top of page

The subtleties of lending discrimination

Updated: Mar 3

“Discrimination” is a nuanced and culturally-charged word. This article begins by providing background and economic research in the service of building a discrimination framework. We then explore the history and intent of the U.S. Fair Lending regulatory regime with respect to alleviating discrimination. We provide an auto finance example and an opportunity to advance intended Fair Lending-based outcomes.

Finally, we observe that the political and regulatory support for Fair Lending or related economic decisions is ever-changing. Beyond this variable government support, there are tools and processes available to help people make the best choices. Enhanced consumer decision-making is the ultimate solution for reducing economic discrimination impact.

Table of contents:

  1. What does "discrimination" mean to you?

  2. Discrimination and the impact of Fair Lending

  3. The Fair Lending opportunity in auto finance

  4. Conclusion

  5. Resources

  6. Appendix - A1) Theories of discrimination and A2) Deeper dive - Economic mechanics

  7. Notes


1. What does "discrimination" mean to you?

“Discrimination” is a tricky word. It is a word carrying emotion and history. It has different meanings to different people in different contexts. In the following graphic, we provide five distinct examples that all relate to discrimination:

discrimination examples

These are real examples. There are certainly more. You may have noticed that the 3 example categories are in a discrimination “badness” order. The discrimination categories are:

  • Operational discrimination

  • Economic discrimination

  • Animus-based discrimination

Discrimination category description

Operational discrimination - For this article, we accept operational discrimination as a normal part of our life. Via evolutionary biology, our brains naturally seek to separate, categorize, and compare sensory information. We are naturally good at making pairwise comparisons. That is why eye doctors have you compare only 2 lenses at a time. Pairwise comparisons are a natural process, instinctive, and involuntary.... it just happens. Why do we like the taste of sweet foods as compared to bitter foods? It is our evolutionary biology replaying the same genetic algorithms that kept our ancestors alive by avoiding poisonous foods. To be fair, operational discrimination is not without some challenges. First, our natural discrimination abilities are best suited for binary comparisons. As such, we find decisions involving more than two comparisons to be significantly more challenging. Also, these naturally occurring causal processes are subject to decision-impacting cognitive biases, such as confirmation bias. [i] But, on balance, our cause-seeking cognitive processes are a very helpful and necessary part of our life. Thus, operational discrimination is considered the least "bad."

Economic discrimination - We dedicate most of this article to the middle "bad" category called "economic discrimination." There is nuance and subtlety regarding the degree to which economic discrimination is "good" or "bad." There are significant interaction differences between market environments and market outcomes as to whether we consider discrimination as "good" or "bad." Economists John List and Uri Gneezy suggest economic discrimination is a rising problem today. They say it is "increasingly widespread, multifaceted, difficult to parse, and often quite nefarious. And it's based entirely on self-interest..." [ii] In today's world, economic discrimination is a fact of life. It often cannot be avoided and must be actively managed. More to come on this!

Animus-based discrimination - The 5th example describing the "animus" category is considered the most “bad” discrimination. We assume most people agree that animus (hatred) is generally a social "bad." There are many historical examples, such as the Jim Crow laws of the U.S. South or Adolph Hitler and Germany's Nazi party. There are certainly many more. In the last half-century, animus-based discrimination has been retreating. [iii] This is owing to a cultural revolution that included the Civil Rights Act and other changes we will discuss later.

Discrimination categories may present very different financial outcomes

Animus-based discrimination noted in example 5 is different from economic discrimination found in examples 3 and 4. Animus-based discrimination is generally considered "more bad." In some ways, economic and animus-based discrimination types are the opposite. Economic discrimination may provide a financial benefit to those discriminating. For example - A car dealer sales process and incentives that drive a higher revenue benefit for the dealer. (In section 3, a more fulsome description of the car sales process is provided.) However, animus-based discrimination may provide a financial detriment to those discriminating. For example - A business earns a lower revenue detriment because it discourages a willing-to-pay customer based on the color of their skin.

Discrimination is a matter of degree and these three discrimination categories have some dynamic nuance.  For instance, transitioning to the “more bad” animus-based discrimination category from the “less bad” economic-based discrimination is generally enabled by 3 theories of discrimination:

  1. Taste-based discrimination -> Becker (1957),

  2. Statistical discrimination -> Arrow (1972), and

  3. Structural discrimination -> Sociology literature. 

Please see the A1 appendix for the theories. These theories provide for this article's foundation.

