The Truth in Lending Act of 1968 stated the aim of federal disclosure law in a single chain of reasoning. Meaningful disclosure of credit terms would create consumer awareness of the cost of credit, awareness would produce the informed use of credit, and informed use would strengthen competition among the firms extending credit. Federal consumer credit law rested on that chain. Congress did not set the price of a loan, cap an interest rate, or forbid a term in 1968. Congress standardized the description of the transaction, through the finance charge and the annual percentage rate, and assigned the discipline of the market to borrowers equipped with a common measure. The promise was precise enough to test. A borrower who received the same price language from every creditor could compare offers, reject expensive credit, and reward cheaper credit, and competition would perform the regulatory work that price control performed under older state usury regimes.

Whether the aim was met is a question about outcomes rather than intentions, and the federal record assembled between 1968 and the end of 2018 permits an unusually direct answer. Four statutory regimes generated measurable evidence. The Truth in Lending Act generated studies of reading, comprehension, and comparison. The Fair Credit Reporting Act generated a congressionally ordered measurement of error inside the credit file. The Fair Debt Collection Practices Act generated a complaint record of large volume and limited evidentiary meaning. The Credit Card Accountability Responsibility and Disclosure Act generated account level evidence on the effect of prohibiting charges that disclosure had described for decades. The ability to repay rules for mortgages generated a further record in which contraction was plain and causation remained disputed.

The sequence in which Congress acted supplies the article’s central claim, named here as the tradition backwards. A legislature designing consumer credit law from evidence alone might have begun with conduct rules for practices that information could not reach, added substantive limits for pricing that attached after a contract was signed, and used disclosure as supporting infrastructure for comparison, supervision, and enforcement. Congress proceeded in the opposite order. Disclosure came first in 1968 and carried the entire federal policy for the price of credit. A reporting statute followed in 1970 and mixed disclosure with procedural duties on accuracy. Conduct prohibitions arrived for debt collection in 1977, in a field where a clearer statement of price had no plausible application to a debt already incurred. Substantive pricing rules arrived last, in 2009, after four decades in which card pricing had developed methods that formal disclosure described without restraining. The tradition built from the instrument that demanded the most from borrowers toward the instruments that demanded the most from firms.

The first two findings frame the contrast. Finding one examines disclosure against its own premise and records a consistent failure at the human links of the chain: mandated forms frequently went unread, the annual percentage rate was widely misunderstood as a measure of relative cost, and comprehension weakened as products grew more complex, a pattern documented from the Lee and Hogarth survey of 1999 through the Federal Trade Commission form testing of 2007 to the credit card experiments reported in 2018. Finding two examines the substantive counterexample inside the same statutory home. The CARD Act fee limits reduced borrowing costs by an annualized 1.6 percent of average daily balances, an estimated 11.9 billion dollars a year in the Agarwal, Chomsisengphet, Mahoney, and Stroebel study published in 2015, with the largest gains falling to borrowers with FICO scores below 660 and no measured offset in interest charges.

The remaining findings grade three different kinds of regime. Finding three measures accuracy inside the credit file through the Federal Trade Commission section 319 study published in 2012 and the followup report of 2015: 26 percent of participants identified a potentially material error, 21 percent obtained a correction after disputing, and 5 percent gained a score change large enough to move a credit tier, a record of high average accuracy joined to a concentrated minority burden. Finding four reads the collection complaint record with discipline: debt collection ranked as the largest complaint category handled by the Consumer Financial Protection Bureau in 2016 and second in 2017, and the totals measured allegations carried to a federal channel rather than adjudicated violations. Finding five addresses mortgage credit access after the ability to repay and qualified mortgage rules took effect in 2014: availability for weaker credit profiles stood far below pre crisis levels on the Urban Institute Housing Credit Availability Index, while the causal share of the rules stayed contested among Federal Reserve studies, lender surveys, and analyses that assigned greater weight to lender overlays and litigation risk.

A counter reading closes the evidence before the verdict. Disclosure performed weakly as a primary instrument and performed measurably better as supporting infrastructure. The Kleimann Communication Group validation testing of 2013 recorded consumers answering 28.8 percent more questions correctly with integrated mortgage forms, and the disclosure record sustained supervision, enforcement, and private litigation under the Truth in Lending Act regardless of whether a borrower had read a form on the day of signing. The sections that follow present the statutory design first, then the five findings in order, then the counter reading, and a closing synthesis states the conclusions the combined record supports, with ratings that separate settled measurements from contested attributions.

How Consumer Credit Law Was Supposed to Work

Federal consumer credit law was assembled in an order that inverted the logic of its own design. Congress legislated first for the problem that information could solve. The governing theory held that a borrower who knew the price of credit would shop as deliberately as a buyer comparing any other price, and that competition for informed borrowers would discipline prices and terms without federal price setting. Congress legislated last, after four decades of experience, for the problems information had not solved: collection conduct that depended on intimidation rather than misunderstanding, and card pricing that changed after the account was opened and after disclosure had been delivered. The statute book therefore records a premise adopted in 1968, extended in 1970, qualified in 1977, and revised in 2009.

Before any federal disclosure statute existed, state law governed the price of consumer credit, and state law governed through ceilings rather than through information. Usury statutes descended from colonial enactments that had carried a ceiling of 6 percent from British practice into American law, and by the twentieth century every state regulated the price of at least some categories of lending. Ceilings, penalties, and exemptions nevertheless differed sharply across states, with higher ceilings and weaker penalties where capital was scarce. A ceiling regime controlled price directly where it applied, but application was uneven: lenders restructured transactions to fall outside a given statute, legislatures carved out exceptions for particular classes of lender, and small loan legislation authorized licensed lenders to charge rates far above the general ceiling in exchange for supervision. By the 1950s and 1960s, protection depended on the legal category of a transaction rather than on anything a borrower could observe at signing.

State disclosure law was thinner and even less uniform. Retail installment sales acts in many states required a written contract and a statement of charges in dollars, and Massachusetts enacted a state truth in lending statute in 1966. No state regime imposed a single national unit of price, and a borrower crossing from an automobile installment contract to a revolving charge account to a small loan never encountered the price of credit expressed in the same language twice. The Federal Deposit Insurance Corporation, in the Truth in Lending chapter of the Consumer Compliance Examination Manual, summarized the position directly: consumers faced a bewildering array of credit terms and rates, and comparison was difficult because loans were seldom presented in the same format. The obstacle was not secrecy, since most creditors stated some figure for the charge, but the absence of a common measure for converting stated figures.

That absence was a concrete, mechanical problem. An add-on rate computed interest on the original balance for the full term, even as the balance declined, and a discount rate deducted interest in advance from the amount the borrower received. A monthly rate on a revolving account stated one and one-half percent per month, without the corresponding annual figure of 18 percent appearing anywhere in the transaction. President Lyndon Johnson, in remarks upon signing the Act on May 29, 1968, used the monthly rate as the illustrative case: a householder opening a charge account would be told the annual rate of 18 percent rather than left to compute one and one-half percent a month. The defect was one of translation rather than of honesty: the monthly figure was accurate, and it was useless for comparison.

The theory Congress built on that diagnosis appeared in the statutory findings. Section 102 of the Truth in Lending Act, codified at section 1601 of title 15 of the United States Code, declared that competition among firms extending consumer credit would be strengthened by the informed use of credit, that the informed use of credit resulted from consumer awareness of the cost of credit, and that the purpose of the title was to assure a meaningful disclosure of credit terms so that the consumer could compare available terms and avoid the uninformed use of credit. Competition was assigned the regulatory work, information was identified as the mechanism, and the federal intervention was defined narrowly as standardization of information itself. Anne Fleming, in The Long History of Truth in Lending, published in the Journal of Policy History in 2018, traced how that framing displaced an earlier framing in which disclosure had complemented usury law. Disclosure as a substitute for price control carries a heavier burden than disclosure as a supplement, and the 1968 settlement imposed the heavier burden.

Senator Paul Douglas of Illinois introduced the first federal credit price disclosure bill in 1960, and the eight years between introduction and enactment turned less on whether cost should be disclosed than on the unit in which cost had to be stated. Fleming documented that creditor witnesses generally accepted dollar disclosure while resisting the annual percentage rate, on the stated ground that households budgeted in dollars and did not reason in percentages. Douglas insisted on the annual percentage rate because a dollar figure without a rate and a term could not be compared across offers, and described the objective as removing the disguises that hid or distorted the true price of credit, locating the harm in presentation rather than in the level of price itself. Senator William Proxmire of Wisconsin carried the measure after Douglas left the Senate following the election of 1966, and the enacted text adopted the Douglas metric, the device supposed to convert separate offers into a single ranked market.

The Truth in Lending Act of 1968 was enacted on May 29, 1968, as Title I of the Consumer Credit Protection Act, Public Law 90-321, and took effect on July 1, 1969, through Regulation Z, issued by the Board of Governors of the Federal Reserve System. Creditors were required to disclose the finance charge and the annual percentage rate in uniform terminology, for closed end credit, such as installment loans, and for open end credit, such as revolving accounts, with timing, form, placement, and advertising also prescribed. A detailed treatment of that machinery appears in the Truth in Lending Act guide, and the design point for the federal sequence is the deliberate omission at the center of the statute. The Federal Deposit Insurance Corporation manual stated the omission plainly: the Act and Regulation Z did not tell financial institutions how much interest could be charged, and did not tell financial institutions whether a loan had to be granted. Federal law regulated the description of the transaction and left pricing substance to state usury law and to the discipline of informed comparison.

The second disclosure statute extended the same informational logic from the price of credit to the record of the borrower. By the late 1960s, consumer reporting agencies had built an elaborate mechanism for investigating and evaluating creditworthiness, credit standing, credit capacity, character, and general reputation, in the language of the congressional findings Congress later adopted. Reports assembled through that mechanism determined access to credit, insurance, and employment for individuals with no access to the contents and no procedure for correcting errors. Congressional hearings on the Fair Credit Reporting Act recorded cases in which inaccurate or unverifiable material, including material about personality and habits gathered through investigative reports, determined an adverse decision that the affected individual could neither inspect nor answer. Christopher Peterson, in Truth, Understanding, and High-Cost Consumer Credit, published in the Florida Law Review in 2003, placed the reporting statute in the same frame: mass borrowing had outrun the informational institutions supposed to make borrowing intelligible, and Congress regulated the flow and visibility of information rather than the underlying decisions.

The Fair Credit Reporting Act was enacted on October 26, 1970, as Title VI of Public Law 91-508, added Title VI to the Consumer Credit Protection Act, and took effect on April 25, 1971, after the statutory interval of 180 days. Section 602 of the added title, codified at section 1681 of title 15, found that the banking system depended on fair and accurate credit reporting and required agencies to exercise a vital role in assembling information with fairness, impartiality, and respect for privacy. The operative provisions converted the findings into procedural rights and duties: reports could be furnished only for permissible purposes defined in the statute; agencies were required to follow reasonable procedures to assure the maximum possible accuracy of reports; and obsolete information was barred after defined periods, generally seven years for adverse information and ten years for bankruptcies, as the Federal Trade Commission summarized the original scheme. A consumer was entitled to disclosure of the file contents, to notification after an adverse decision, and to a dispute procedure, with an agency duty to reinvestigate and to correct or delete inaccurate or unverifiable information.

The reporting statute fitted the disclosure model in its central device, the consumer right to see the file, and departed from the disclosure model in two respects that anticipated the later federal direction. Accuracy was imposed as a duty on the agency, not left to the corrective effect of consumer inspection alone, and obsolescence was imposed as a flat rule excluding defined categories of truthful but aged information, regardless of the preference of the agency, the user of the report, or the consumer. Neither duty depended on a consumer shopping more wisely, and each restricted what a firm could do even when every party had been fully informed. The departure was limited and procedural in 1970, but substantive restriction had been accepted in the same omnibus statute within two years of the Truth in Lending Act.

The Fair Debt Collection Practices Act of 1977, approved on September 20, 1977, as Public Law 95-109 and added to the Consumer Credit Protection Act as Title VIII, carried that acceptance into a field where the disclosure theory had no plausible application. A debtor in default gained nothing from a clearer statement of the price of a debt already incurred, and the abuses cataloged in section 802 were abuses of conduct rather than presentation: the findings tied abusive practices to personal bankruptcies, marital instability, loss of jobs, and invasions of privacy, and found existing redress inadequate. The operative sections prohibited communication at unusual or inconvenient times or places, with communication before 8 in the morning or after 9 in the evening presumptively inconvenient; prohibited harassment and abuse, including threats of violence, obscene language, and repeated calls intended to annoy; prohibited false representations about the character, amount, or legal status of a debt; and prohibited unfair practices, including collection of amounts not authorized by the agreement creating the debt or permitted by law. Debt validation and a cease communication right supplied procedural elements, but the core was a list of acts prohibited whether the debtor understood the acts, had consented in a contract, or had received perfect disclosure in advance.

Coverage limits in the 1977 statute qualify the pivot without reversing it. The Act regulated debt collectors, defined principally by third party collection, reaching agencies and collection attorneys while generally leaving a creditor collecting an original debt in the creditor name to state law. Remedies relied on private actions for actual and statutory damages, plus administrative enforcement assigned to the Federal Trade Commission. The regulatory form was the point of departure: Congress moved from standardizing the terms on which firms competed to specifying conduct that competition was not permitted to reward, on the stated justification that means other than misrepresentation and abuse were available for effective collection. A market expected to function after defined practices are removed from it is the structure of a substantive rule, not of a disclosure rule.

Between 1977 and 2009, the largest growth in consumer credit occurred in open end revolving credit, governed by the Truth in Lending disclosure framework as amended. Card pricing developed during the 1980s and 1990s, and into the 2000s, so that the rate disclosed at solicitation described only part of the eventual cost. Issuers applied different rates to different balance categories on the same account, raised rates on balances already incurred after a late payment on an unrelated account under universal default clauses, computed charges on preceding cycle balances under double cycle billing, allocated payments to the lowest rate balance first, and set penalty and over limit fees bearing no necessary relation to the cost of the triggering event. Each element could be, and generally was, described somewhere in the account documents, so formal disclosure could be satisfied while the operation of the account depended on interactions among terms that a cardholder comparing solicitations could not project. The Senate Committee on Banking, Housing, and Urban Affairs cataloged the practices as objects of prohibition rather than of clearer explanation.

