The CFPB and the Equal Credit Opportunity Act

How Regulators Can Improve Consumer Protection and Access to Credit


View Full Document as PDF

Freedom, equality, and justice are all bedrock principles of the American experiment. Yet it is no secret that the U.S. has not always lived up to these values. Discriminatory treatment has plagued the United States since its founding, whether toward women, immigrants, or African Americans.

A series of landmark civil rights laws were enacted during the 1960s to address such discrimination. One of these laws, the Equal Credit Opportunity Act (ECOA), was intended to ensure that all consumers have equal opportunity in credit applications. Credit discrimination based on race and gender was widespread in the early 20th century. The ECOA attempted to fix that.

Unfortunately, decades later, regulators have begun to stretch their authority beyond that noble goal to enact a policy agenda that Congress never intended. The Consumer Financial Protection Bureau (CFPB), a federal regulator established under the 2010 Dodd-Frank Wall Street Reform and Consumer Protection Act, has been highly aggressive in enforcing ECOA. It has done so using dubious statistics and unfounded legal theories to bully firms into settlements rather than proving a case in federal court. In doing this, the Bureau’s focus has been not on rooting out material discrimination, but on socially engineering outcomes for individuals who are members of protected groups under civil rights laws.

The CFPB has shifted from the long-held belief that companies should be prosecuted for actual discriminatory treatment to prosecuting companies for perceived discriminatory effect. Specifically, the CFPB has gone after firms whose policies result in outcomes that diverge statistically from the average. Under this novel theory of what constitutes discrimination, the government does not need to prove that a firm had any intent to discriminate. Claims can be brought based solely on statistical calculations that suggest a facially neutral policy disparately affects individuals in protected classes—even if that company had no intent to discriminate and undertook no action to do so.

Ultimately, this harms both businesses and consumers. New technologies, such as alternative credit scoring systems, have the potential to extend affordable credit to consumers who have long been shut out of the financial system. But overzealous enforcement of ECOA threatens these developments. If startup ventures risk prosecution or litigation because their otherwise neutral algorithms have unintended and largely unpredictable effects, innovation in this area will be chilled.

On May 8, 2018, Congress voted to overturn one of the Bureau’s most harmful guidance documents relating to the ECOA liability of auto lenders. While this was a step in the right direction, to date, there has been no Congressional or regulatory action to address the underlying problems with how the CFPB has interpreted the ECOA. With new leadership at the Bureau, the regulator has an opportunity to correct its course. Fair lending is a noble goal, but it must be enforced appropriately.

The Equal Credit Opportunity Act. The Equal Credit Opportunity Act was enacted during a period that saw two seemingly unrelated but extremely significant events—the explosion of consumer credit and the civil rights movement. Beginning largely in the 1950s, consumer credit became an essential tool of economic mobility for working and middle class Americans. But not all groups shared in this prosperity, with members of certain minority groups subject to discrimination in the provision of credit.

In response to these concerns, Congress established the National Commission on Consumer Finance in 1968 to investigate the industry. At its first hearings, the Commission heard largely anecdotal evidence of prejudice against single women and racial minorities. For example, witnesses testified that lenders were often unwilling to extend credit to divorced women. Interestingly, the Commission’s final report in 1972 did not recommend federal regulation as the answer. Instead, its members concluded that cultural change and market forces would largely do away with most discrimination.

Nevertheless, Congress established the Equal Credit Opportunity Act in response to the report. Originally passed in 1974 and subsequently amended in 1976, the ECOA prevents discrimination in the granting of credit on the basis of race, color, religion, national origin, sex, marital status, age, receipt of income from any public assistance program, or because a person has exercised certain legal rights under a consumer protection statute. Under ECOA, a lender is generally prohibited from inquiring about these variables or using them in making lending decisions if learned. ECOA’s implementing regulation requires that credit scoring systems be “empirically derived, demonstrably and statistically sound.”

What ECOA Does Say, and What It Does Not Say. The premise of ECOA—to ensure fair, equitable, and nondiscriminatory access to credit—is clear. But there is debate as to how this premise is to be applied in practice. Does the law require only equal opportunity or equal outcomes? Two dueling models of how to enforce the ECOA, the legal theories of disparate treatment and disparate impact, lead to different approaches.