Disparate impact is the standard for determining the degree of discrimination badness.  Disparate impact refers to practices in employment, housing, and other areas that adversely affect one group of people more than another, regardless of whether rules applied by employers or landlords may be intended to be neutral.  Harvard economist and discrimination scholar Roland Fryer [xxiii] said:

"Not every disparity is discrimination." and "It's just really hard to pinpoint discrimination {disparate impact} without really good data and really great techniques."

In section 2 - Discrimination and the impact of Fair Lending – disparity will be explored in the context of consumer lending and the legal framework seeking to lessen disparate impact.

When is economic discrimination more ok?

Regarding economic discrimination, there is empirical evidence helping us understand the subtle differences, using well-narrated randomized control trial (RCT) results. [ii] The evidence reveals economic discrimination nuances, as related to examples 3 and 4. The testing outcomes show how economic discrimination may be "bad" but is not always "bad." The degree of perceived economic discrimination “badness” depends on whether the subject of economic discrimination had some control over that being discriminated against.

nuance of economic discrimination

For example, compare how you feel about these two economic discrimination impacts:

  1. A person gets a higher loan rate because they didn’t make past loan payments.

  2. A person gets a higher car repair price because they are disabled.

As is consistent with the banking tradition, most people believe pricing based on past payment behavior is more ok than pricing based on a disability. This is because people generally do not choose to be disabled, but, many perceive someone's ability to make a payment as a choice. [iv] To summarize this point, many economists suggest a related conclusion. Uri Gneezy is a professor of Economics & Strategy at the University of California, San Diego. Dr. Gneezy concludes:

“…individuals hold more prejudice toward those when they feel they have a choice in conditions...”

As such, economic discrimination is not always considered “bad.” In the next"Agents of discrimination" section, we summarize why those economic mechanics are not always considered "bad." In the A2 appendix, we provide a deeper dive into those economic mechanics. As a helpful analogy, we suggest that:

The supply/demand negotiation is the battleground for consumer surplus, producer surplus [v], and economic discrimination. The tools for this battle are information.

The next graphic shows this economic discrimination "battleground" in the context of a standard supply and demand curve.

negotiation and economic discrimination

Agents of discrimination

Next are the primary agents of the economic transaction, described in a way that minimizes negative economic discrimination outcomes.

The seller: The salesperson is generally well-informed. Many businesses employ well-trained, experienced sales professionals. We assume most businesses have significant support for their sales process and may deploy sales incentives. An uninformed buyer is an easy mark for a good salesperson. The degree to which economic discrimination impacts the price outcome is influenced by signals from the buyer. Those sales signals include a) how informed the buyer is and b) the degree to which the buyer is willing or able to use that information. [vi] Then, the salesperson attempts to leverage this knowledge to extract consumer surplus. (i.e., a higher price)

The buyer: On the other side of the transaction, a well-informed buyer will also use information and process discipline to pursue their own self-interest. [vii] The savvy buyer has a very developed sense of the value of their preferences (aka, utility), this buyer will also activate multiple purchase option alternatives from which to choose. (aka, BATNAs) This buyer uses processes and tools to assist with the purchase decision. This buyer has another essential tool, they may always walk away. (aka, No deal option) [viii] The savvy buyer attempts to leverage this knowledge and process discipline to extract producer surplus. (i.e., a lower price)

As an important practical point: At the time of negotiation, the actual market clearing price is unknown. The buyer and the seller may be armed with historical reference prices. In the case of cars, this may be the Manufacturer's Suggested Retail Price ("MSRP") or internet-sourced actual prices from recent sales. The buyer and seller are negotiating to make self-interested improvements to the final, agreed-upon price. In this fully-informed economic transaction situation, the final price fully expresses the self-interested utility of both the buyer and seller. The market-clearing price is revealed as an average of sales prices after the fact.

The free enterprise ideal

Certainly, this description of economic discrimination seems more reasonable. (not too "bad") This vision suggests two equally informed and decision-empowered parties sharing their preferences in a way that leads to a mutually beneficial agreement. This is the vision of the U.S. economic system based upon the free enterprise ideal. Next, this ideal is described by the upper left quadrant of the graphical framework - "High potential discrimination environment / Low discrimination impact."