The Credit Card Accountability Responsibility and Disclosure Act of 2009, approved on May 22, 2009, as Public Law 111-24, amended the Truth in Lending Act with operative provisions of a kind the 1968 statute had withheld. Section 101 prohibited an increase in an annual percentage rate, fee, or finance charge on an existing balance except under specified conditions, covering expiration of a disclosed promotional period, movement in an index outside creditor control, completion or failure of a workout arrangement, and a payment more than 60 days late, with restoration of the prior rate after six consecutive timely minimum payments in the delinquency case. Section 101 also required 45 days of advance notice of a rate increase or significant change in terms, protected the right to cancel, and barred increases during the first year after opening and during a promotional period of at least six months. Section 102 prohibited double cycle billing, prohibited penalties for on time payment, required amounts paid above the minimum to be allocated to the highest rate balance, and barred an over limit fee without express consumer opt-in. Related provisions required penalty fees to be reasonable and proportional to the violation. The principal pricing provisions took effect on February 22, 2010, nine months after enactment, under section 3 of the Act.

Disclosure had not been abandoned in 2009. The Credit CARD Act added disclosures, including a minimum payment warning stating the months, the total cost at the minimum payment, and the monthly amount required to retire the balance in 36 months. The change lay in the division of labor between the two instruments. In 1968, disclosure carried the entire federal policy for price and substantive restriction was absent. In 2009, disclosure was assigned warning, notice, and comparability at account opening, while practices shown to operate independently of cardholder attention were removed from the terms an issuer could offer and a cardholder could accept. Consent ceased to be a defense to a retroactive rate increase or to double cycle billing, just as consent had ceased to be a defense to a harassing collection call after 1977. The federal government first standardized the description of credit, then standardized access to the record behind credit decisions, then prohibited defined conduct in collection, and finally prohibited defined pricing practices in the largest consumer credit market.

The question for the remaining evidence sections follows from that sequence and from the purposes Congress wrote into the statutes. The Truth in Lending Act promised that meaningful disclosure would produce the informed use of credit and strengthen competition among the firms extending credit. The Fair Credit Reporting Act promised accuracy and fairness in reporting, through reasonable procedures, consumer access, and dispute rights. The Fair Debt Collection Practices Act promised the elimination of abusive collection practices. The Credit CARD Act, in the purpose carried in the long title of Public Law 111-24, promised fair and transparent practices for open end consumer credit, with fairness assigned to prohibitions and transparency to disclosures. Judged against those self declared aims, the federal record offers an unusually clean comparison, because the two strategies were applied to related problems, in the same statutory home, over a period long enough for each to generate an observable record. Which kind of rule, the rule that informed the consumer or the rule that restricted the firm, produced the outcome the enacting Congress had announced? Evidence on pricing, reporting accuracy, collection conduct, and card practices after 2010 supplies the material for an answer.

Finding One: Disclosure and the Limits of Comprehension

The Truth in Lending Act of 1968 rested on a clear premise. Enacted as Title I of the Consumer Credit Protection Act and implemented through the Federal Reserve Board Regulation Z, the statute required creditors to disclose the cost of credit in a standardized form before a consumer became bound. The annual percentage rate, the finance charge, the amount financed, the total of payments, and the payment schedule were to replace a confusion of contract rates, fees, and points with a common measure that a borrower could compare across competing offers. Congress described the purpose in the statutory text as the informed use of credit. Disclosure was the mechanism of protection itself, and the mechanism depended on a sequence of events in the mind and conduct of the borrower.

That sequence contained several distinct links, and the evidence assembled between 1968 and 2018 showed weakness at nearly every link. A disclosure had to reach the borrower before commitment. The borrower had to notice the disclosure, read the disclosure, understand the terms presented, translate a rate into consequences for a household budget over months or years, compare that translation with alternatives, and act on the comparison by rejecting an expensive offer or choosing another creditor. Failure at an early link made later links largely irrelevant. An accurate annual percentage rate in a document that arrived at a closing table, received a signature without reading, or presented a calculation that the borrower could not interpret delivered legal compliance without the informed choice that the 1968 statute had promised.

The most systematic statement of that failure came from Omri Ben-Shahar of the University of Chicago Law School and Carl E. Schneider of the University of Michigan Law School. Ben-Shahar and Schneider first joined their arguments in an article titled The Failure of Mandated Disclosure, published in the University of Pennsylvania Law Review in 2011 at volume 159, pages 647 to 749. The authors expanded the article into a book titled More Than You Wanted to Know: The Failure of Mandated Disclosure, published by Princeton University Press in 2014. Ben-Shahar brought a research program in contracts and consumer protection. Schneider brought a research program in health law and bioethics, including informed consent.

Ben-Shahar and Schneider defined mandated disclosure as a legal requirement that one party to a transaction provide information to the other party so that the recipient could make a better decision. The 2014 book surveyed mandated disclosure across consumer credit, contracts, medicine, and privacy, and the authors found a technique that lawmakers reached for repeatedly because the technique appeared inexpensive, respectful of choice, and compatible with markets. The authors argued that the appearance misled. In the account developed across the 2011 article and the 2014 book, mandated disclosure ordinarily failed to inform, failed to improve decisions, and imposed costs in time, attention, and misplaced regulatory effort. The failure was structural rather than accidental, and better drafting alone could not repair the structure.

Complexity formed the first structural problem. A mortgage, a credit card account, or an adjustable rate loan combined a price with contingencies that unfolded over time. Interest could be fixed or variable. Fees could arise at origination, during servicing, after a late payment, after a transfer, or at payoff. A minimum payment rule could extend repayment for years and multiply interest cost in ways that a monthly figure concealed. A disclosure faithful to such a product had to describe a branching set of possible futures, and a disclosure simple enough for rapid reading had to omit or compress many of those branches. Ben-Shahar and Schneider argued that plain language addressed only the surface of that problem, because the difficulty lay in the number of interacting terms and in the mathematics required to weigh those terms against each other.

The annual percentage rate illustrated the difficulty in a single figure. Regulation Z treated the annual percentage rate as a broader measure of credit cost than the contract interest rate, because the annual percentage rate incorporated many fees and points in addition to interest. A higher fee load therefore produced an annual percentage rate equal to or greater than the contract rate, even where the stated interest rate appeared identical between two offers. The comparison value of the figure depended on a borrower grasping that relationship and the sensitivity of effective cost to the period for which a loan was held. A borrower who read the annual percentage rate as another name for the interest rate, or who treated a small difference as trivial without converting the difference into dollars over the life of a loan, possessed the mandated number without the mandated understanding.

Accumulation formed the second structural problem. No single legislature, agency, or court controlled the total volume of disclosure that federal and state law imposed, and each mandate entered the world with a justification framed around a single transaction or a single risk. The Truth in Lending Act of 1968 generated disclosures for closed end and open end credit. The Real Estate Settlement Procedures Act of 1974 added settlement disclosures for many mortgage transactions. The Fair Credit and Charge Card Disclosure Act of 1988 amended the Truth in Lending Act and produced the standardized summary table for card applications that became known by the name of Representative Charles Schumer. Later statutes added disclosures for high cost mortgages, billing errors, rate increases, and minimum payment consequences. The combined package, however, confronted a borrower with overlapping forms and repeated definitions, with no institution charged with subtracting material as additions accumulated.

Ben-Shahar and Schneider treated accumulation as a problem of attention as a finite resource. A new disclosure competed with every earlier disclosure, with the sales presentation, with the contract, and with the time pressure of a closing. Aleecia M. McDonald and Lorrie Faith Cranor of Carnegie Mellon University gave that competition a concrete measure outside consumer credit in a study titled The Cost of Reading Privacy Policies, published in I/S: A Journal of Law and Policy for the Information Society in 2008. McDonald and Cranor estimated that an average privacy policy contained about 2,500 words and that reading the privacy policies encountered on new websites in a year would require about 244 hours for an average American internet user. Privacy policies represented only one category among many, and the broader pattern supported the conclusion that complete reading of all disclosures had ceased to be a plausible expectation by the second decade of the twenty first century. Rational recipients triaged, and most documents received a signature or a glance rather than study.

Timing formed the third structural problem. Disclosure law often keyed delivery to application, account opening, or consummation, and those legal moments did not necessarily correspond to the moment of genuine deliberation. A home buyer who had negotiated a purchase price, arranged moving plans, and paid application and appraisal fees arrived at a mortgage closing with powerful commitments to completion. Delivery before signature satisfied a legal sequence while arriving after the decision had hardened. The Consumer Financial Protection Bureau integrated mortgage disclosure rule, effective for applications received on or after October 3, 2015, recognized the timing defect in part, because the five page Closing Disclosure generally had to be received at least three business days before consummation. A pause of three days addressed only the final document in a long sequence of earlier commitments.

The demands placed on the recipient formed the fourth structural problem. Effective use of credit disclosure required more than reading skill. A borrower had to understand percentages, compounding, amortization, variable rate indexes and margins, and the difference between a rate and a total dollar cost. A borrower also had to forecast personal conduct and circumstances, including income stability, length of residence, speed of repayment, and likelihood of carrying a card balance rather than paying in full. Ben-Shahar and Schneider argued that mandated disclosure amounted, in effect, to a program of mass education delivered in fragments at moments of transaction, and that no legislature had supplied the education that the program presupposed.

Political and commercial incentives reinforced the mismatch. Ben-Shahar and Schneider argued that disclosure appealed to lawmakers because a disclosure mandate allowed action after a scandal or a complaint without a decision to forbid a product, cap a price, or impose a substantive term. Disclosure also appealed across ideological divisions, because one side could present information as consumer empowerment and another side could present information as a market preserving alternative to command regulation. Disclosers, for their part, had incentives to comply defensively. A creditor concerned with liability had reason to include every term that a regulator, a court, or a future plaintiff might consider material, and defensive completeness pushed documents toward length and legal precision at the expense of readability. The resulting document protected the discloser in litigation more reliably than the document protected the recipient in a transaction.

Do consumers read the disclosures the law requires?

Evidence available by 2018 indicated that most consumers did not read most mandated credit disclosures closely, and many skimmed or signed without reading. Studies of mortgage and card disclosures found limited recall, weak comprehension of cost terms, and little use of disclosed figures in choosing among competing offers.

The mortgage evidence gave the general claim a specific foundation. James M. Lacko and Janis K. Pappalardo of the Bureau of Economics of the Federal Trade Commission conducted in depth interviews with 36 consumers who had obtained a mortgage within the preceding four months, with interviews conducted between September 2004 and February 2005 in Montgomery County, Maryland. The interview record, published as part of the 2007 Federal Trade Commission staff report titled Improving Consumer Mortgage Disclosures: An Empirical Assessment of Current and Prototype Disclosure Forms, showed borrowers who had completed transactions without a secure grasp of the terms of the loans obtained. Borrowers described shopping behavior that centered on monthly payment, on trust in a broker or lender, or on a quoted interest rate, rather than on a systematic comparison of total cost using the mandated forms. The interviewees had received disclosures. Receipt and use proved to be different events.

Comprehension of the annual percentage rate supplied the clearest test of the original design, because comparison through a standardized rate stood at the center of the 1968 scheme. Jinkook Lee and Jeanne M. Hogarth examined that comprehension in an article titled The Price of Money: Consumers Understanding of APRs and Contract Interest Rates, published in the Journal of Public Policy and Marketing in 1999 at volume 18, pages 66 to 76. Lee and Hogarth analyzed survey responses from consumers who had applied for a mortgage or a refinance in the preceding five years. Among those respondents, 40 percent did not understand that the annual percentage rate, as the broader cost measure, is the same as or greater than the contract rate. Income had no meaningful association with understanding, and education had almost no association, with graduate education forming the exception. The misunderstanding concerned a threshold point rather than an advanced calculation. A borrower who could not order the two figures lacked the basic map that made comparison shopping through Regulation Z figures possible.

The Federal Trade Commission testing by Lacko and Pappalardo extended the point from one figure to whole forms. In the quantitative phase of the 2007 study, Lacko and Pappalardo tested disclosure forms with 819 mortgage customers and found that both prime borrowers and subprime borrowers failed to understand key loan terms when viewing the disclosures in use during the study period, including costs and features central to the price and risk of a loan. Confusion crossed the prime and subprime division, and difficulty increased for more complex loans, the category in which accurate comprehension carried the largest financial consequences.

Research on the mathematics of interest deepened the explanation. Victor Stango of the University of California, Davis, and Jonathan Zinman of Dartmouth College published a study titled Exponential Growth Bias and Household Finance in the Journal of Finance in 2009 at volume 64, pages 2807 to 2849. Stango and Zinman documented a tendency to linearize exponential growth, meaning a tendency to treat compound growth as if growth proceeded in straight line increments. The bias produced a systematic underestimation of an interest rate implied by other loan terms and of a future value implied by investment terms. Households showing greater bias also borrowed more, saved less, and favored shorter maturities in the data examined by Stango and Zinman. The finding mattered for disclosure design because an annual percentage rate is a compact expression of a compounding process. A borrower subject to exponential growth bias could read the rate correctly as a printed number and form an incorrect judgment about the burden that the rate would impose over time.

Credit card research available by 2018 separated recognition from evaluation in a revealing way. Alycia Chin and Wändi Bruine de Bruin reported experiments in a manuscript titled Helping Consumers to Evaluate Annual Percentage Rates on Credit Cards, dated August 27, 2018 and accepted for the Journal of Experimental Psychology: Applied. In the first study, participants using standard credit card disclosures proved able to identify the absolute annual percentage rate stated in a disclosure at high rates of accuracy, and that recognition success concealed a large failure of relative judgment. Estimates of the cost of a card annual percentage rate relative to the credit card market differed from actual relative costs by 47 percentage points on average, and participants underestimated the relative cost of expensive cards. Participants who correctly identified the absolute rate performed no better at estimating relative cost than participants who missed the figure. A disclosed number without a distribution or a benchmark gave little basis for judging whether a rate was cheap, ordinary, or expensive.

Placement and salience produced a related failure in mortgage experiments reported in 2013. Debra Pogrund Stark, Jessica M. Choplin, Mark LeBoeuf, and Andrew Pizor reported two experiments in a paper titled When Is Consumer Understanding Necessary to Make Wise Home Loan Decisions, dated July 16, 2013. Stark, Choplin, LeBoeuf, and Pizor found that participants made much better use of an enhanced annual percentage rate disclosure developed by the authors to identify the lower cost loan than participants made of the Consumer Financial Protection Bureau form under consideration during the study period, in which the annual percentage rate appeared in an inconspicuous position on page three of the Loan Estimate. The same experiments included financial literacy questions, and the authors reported that a vast majority of participants failed to identify problematic terms, risks, and lender remedies upon default in loan materials.