Disparate treatment occurs when a lender treats a consumer differently because of a characteristic that defines a protected category. Under this theory, a firm must have the intent to discriminate against a protected class. For example, a bank cannot exclude customers who meet the qualifications of a credit card promotion offer simply because they are Hispanic.

Disparate impact occurs when a lender’s policy has a disproportionate effect on a certain class, even if the lender had no intent to discriminate and the practice appears to be neutral—unless that policy meets a legitimate business necessity like cost or profitability. In this case, even companies with policies explicitly designed to prevent discrimination could be charged with discrimination if there is an adverse statistical result. Unsurprisingly, aggrieved parties are often difficult to find in these cases.

The ECOA itself, however, does not address disparate impact. Its language only includes the prohibition of discriminatory treatment, not the discriminatory effect of a certain policy. Instead, the ECOA broadly prohibits discriminatory treatment, stating:

It shall be unlawful for any creditor to discriminate against any applicant, with respect to any aspect of a credit transaction—on the basis of race, color, religion, national origin, sex or marital status, or age …

In defining the scope of the law, Congress used the term “discriminate” to prohibit discriminatory treatment or intent. The ECOA includes no language relating to the consequences of a given policy, which would impose disparate impact liability. As the late Supreme Court Justice Antonin Scalia commented regarding the Constitution, the ECOA “says what it says and doesn’t say what it doesn’t say.”

Nevertheless, lower courts have not always focused on the statutory construction of ECOA, with certain cases upholding disparate impact liability. Instead, they have relied principally on excerpts from committee reports developed in connection with the 1976 ECOA amendments, which reference an “effects test” for credit discrimination. There are two problems with this analysis.

First, while the Senate Banking Committee may have made reference to the effects of credit discrimination in their deliberations, in no place in the statute did Congress actually implement this language when it had ample opportunity to do so. The 1976 amendments merely added a broader list of protected classes. It did not alter ECOA to explicitly cover disparate impact claims.

Second, the Supreme Court has consistently reaffirmed the primacy of statutory text in interpreting law. In other words, it must interpret the meaning of words that are actually there when deciding what Congress intended. Therefore, the failure of Congress to enact disparate impact proscriptions must be interpreted to mean that Congress decided to limit the scope of the ECOA to disparate treatment.

The Supreme Court has never decided whether the disparate impact theory is valid under ECOA. But the Court has ruled on many other anti-discrimination laws to set some precedent. The most recent was a dispute under the Fair Housing Act (FHA), Texas Department of Housing and Community Affairs v. Inclusive Communities Project, Inc. In that decision, Justice Anthony Kennedy announced that “antidiscrimination laws should be construed to encompass disparate-impact claims when their text refers to the consequences of actions and not just to the mindset of the actors, and where that interpretation is consistent with the statutory purpose.”

The FHA includes this “effects” language. The Court thus concluded that “this results-oriented language counsels in favor of recognizing disparate-impact liability.” Further, in other statutes, such as the Age Discrimination in Employment Act (ADEA), Congress purposefully included “effects” language that signifies a deliberate decision to permit disparate impact claims.ECOA’s implementing regulation, Regulation B, originally recognized this and prohibited only disparate treatment. But in 1994, the Federal Reserve issued an interagency “Policy Statement on Discrimination in Lending,” claiming that Regulation B could prohibit both disparate treatment and disparate impact, effectively skirting the rulemaking process to pursue an agenda Congress never intended. When enforcement and rule-writing responsibilities of the statute and implementing regulation were transferred to the CFPB under the Dodd-Frank Act, the CFPB confirmed that it would persist with this interpretation of ECOA. Unlike other credit discrimination claims, disparate impact claims do not require the CFPB to prove a firm’s intent to discriminate. This has enabled the CFPB to aggressively enforce the ECOA against completely neutral practices that it disfavors, largely for policy-based reasons—not legal ones.

