From Broad Rule to Delayed Launch
Section 1071’s 2026 reset is easiest to understand as a legal and implementation lifecycle, not as a sudden decision to collect less information. This chapter traces how Congress created the reporting mandate, how the CFPB translated it into a broad 2023 rule, what the Texas courts actually did, and why the Bureau now treats the initial collection as a staged launch. The practical value of the timeline is precision: it separates statutory authority, agency design, judicial review, and procedural delay.
The original design: a statutory floor with a broad regulatory layer
Congress added Section 1071 to the Equal Credit Opportunity Act through the Dodd-Frank Act of 2010. The provision requires financial institutions to collect and report information about applications for credit to women-owned, minority-owned, and small businesses. Its purposes are not limited to enforcement after a loan is made. The data are meant to help enforce fair-lending laws and to help communities, governmental entities, and creditors identify business and community-development needs and opportunities.
That combination creates a design problem. A simple approval or denial count might show that two groups received different outcomes, but it would not necessarily show whether they applied for different products, sought different amounts, operated in different industries, or encountered different underwriting conditions. Congress therefore listed core information and authorized the CFPB to require additional data when the Bureau determines that the data would help fulfill Section 1071’s purposes.
The statutory language is the floor. The additional-data authority is the space in which the CFPB had to exercise judgment. The Bureau could ask not only, “What did Congress expressly require?” but also, “What context is needed to interpret those required fields?” That is why the dispute over the 2023 rule was not simply a dispute over the existence of the mandate. It was a dispute over the breadth of the first regulatory implementation.
The 2023 final rule, published in May of that year, used that discretionary space expansively. It added Subpart B to Regulation B and established a broad collection regime for small-business credit. The later district-court opinion described the rule as containing 20 rule-level data points, nine beyond the statutory set. The CFPB’s 2026 preamble identifies the discretionary additions as including pricing, denial reasons, application method, application recipient, number of workers, industry code, time in business, number of principal owners, and LGBTQI+-owned business status.
The rule also made broad choices about the population from which data would be drawn. It used a threshold of at least 100 covered credit transactions in each of two preceding years to define a covered financial institution, after increasing the original proposed threshold of 25. It defined a small business using annual gross revenue of $5 million or less. It did not initially exclude merchant cash advances or agricultural lending simply because their structures and underwriting differed from conventional loans. The intended result was a dataset with wide coverage and enough contextual fields to support analysis.
The numbers became a point of contention. The plaintiffs in the Texas case referred to 81 separate pieces of information. The Southern District of Texas explained that this presentation conflated data fields with data points. One data point can require several fields: a name, for example, may need separate first, middle, and last-name fields. That distinction does not make the compliance burden imaginary. It changes the nature of the claim. The question is not whether the rule created 81 separate statutory commands, but whether the CFPB reasonably converted a smaller set of statutory and discretionary categories into a system that lenders could implement accurately.
Implementation is where an apparently neat regulatory list encounters the lending market. A high-volume lender may collect standardized information through software. A community lender may depend on a loan officer who gathers information through several conversations, knows the applicant, participates in underwriting, and serves on a credit committee. A merchant cash advance may price and repay itself differently from a conventional loan. Agricultural credit may depend on crops, livestock, weather, seasonal income, and unusual collateral. Indirect lending may place the applicant in front of a dealer or vendor rather than the financial institution that ultimately sets material terms.
Those differences matter because data quality depends on the path by which a field is produced. A field can be legally required and still be difficult to define at the application stage. A field can be analytically valuable but unreliable if applicants interpret the question differently, if lenders cannot determine which interaction counts, or if the data omit the surrounding facts needed to interpret a price or denial. The CFPB’s later explanation of the 2026 rule repeatedly returns to this gap between the value of information in an established dataset and the cost of demanding it at the beginning of a new one.
What the Texas litigation did—and did not do
The litigation began soon after the 2023 rule was issued. The plaintiffs challenged the rule on constitutional and Administrative Procedure Act grounds. Their claims included an argument that the CFPB’s funding structure was unconstitutional, an argument that the Bureau exceeded its statutory authority, and arguments that the rule was arbitrary and capricious because the agency had not adequately addressed significant comments or the costs and effects of the collection regime.
The district court’s preliminary-injunction orders changed the implementation environment before the merits were resolved. In July 2023, the court granted relief to the plaintiffs and their members while declining at that point to extend it nationwide. After intervenors joined and sought the same protection, the court later extended the injunction to all covered financial institutions. A preliminary injunction is temporary relief designed to preserve the situation while a case proceeds; it is not a final determination that the challenged rule is unlawful.
A separate part of the legal story ended in 2024 when the Supreme Court rejected the challenge to the constitutionality of the CFPB’s funding structure. That decision removed the specific constitutional premise on which the district court’s original injunction had rested. The CFPB then issued an interim final rule extending compliance dates, and the parties proceeded to brief the remaining claims about the rule’s statutory authority and APA reasoning.
On August 26, 2024, the Southern District of Texas ruled on those then-live merits claims. It denied the plaintiffs’ motion for summary judgment and granted the CFPB’s cross-motion. In practical terms, the court held that the plaintiffs had not shown the 2023 rule to be unlawful under the arguments before it. The court concluded that Section 1071 authorized the CFPB to require additional data collection and that the Bureau had reasonably considered the rule’s costs, benefits, and comments.