Taken together, the studies from 1999 to 2018 described a pattern more specific than a general claim about inattention. Consumers often completed transactions without reading long forms closely, and consumers who did attend to a form often attended to monthly payment, headline interest rate, cash required at closing, or a single salient fee rather than to total cost. Consumers facing complex products, including interest only payments, balloon payments, and adjustable rates, faced the greatest comprehension burden and, in the Lacko and Pappalardo testing of 2004 to 2007, showed the greatest difficulty. Disclosure therefore changed decisions weakly not because information had no value in principle, but because the mandated form of information repeatedly missed the cognitive and practical conditions under which a decision could absorb information.

The record also contained a genuine counterpoint. Lacko and Pappalardo did not conclude in 2007 that disclosure design was hopeless. The prototype disclosures developed for the Federal Trade Commission study significantly improved recognition of mortgage costs in testing with 819 mortgage customers, and improved disclosures produced the greatest benefit for complex loans. Chin and Bruine de Bruin reported in 2018 that a histogram showing the distribution of annual percentage rates in the credit card market improved estimates of relative cost and produced more negative evaluations of an expensive card offer, without reducing identification of the absolute rate. Stark, Choplin, LeBoeuf, and Pizor reported in 2013 that an enhanced annual percentage rate disclosure improved identification of a lower cost home loan. Each result pointed toward design, placement, and context as variables with real effects.

A field experiment in payday lending pointed in the same qualified direction. Marianne Bertrand and Adair Morse published a study titled Information Disclosure, Cognitive Biases, and Payday Borrowing in the Journal of Finance in 2011 at volume 66, pages 1865 to 1893, reporting a field experiment at a national chain of payday lenders. A treatment presenting the accumulated dollar cost of payday borrowing over repeated pay cycles, in comparison with other credit products, made participants 5.9 percentage points less likely to borrow in the pay cycles after the intervention, a decline of 11 percent relative to the comparison group. A review by George Loewenstein, Cass R. Sunstein, and Russell Golman titled Disclosure: Psychology Changes Everything, published in the Annual Review of Economics in 2014, likewise approached disclosure as a psychological intervention rather than as a neutral transfer of facts. The counterpoint, stated at the level supported by those sources, was that the critique struck hardest at accumulated volume, legalistic design, poor timing, and context free numbers, and struck less decisively at the concept of informing a choice through tested, salient, comparative information. A later section develops that distinction at length.

Finding one therefore ended, on the evidence available through 2018, with a measured conclusion. The Truth in Lending Act of 1968 succeeded in standardizing the vocabulary of credit cost and in forcing cost figures into transactions, and standardization created a record that regulators, researchers, and competing creditors could inspect. The statute and successor regimes succeeded far less reliably at the human task assigned to disclosure, because borrowers frequently did not read, often did not understand, and rarely used the annual percentage rate and total cost figures in the comparative manner that the statutory design contemplated. Ben-Shahar and Schneider gave that gap a theory in 2011 and 2014, grounded in complexity, accumulation, timing, and the demands placed on recipients, and the empirical record from Lee and Hogarth in 1999, Lacko and Pappalardo in 2007, Stango and Zinman in 2009, Stark and coauthors in 2013, and Chin and Bruine de Bruin in 2018 supplied the gap with measurements. Disclosure delivered information onto paper and screens across the American credit market. Comprehension arrived far less dependably.

Finding Two: The Card Act and the Substantive Counterexample

The second finding rests on a distinction that runs through federal consumer credit law. A disclosure rule changes the information attached to a transaction and leaves the transaction available on the same substantive terms. A substantive rule removes a term, a fee practice, or a method of calculating charges from the range of choices open to the creditor, regardless of how clearly that practice might have been described in an application or a periodic statement. The Credit Card Accountability Responsibility and Disclosure Act of 2009, known as the CARD Act and enacted as Public Law 111-24, contained both kinds of provision. The provisions that produced the largest measured savings did not depend on a cardholder reading a statement, comparing offers, or changing payment behavior. Those provisions prohibited or capped charges that issuers had imposed through the back end of the contract, after account opening. When a prohibited fee disappears, a saving can be recorded even for an account holder who never read the disclosure that announced the change.

President Barack Obama signed the CARD Act on May 22, 2009, and Congress placed the statute largely inside the Truth in Lending Act. The first stage of implementation, covering advance notice of rate increases and the time allowed for payment, took effect on August 20, 2009. The principal body of rules took effect on February 22, 2010, and the provisions governing the reasonableness and proportionality of penalty fees, together with the periodic reevaluation of rate increases, took effect on August 22, 2010. A reader tracing the sequence from statute to regulation to measured outcomes can follow the CARD Act provisions explained as a chronological map of that sequence. The Federal Reserve Board issued the Regulation Z amendments that carried most of the statutory commands into operating rules for issuers.

The rate provisions illustrate the difference. Before the CARD Act took effect, card agreements commonly permitted an issuer to raise the annual percentage rate on an existing balance at any time and for any reason, and universal default permitted a rate increase on one account after a default connected with an entirely different obligation. The Act prohibited an increase in the annual percentage rate, a fee, or a finance charge applicable to an outstanding balance except within stated exceptions. A rate could rise at the expiration of a promotional period disclosed clearly before the period began, provided that the higher rate did not exceed the rate disclosed for the period after expiration. A rate could move with a variable index outside the control of the creditor. A rate could rise after a payment had become more than sixty days late, with restoration available after six months of timely payments. A rate could also change after a workout or temporary hardship arrangement ended or was breached, capped at the rate that had applied before the arrangement began. Outside those channels, the price attached to money already borrowed was fixed by law rather than by issuer discretion. The Act barred most rate and fee increases during the first year after an account opened, and it required forty-five days of written notice before a permitted increase or a significant change in terms could take effect on future transactions.

A second group of provisions governed how money moved through an account and how penalties attached to mistakes. The Act required an issuer to apply any amount paid in excess of the minimum payment first to the balance bearing the highest rate of interest, and then to each successive balance in descending order of rate. The displaced practice had applied excess payments to the lowest rate balance first, a method that left high rate balances outstanding for longer and increased the interest collected over the life of the debt. The Act prohibited double cycle billing, under which an issuer computed finance charges by reaching back into a preceding billing cycle after a grace period had been lost, and it prohibited a fee for paying by mail, telephone, or electronic transfer unless the payment involved an expedited service performed by a representative of the creditor. The Act also required statements to be sent at least twenty-one days before the due date.

The fee provisions followed the same pattern and reached the charges that later measurement found to be most consequential. The Act barred an over-limit fee unless the cardholder had affirmatively opted in to the completion of transactions that would exceed the credit limit, and an account holder who declined that election faced a declined transaction rather than an approved transaction paired with a fee. The Act directed that any penalty fee or charge be reasonable and proportional to the omission or violation involved, and it directed the Federal Reserve Board to establish standards for assessing that relationship by reference to the cost incurred by the creditor, the deterrence of omissions or violations, and the conduct of the cardholder. The Act limited the fees that could be absorbed by the credit line on a low limit, high fee card during the first year of an account to twenty-five percent of the initial credit limit, a response to fee harvester products in which account opening charges consumed much of the nominal credit before a purchase had been made. The statute set no general ceiling on credit card interest rates. The statute instead fenced off a set of pricing methods, which makes the later evidence traceable to identified methods that ceased to be available.

The economics of the pre-Act market explain why fencing off those methods could lower total cost rather than merely relabel it. Credit card solicitations competed most visibly on dimensions that an applicant could compare at the moment of application, including the annual percentage rate, an introductory rate, and rewards. Late fees, over-limit fees, penalty rates, and payment allocation rules operated later, after spending and repayment difficulty had placed an account in the position where a fee could attach. Sumit Agarwal, Souphala Chomsisengphet, Neale Mahoney, and Johannes Stroebel described that structure in their published study as a setting of low fee salience combined with limited market competition. If fees escaped attention at the point of purchase, a legal limit could reduce the total price rather than forcing an equal increase elsewhere, because the pre-Act price contained a margin that competition on salient terms had failed to remove. The profitability figures in the same study gave that account empirical weight. In the period before the CARD Act provisions took effect, accounts held by borrowers with FICO scores below 620 generated realized profits of 7.9 percent of average daily balances, against an average of 1.6 percent of average daily balances across all accounts.

Opposition to the Act during the 2009 debate predicted a different result. Edward Yingling, president and chief executive of the American Bankers Association, which lobbied against the legislation, argued in May 2009 that the new rules would limit the ability of card companies to price according to risk, and that less credit would be available generally, with younger consumers and startup small businesses among those likely to lose access to cards entirely. A lender denied revenue from penalty fees and repricing would recover the revenue through higher interest charges, through annual fees and other front end charges, or through tighter approval standards that withdrew credit from higher risk applicants. The same concern appeared from a different position in the regulatory debate. Barr, Mullainathan, and Shafir, writing in 2009 as proponents of regulating late fees, recorded a worry that a reduced revenue stream to lenders from those fees would lead to adjustment in other rates and fees as compensation. The hypothesis expressed the standard expectation for a competitive market in which every component of price is visible, and the CARD Act created the conditions for testing that expectation against account level records rather than against competing forecasts.

Did the CARD Act actually save consumers money?

Yes. Agarwal, Chomsisengphet, Mahoney, and Stroebel estimated in the Quarterly Journal of Economics in 2015 that fee limits reduced borrowing costs by an annualized 1.6 percent of average daily balances, saving consumers about 11.9 billion dollars a year, without an offsetting rise in interest charges.

The study behind that answer appeared as “Regulating Consumer Financial Products: Evidence from Credit Cards” in the Quarterly Journal of Economics in 2015, in volume 130, number 1, at pages 111 to 164. A working paper version circulated through the National Bureau of Economic Research in 2013 with different figures, so the journal version controls any statement of the result. The data came from the regulatory reporting system administered by the Office of the Comptroller of the Currency. The panel covered about 160 million consumer and small business credit card accounts held at the eight largest banks in the United States, with monthly account level information on contract terms, utilization, payments, fees, interest charges, credit limits, and balances running from January 2008 through December 2012.

The research design exploited an exemption written into the statute itself. The CARD Act regulated consumer credit cards and did not cover small business credit cards, even though the two products shared much of their institutional structure. Both kinds of account rested on the personal financial position of the account holder, and applicants for both kinds were screened on personal FICO scores. Agarwal, Chomsisengphet, Mahoney, and Stroebel therefore compared changes in outcomes for regulated consumer accounts with changes in outcomes for unregulated small business accounts over the same months, a difference in differences method that subtracts the movement in the comparison group from the movement in the regulated group. The method addressed the central difficulty of the period, which was that the fee limits took effect in February and August 2010, during the aftermath of the financial crisis of 2008 and 2009, when delinquency, charge offs, unemployment, and issuer risk appetite were all moving for reasons unrelated to the statute. The authors reported that fees, interest charges, average daily balances, credit limits, and new account formation had moved together for consumer and small business accounts during the period before the CARD Act took effect, which supported the assumption that the two series would have continued on parallel paths in the absence of the law. That assumption cannot be observed directly, and the finding rests on it, but the pre-Act record offered a check on the assumption.

Measured against that comparison, the fee limits reduced total fees by an annualized 1.6 percent of average daily balances across all consumer accounts, a reduction of 22.58 dollars per account per year. The distribution of the reduction was sharply uneven. For account holders with FICO scores below 660, total fees fell by 5.3 percentage points of average daily balances, or 57.69 dollars per account per year, led by a decline of 3.3 percentage points in over-limit fees and a decline of 1.5 percentage points in late fees. For account holders with FICO scores above 660, who had carried lower fee levels before the Act took effect, fees fell by 0.5 percentage points of average daily balances, or 7.59 dollars per account per year. The pattern is the signature of a substantive rule rather than an information rule. A disclosure tends to deliver the largest gains to the account holders best equipped to read, compare, and act, generally account holders with higher credit scores and more alternative offers. The CARD Act fee limits delivered the largest gains to the group with the lowest scores, because that group had been paying the fees that the statute removed, and receipt of the gain required no reading, no comparison, and no application for a replacement card.

The aggregate estimate followed from the per account result and from the size of the market. Agarwal, Chomsisengphet, Mahoney, and Stroebel extrapolated the 1.6 percent reduction to total outstanding credit card balances of 744 billion dollars in the first quarter of 2010, a figure drawn from the Federal Reserve Bank of New York, and arrived at annual savings for credit card users in the United States of 11.9 billion dollars. The estimate also covers the fee provisions measured in the study rather than every provision of the Act, and the authors separated that estimate from the much smaller effect attributed to the new statement disclosure.

The second half of the study tested the offset hypothesis on the same data and with the same design, and the result was negative across each channel through which an offset could have appeared. The authors found no evidence of an anticipatory increase in interest charges before the fee limits took effect, no evidence of a sharp increase in interest charges after implementation, and no evidence of a gradual increase that accumulated over the months that followed. The point estimate for the offset was approximately zero, and the authors reported that an offset greater than sixty-one percent of the fee reduction could be ruled out with ninety-five percent confidence. Interest charges on new accounts, where issuers faced the fewest constraints on resetting contract terms, showed no offsetting increase either. On the quantity side, the authors estimated an effect of precisely zero on credit limits and on average daily balances, and an estimate of zero for the number of new accounts, although the standard errors on new accounts were too large to rule out meaningful effects in either direction. The statement on existing accounts was therefore firmer than the statement on new account formation.

Corroboration from the regulator charged with reviewing the Act pointed in the same direction and rested on a different method. The Consumer Financial Protection Bureau published its first review of the consumer credit card market in October 2013. The Bureau found that the total cost of credit, defined to include all fees, interest, and finance charges paid by the consumer to the issuer, had declined by two percentage points between 2008 and 2012. The Bureau reported that over-limit fees had been effectively eliminated as a cost to consumers and as a source of revenue to issuers after the opt-in requirement took effect, and that consumers had paid about 2.5 billion dollars less in over-limit fees than in 2008. The Bureau estimated that the average late fee had fallen by six dollars after the Act took effect, producing a decrease of 1.5 billion dollars in late fees paid by consumers in 2012. Two institutions, using different data treatments and answering to different mandates, recorded the same central fact of a lower total cost driven by the disappearance of penalty fees.