Language of Anti-Discrimination Statutes (emphases in original)


Disparate Treatment Language

Disparate Impact Language


(a) It shall be unlawful for any creditor todiscriminate against any applicant, with respect to any aspect of a credit transaction— on the basis of race, color, religion, national origin, sex or marital status, or age …



805(a): (a) In general It shall be unlawful for any person or other entity whose business includes engaging in residential real estate-related transactions todiscriminate against any person in making available such a transaction, or in the terms or conditions of such a transaction, because of race, color, religion, sex, handicap, familial status, or national origin.

804(a): It shall be unlawful…To refuse to sell or rent after the making of a bona fide offer, or to refuse to negotiate for the sale or rental of, or otherwise make unavailable or deny, a dwelling to any person because of race, color, religion, sex, familial status, or national origin.


It shall be unlawful for an employer—(1) to fail or refuse to hire or to discharge any individual or otherwise discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because ofsuch individual’s age;

(2) to limit, segregate, or classify his employees in any way which would deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect his status as an employee, because of such individual’s age;

Source: House Financial Services Committee, Unsafe at Any Bureaucracy Part I

Problems with the CFPB’s Enforcement of ECOA

Overreaching Authority: The Case of Auto Finance Companies. The CFPB is one of the most powerful regulatory agencies in United States history. It alone has rulemaking, supervision, and enforcement authority over nearly every consumer financial product in the economy. And yet, even such a powerful agency faces some limits. The Dodd-Frank Act, which created the CFPB, included a bipartisan amendment that explicitly exempted auto dealers from the Bureau’s broad authority. However, it did not exempt auto dealers from regulatory oversight altogether, as Dodd-Frank granted authority to regulate auto dealers to the Federal Trade Commission, which historically has expertise in policing car sales. Despite this, the Bureau has sought to charge auto finance companies that work with auto dealers with disparate impact discrimination under ECOA. This was done not through the public notice-and-comment process of a rulemaking, as required by the Administrative Procedure Act, but via regulatory “dark matter”—agency issuances such as guidance documents, memos, and interpretive bulletins that are not officially “rules” but nevertheless carry regulatory weight.

On March 21, 2013, the CFPB issued a bulletin detailing the liability that auto finance companies may face under ECOA. The Bureau alleged that these third-party auto finance companies’ markup and compensation policies trigger disparate impact liability if the auto dealers’ credit decisions result in discriminatory outcomes. The bulletin provided no evidence of discrimination. Rather, it claimed that the way in which consumers bargain with auto dealers for loans creates a situation ripe with potential for discrimination.

The bulletin advised auto finance companies to impose controls on compensation and dealer discretion policies that allow auto dealers to mark up interest rates. As attorney Kim Perez, then a J.D. candidate at the University of North Carolina law school, noted for the North Carolina Banking Institute, the bulletin “therefore seeks not to regulate auto dealers directly, but rather to effectively enlist indirect auto lenders under its jurisdiction; in order to police auto dealer transactions that otherwise would be outside its jurisdiction.”

Guidance documents are not legally binding and cannot be relied upon in an enforcement proceeding (although they may be binding in practice.) While interpretive rules are merely designed to clarify what the law says, the auto lending bulletin stretched the ECOA to cover an entirely new arena that was not prescribed in the statute. Not only did this completely change the common understanding of the law, it imposed a new compliance burden on lenders that was not subject to the Administrative Procedure Act’s notice-and-comment rulemaking process. In many cases, this would require completely changing their business model. As Perez has further written: “This uncertainty on the Bulletin’s binding authority puts lenders between a rock and a hard place: weighing the risks of noncompliance with the benefits of continuing to allow dealer reserve compensations.”

Based on that guidance document, the CFPB, in conjunction with the Department of Justice, pursued enforcement actions against four entities—Ally Financial, American Honda Finance Corporation, Fifth Third Bank, and Toyota Motor Credit. The issue at hand was a standard compensation practice whereby a dealer assists in financing a car through a third-party lender and is compensated by “marking up” the interest rate. Consumer groups, such as the Center for Responsible Lending, have lobbied against this practice for years. The CFPB relied on analysis by these consumer groups, not consumer complaints or internal research, to justify its enforcement action.