The court’s reasoning illustrates the difference between policy disagreement and legal invalidity. Under the APA’s arbitrary-and-capricious standard, an agency must act reasonably and explain its decision, but a court may not simply replace the agency’s policy judgment with its own. The court therefore asked whether the CFPB had considered relevant issues and offered a rational explanation. It did not ask whether the broadest possible collection would have been the best policy.
That distinction shaped several issues. The plaintiffs argued that the rule would reduce access to credit, impose disproportionate costs on smaller lenders, and produce unreliable data. The court found that the CFPB had addressed those concerns in the administrative record, even if it had reached conclusions the plaintiffs disliked. The Bureau had discussed one-time and ongoing costs, the effects on small businesses, rural areas, and financial institutions, and the expected benefits of more complete information. The court’s conclusion was not that the 2023 design was guaranteed to work. It was that the agency had not failed its legal duty to consider and explain the choice.
The court’s analysis separated two kinds of challenge that can sound similar in ordinary language. Under Section 706(2)(C) of the APA, the question was whether the CFPB had acted beyond the authority Congress delegated. The plaintiffs argued that the additional-data provision was narrower than the Bureau understood it to be. The court read the statutory scheme as a whole. It noted that some information the statute expressly requires—such as a business’s location or revenue—must be collected even if a lender would not otherwise possess it during an application. That made it difficult to draw a categorical line between permissible collection under the statute and the additional data authorized by the Bureau. The court therefore treated the disputed additions as a question of delegated authority, not as a referendum on whether each field was a wise policy choice.
The arbitrary-and-capricious claims asked a different question: whether the Bureau had examined the relevant problems and supplied a reasonable explanation. The court did not require certainty about future lending behavior or a perfect cost model. It looked for engagement with the concerns that could affect the decision. The opinion points to the Bureau’s treatment of one-time system costs, ongoing reporting costs, rural effects, credit availability, and the expected benefits of more complete data. A disagreement with the Bureau’s assumptions, or a belief that another estimate was better, did not by itself show that the agency had ignored the issue. This is the legal boundary that makes the later policy reset possible: an agency can lawfully reconsider a rule that a court previously upheld, provided the new rule is supported by its own reasoned explanation.
These procedural categories also explain why the public timeline contains apparently conflicting signals. A preliminary injunction can pause implementation before a court has resolved the merits. Summary judgment resolves the legal claims on the record before the court. A stay pending appeal can pause an order or a compliance obligation for a defined group while the appeal continues. The Fifth Circuit’s February 2025 order did the last of these. It changed what the named plaintiffs and intervenors had to do while the appeal was pending, but it did not convert the district court’s merits ruling into a reversal. The 2026 rule must therefore be read as a later agency redesign in a changing legal environment, not as a court-drafted remedy.
The appellate order that followed was narrower than the district-court merits ruling. On February 7, 2025, the Fifth Circuit granted the Texas Bankers plaintiffs’ motion for a stay pending appeal. The order tolled the compliance deadlines, but only for the plaintiffs and intervenors in that case and pending further order. It did not announce that the rule was invalid. Indeed, the order expressly stated that nothing in it should be construed as a comment on the ultimate merits of the underlying litigation.
That procedural posture is easy to flatten into a misleading story: the rule was issued, a court struck it down, and the CFPB wrote a smaller replacement. The record does not support that sequence. The district court upheld the challenged claims on the merits; the Fifth Circuit paused compliance for a specified group without deciding the merits; and the CFPB continued to revise the rule through its own rulemaking process. Litigation delayed and complicated implementation, but it did not itself erase the statutory mandate or produce a merits ruling that compelled the 2026 design.
The sequence also explains why the 2026 rule gives unusual attention to compliance timing. The CFPB’s preamble recounts interim and final extensions in 2024 and 2025, the Fifth Circuit’s limited stay, and continued challenges in other jurisdictions. By the time the Bureau proposed revisions in 2025, institutions had faced a rule whose requirements, deadlines, and legal environment had shifted repeatedly. Some had begun preparing systems and forms. Others had not known whether they would ultimately be covered. A revised rule would itself require changes to preparation already undertaken.
The Bureau describes that history as part of a broader reconsideration. It points to reactions after the 2023 rule, implementation feedback about operational difficulty, ongoing litigation, and executive-branch efforts to reduce regulatory burden. It now says that Section 1071 should begin with core lenders, products, and data, then develop as the Bureau and industry learn from actual collection. That is an agency policy choice made through a new rulemaking, not a judicial command disguised as one.
A second timing distinction matters. Delayed compliance does not mean delayed importance. The purposes of Section 1071 remain the same even when the mechanism for pursuing them changes. The 2026 rule sets a single initial compliance date of January 1, 2028 for institutions that meet its revised coverage test, with a transition option for using an earlier look-back period to determine coverage. The delay gives remaining covered institutions time to alter their systems and workflows, but it also postpones the public evidence that would show whether the narrower design works.
The legal timeline therefore leaves the central empirical question open. The courts addressed authority, explanation, and procedural relief at particular moments. They did not measure whether the 2023 collection would have produced reliable data, whether it would have changed prices or access, or whether the 2026 collection will better reveal discrimination. Those questions belong to implementation and later review.
The next question is consequently concrete rather than doctrinal: who and what will enter the revised data stream? The answer is spread across several gates. Product type, transaction amount, lender volume, Farm Credit System status, and applicant revenue each remove some activity from the initial collection. Once those gates are applied, the headline claim that the CFPB is collecting “less” can be tested against the actual coverage that remains.