The same statute supplied an internal comparison that sharpens the interpretation, because the CARD Act also contained a prominent disclosure intervention whose measured effect was small. The Act required each periodic statement to show the time and total cost involved in repaying a balance through minimum payments alone, together with the monthly payment and total cost required to repay the balance within thirty-six months. Agarwal, Chomsisengphet, Mahoney, and Stroebel found that the share of account holders paying at a rate sufficient to repay the balance within thirty-six months rose by 0.4 percentage points, on a base of 5.3 percent, after the disclosure took effect in February 2010. The authors found no evidence of a change in overall payments, and the authors estimated as an upper bound that the disclosure reduced aggregate interest payments by no more than 0.01 percent of average daily balances, or about 57 million dollars a year. Set against savings of 11.9 billion dollars a year from the fee limits, the disclosure component accounted for a fraction of one percent of the measured benefit. The contrast shows instead where the money was. Money moved when the law removed charges, and money barely moved when the law improved the description of charges that remained lawful.

The limits of the finding are stated in the study and bound the claim that can be drawn from the finding. The account data ended in December 2012, which left about two years of observation after the August 2010 penalty fee provisions took effect, and the authors stated that the data did not permit investigation of longer run effects on entry into the industry, exit from the industry, or margins governed by contracts extending over several years. The American Bankers Association, in a February 2013 letter to the Consumer Financial Protection Bureau signed by senior vice president Nessa Feddis, maintained that average credit card interest rates had increased and that credit card credit had become less available, particularly for subprime borrowers, after the Act and the associated regulations took effect. That letter relied on market aggregates from a period dominated by the financial crisis and the recession, and the difference in differences design was constructed precisely to separate a statutory effect from those aggregates. The disagreement therefore turns on method rather than on arithmetic, and the published study disclosed the assumption on which the method depends.

The implication for the wider account of consumer credit law is structural. Federal law after 1968 built an elaborate apparatus for telling borrowers the price of credit, and that apparatus rested on a theory in which an informed borrower disciplines an issuer by taking business elsewhere. The CARD Act departed from that theory at the points where the theory had failed in practice, because the charges at issue attached after borrowing, depended on events that a borrower could not reliably forecast at application, and fell most heavily on borrowers with the least capacity to move to another issuer. Regulation of those charges produced savings that were large in aggregate, progressive in distribution across the credit score range, and unmatched by any measured compensating increase in interest charges or contraction in credit limits during the period studied. The counterexample does not establish that substantive regulation succeeds in every consumer market. The counterexample establishes a narrower and more useful proposition, grounded in 160 million accounts observed month by month. Where a price component escapes attention at the point of sale and competition leaves the revenue intact, a legislature can reduce the total price of credit by prohibiting the component, and disclosure of the component is not an equivalent instrument.

Finding Three: Accuracy Inside the Credit File

The third finding concerns the distance between the accuracy standard that Congress wrote into law in 1970 and the accuracy that the finished credit file delivered in practice. The Fair Credit Reporting Act did not promise any consumer an error free file. Section 607(b) of the Act directed every consumer reporting agency, whenever the agency prepared a consumer report, to follow reasonable procedures to assure maximum possible accuracy of the information concerning the individual about whom the report related. The operative words were procedural. The statute regulated the care with which a bureau assembled, matched, and maintained information, and it attached a separate set of duties to the reinvestigation of a disputed item after a consumer gave notice of a possible error. A single mistaken balance, a misattributed account, or a collection item belonging to another person did not, by itself, establish a violation, because the legal question centered on the reasonableness of the procedures behind the file rather than on perfection in the outcome for that file. The distinction shaped everything that followed in consumer credit law, from enforcement practice to private litigation, and it explains why the most revealing evidence about the system came not from a courtroom but from a measurement exercise that Congress ordered three decades after enactment.

The measurement exercise arrived through the Fair and Accurate Credit Transactions Act of 2003. Section 319 of that statute required the Federal Trade Commission to conduct a study of the accuracy and completeness of consumer reports and to report the results to Congress. The Commission answered with the Report to Congress Under Section 319 of the Fair and Accurate Credit Transactions Act of 2003, published in December 2012, followed by a sixth interim and final report in January 2015 that carried a follow up investigation of unresolved disputes. Between those two publications, the Commission produced the most systematic account available before 2019 of what a consumer found upon opening the file, what happened after the consumer complained, and how often a correction changed the price of credit. Readers seeking the statutory framework behind the study can consult the Fair Credit Reporting Act guide, which sets out the duties that the 1970 Act imposed on the three nationwide bureaus, Equifax, Experian, and TransUnion, and on the businesses that furnished information to those bureaus.

The design of the 2012 study deserves attention before the results, because the design determined what the numbers could and could not show. The study worked with 1,001 participants, selected, according to the Commission, to match the demographic and credit score characteristics of the general public, and the participants reviewed a combined total of 2,968 credit reports covering the three national bureaus. Each participant examined the reports with the help of a study associate, whose role was to assist in identifying information that appeared wrong and in separating trivial discrepancies from errors with a plausible effect on credit standing. The study counted only the second category. Its central unit was the potentially material error, defined by the prospect that the information, if wrong, could affect a credit score or a decision based on the report. An error became a confirmed material error in the vocabulary of the study when the bureau modified the report in response to the participant’s dispute, so confirmation rested on the outcome of the same dispute process that an ordinary consumer would have used outside the study. After the dispute process ended, participants received fresh reports and fresh scores, the Fair Isaac Corporation rescored the files, and the Commission compared the original and corrected versions to measure the effect of each confirmed error on the consumer’s score and credit risk classification.

That design carried a built in qualification that any honest reading of the results has to preserve. The participants did not face the file alone. A trained associate sat with each participant, worked through the entries, and helped frame disputes under the Fair Credit Reporting Act process. The measured error rates therefore described the errors that assisted review could surface, and the dispute outcomes described the results that a guided dispute could obtain. No equivalent assistance awaited the consumer who requested a report through the annual disclosure system created by the 2003 amendments, read the entries without expert help, and had to decide alone whether an unfamiliar account name reflected an error, an unfamiliar corporate parent, or a legitimate obligation recorded under a different name. The Commission’s figures set a floor and a benchmark rather than a portrait of unaided experience, and the gap between assisted detection and unaided detection forms part of the finding itself.

How many credit reports contain a material error?

In the Federal Trade Commission study published in December 2012, 262 of 1,001 participants, or 26 percent, identified a potentially material error on at least one of three reports, and 206 participants, or about one in five, obtained a correction after disputing the information through the statutory process.

The headline figure that traveled through press coverage in February 2013, one in five, referred to the 206 participants whose reports the bureaus changed, a group equal to 21 percent of the full sample. The larger figure, 26 percent, measured allegation rather than confirmation: 262 participants identified at least one potentially material error on at least one of the three reports and filed a dispute with at least one bureau. The difference between the two numbers, 56 participants, represented disputes that produced no modification at all, and the Commission counted 56 consumers, 5.6 percent of participants, in that unresolved category at the close of the initial process. Other participants obtained partial relief, with some requested changes made and others refused, so the outcome distribution formed a gradient rather than a clean division between success and failure. Four out of five participants who filed disputes obtained at least some modification, a ratio that cut in two directions at once. The ratio testified that the dispute channel worked for a substantial majority of the consumers who entered it, and the same ratio confirmed that entry into the channel was the precondition for relief, since none of the corrections measured in the study occurred before a consumer detected an error and demanded an investigation.

The character of the errors mattered as much as the count. The Commission reported that the most common confirmed errors involved tradeline information, the account level records of balances, payment histories, and account status, which accounted for close to half of the alleged errors, and collection accounts, which represented almost one third. Misidentification ran through both categories. Roughly 35 percent of the allegations concerning tradeline information asserted that the account did not belong to the consumer at all, and the share rose above 80 percent among disputes over collection items, where the dominant claim was that the debt belonged to someone else. Header information produced a third class of error, covering names, addresses, dates of birth, and employment entries, the very fields on which the matching of incoming data to the correct file depended. A file could therefore be wrong in ways that compounded. A mistaken identifier invited mistaken matching, mistaken matching imported another person’s tradeline or collection record, and the imported record then fed the scoring model as though the record described the consumer named at the top of the report.

Error counts alone understated the stakes, because the consequences of an error depended on the distance between the erroneous score and the corrected score, and on whether that distance crossed a boundary that a lender used in pricing. The Commission measured both dimensions. Of the full sample, 129 participants, 13 percent, experienced a change in credit score after the bureaus modified their reports. For the majority of that group, the change was modest: more than half of the 129 participants recorded a maximum change of fewer than 20 points. At the report level, half of the positive score changes fell below 15 points. The tail of the distribution, however, carried the policy weight. Approximately one in 20 consumers in the sample recorded a maximum score change of more than 25 points, only one in 250 recorded a maximum change of more than 100 points, and 27 participants saw at least one of their three scores rise by more than 50 points after correction. The Congressional Research Service, summarizing the study in its reporting on consumer credit policy, noted that for consumers whose scores rose, over 40 percent of the increases exceeded 20 points, a magnitude capable of moving a consumer across the boundary of a pricing bracket when the original score sat near the edge of that bracket.

The tier analysis converted those point changes into economic terms. Lenders and insurers sort applicants into credit risk tiers and attach prices, deposits, and eligibility rules to the tiers, so a score change acquires practical meaning when the change carries the consumer from one tier into another. The Commission found that 52 participants, 5 percent of the full sample, experienced a score increase after correction large enough to move the consumer into a lower credit risk tier, the kind of movement that made a lower auto loan interest rate more likely. One in twenty consumers in a nationally drawn sample therefore carried an error serious enough, under the study’s reasoning, to affect the terms or the eligibility attached to a credit application if the error went uncorrected. The conditional figure sharpened the point. Among the 262 participants who identified and disputed an error, 52, or 20 percent, obtained the tier improving increase, so a consumer who found a material error faced a one in five prospect that the error was suppressing the consumer’s risk classification. The converse also held and deserves equal weight: most errors, even confirmed ones, changed no tier, and most files in the sample contained no potentially material error that assisted review could find. The study portrayed a system that performed accurately for the large majority of consumers and inaccurately, with consequences concentrated in price, for a minority measured in millions once the percentages were projected across the national population of credit holders.

The follow up report of January 2015, titled as the sixth interim and final report under section 319, returned to the participants whose disputes had failed to move the bureaus, a population that the December 2012 findings had left in shadow. The follow up study surveyed 121 consumers who had at least one unresolved dispute from the original study. Thirty seven of those consumers, 31 percent, reported that the disputed information was, upon reflection, correct, and that acceptance closed the matter. The remaining 84 consumers, nearly 70 percent of the group, maintained that at least some of the disputed information remained inaccurate. Within that group of 84, thirty eight consumers, 45 percent, said the consumer intended to press the dispute further, while 42 consumers, 50 percent, said the consumer intended to abandon the dispute, and four consumers remained undecided. The abandonment figure carried the heaviest implication of the follow up work. Half of the consumers who believed a bureau held wrong information about them, after a completed round of the statutory dispute process, planned to stop contesting the record, which left the contested information in place as the operative version of the consumer’s history for lenders, insurers, landlords, and employers who later requested a report. The follow up study also examined reinsertion, the reappearance of negative information previously removed after a successful dispute, and found the practice rare, with two instances amounting to about 1 percent of the consumers whose reports had been modified.

Read together, the 2012 and 2015 reports located the weakness of the accuracy regime with unusual precision. The weakness sat less in the correction of detected errors, where four out of five disputes produced a change, than in the allocation of the work of detection and persistence. The statute’s reasonable procedures standard governed the bureau at the moment of report preparation, and the reinvestigation duty governed the bureau after notice, but no provision of the 1970 Act, even after the 2003 amendments added duties for furnishers and a right to free annual disclosures, assigned any institution the task of auditing an individual file for the consumer’s benefit before the consumer complained. Detection depended on the consumer obtaining the report, reading entries that often appeared in coded or abbreviated form, recognizing an error among legitimate entries recorded under unfamiliar names, and filing a dispute with documentation. Persistence depended on the same consumer after an adverse outcome, since the follow up findings showed that a consumer who lost the first round and accepted the result left the disputed item in place by default. The assistance built into the study design threw that allocation into relief: assisted participants found potentially material errors at a rate of 26 percent, and the study offered no basis for assuming that unaided consumers, working without a trained associate, matched that rate of detection across the same files.

The procedural character of the standard produced a second consequence, visible in the way accountability operated. Because section 607(b) asked whether procedures were reasonable rather than whether a particular report was correct, an isolated error, even a tier changing error of the kind that affected 52 participants, established little on its own in an enforcement action or a private suit. The inquiry moved to the bureau’s matching rules, its treatment of mixed files, its handling of data from furnishers with known error patterns, and the adequacy of its dispute investigations. Systemic patterns could support liability where a single file could not, which meant that the consumers most visibly harmed by inaccuracy, the one in twenty with a tier changing error, had no automatic remedy flowing from the harm itself, and instead depended either on the dispute process or on litigation framed around procedures. The Commission’s study, ordered precisely because Congress wanted outcome evidence, supplied the outcome evidence that the statute’s own enforcement logic treated as secondary: the files themselves, examined one by one, contained material errors at a rate of roughly one consumer in four by allegation and one consumer in five by confirmed correction.

A fair reading of the findings also requires the qualifications that the Commission attached to the work. The study measured errors that participants alleged and bureaus confirmed through modification, and modification was an imperfect test, since a bureau might modify a report to close a dispute without conceding the original entry was false, and might refuse modification of an entry that was in fact false. The Commission framed confirmed rates as resting on that process outcome, and quantities such as the true error rate in the national file system could only be bounded, not read directly from the dispute results. Score effects were measured through one scoring system, the Fair Isaac model that the study employed for rescoring, and other models weighted tradelines and collections differently, so the 13 percent and 5 percent figures described effects within that model rather than universal constants. The sample, at 1,001 participants and 2,968 reports, supported national estimates of consumer level rates, yet the absolute counts behind the most consequential findings were small, 52 consumers for the tier change result, and small counts produced wide margins around the projected national figures. None of those qualifications dissolved the central result. Under assisted review, a quarter of consumers found a potentially material error, a fifth obtained a correction, an eighth gained points, and a twentieth crossed into a better risk tier, and a follow up inquiry found that most consumers whose disputes failed believed the error remained, with half of that group prepared to give up the contest.