However, in December 2017, the Government Accountability Office (GAO) determined that the auto lending bulletin was in fact a “rule” that was never submitted to Congress as required under the Congressional Review Act. In May 2018, Congress voted to disapprove  the auto lending bulletin, rescinding the rule and prohibiting the CFPB from issuing a similar one. This action may provide needed regulatory relief for auto lenders, but the whole controversy highlights the danger of the CFPB’s capability to use its enormous power arbitrarily.

How Dealer Markup Works. The compensation practice works like this. As part of an auto sale, a dealer may offer financing for the vehicle through a creditor. The creditor offers a wholesale interest rate to the dealer based on factors like the customer’s creditworthiness. The dealer then negotiates with the customer an interest rate above the wholesale rate in order to cover the dealer’s costs and provide a return. The difference between the rate accepted by the buyer and the rate quoted by a creditor is the dealer’s “markup.” Because the CFPB’s action targeted auto finance companies’ practice of allowing auto dealers to mark up interest rates, it is a de facto regulation of auto dealers, which are supposed to be exempt from the CFPB’s authority under Dodd-Frank. 

This dealer markup is a common retail practice, but consumer advocacy groups and the CFPB argue that the auto lender “discretion” in negotiating rates means that consumers, particularly members of ethnic minority groups, pay a higher rate than they would otherwise. If the Bureau could indicate that minorities paid higher interest rates on average, this would constitute disparate impact discrimination under the CFPB’s interpretation of ECOA.

There are numerous problems with the CFPB’s claims.

First, the CFPB had no data on race or national origin to validate its claims. The law prohibits auto lenders recording any form of racial characteristics, to ensure that such factors do not influence a lending decision. Instead, the Bureau relied on a proxy methodology, known as Bayesian Improved Surname Geocoding (BISG), which involved taking a customer’s surname and zip code and attempting to estimate his or her likely ethnic background.

This methodology proved to be highly erroneous. It significantly overestimated the number of individuals allegedly discriminated against. In fact, internal Bureau emails reveal that the CFPB knew that its controversial statistical approach was less accurate than other available methods and was prone to significant error. For example, for every 100 African Americans, the Bureau’s proxy methodology could only identify roughly 19 of them correctly. Accordingly, CFPB staff themselves did not believe their case would hold up in federal court.

Even worse, the BISG methodology is entirely inappropriate for the kind of analysis the CFPB attempted to conduct. BISG is designed to try to predict the relative proportions of different groups in a population sample, but not whether any particular person is of a particular ethnicity. Had the Bureau gone through the notice-and-comment process of a rulemaking, the methodology likely would not have survived public scrutiny. Indeed, even the most prominent BISG scholars, including its inventor, Marc Elliot of the RAND Corporation, have criticized the CFPB’s use of the methodology. The CFPB’s gross overestimation of the supposed discrimination became clear when it distributed the refund checks to many white Americans as compensation for supposed discrimination against them as African Americans.

Second, the Bureau’s findings omit numerous factors that can explain the perceived disparate impact on protected classes. The Bureau’s methodology never accounted for:

·         Consumers’ credit risk, such as different income levels or credit scores;

·         Whether the car was new or used;

·         Promotional financing rates for certain vehicles and differences in demand for those vehicles;

·         The location of different showrooms and their customers’ demographics;

·         Whether the customer had shopped around for a better deal;

·         Customers’ negotiating ability;

·         Customer commitment to the transaction; or

·         The time frame of the transaction.

Without accounting for these many variables, it is difficult to identify discrimination as the sole or main cause of higher interest rates. When these factors are taken into account, as they were by many industry studies criticizing the CFPB’s findings, the racial disparity all but disappears. The supposed disparate impact found by the CFPB does not hold up under empirical scrutiny.

Third, claims that consumers are harmed by such markup compensation policies are off the mark. The policies enable dealers to obtain financing that consumers otherwise could not obtain on their own. Such financing enables more vehicles to be sold to more people across more income levels in a cheaper and quicker way. Moreover, the interest rate is only one of many points negotiated in an automobile transaction, and a marked up interest rate is often used to offset other factors, such as the down payment. If the law were to ban the negotiated interest rate, all other components of the transaction would likely become more expensive to compensate for the dealer’s loss of revenue.

