rural census tracts would lose more than 20% of their data, 2020 Rule, 85 Fed. Reg. at 28373—but again stated that it “lack[ed] sufficient data to quantify the impact of this decrease in data,” id. at 28403. As required under the Dodd-Frank Act, the 2020 Rule further included a specific discussion of the impact of the changes on consumers in rural areas, but discounted the new thresholds as having no “direct[] impact [on] consumers in rural areas,” in part, due to “[r]ecent research suggest[ing] that financial institutions that primarily serve rural areas are generally not HMDA reporters” because “the current asset and geographic coverage criteria already in place disproportionately exempt small lenders operating in rural communities” which “may represent one of the few sources of credit for many rural areas.” Id. While the 2020 Rule’s closed-end threshold may have only a small effect on overall “visibility into rural and low-to-moderate income tracts,” this agency finding bypasses the key issue of whether sufficient data remain in affected low-to-moderate income and rural areas to “permit the public and public officials to identify patterns and trends at the local level,” id., especially given that—as CFPB explicitly acknowledged, see id.—rural lenders are already disproportionately exempted from HMDA requirements. Put another way, given that so many small lenders operating in rural areas 64 already enjoy full exemptions from HMDA reporting due to the $50,000,000 asset qualifier, the 2020 Rule increased the opacity in understanding home lending practices in these areas yet further, thereby undermining an express statutory purpose, see 12 U.S.C. § 5512(b)(2)(A)(ii) (requiring CFPB, when exercising its rulemaking authority, to consider, inter alia, “the impact on consumers in rural areas.”), without any effort by the agency to measure or assess this impact. Moreover, regarding the disproportionate impact of the 2020 Rule on the multi-family housing market, CFPB estimated that 13% of multifamily loan transactions would no longer be reported under the 2020 Rule, but concluded that this decrease was “justified by the benefits of relieving smaller-volume institutions of the burdens of HMDA reporting.” 2020 Rule, 85 Fed. Reg. at 28374. The 2020 Rule makes no effort to explain why, or to acknowledge who would be most impacted by this loss. Likewise, nowhere in the 2020 Rule did the agency attempt to explain or defend the Rule’s effect on data from credit unions, despite comments bringing this issue to CFPB’s attention. CFPB is, of course, correct that “any loan-volume threshold will affect individual markets differently, depending on the extent to which smaller creditors service individual markets and the market share of those creditors,” id. at 28373, but who is affected matters, especially when an agency is implementing a statute put in place by Congress to remedy the failures of “some depository institutions,” on their own, “to provide adequate home financing to qualified applicants on reasonable terms and conditions,” 12 U.S.C. § 2801(a). Furthermore, when a market is dominated by small lenders, such as a rural market, or a market replete with low-to-moderate income census tracts, multifamily housing, and credit unions, a modest decrease in the number of covered institutions may leave the public, local officials, and housing advocates with a substantially blurrier picture of who is being able to access credit, and on what terms, than 65 a market which is generally served by more lenders of any and all sizes. The agency’s failure to engage further with the data that would be lost, despite acknowledging that the brunt of that loss would be borne disproportionately by the communities already least likely to have access to credit, represents a failure “to consider an important aspect of the problem” before the agency, and therefore was arbitrary and capricious. State Farm, 463 U.S. at 43. Yet again, as with the agency’s treatment of nonquantifiable costs of the increased threshold to consumers, CFPB acknowledged a problem and dismissed its significance without further consideration. The agency’s recognition of the disparate impact on protected groups, without any meaningful discussion of the issue in the context of alternatives to the 2020 Rule’s policy choices, and in the face of comments noting the disproportionate impact of the 2020 Rule on the very populations HMDA is designed to protect, points to the agency’s failure to “consider an important aspect of the problem” and “articulate a satisfactory explanation for its action,” id., and is grounds for remand of this portion of the 2020 Rule. See Am. Bankers Ass’n, 934 F.3d at 656, 668–71 (remanding, as arbitrary and capricious, part of agency rule intended to promote credit union growth, because agency did not adequately respond to comments objecting that lowincome and minority residents in urban core areas would be disparately impacted and excluded). 23 * * * To sum up, plaintiffs have accurately identified several flaws in the 2020 Rule’s cost- 23 CFPB’s consideration of the costs to consumers of the 2020 Rule is further arbitrary and capricious because, despite its statutory authorization to “provide for such adjustments and exceptions” to classes of transactions as “are necessary and proper to effectuate the purposes of [HMDA] . . . or to facilitate compliance therewith,” 12 U.S.C. § 2804(a), nowhere in the 2020 Rule does CFPB consider “adjusting” the collection and reporting requirements on smaller-volume lenders to reduce their burden, as opposed to fully “excepting” them from their previous disclosure obligations, which alternative could preserve more of the data available to the public and public officials. “State Farm teaches that when an agency rescinds 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 (alterations in original)). The agency’s failure to consider a less harmful alternative flies in the face of that obligation and confirms that the action taken as to closed-end mortgage loans is arbitrary and capricious. 66 benefit analysis supporting the increased closed-end mortgage loan threshold, rendering this aspect of the Rule arbitrary and capricious and requiring vacatur. In the “benefits to covered persons” discussion, CFPB failed to respond appropriately to comments questioning the magnitude of the estimated savings, and in the discussion of the “costs to consumers,” CFPB failed adequately to consider the nonquantifiable harms stemming from the loss of data under the heightened reporting threshold, and the disparate impact of that loss on traditionally “underserved” populations. Vacatur and remand of the closed-end mortgage loan reporting threshold under the 2020 Rule is appropriate to allow CFPB to address the flaws in the costbenefit analysis supporting the threshold. IV. CONCLUSION For the reasons set forth above, plaintiffs’ motion for summary judgment, ECF No. 14, is GRANTED IN PART and DENIED IN PART, and defendant’s cross-motion for summary judgment, ECF No. 18, is GRANTED IN PART and DENIED IN PART. Plaintiffs’ motion for summary judgment is granted to the extent that CFPB’s justification for the loan-volume reporting threshold for closed-end mortgage loans under the 2020 Rule is arbitrary and capricious, requiring that this aspect of the rule be vacated and remanded to CFPB for further proceedings consistent with this decision, but the motion is otherwise denied. CFPB’s crossmotion for summary judgment is granted as to the 2020 Rule’s treatment of open-end lines of credit, but is otherwise denied. An Order consistent with this Memorandum Opinion will be filed contemporaneously. Date: September 23, 2022 __________________________ BERYL A. HOWELL Chief Judge 67

Some case metadata and case summaries were written with the help of AI, which can produce inaccuracies. You should read the full case before relying on it for legal research purposes.

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.