The finding therefore redefined what the 1970 settlement on accuracy had actually delivered by the time of measurement in 2012. Congress had chosen a process standard over an outcome guarantee, had placed correction machinery behind a consumer initiated trigger, and had supplemented that machinery in 2003 with free annual access and a mandated federal measurement. The measurement showed a reporting system of high average accuracy and a persistent minority burden, distributed without regard to the consumer’s capacity to detect or contest an error. For the consumer inside the 5 percent, the distinction between a procedural standard and an accuracy guarantee was not abstract doctrine. The distinction arrived as a higher interest rate on an automobile loan, a larger deposit, or a denied application, priced from a file that the consumer had never seen and that only the consumer could set in motion to repair.

Finding Four: Collection Complaints and What They Measure

Debt collection generated a large and persistent volume of consumer complaints in the federal reports published between 2013 and 2018, and complaint volume became a frequently cited measure of conditions under the Fair Debt Collection Practices Act. The Fair Debt Collection Practices Act of 1977, Public Law 95-109, approved on September 20, 1977, stated a purpose of eliminating abusive debt collection practices, insuring that collectors who refrained from abusive practices were not competitively disadvantaged, and promoting consistent state action against collection abuses. A complaint count, however, entered the public record through a different process from a judicial decision, an examination finding, or an enforcement order. A complaint appeared in published totals only after an agency defined the category, the period, and the treatment of duplicates, referrals, and incomplete submissions.

Complaints about collection also predated the federal system that made the figures most visible. In a message accompanying the annual report on the Fair Debt Collection Practices Act for 2013, Consumer Financial Protection Bureau Director Richard Cordray wrote that federal agencies had received more than 200,000 consumer complaints in 2013 about the conduct of debt collectors. That total referred to federal agencies collectively, rather than to a single database, and the report did not convert the complaints into findings.

The Consumer Financial Protection Bureau began operating on July 21, 2011, and the bureau accepted consumer complaints about credit cards on the same day, according to the bureau announcement of the 2013 Consumer Response Annual Report. In 2013, the bureau began accepting complaints about money transfers, debt collection, and payday loans, and the bureau specified July 2013 as the starting point for debt collection complaints. The early debt collection total therefore reflected a partial year and a new channel. In the annual report on the Fair Debt Collection Practices Act for 2013, Director Cordray stated that the bureau had handled approximately 30,300 debt collection complaints between the start of intake in July 2013 and the end of that calendar year, and the report described debt collection as the bureau largest source of complaints per month during that interval. The statement identified reporting intensity within a new system, rather than a rate of unlawful conduct.

Published snapshots enlarged the cumulative picture while preserving the same limitation. In the monthly complaint report spotlighting debt collection in March 2016, the Consumer Financial Protection Bureau reported that the bureau had handled approximately 219,200 debt collection complaints as of March 1, 2016, and the bureau described debt collection as the most complained about financial product in the bureau system at that point. By December 1, 2016, according to the bureau monthly complaint report for that period, the cumulative total had reached approximately 285,000 complaints. In June 2016 alone, 7,032 of 24,500 complaints handled concerned debt collection, placing the category first for that month, according to the bureau monthly complaint snapshot for July 2016. None of the figures, standing alone, supplied a denominator of collection calls, letters, accounts placed for collection, or consumers contacted.

Annual totals added a denominator of a limited kind, namely the total number of complaints handled by the same agency during the same year, and the annual totals showed both persistence and a change in rank. The Consumer Response Annual Report for 2016 recorded 291,400 complaints handled by the bureau during 2016. A bureau summary stated that debt collection accounted for 30 percent of complaints submitted in 2016, ahead of credit reporting at 19 percent and mortgages at 18 percent. The Consumer Response Annual Report for 2017 recorded 320,200 consumer complaints in total, and a published summary of that report in Mondaq, based on the bureau report, placed credit or consumer reporting first at 31 percent, debt collection second at 26 percent, and mortgages third at 12 percent. The annual report on the Fair Debt Collection Practices Act for 2017, issued jointly by the Consumer Financial Protection Bureau and the Federal Trade Commission in March 2018, recorded approximately 84,500 debt collection complaints handled by the bureau during 2017. The fall 2018 semiannual report of the bureau recorded 329,000 complaints received between October 1, 2017 and September 30, 2018, with credit or consumer reporting accounting for 37 percent and debt collection accounting for 25 percent. Debt collection complaints therefore remained among the largest categories throughout the period from July 2013 to September 2018, while the largest single category changed as credit reporting complaints grew. Neither movement, without additional evidence, proved a corresponding movement in underlying conduct.

What do complaint numbers actually measure?

Complaint numbers measure the complaints a particular channel received, classified, and published during a defined period. The numbers record consumer allegations, filing awareness, and collection activity at scale, rather than adjudicated violations. Their subjects indicate recurring points of dispute; their totals alone cannot establish how often unlawful collection conduct occurred in that period.

The channel itself formed part of the measurement. The Consumer Financial Protection Bureau launched the Consumer Complaint Database in June 2012, before debt collection entered the bureau complaint system, and bureau reports described complaint data made public by product, date, geography, and company. The annual report on the Fair Debt Collection Practices Act for 2017 recorded that five sample letters for consumers corresponding with debt collectors had been downloaded more than 517,000 times as of December 2017, and the report described bureau debt collection educational materials as among the most viewed categories in the Ask CFPB tool. The evidence showed an active effort to make consumer rights and complaint routes known, operating alongside the complaint totals. A rise in complaints after a new intake category opened, a database launched, or educational material circulated could reflect greater awareness of a remedy, greater incidence of the disputed conduct, growth in the underlying market, or a combination of the three, and the published total did not separate those components.

Market size supplied a second necessary context. The Consumer Financial Protection Bureau released a survey of consumer experiences with debt collection on January 12, 2017, and the bureau estimated that more than 70 million consumers had been contacted by a creditor or collector about a debt in collection during the prior twelve months. The bureau also estimated in January 2017 materials that more than 6,000 debt collection firms operated in the United States. Against a contact population measured in tens of millions during a twelve month period, a bureau total of approximately 84,500 debt collection complaints handled during 2017 represented the complaints that reached and completed the bureau intake process, rather than a census of disputed contacts. The comparison did not minimize the complaints, because cost, time, confusion, or lack of information could suppress filing. The comparison prevented treatment of an intake total as if the total counted every disputed event in the market.

Classification decisions supplied a third context, because a total depended on product assignment, duplicate treatment, and the handling of incomplete submissions and referrals. The Consumer Response Annual Report for 2017, as summarized in Mondaq, reported that 78 percent of the 320,200 complaints in 2017 were sent to companies for review and response, 16 percent were directed to other regulatory agencies, and 5 percent were classified as incomplete. The fall 2018 semiannual report made the verification limit explicit, in a sentence quoted in an insideARM summary of the report, by stating that the bureau did not verify all the facts alleged in complaints and instead gave companies an opportunity to confirm a commercial relationship with the consumer before providing a substantive response. A published complaint therefore carried the authority of a consumer allegation recorded by a federal agency. A published complaint did not carry the authority of an agency determination that the alleged act occurred or that the alleged act violated the Fair Debt Collection Practices Act.

The subjects of complaints carried more analytical weight than the raw ranking, because the subjects repeated across separate reports, periods, and counting methods. In the March 2016 monthly complaint snapshot, the Consumer Financial Protection Bureau reported that complaints about attempts to collect a debt that the consumer reported was not owed accounted for 38 percent of debt collection complaints submitted to the bureau up to March 1, 2016. The December 2016 monthly complaint report placed the corresponding figure at 39 percent. The annual report on the Fair Debt Collection Practices Act for 2017 gave a more detailed breakdown of the complaints about debts reported as not owed, with 57 percent reporting that the debt was not the consumer debt, 27 percent reporting that the debt had been paid, 11 percent reporting that the debt resulted from identity theft, and 5 percent reporting that the debt had been discharged in bankruptcy, according to a summary of the report published by John C. Hubbard LLC in 2018. The same 2017 report identified written notifications about a debt as the second most common complaint issue and communication tactics as the third most common issue, according to a Consumer Finance Monitor summary of the report published on March 22, 2018. The March 2016 snapshot also described complaints about repeated calls, calls to places of employment, and insufficient information to verify a debt. The recurrence of debts reported as not owed, verification information reported as insufficient, and communication tactics reported as improper across the 2016 and 2017 reports supported a finding about the recurring structure of disputes between consumers and collectors. The recurrence did not establish the share of such disputes in which a collector had acted unlawfully, because a debt could be disputed in good faith by both parties and records could conflict.

Company response data added process information without resolving merits. The summary of the Consumer Response Annual Report for 2017 in Mondaq reported that companies responded to 95 percent of complaints referred to companies during 2017, that 89 percent of complaints referred to companies were closed, and that, across product categories, 75 percent of cases were closed with an explanation, 10 percent were closed with nonmonetary relief, and 4 percent were closed with monetary relief. No closure category functioned as a judgment for the consumer or for the company, and the distribution described how the process ended, rather than how a court or an examiner would have decided the disputed facts.

Enforcement and supervision data differed from complaint data in method and consequence, because an enforcement action rested on an agency investigation and evidence assembled by the agency. On September 9, 2015, the Consumer Financial Protection Bureau announced enforcement actions against Encore Capital Group and Portfolio Recovery Associates, described by the bureau as the two largest debt buyers in the country. The bureau ordered Encore Capital Group to refund consumers up to 42 million dollars, to stop collection efforts on 125 million dollars in debts, and to pay a 10 million dollar penalty. The bureau ordered Portfolio Recovery Associates to refund consumers about 19 million dollars, to cease collection on 3 million dollars in debts, and to pay an 8 million dollar penalty. Those figures measured relief and penalties in two resolved actions, rather than prevalence of the same conduct across the market, and complaint totals for the same period did not prove the conduct in the two actions.

The contrast became clearer at annual scale. The Federal Trade Commission and Consumer Financial Protection Bureau joint release of March 20, 2018, concerning 2017 activities, reported that the Federal Trade Commission filed or resolved ten cases under the Fair Debt Collection Practices Act or the Federal Trade Commission Act against 42 defendants during 2017, obtained more than 64 million dollars in judgments, and banned 13 companies and individuals from working in debt collection. A Consumer Finance Monitor summary of the annual report on the Fair Debt Collection Practices Act for 2017 listed examination findings that included impermissible communications with third parties, deceptive implication that authorized users were responsible for a debt, false representation about the effect of payment on credit scores, and communication outside permitted hours or at times identified as inconvenient. Examination findings came from review of company records by agency examiners, and agreement between the two records carried meaning precisely because the records were generated independently.

The same complaint record supported different institutional readings, and attribution matters for each reading. ACA International, the trade association for the collection industry, published analyses of bureau complaint data that emphasized communication and information failures. A Lexology summary of an ACA International white paper on 2017 complaints reported that ACA International attributed a majority of complaints about debts reported as not owed, and about insufficient written notification, to lack of effective communication and information sharing between lenders and debt collectors, rather than to intentional misconduct by debt collectors. The same summary reported that ACA International placed complaints alleging harassing or illegal practices at 12 percent of debt collection complaints in the 2017 analysis, interpreted the composition as evidence that a majority of debt collectors followed consumer protection laws, and argued that the Consumer Financial Protection Bureau should state the limitations of complaint data more clearly. An earlier ACA International research summary, reported by Auto Remarketing, recommended distinguishing first party from third party collections to avoid double counting and limiting the definition of a complaint to allegations of wrongful conduct.

A consumer advocacy reading of overlapping evidence placed weight on repetition and on the nature of the most common dispute. In a 2016 letter to Consumer Financial Protection Bureau Director Richard Cordray, a coalition of 59 national and state organizations led by the National Consumer Law Center, the National Association of Consumer Advocates, Consumers Union, Americans for Financial Reform, and the Center for Responsible Lending described attempts to collect a debt from the wrong person or in the wrong amount as a pervasive problem, according to a summary of the letter published by Americans for Financial Reform. The coalition urged the bureau to require collectors to review original account level documentation before initiating collection and to adopt stricter standards against lawsuits and default judgments based on faulty documentation. The advocacy reading treated the repeated subjects as evidence of a systemic documentation problem in the transfer and collection of debts. Neither reading could derive a violation rate from complaint totals without evidence drawn from examinations, litigation, surveys, or company records.

The durable finding from the complaint record, stated within the limits of the record, concerned volume, persistence, and the location of friction. Between July 2013 and September 2018, consumers filed debt collection complaints with the Consumer Financial Protection Bureau in numbers that placed the category first in several monthly and annual tabulations and second in the annual tabulations for 2017 and for the twelve months ending September 30, 2018. The most repeated subjects, attempts to collect debts that consumers reported were not owed, insufficient information to verify debts, and communication tactics that consumers reported as improper, appeared in bureau reports for 2016 and 2017 under different counting frames. The repetition identified where the collection system generated disputes that consumers carried to a federal intake channel. The totals did not count violations and did not distribute responsibility across creditors, collectors, debt buyers, and collection law firms. Complaint volume measured a real phenomenon, the filing of complaints at scale over a sustained period, and the phenomenon mattered for consumer credit law because the filings persisted under a statute that had prohibited abusive, deceptive, and unfair collection practices since 1977. The complaint totals, read with discipline, marked the questions that such evidence had to answer.

Finding Five: Mortgage Credit and the Access Question

Finding five is the contested finding, and the contest concerns causation rather than the direction of the underlying facts. By every broad measure available through the end of 2018, mortgage credit was far harder to obtain for a borrower with a weaker credit profile than before the financial crisis, and the contraction for such borrowers was deeper and longer lasting than the contraction for borrowers with strong profiles. The Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010, Public Law 111-203, imposed an ability to repay requirement on mortgage lenders, and the Consumer Financial Protection Bureau rule implementing that requirement took effect on January 10, 2014, years after underwriting had begun to tighten in response to rising defaults. Because the rule entered a market already changed by lender, investor, guarantor, and borrower behavior independent of the statute, the evidence resists clean attribution.

The statutory command answered a feature of the boom preceding the crisis of 2007 to 2009, in which substantial mortgage credit had been extended on terms depending for repayment on refinancing, on house price appreciation, or on a low introductory payment rather than on borrower income. Sections 1411 and 1412 of the Dodd-Frank Act amended the Truth in Lending Act to prohibit a creditor from making a residential mortgage loan without making a reasonable and good faith determination, based on verified and documented information, that the consumer had a reasonable ability to repay the loan according to its terms, together with applicable taxes, insurance, and assessments. The duty was enforceable through borrower damages claims and an ability to repay defense in foreclosure.