This was a common practice among the auto lenders that settled with the CFPB. For example, American Honda Finance Corporation lowered one portion of its markup as a result of the CFPB consent order. But it also raised the price of other less negotiable aspects of its rates, thereby increasing the overall cost of the loan to the consumer by 1.1 percent, or around $586 in extra interest payments over the life of a typical four-year $25,000 loan.

As has often been the case with its rulemakings, the CFPB never conducted a rigorous cost-benefit analysis of any of the proposals it put forward, such as looking at the effects that a change in compensation policy would have on lenders and consumers. Nevertheless, the CFPB pursued settlements with the four companies totaling $200 million in fines. No case was proven in a federal court. In many cases, the Bureau used its substantial leverage over these companies to force them into settling. Some of the consent orders stipulated that the auto finance companies could no longer engage in the kind of markup or compensation practices that consumer groups have lobbied against for years. Internal emails reveal that the Bureau believed that by issuing hefty enforcement actions against prominent companies, it would be able to “tip” the market away from the markup and compensation policies it targeted in its enforcement action.

The problems with a disparate impact theory of liability have been apparent since its inception—in particular, its lack of reliability as an identifier of actual violations of law. The CFPB relied on questionable statistical correlations to assert discrimination in order to justify its efforts to shape private industries practices to its liking, as opposed to rooting out actual discrimination.

Stifling Innovation. The most noticeable impact of aggressive CFPB enforcement is the large fines issued against leading auto lenders. But the greatest “unseen” impact is the chilling effect it is likely to have on financial innovation.

Large, established firms with armies of lawyers and compliance officers can more easily deal with increased regulatory burdens. In the case of Ally Financial, it can also better afford an expensive, even if unjustified, settlement out of court. By contrast, new startup ventures with little capital cannot run the risk of hefty enforcement actions from overzealous officials.

Financial technology, or “FinTech,” firms have an enormous potential to revolutionize the consumer finance industry, by leveraging new data and machine learning to extend credit to more consumers through new, efficient business models that rely on algorithms instead of loan officers.

As consumer credit is now largely guided by statistical models, the underwriting of a loan relies heavily on an applicant’s credit scores. Yet approximately 35.4 million Americans, known as no-hit or thin-file consumers, are either “credit invisible” or lack sufficient data to generate a credit score.

From the perception of a lender, a consumer’s lack of credit history is a sign of that individual being a higher credit risk, even if in reality he or she has the ability to repay the loan and has a good track record of paying bills on time. The problem is that the information lenders obtain from credit reports does not present a complete picture of a consumer’s credit risk.

Minority groups are seriously overrepresented among the credit invisible and thin-file population, with only 51 percent of African-Americans and 58 percent of Hispanics having a credit score, compared to 75 percent of all American adults. By including alternative data such as utility bill payments, electronic transactions, educational attainment, and other information loosely tied to a person’s financial behavior, the accuracy of existing credit scores and the amount of credit visible individuals could both improve. One early study from 2006 found that including just two alternative data points—utility and landline and mobile phones bill payments—allowed the creation of credit scores for nearly two thirds of consumers in the study’s thin-file sample. Minorities benefited most, with acceptance rates rising by 22 percent for Hispanic borrowers and 21 percent for African-American borrowers.

A shift toward greater automation in credit scoring and lending decisions may also reduce opportunities for prohibited bias. As University of California, Berkeley, researchers found in their study of conventional versus FinTech mortgage lending markets, FinTech lenders, on average, had virtually no disparate impact pricing on minority groups. As the authors concluded:

Our findings for more algorithmic FinTech mortgage lending suggest that, in addition to the efficiency gains of these innovations, they may also serve to make the mortgage lending markets more accessible to African-American and Hispanic borrowers and provide these borrowers with fairer pricing.

However, overzealous regulators threaten all of these innovations when they pursue prosecution or litigation based on the disparate impact theory of liability. For example, credit scores derived from alternative data containing no records based on prohibited characteristics may be shown to have a similar statistically divergent outcome in certain instances, such as in auto loans, regardless of whether actual discrimination has occurred. A small startup firm cannot weather the kind of regulatory risk this entails. Where such risk is high, lenders will resort to more standardized services that play it safe—in other words, they fail to innovate and serve new markets.