The Consumer Financial Protection Bureau issued the final rule on January 10, 2013, with an effective date of January 10, 2014. The rule required a creditor, before consummation, to consider and verify eight subjects: income or assets; employment status; the monthly payment on the loan being made; the monthly payment on any simultaneous loan secured by the same dwelling; monthly payments for mortgage related obligations such as property taxes and insurance; debts, alimony, and child support obligations; the monthly debt to income ratio or residual income; and credit history. The rule prescribed the subjects of inquiry without prescribing a single underwriting model, a minimum credit score, or a maximum ratio for every loan under the general standard. A creditor could comply through individualized assessment outside the qualified mortgage category, and nonqualified lending remained lawful after January 2014, so the legal risk of that route, rather than any prohibition, formed the center of the later argument about access.

The qualified mortgage category supplied the compliance center of the design. A loan generally qualified if income and assets were verified, if the term did not exceed thirty years, if payments were regular and periodic, and if the loan lacked negative amortization, interest only payments, and, subject to limited exceptions for certain small creditors, balloon payments. Points and fees generally could not exceed three percent of the loan amount for a larger loan, with different limits for smaller loans, and for a general qualified mortgage the monthly debt to income ratio could not exceed forty three percent. A qualified mortgage that was not higher priced, meaning in general a first lien loan priced no more than one and a half percentage points above the average prime offer rate, received a safe harbor: the lender was conclusively presumed to have satisfied the ability to repay requirement. A higher priced qualified mortgage received a rebuttable presumption of compliance, which a borrower could overcome by showing insufficient residual income or assets at origination. For a fuller account of the statute across banking markets, the Dodd-Frank banking impact assessment traces effects beyond mortgage origination.

One structural feature softened the forty three percent limit across the bulk of originations and complicates every aggregate estimate. A temporary provision, known as the GSE patch, treated a loan as a qualified mortgage despite a ratio above forty three percent if the loan was eligible for purchase by Fannie Mae or Freddie Mac or for insurance or guarantee by a federal agency, provided the other conditions were met. Agency insured loans and portfolio loans by small creditors received separate treatment under agency and small creditor rules. The patch was temporary by design, scheduled to run for seven years from the effective date, and agency backed lending accounted for a large majority of originations in the years after 2008. The strictest version of the definition therefore governed mainly the nonconforming segment, largely jumbo and other agency ineligible loans, while the segment serving most first time buyers operated under agency standards incorporated by reference.

The measured record begins well before the rule. The Federal Reserve Senior Loan Officer Opinion Survey on Bank Lending Practices recorded the initial turn: in the survey released in November 2007, about forty percent of domestic respondents reported tightening standards on prime residential mortgages over the preceding three months, and sixty percent of banks originating nontraditional mortgages reported tightening on those loans. By February 2010, banks had generally ceased tightening on many loan types but had yet to unwind the considerable tightening of the preceding two years. Tightening therefore preceded enactment in July 2010, issuance of the final rule in January 2013, and the effective date in January 2014, and a borrower denied credit in 2015 faced layers of revised agency standards, investor and guarantor requirements, lender overlays, and the ability to repay rule.

The Urban Institute Housing Credit Availability Index supplied the most widely cited summary measure. Researchers at the Housing Finance Policy Center of the Urban Institute defined the index as the percentage of owner occupied home purchase loans likely to default, meaning to go unpaid for more than ninety days past the due date. A lower reading indicates less tolerance of default risk and tighter standards, and a higher reading indicates greater tolerance and easier access. The release covering the third quarter of 2016, published in January 2017, reported overall availability of 5.1 percent, edged up from 5.0 percent in the preceding quarter. The same releases set the benchmark for 2001 to 2003, treated as a period of reasonable standards before the loosest boom years, at 12.5 percent for the whole market, so availability in the third quarter of 2016 stood at less than half that benchmark. Risk taken in the government channel and the portfolio and private label securities channel remained close to record lows in the same release.

The distribution of lending by credit score showed where the contraction fell. Laurie Goodman of the Urban Institute, writing in 2017 in the Boston College Journal of Law and Social Justice, summarized calculations by Goodman with Jun Zhu and Bing Bai comparing lending in 2015 with lending in 2001. The number of loans to borrowers with scores above 700 had fallen 1.4 percent, the number to borrowers with scores from 660 to 700 had fallen 20.3 percent, and the number to borrowers with scores below 660 had fallen 64.9 percent. In a separate analysis published in February 2016, Goodman, Zhu, and Bai estimated 5.2 million additional mortgages from 2009 to 2014 if standards similar to those of 2001 had applied, and 1.1 million additional loans in 2015 alone, bringing the cumulative estimate for 2009 to 2015 to 6.3 million loans not made. Goodman acknowledged a limitation that matters for attribution: the calculation likely overstated the supply effect because the method conflated supply and demand, since households that had watched foreclosures among relatives and neighbors may have sought credit less readily regardless of lender standards.

Did the ability to repay rule cut off mortgage credit?

The evidence available by the end of 2018 did not establish a marketwide cutoff. Federal Reserve research found small aggregate effects in 2014, concentrated effects for nonconforming high debt loans thereafter, and a temporary GSE exemption that shielded most originations from the strictest debt limit.

The most precise Federal Reserve evidence cut in two directions, and the divergence between the two leading studies is itself informative. Neil Bhutta, Nathaniel Popper, and Daniel Ringo, in a FEDS Notes analysis dated December 29, 2015, used Home Mortgage Disclosure Act data to test for discontinuities at the January 10, 2014 application date, at lender size thresholds, and at the pricing threshold for safe harbor treatment. Bhutta, Popper, and Ringo found that the higher priced share of refinance loans fell at the threshold, with a larger decline for small creditors, and that small creditor lending volume dropped at the elevated safe harbor threshold of three hundred fifty basis points above the average prime offer rate in 2014, with no comparable drop in 2013. The magnitudes were small marketwide: among creditors other than small creditors, the higher priced share of refinance loans fell by less than one percentage point, the higher priced share of home purchase loans showed no discernible drop, no case showed a discontinuity in the rate of high debt to income loans, and other creditors showed no evidence of increased denial rates. Small creditors, where effects were larger, accounted for seven percent of the market, and the authors characterized the response as minor in economic significance.

Aurel Hizmo and Shane Sherlund, in a FEDS Notes analysis dated November 16, 2018, reached a stronger conclusion for the segment where the limit actually applied, using a comparison that exploited the patch. Hizmo and Sherlund used Optimal Blue data from January 2013 through September 2018, covering more than one fourth of home purchase applications, and compared nonconforming loans, subject to the forty three percent criterion, with conforming loans, exempt under the patch. The share of nonconforming applications above forty three percent that proceeded to rate lock fell by about sixty percent in the year of implementation and remained low through the middle of 2017, while the conforming share showed no comparable change. On pricing, accepted nonconforming applications above the threshold carried rates as much as thirty to forty basis points higher than applications at or below the threshold, against one to two basis points for conforming loans. Hizmo and Sherlund interpreted the sequence as rationing followed by pricing: lenders initially restricted credit to nonqualified borrowers, and as lenders learned the costs and risks of the rule, the premium rose and the high debt share partly recovered.

Survey evidence supplied a third kind of measurement, closer to the decision point and weaker in causal discipline. The Federal Reserve put specialty questions on the new rules to respondents in the Senior Loan Officer Opinion Survey conducted in July 2014. More than half of loan officers responding on the subject reported that the rules had reduced approval rates for prime conforming, prime jumbo, and nontraditional mortgages. Among respondents reporting no effect, about half reported that policies would have been tighter in any event, a response the survey reporting attributed mainly to adjustments Fannie Mae and Freddie Mac had made to automated underwriting models in 2008. The American Bankers Association Real Estate Lending Survey, released in June 2015 and based on 182 banks, seventy seven percent holding assets below one billion dollars, reported that eighty percent of respondents expected the Bureau mortgage rules to continue to constrict credit, that ninety percent of loans made by the typical bank in 2014 were qualified mortgages, and that high debt to income levels were the most likely reason a loan failed to meet qualified mortgage standards, followed by lack of required documentation.

The Urban Institute reading, at its strongest, held that the rules explained less of the tightness than the incentives surrounding the rules, and that the credit box could expand substantially without a return to boom era risk. Goodman argued that credit was tight in large part because originators imposed overlays above the underwriting boxes of Fannie Mae, Freddie Mac, and the Federal Housing Administration, so that an agency willing to underwrite a borrower with a score of 620 faced an originator requirement of 660 in the example Goodman used. Goodman identified three sources of that reluctance: representation and warranty risk, under which a guarantor or insurer discovering a documentation defect after default could force repurchase or refuse to honor insurance; litigation risk, particularly enforcement under the False Claims Act; and the high and uncertain cost of servicing delinquent loans. Goodman supported the implication with performance evidence: loans originated from 2011 through the second quarter of 2015 defaulted at exceptionally low rates in the data Goodman examined, and loans with the same characteristics performed better than comparable loans from 1999 to 2003, a pattern attributed to extraordinary underwriting attention.

The lender side reading, at its strongest, held that the legal architecture predictably produced exactly the behavior the surveys recorded, and that stated lender policy deserved weight as evidence. A nonqualified mortgage was lawful in theory, but lawfulness did not make the product commercially equivalent to a qualified mortgage, because the lender lost the safe harbor, accepted exposure to borrower litigation for the life of the loan through the foreclosure defense, and faced uncertainty about how a court would evaluate residual income and documentation years after origination. Faced with that asymmetry, a bank could rationally confine lending to the qualified mortgage box or restrict nonqualified lending to borrowers the bank knew well, and the association surveys reported each response in substantial proportions. Hizmo and Sherlund supplied independent support where the rule bound without the patch: a sixty percent fall in high debt nonconforming applications reaching rate lock in the implementation year is difficult to explain as coincidence, because the conforming comparison group, subject to the same market and interest rates, showed no comparable fall. On this reading, the absence of a large aggregate effect in 2014 proved little, because the patch deferred the binding constraint for most of the market.

Several confounders prevent either reading from converting shared facts into a settled causal share. Demand changed alongside supply, as Goodman acknowledged: households that had lost homes, or that carried damaged credit records from crisis era delinquency and foreclosure, were in many cases ineligible under agency waiting periods or unable to accumulate a down payment, and other households chose renting or delayed purchase after observing crisis losses. Investor requirements changed alongside regulation as well. Putback risk operated as an independent force with a timeline overlapping the rule, and clarification efforts by the Federal Housing Finance Agency and the government sponsored enterprises unfolded over several years without fully removing originator uncertainty.

The asymmetry with the credit card evidence in the earlier findings deserves emphasis, because the difference lies in experimental design rather than in researcher diligence. A card rule that changes a price or practice across a defined portfolio at a defined date, against a background market that had not already collapsed, leaves a comparatively legible before and after comparison. The mortgage rule offered no such baseline. Mortgage credit had contracted from 2007 onward, the statute passed in the middle of that contraction in 2010, the rule issued in 2013 after lenders had anticipated its broad shape for years, and the effective date in 2014 fell in a market where agency standards, investor standards, guarantor enforcement, and lender planning had each moved repeatedly since 2007. Anticipation blurred the implementation date as a test, and the patch blurred the treatment itself, because most loans were exempt from the defining debt limit while a minority segment bore the full design.

The finding therefore stands as contested in a precise sense. The contraction for borrowers with weaker profiles, its persistence through 2018 at less than half the default risk tolerance recorded for 2001 to 2003 on the Urban Institute index, and the concentration of the fall among borrowers below a score of 660 are established measurements no party disputed. The proposition that the rules caused some restriction at identifiable margins, in pricing above the forty three percent threshold in the nonconforming segment, in safe harbor pricing among small creditors, and in stated policies of banks confining lending to the qualified mortgage box, rests on converging Federal Reserve research and lender surveys. The proposition that the rules caused most of the contraction, or that credit would have returned to the standard of 2001 in their absence, rests on no comparable convergence, because the Urban Institute analysis attributed the larger share of tightness to overlays driven by putback risk, litigation risk, and servicing costs, and because the contraction predated the rules by years. Between those poles, the causal share attributable to the rules, as distinct from post-crisis lender risk appetite, investor requirements, and putback risk, remained genuinely unresolved on evidence published through the end of 2018.

The Counter-Reading: What Disclosure Still Does

The findings assembled in the preceding sections point in one direction with unusual consistency. Mandated disclosure, as practiced across consumer credit markets from the enactment of the Truth in Lending Act in 1968 onward, failed to produce the informed borrower that the statutory design presupposed. Borrowers signed forms without reading them in full and read forms without grasping the price terms that mattered most. Taken alone, that record invites a blunt verdict, namely that disclosure itself was a mistaken instrument. The evidence, examined in full, does not support that verdict. A narrower conclusion emerges when three functions of disclosure are separated from the inflated claim that disclosure alone could discipline a credit market: comparison shopping in standardized markets, the creation of an enforcement record, and the distinction between a flawed design and a flawed concept. In those three functions, the record contains positive results that the negative findings do not cancel.

The first function concerns the borrower who does shop. Much of the critical literature measured the average borrower, who faced time pressure, limited numeracy, and a seller with no interest in slowing the transaction for study. Averages conceal distribution. In any large market a minority compares offers or uses a disclosed price term in negotiation, and the competitive effect of that minority extends beyond its own transactions. A lender quoting terms in a standardized format faces the knowledge that the quotation can be carried to a competitor in legible form. The relevant question is therefore not whether disclosure transformed the conduct of the typical borrower at the closing table. The question is whether standardized disclosure measurably improved performance in the act of comparison when comparison was attempted under controlled conditions. On that question, the mortgage disclosure testing conducted for the Consumer Financial Protection Bureau supplied the strongest evidence available before 2019.

The Bureau inherited a concrete assignment. The Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010 directed the Bureau to combine the disclosures required under the Truth in Lending Act and the Real Estate Settlement Procedures Act for mortgage transactions, two regimes that had produced overlapping forms with duplicative and sometimes inconsistent information. Beginning in May 2011, the Bureau and the Kleimann Communication Group, the contractor engaged to conduct consumer testing for the Bureau, subjected successive prototypes to qualitative testing with consumers and industry participants in rounds held in Baltimore, Los Angeles, Chicago, Springfield, Albuquerque, Des Moines, Birmingham, Philadelphia, and Austin, with a final confirmation round in Baltimore in March 2012. The Kleimann Communication Group documented that evolution in a report issued in July 2012 under the title Know Before You Owe: Evolution of the Integrated TILA-RESPA Disclosures. Participants worked through sample transactions and compared prototype Loan Estimate and Closing Disclosure forms. The Bureau reported that participants comprehended key features of a loan, compared similar Loan Estimates, and selected the loan that fit a stated personal situation, and that experienced and inexperienced participants alike identified increased costs, new fees, and fees that had shifted categories when an estimate and a closing disclosure were placed side by side.