Among banks, there is considerable uncertainty as to how to comply with the disparate impact standard. As a result, many banks have implemented costly compliance programs as a precaution to overzealous regulators. According to the American Bankers Association:

Under these novel approaches to fair lending enforcement, banks have been struggling to know what is expected of them and how these theories affect their efforts to tailor lending products to reach the variety of credit-worthy borrowers in their communities.

Aggressive regulators skirting the rule of law will only hinder the efforts of firms working to extend credit to currently underserved communities, including those of protected classes. Both lenders and customers will be made worse off, including those who would benefit from potential innovations that would extend affordable credit to consumers that have long been shut out from the financial system. As a recent Treasury Department report concluded: “The CFPB’s approach to enforcement and rulemaking has hindered consumer choice and access to credit, limited innovation, and imposed undue compliance burdens, particularly on small institutions.” Clear rules are required for financial innovation to flourish. A disparate impact theory of liability has the opposite effect.

The CFPB’s Own Disparate Impact Problem. Ironically, the CFPB itself is a fitting example of the problems with the theory of disparate impact. In 2013, American Banker obtained data on the CFPB’s job performance ratings that outlined stark racial disparities. According to the report, “whites were twice as likely to receive the agency’s top grade than were African-American or Hispanic employees.” That is exactly the kind of evidence the Bureau has relied on to pursue disparate impact charges against companies. One CFPB employee told American Banker: “If it was a lender and had similar statistics, it would be written up, immediately referred to the Justice Department, sued and publicly shamed.”

Broad statistical correlations do not prove discrimination. They may flag an underlying problem, but correlations alone are not proof of wrongdoing. Establishing evidence of discriminatory intent requires further investigation of instances of demonstrated discrimination, such as examining individual loan portfolios, beyond circumstantial evidence.

Opportunity for Reform. In November 2017, Consumer Financial Protection Bureau Director Richard Cordray resigned. President Trump then temporarily appointed Office of Management and Budget Director Mick Mulvaney as acting director. In January 2018, Mulvaney sent a memo to all CFPB staff outlining his vision for the agency. The “Mulvaney Memo,” as it became known, established three priorities. Going forward, the CFPB would:

1)      Faithfully enforce the consumer protection laws as written, but not attempt to regulate beyond that mandate.

2)      Focus on quantifiable and unavoidable harm to the consumer and only pursue lawsuits where harm can be found to have occurred.

3)      Pursue formal rulemaking and less regulatory “dark matter,” such as guidance documents.

The CFPB’s enforcement of the ECOA against auto lenders violates all of these principles.

First, the CFPB, to paraphrase Mulvaney, “pushed the envelope,” as Dodd-Frank explicitly prevents the Bureau from regulating auto dealers.

Second, rather than focusing on quantifiable and unavoidable harm, the Bureau used flawed statistical methods and questionable legal theory to stamp out certain compensation policies it did not like on the basis of policy preferences, not law.

Third, instead of going through the appropriate procedures to write a new rule, the Bureau relied on a guidance document to rewrite existing law.

The new leadership and new priorities at the CFPB present a valuable opportunity to reform the Bureau’s enforcement of the ECOA. Acting CFPB Director Mulvaney has already shown a willingness to implement reform, with his decision to reorganize the Office of Fair Lending and Equal Opportunity (OFLEO), which oversaw the auto lending cases. In February 2018, Mulvaney shifted OFLEO from the CFPB’s Supervision, Enforcement, and Fair Lending Division to the CFPB director’s office, effectively stripping OFLEO of supervision and enforcement functions, while continuing to focus on advocacy, coordination, and education.

Focus on Intent. Acting Director Mulvaney’s priority for the CFPB to “faithfully enforce the consumer protection laws as written” means the CFPB should enforce the ECOA to prohibit disparate treatment of protected classes, not disparate impact. As noted, nowhere does the language of the ECOA give authority to regulators to prosecute firms for the disparate effect of a facially neutral policy. While the Supreme Court has not settled this question, a number of other Supreme Court precedents do not support the current interpretation of the law under Regulation B.

In the auto lending cases, the CFPB used the ECOA not to root out actual discrimination, but to try to eliminate a practice that consumer advocacy groups had long lobbied against. This does nothing to advance equality, and may actually work to inhibit financial innovations that would extend credit to underserved communities.

The CFPB should correct Regulation B’s focus on disparate impact by rescinding any guidance documents that recognize the theory or through a new rulemaking to bring the rule in line with the underlying statute.

Expand “No-Action Letter” Program. Congress gave the CFPB the tools to foster financial innovation through providing regulatory certainty. Project Catalyst is a program of the CFPB that enables the Bureau to issue “no-action letters,” which signify that the Bureau “has no present intent to recommend initiation of supervisory or enforcement action” against a firm based on a certain statute the CFPB administers. Since its founding in 2010, the CFPB has only issued one no-action letter, to Upstart, an online platform for consumer loan services that evaluates applications using certain non-traditional factors such as education and employment history.

As noted, alternative credit scoring systems have enormous potential to bring thousands of credit invisible consumers into the formal financial system. But disparate impact liability under ECOA will remain a major impediment as long as companies like Upstart continue to have to ask for permission to do business.

To fulfill the promise of FinTech, the CFPB should expand the no-action letter program under Project Catalyst to issue more letters to innovative firms that require regulatory certainty before they experiment with alternative methods to extend credit to underserved communities.

ECOA and Small Business Lending. When the Dodd-Frank Act established the CFPB in 2010, Section 1071 of the Act amended the ECOA to require the collection of women- and minority-owned small business loan data. Such a requirement raises two major concerns, particularly regarding potential disparate impact enforcement.

First, unlike consumer credit products that typically involve a limited number of variables to be considered in the underwriting process, lending to small businesses is highly tailored and dependent on any number of relevant variables. Statistical correlations that typically fail to capture nuanced explanatory variables will only become less reliable for small business lending than they are for proving consumer-lending discrimination, particularly if they require the use of proxy metrics.

Second, because the loan underwriting may be more individualized and complex, the temptation to use statistical shortcuts to allege discrimination may increase. As with innovative financial technology firms, this gives financial institutions little incentive to expand their lending portfolios. If anything, the increased burden of data collection and compliance programs, combined with the risk of potential enforcement actions, will likely drive lenders out of the business.

Congress should repeal section 1071 of Dodd-Frank entirely. Short of repeal, the CFPB should tread carefully in its rulemaking process so as to not inadvertently discourage lending to the small businesses it seeks to protect.

Conclusion. Fair lending laws like the Equal Credit Opportunity Act are primarily focused on expanding credit opportunity, but problematic theories of fair lending liability undermine this goal. Disparate impact, as opposed to disparate treatment, does not require an agency to prove that a company discriminated against individual members of a group designated as a protected class under civil rights laws. Instead, it merely requires for a facially neutral policy to have a statistically disparate impact on a protected class.

Such a low standard of proof has allowed aggressive regulators like the Consumer Financial Protection Bureau to bring dubious enforcement actions against financial institutions, as demonstrated by a string of auto lending cases beginning in 2013. These cases had little to do with stamping out demonstrated discrimination and more to do with achieving policy objectives of the Bureau, such as eliminating compensation practices its leadership dislikes.

Worst of all, such aggressive enforcement of the ECOA may lead to fewer credit options for working and middle class consumers, as innovative financial services firms with the potential to extend credit to currently underserved markets are snuffed out. Such firms may not be willing to take a risk on alternative business models that could trigger disparate impact liability. As noted, FinTech firms and other innovative enterprises are working to extend credit to new, underserved communities. Aggressive disparate impact enforcement undermines these opportunities.

While fair lending remains a noble goal, the CFPB must take steps to ensure that it no longer relies upon unsound methods like the disparate impact standard. The ECOA was conceived as a means of ensuring equal opportunity to credit. As Acting Director Mulvaney has promised, the CFPB should faithfully enforce ECOA as written, and not go beyond the mandate Congress intended.