Qualitative results establish usability rather than superiority over the forms in force, and the Bureau sought a controlled comparison after the proposed rule was published. Between April 23, 2013 and June 13, 2013, the Bureau, with assistance from the Kleimann Communication Group, conducted a quantitative validation study with 858 consumers in 20 locations across the country. One group used the forms then in force; another used the new Loan Estimate and Closing Disclosure. The Bureau reported in November 2013 that participants using the new forms provided more correct answers than participants using the prior forms, an improvement of 28.8 percent with a margin of error of plus or minus 4.7 percentage points, across experienced and inexperienced consumers and across fixed rate and interest only adjustable rate loans. Detail published by the Bureau on the Loan Estimate showed the same pattern for individual items. In the post proposal testing, 99 percent of consumers using the Loan Estimate identified the loan amount, compared with 61 percent using the existing forms, and 90 percent stated correctly when the interest rate could change, compared with 81 percent using the existing forms. In its announcement of the final rule, the Bureau stated that the new forms performed better than the prior forms on comparing competing offers by as much as 42 percent, and better on comparing estimated terms with final terms by as much as 28 percent.

Those figures require disciplined reading. A test in which a participant sits with a form does not reproduce a closing conducted under time pressure, and a gain in correct answers does not convert dollar for dollar into savings in a market transaction. The testing also compared a redesigned form against predecessors that had accumulated requirements under two statutes; the result demonstrated what redesign could achieve, not what the average mandate achieved without such investment. Within those limits, the Know Before You Owe testing remains the clearest verified demonstration in the federal record before 2019 that standardized disclosure can help the consumers who use it for its central intended purpose. The final rule issued on November 20, 2013 produced a three page Loan Estimate, delivered within three business days after application and replacing the early Truth in Lending statement and the Good Faith Estimate, and a five page Closing Disclosure, delivered at least three business days before closing and replacing the final Truth in Lending statement and the HUD-1 settlement statement, with the rule taking effect for applications on October 3, 2015. A borrower holding two Loan Estimates held documents in identical structure, with the same terms in the same positions on the page. That structural identity was the mechanism the testing validated.

Is disclosure therefore worthless?

No. The testing and enforcement record shows a narrower failure. Disclosure performed poorly as a sole protection against complex, high pressure selling, and performed measurably better as a standardized record for comparison, supervision, and litigation. The distinction preserves the criticism without discarding a tool that other protections depend upon in practice.

The second function explains why the distinction matters outside a testing room. Disclosure creates a documentary record, and a record has legal consequences that a sales conversation does not have. An oral statement leaves the parties to contest recollection, emphasis, and context. A standardized statement on a prescribed form, retained under a retention requirement and reproducible in an examination or a courtroom, fixes the terms the creditor chose to present at a defined moment. Supervisors can compare disclosed rates and charges against the underlying notes without reconstructing each negotiation. Enforcement attorneys and private plaintiffs can plead a violation from the face of a document, and under the Truth in Lending Act the private action was central to the design rather than incidental to it.

The statutory remedies attach to the documents. Under Section 1640 of Title 15 of the United States Code, a creditor that failed to comply with the disclosure requirements of the Act faced liability for actual damages, for statutory damages in individual and class actions within limits set by the statute, and for the costs of a successful action together with a reasonable attorney fee as determined by the court. Statutory damages and attorney fees altered the economics of enforcement, because a violation causing a small provable loss could support a claim that an unassisted borrower would never have pursued on actual loss alone. The rescission remedy in Section 1635 supplied a sharper example. For certain transactions secured by a principal dwelling, a borrower held a right to rescind for three business days after consummation, and that period could extend to as long as three years when the required notice of the right to rescind or the material disclosures defined by the statute were not delivered or were inaccurate. A rescission claim therefore turned on a documentary question, namely what was handed over, in what form, and on what date. The enforcement system that Congress constructed, combining agency supervision, agency enforcement, and private litigation, ran on the disclosure record. Weakening the record would not have strengthened the other protections; the other protections drew evidence from the record.

The enforcement function also reframes a familiar objection. Much criticism observed correctly that few borrowers read the forms, and inferred that the forms served no one. The inference overlooked other readers: the examiner comparing a sample of files, the enforcement attorney testing a pattern across files, the legal aid lawyer advising a borrower after a default notice arrived, and the competitor deciding whether a quoted price could be matched. A disclosure regime produces an archive of transaction terms as a byproduct of each private delivery, and terms stated in divergent formats cannot be aggregated, audited, or compared at scale. On this function the relevant measure is not comprehension by the recipient on the day of delivery but verifiability by a third party after the fact, and on that measure the disclosure system performed work that no purely substantive rule performed on its own.

The third function requires a separation between design and concept, and the leading critical work supplies the terms for that separation even where its authors pressed the argument further. Omri Ben-Shahar and Carl E. Schneider developed their critique in an article published in the University of Pennsylvania Law Review in 2011 and in the book More Than You Wanted to Know: The Failure of Mandated Disclosure, published by Princeton University Press in 2014. The core of that critique, in summary, was accumulation and mismatch. The law mandated disclosures in transaction after transaction until no recipient could attend to the total volume; disclosures arrived when a decision had in substance been made or when attention was scarcest; the information reflected categories lawyers and regulators considered material rather than the few facts a decision maker could absorb; and complexity placed a ceiling on what clearer expression could achieve, because a complex product cannot be rendered simple by plainer sentences alone. Ben-Shahar and Schneider further contended that the political convenience of disclosure displaced harder regulatory choices, and that many recipients made a rational decision to ignore material that experience had taught them rarely changed an outcome. Each element described mandated disclosure as accumulated and administered in practice. None established that informing a transaction was in itself a worthless aim.

The distinction has practical consequences, because failures of accumulation, timing, and complexity admit of different remedies than failure of the underlying idea would admit. The repair tradition inside the federal regime worked on exactly those variables. That tradition shortened forms and removed duplication, as the integration of the two mortgage regimes removed an entire layer of overlapping paperwork. That tradition required testing with consumers before adoption, as the Bureau did in the qualitative rounds from 2011 to 2012 and in the quantitative validation study in 2013, rather than treating a drafting committee judgment as sufficient evidence that a form communicated. That tradition favored layered and summary presentation, in which a short statement of the price terms and risk features that governed comparison appeared at the front and supporting detail followed, a structure embodied in the first page of the Loan Estimate. That tradition also treated timing as a design variable, most visibly in the requirement that the Closing Disclosure arrive at least three business days before closing, a rule that converted disclosure from a document signed in a stack at the closing table into a document that could be questioned and compared against the earlier estimate. Ben-Shahar and Schneider expressed skepticism that simplification could overcome the complexity of the products themselves, and the record gives that skepticism weight where product complexity was itself the risk. Skepticism about the sufficiency of a repair, however, differs from proof that repaired and unrepaired versions performed identically, and the Bureau validation study measured a difference between them.

Weighed together, the five findings that structure this article and the counter evidence in this section leave the question in a specific place. Disclosure as a primary instrument, asked to protect borrowers on its own against complex products and unequal bargaining power, performed weakly across five decades of federal consumer credit law, and attempts to rescue that role by adding further disclosures generally deepened the accumulation that undermined it. Disclosure as supporting infrastructure performed a different and verifiable set of tasks: the Bureau testing of 2011 to 2013 showed standardized forms improving comparison by the consumers who compared; the Truth in Lending Act remedies showed a disclosure record sustaining supervision, public enforcement, and private litigation that substantive standards alone would have left without documentary footing; and the repair tradition showed failures of length, timing, duplication, and presentation yielding to redesign and testing even where the deeper problem of product complexity remained. That division, weak as a primary instrument and valuable as supporting infrastructure, is the honest finding the full record supports, and it counsels neither the abandonment of disclosure nor a return to reliance upon it as a substitute for substantive protection.

The five-finding evidence table gathers each finding with the measured outcome, the direction of the effect, the evidence period, the principal sources, and the rating the record supports.

Finding	Measured outcome	Direction	Evidence period	Principal sources	Rating
Disclosure comprehension	Mandated disclosures frequently unread and weakly understood	Weaker than the theory predicted	Studies from 1999 to 2018	Ben-Shahar and Schneider (2014), Lacko and Pappalardo (2007)	Settled in direction, contested in degree
CARD Act cost effect	Borrowing costs fell an annualized 1.6 percent of average daily balances, an estimated 11.9 billion dollars a year, concentrated among borrowers with FICO scores below 660	Consumer saving, no measured offset	2008 to 2012 account data, study published 2015	Agarwal, Chomsisengphet, Mahoney, and Stroebel (2015)	Settled for the period studied
Credit report accuracy	26 percent of study participants identified a potentially material error, 21 percent had a correction after disputing, 5 percent had a score change large enough to move credit tier	Material error common, severe error uncommon	Study published 2012, follow-up 2015	Federal Trade Commission section 319 study (2012) and follow-up (2015)	Settled
Collection complaints	Debt collection was the largest complaint category in 2016 at 30 percent of complaints handled and second in 2017 at 26 percent	Complaint volume high and persistent, not a violation count	2013 to 2018	Consumer Financial Protection Bureau complaint snapshots and annual reports	Settled as complaint data
Mortgage credit access	Availability for weaker credit profiles far below pre-crisis levels, causal share of the ability-to-repay and qualified mortgage rules unresolved	Contraction measured, attribution disputed	2014 to 2018	Urban Institute Housing Credit Availability Index, Federal Reserve research notes	Contested

What the Record Supports

The combined record supports a division of labor that the federal sequence reached only after four decades. Disclosure, assigned the primary protective role in 1968, performed weakly in that role across every direct measurement assembled in the findings. Mandated forms went unread in large numbers, the annual percentage rate was misunderstood at the threshold level that comparison required, and comprehension fell as product complexity rose. The failure had structural causes rather than a single drafting defect: volume accumulated across statutes, delivery arrived after decisions had hardened, and the mathematics of compounding exceeded the preparation that a transaction document could supply. A legislature relying on disclosure alone to discipline the price of credit relied on the weakest link in the chain, the attention and calculation of a borrower at the moment of commitment, and the measurements from 1999 through 2018 recorded the consequence with unusual consistency.

Substantive rules, assigned a major role only in 2009, delivered the largest measured benefit in the record. The CARD Act fee limits saved an estimated 11.9 billion dollars a year on the Agarwal, Chomsisengphet, Mahoney, and Stroebel estimate, reduced costs most for borrowers with the lowest credit scores, and produced no measured offset in interest charges, credit limits, or balances during the period studied. The mechanism explains the distribution. A prohibition removes a charge for every account holder subject to the charge, including account holders who never read a statement, never compared an offer, and lacked the credit standing to move an account elsewhere. The internal comparison inside the same statute reinforces the point: the minimum payment disclosure improved a description and moved aggregate payments by an amount bounded near zero, while the fee limits removed charges and moved billions of dollars. Where a price component escapes attention at the point of sale, the evidence favors removing or capping the component over describing the component more clearly.

The middle findings counsel precision about what other instruments measure. The accuracy regime of the Fair Credit Reporting Act delivered high average accuracy and left detection and persistence to the consumer, so a minority carrying a material error bore costs in price and eligibility that the procedural standard did not prevent. The complaint regime delivered a map of friction rather than a count of violations, valuable for locating recurring disputes over debts reported as not owed and valuable for supervision, and misleading when read as a violation rate. Researchers, students, and practitioners tracing claims of this kind across statutes, studies, and enforcement records face the same accumulation problem in miniature, and an orderly working method helps: keep your statute notes, citations, and case chronologies together free on VaultBook. Order in the evidence matters because each source in the record answers a different question, and a rating earned by one kind of evidence cannot be transferred to another kind without loss.

The limits of the verdict deserve equal emphasis. The card evidence, the strongest causal evidence in the record, covers one market and one period: consumer accounts at eight large banks observed from 2008 through 2012, with small business accounts supplying the comparison group and a parallel trends assumption carrying the design. Entry, exit, and contract margins over longer horizons lay outside the data, and a result measured in a market with low fee salience does not transfer automatically to markets where prices are salient at purchase. The mortgage question stays contested for structural reasons: tightening predated the ability to repay rule by years, a temporary exemption shielded most originations from the strictest debt limit, and lender overlays driven by repurchase and litigation risk supplied an independent cause that aggregate figures cannot separate from the rule. The honest rating for mortgage access remains contested, and the honest rating for disclosure comprehension remains settled in direction and contested in degree.

The tradition backwards therefore ends with an assignment rather than a slogan. Disclosure belongs in the supporting position: standardizing the vocabulary of price, enabling comparison for borrowers who shop, and creating the documentary record on which supervision, enforcement, and litigation depend. Substantive rules belong in the primary position for conduct and for pricing practices that operate after signing and outside borrower attention. Accuracy belongs to institutional duties with audited outcomes rather than to consumer detection alone, and complaint data belongs in supervision as a signal rather than in argument as a verdict. Congress built the structure in reverse order between 1968 and 2009, and the measured record, read within its limits, shows the cost of the order and the value of the correction.

Frequently Asked Questions

Q: Did consumer credit law disclosure actually work?

Disclosure worked at standardization and failed at the task the Truth in Lending Act assigned to informed choice. Creditors across the market came to state the finance charge and the annual percentage rate in a common vocabulary, and that vocabulary created a record that regulators, researchers, and competitors could inspect. The human chain behind the statutory design performed far less reliably. Ben-Shahar and Schneider, in More Than You Wanted to Know, published in 2014, concluded that mandated disclosure ordinarily failed to inform and failed to improve decisions. Lacko and Pappalardo, in Federal Trade Commission testing published in 2007, found borrowers completing mortgage transactions without a secure grasp of loan terms. The measured verdict separates the two functions: disclosure built a common language for credit cost, and comprehension of that language arrived too unevenly to discipline prices on its own.

Q: Do people understand APR under consumer credit law?

Understanding of the annual percentage rate fell short of the level that comparison shopping required. Lee and Hogarth, in a study published in the Journal of Public Policy and Marketing in 1999, analyzed survey responses from consumers who had applied for a mortgage or a refinance and found that 40 percent did not understand that the annual percentage rate, as the broader cost measure, is the same as or greater than the contract rate. Recognition performed better than judgment in later testing. Chin and Bruine de Bruin reported in a 2018 manuscript that participants identified the absolute rate stated in a card disclosure accurately, while estimates of the cost of that rate relative to the card market differed from actual relative costs by 47 percentage points on average. A borrower could therefore read the mandated figure correctly and misjudge whether the figure represented cheap or expensive credit.

Q: How many credit reports contain errors under consumer credit law?

The Federal Trade Commission section 319 study, published in 2012, supplied the most systematic measurement. In that study, 26 percent of participants identified at least one potentially material error on at least one of three reports. After participants disputed the information, 21 percent had a modification made to a report. A smaller group faced consequences for price and eligibility: 5 percent had a score increase large enough to move into a better credit risk tier after correction. The study worked with 1,001 participants and 2,968 reports from the three national bureaus, and trained associates helped participants review the files, so the figures describe errors that assisted review could surface. The pattern that emerges is one of high average accuracy joined to a concentrated burden, because a material error affected roughly one consumer in four by allegation and changed a risk tier for one consumer in twenty.

Q: Did the credit card rules reduce costs under consumer credit law?

The credit card rules reduced measured borrowing costs, and the strongest evidence came from Agarwal, Chomsisengphet, Mahoney, and Stroebel in the Quarterly Journal of Economics in 2015. Using account data from about 160 million accounts at eight large banks between 2008 and 2012, and comparing regulated consumer cards with unregulated small business cards, the authors estimated that the CARD Act fee limits reduced borrowing costs by an annualized 1.6 percent of average daily balances. The reduction exceeded 5.3 percent of average daily balances for account holders with FICO scores below 660. Extrapolated across the market, the reduction equaled an estimated 11.9 billion dollars per year. The authors found no evidence of an offsetting increase in interest charges, so the saving represented a lower total price for credit rather than a shift from fees into interest during the period studied.

Q: Did consumer credit law reduce access to credit?

The mortgage evidence records a contraction in access and leaves the causal share of the rules contested. The Urban Institute Housing Credit Availability Index stood at 5.1 percent for the third quarter of 2016, against a benchmark of 12.5 percent for 2001 to 2003, and lending to borrowers with scores below 660 fell far more sharply than lending to borrowers with strong scores. Federal Reserve research, however, found small marketwide effects from the ability to repay rule in 2014, and a temporary exemption shielded most originations from the strictest debt limit. Tightening also predated the rule by years, and analyses from the Urban Institute assigned substantial weight to lender overlays driven by repurchase risk, litigation risk, and servicing costs. Contraction is a settled measurement. Attribution of that contraction to the rules, as distinct from post crisis lender behavior, remains contested.

Q: Which works better in consumer credit law, disclosure or substantive rules?

The CARD Act supplied an internal comparison because the statute combined both instruments. The fee limits, a substantive instrument, reduced borrowing costs by an annualized 1.6 percent of average daily balances, an estimated 11.9 billion dollars a year, in the Agarwal, Chomsisengphet, Mahoney, and Stroebel study published in 2015. The minimum payment disclosure in the same statute, an information instrument, raised the share of account holders paying at a rate sufficient to retire a balance within thirty six months by 0.4 percentage points, and the authors estimated an upper bound for the aggregate interest saving at about 57 million dollars a year, or 0.01 percent of average daily balances. On measured savings in that market and period, substantive rules outperformed disclosure by a wide margin. The comparison addresses price outcomes, and disclosure retained separate value as a record for comparison, supervision, and enforcement.

Q: Is debt collection a top complaint under consumer credit law?

Debt collection ranked at or near the top of the federal complaint record, and the ranking measures allegations rather than violations. The Consumer Financial Protection Bureau Consumer Response Annual Report for 2016 recorded debt collection as the largest complaint category, at 30 percent of complaints handled during that year. The corresponding report for 2017 placed debt collection second, at 26 percent, behind credit reporting. Bureau reports stated that the bureau did not verify all facts alleged in complaints and gave companies an opportunity to confirm a commercial relationship before a substantive response. A complaint therefore records a consumer allegation received and classified by a federal channel. The persistence and volume of collection complaints identify a durable point of friction in the collection system, and adjudicated violations require evidence from examinations, enforcement actions, or litigation rather than from complaint totals.

Q: Is there a federal payday lending law in consumer credit law?

No comprehensive federal payday lending statute governed the market for the general public during the period covered by the federal record through 2018, and state law carried most payday regulation, including licensing, rate limits, and prohibitions that varied across states. Two federal measures reached parts of the market. The Military Lending Act imposed a 36 percent rate cap on covered payday and similar loans made to active duty servicemembers and covered dependents, a protection that applied to a defined population rather than to borrowers generally. The Consumer Financial Protection Bureau issued a payday lending rule in 2017 under federal unfair practices authority, and the main provisions of that rule carried a compliance date set after issuance, so the rule formed part of the federal framework without amounting to a comprehensive payday statute enacted by Congress. State law remained the primary source of payday lending rules for most borrowers.

Q: Why do mandated disclosures fail to change consumer decisions?

Ben-Shahar and Schneider, in the 2014 book More Than You Wanted to Know, located the failure in structure rather than in drafting quality alone. Complexity meant that a faithful disclosure had to describe branching contingencies, while a readable disclosure had to omit or compress those branches. Accumulation meant that each new mandate competed with every earlier mandate for a finite stock of attention. Timing meant that delivery at application or closing often arrived after a decision had hardened through negotiation, fees, and moving plans. The demands on recipients completed the problem, because effective use required skill with percentages, compounding, and amortization, together with forecasts of income, residence, and repayment behavior. McDonald and Cranor estimated in a 2008 study that reading the privacy policies encountered in a year would require about 244 hours, a measure from an adjacent field that illustrates why complete reading across all mandated documents ceased to be a plausible expectation.

Q: Did the CARD Act reduce the availability of credit cards?

The account level evidence found no measured contraction in availability for existing accounts during the period studied. Agarwal, Chomsisengphet, Mahoney, and Stroebel, in the Quarterly Journal of Economics in 2015, estimated an effect of precisely zero on credit limits and on average daily balances after the CARD Act fee limits took effect. The estimate for new account formation was also zero, although the standard errors were too large to rule out meaningful effects in either direction, so the statement about new accounts carries less confidence than the statement about existing accounts. Industry commentary predicted and later asserted a contraction. The American Bankers Association, in a February 2013 letter to the Consumer Financial Protection Bureau, maintained that card credit had become less available, particularly for subprime borrowers, relying on market aggregates from a period dominated by the financial crisis and the recession rather than on a comparison with an unregulated control group.

Q: How did the ability-to-repay rules affect mortgage lending to borrowers with weaker credit?

The effects concentrated where the qualified mortgage debt limit actually applied. Hizmo and Sherlund, in a Federal Reserve FEDS Notes analysis dated November 16, 2018, compared nonconforming loans subject to the forty three percent debt limit with conforming loans exempt under a temporary provision. The share of nonconforming applications above the limit that proceeded to rate lock fell by about 60 percent in the year of implementation and remained low through the middle of 2017, while the conforming share showed no comparable change. Accepted applications above the threshold carried rates as much as thirty to forty basis points higher. Bhutta, Popper, and Ringo, in a 2015 Federal Reserve analysis, found small marketwide effects in 2014, concentrated among small creditors. The combined record supports restriction at identifiable margins for high debt borrowers, inside a wider contraction that predated the rule and had independent causes.

Q: What is the difference between a disclosure rule and a substantive rule?

A disclosure rule changes the information attached to a transaction and leaves the transaction available on the same substantive terms, provided the information is delivered in the prescribed form. A substantive rule removes a term, a practice, or a method of calculation from the range of choices open to the parties, regardless of how clearly that term might have been described. The Truth in Lending Act of 1968 illustrates the first kind: creditors had to state the finance charge and the annual percentage rate, and federal law set no ceiling on the rate itself. The Fair Debt Collection Practices Act of 1977 illustrates the second kind in conduct, prohibiting harassment and false representations even where a debtor had received full information. The CARD Act of 2009 illustrates the second kind in pricing, prohibiting double cycle billing and retroactive rate increases on existing balances even where account documents had described those practices in advance.

Q: How do researchers measure whether a consumer credit law worked?

Researchers match the method to the kind of claim a statute makes. Agarwal, Chomsisengphet, Mahoney, and Stroebel measured the CARD Act in 2015 through a difference in differences design, comparing regulated consumer cards with unregulated small business cards across about 160 million accounts, so that movements caused by the financial crisis could be separated from movements caused by the fee limits. The Federal Trade Commission measured credit reporting accuracy in the 2012 section 319 study by recruiting 1,001 participants, reviewing 2,968 reports with trained assistance, and rescoring files after disputes. The Kleimann Communication Group measured disclosure quality through controlled testing, in which consumers answered questions and compared loan offers using competing forms. Each method answers a different question: price effects require account data and a comparison group, accuracy requires file level review, and comprehension requires testing under controlled conditions rather than inference from market outcomes.

Q: Why do complaint counts not measure the number of violations?

A complaint count records allegations that reached a particular channel, survived classification, and appeared in a published total. The Consumer Financial Protection Bureau stated in the fall 2018 semiannual report that the bureau did not verify all the facts alleged in complaints, and company response categories such as closed with explanation describe how a process ended rather than how a court or examiner would have decided the facts. Awareness shapes the total as well, because a new intake channel, a public database, or educational material can raise filings without any change in underlying conduct. Scale supplies a further limit. The bureau estimated in a January 2017 survey that more than 70 million consumers had been contacted about a debt in collection during the prior twelve months, so an annual complaint total represents the filings that completed intake against a contact population measured in tens of millions, rather than a census of disputed events or unlawful acts.

Q: Which consumer credit rule produced the largest measured savings?

The CARD Act fee limits produced the largest measured savings in the federal record assembled through 2018. Agarwal, Chomsisengphet, Mahoney, and Stroebel, in the Quarterly Journal of Economics in 2015, estimated that the limits reduced borrowing costs by an annualized 1.6 percent of average daily balances across consumer accounts, equal to 22.58 dollars per account per year. For account holders with FICO scores below 660, the reduction reached 5.3 percent of average daily balances, or 57.69 dollars per account per year, led by declines in over limit fees and late fees. Extrapolated to outstanding balances of 744 billion dollars in the first quarter of 2010, the estimate equaled annual savings of 11.9 billion dollars. No disclosure intervention in the record approached that scale: the CARD Act minimum payment disclosure carried an estimated upper bound near 57 million dollars a year in the same study.

Q: What happened to credit report errors after consumers disputed them?

Most disputes produced some correction, and a substantial minority of consumers left the process believing an error remained. In the Federal Trade Commission section 319 study published in 2012, four out of five participants who filed disputes obtained at least some modification of a report. The followup report of January 2015 surveyed 121 consumers whose disputes had remained unresolved after the original study. Of that group, 37 consumers, or 31 percent, reported on reflection that the disputed information was correct. Nearly 70 percent of the surveyed group continued to believe that disputed information was inaccurate. Within that group, about half planned to press the dispute further and about half planned to abandon the dispute, which left contested information in place as the operative version of the record for lenders and other users who requested a report afterward.

Q: Did lenders offset the CARD Act limits by raising interest rates?

The account level evidence found no offset during the period studied. Agarwal, Chomsisengphet, Mahoney, and Stroebel, in the Quarterly Journal of Economics in 2015, tested for an anticipatory increase in interest charges before the fee limits took effect, a sharp increase after implementation, and a gradual increase over the months that followed, and found no evidence of any of the three patterns. The point estimate for the offset was approximately zero, and the authors reported that an offset greater than sixty one percent of the fee reduction could be ruled out with ninety five percent confidence. Interest charges on new accounts, where issuers faced the fewest constraints on resetting contract terms, showed no offsetting increase either. The finding rests on data ending in December 2012, so longer horizon adjustments in entry, exit, or contract structure lay outside the measurement rather than inside the conclusion.

Q: What does the evidence say about disclosure helping comparison shopping?

Controlled testing supplied the strongest positive evidence for disclosure in the federal record. Between April and June 2013, the Consumer Financial Protection Bureau, with assistance from the Kleimann Communication Group, conducted a quantitative validation study with 858 consumers in 20 locations. One group used the mortgage forms then in force, and another group used the integrated Loan Estimate and Closing Disclosure. The bureau reported in November 2013 that consumers using the integrated forms answered 28.8 percent more questions correctly, with a margin of error of plus or minus 4.7 percentage points, across experienced and inexperienced consumers. Detail on the Loan Estimate showed 99 percent of consumers identifying the loan amount, against 61 percent with the existing forms. The result demonstrates that redesigned, tested, standardized forms improved comparison performance under controlled conditions, a narrower and better supported claim than a claim that mandated forms improved decisions in ordinary transactions.

Q: Did the volume of mandated disclosure grow faster than its usefulness?

The accumulation evidence supports that conclusion for disclosure as administered across successive statutes. The Truth in Lending Act of 1968 generated disclosures for closed end and open end credit. The Real Estate Settlement Procedures Act of 1974 added settlement disclosures for mortgage transactions. The Fair Credit and Charge Card Disclosure Act of 1988 added a standardized summary table for card applications, and later statutes added disclosures for high cost mortgages, billing errors, rate increases, and minimum payment consequences. No institution held responsibility for subtracting material as additions accumulated. Ben-Shahar and Schneider argued in 2014 that attention functioned as a finite resource and that each mandate competed with every earlier mandate, with the sales presentation, and with the time pressure of a closing. The counterpoint qualifies rather than reverses the conclusion: prototype testing by Lacko and Pappalardo in 2007 and the Kleimann validation study in 2013 showed that redesign could recover usefulness from a shorter and better organized package.

Q: What would count as strong evidence that a disclosure rule works?

Strong evidence would connect comprehension to behavior and behavior to outcomes in sequence. A controlled test would show that recipients understand the disclosed terms, including relative cost against a market benchmark. Field evidence would then show that recipients act on that understanding by choosing cheaper offers or avoiding expensive terms at rates beyond a comparison group. Account or transaction data would finally show lower total cost, rather than a shift of cost into a different charge. The payday lending experiment by Bertrand and Morse, published in the Journal of Finance in 2011, approached that standard for one intervention: a treatment presenting accumulated dollar cost made participants 5.9 percentage points less likely to borrow in later pay cycles, a decline of 11 percent relative to the comparison group. Gains confined to correct answers on a test, without measured changes in choices or costs, would count as evidence about form design rather than evidence that a disclosure rule protected borrowers.