Economic Discrimination Framework: Market Environment v Impact

economic discrimination framework

Please see the A2 appendix for a deeper dive into the four segments.

We will discuss the Fair Lending regime in the next section. The Fair Lending regulatory framework seeks to protect classes of people that have traditionally fallen into the upper right quadrant.

After the Fair Lending section, we discuss the opportunity and solutions for when economic discrimination turns more "bad." From the framework diagram, this is described by the "High discrimination impact" area found in the upper right quadrant. The suggested solutions are intended to transition a consumer from the right "High Discrimination impact" area to the "Low Discrimination impact" area. In the immediate decision context, we generally accept the market environment as a given. Our hunt is "Given the market environment, how can we help people make the best decision possible in service of reducing economic discrimination impact?"

In summary:

  1. Understanding and influencing the motivation of those with an opportunity to discriminate is key to anticipating a) the likelihood to discriminate and b) the nature of the discrimination.

  2. Generally, animus-based discrimination, like example 5, is on the decline. However, economic discrimination, like examples 3 and 4, is on the rise. Policymakers will better serve society by focusing on economic discrimination.

  3. For the buyer, an act of discrimination may introduce bias to a process that leads to a less-than-accurate outcome.

  4. In the context of economic discrimination, an informed decision process is king. Those a) with more information, plus b) the willingness and ability to leverage an effective decision process are more likely to keep a surplus and benefit from economic discrimination.


2. Discrimination and the impact of Fair Lending

The U.S. Fair Lending mortgage and consumer lending-based regulatory regime (broadly defined) is a useful and exemplary anti-discrimination model. Admittedly, it is not perfect. But given the dynamic and complex nature of the lending environment, its efforts to create a fulsome framework are admirable. The Fair Lending regime was implemented in response to the Civil Right Act of 1964, enacted during the Johnson administration. The framework includes the following:

  1. Product Focus: Historically, Fair Lending primarily focused on the Mortgage lending product. Over time, its lending product scope has expanded. Today Fair Lending's loan product scope includes mortgages along with consumer lending products (personal loans, credit cards, auto, etc) and small business loans.

  2. Protected groups: Fair Lending law defines applicable protected classes as protected from discrimination. As a rule of thumb, protected classes are everyone except younger, white, Christian, ambulatory, and heterosexual men. [ix]

  3. Definition: It defines what discrimination is via disparate treatment and disparate impact [x].

  4. Loan Decision focus: Historically, Fair Lending focused on approve/deny loan decisions, especially where human judgment is involved. Today, there is an expanded appreciation that statistical models may have systemic biases based on the data provided to create credit assessment models. [xi]

  5. Testing: It has a testing regime to identify mortgage lending discrimination via HMDA testing.

  6. Enforcement: It has a discrimination enforcement mechanism via the CFPB and the U.S. Department of Justice, Civil Rights Division. Other bank regulators, like the OCC, the Fed, and the FDIC have Fair Lending responsibility.

Keep in mind that the purpose of Fair Lending is not to eliminate the economic discrimination environment. If that were true, the FICO score would not be permissible to underwrite loan decisions. [xii] Fair Lending does reduce discrimination impact by implementing rules intended to maintain a level-protected class lending playing field. As suggested in the A2 appendix, Fair Lending accommodates the economic discrimination potential existing in some market environments. Fair Lending laws and regulation focuses on impact-based outcomes. It was mentioned in the last section that: "Conceptually, the final price fully expresses the self-interested utility of both the buyer and seller." Thus, you may think of Fair Lending's intent as to ensure that all buyers, regardless of social background, have an equal opportunity to fully express their utility. [xiii]

As such, some economic discrimination is permissible as long as the lending sales process is equalized across protected and non-protected classes. The Fair Lending regulatory regime is asked to walk a very fine line. A line that:

  • Prohibits animus-based discrimination in the lending process, but in a way that

  • Allows economic discrimination as long as the sales process for the applicable loan products is applied evenly across social classes.

In the next section, we describe an auto finance process that 1) demonstrates outcomes sometimes inconsistent with the Fair Lending regime's intent, and 2) opportunities to make disparate impact-based economic discrimination improvements.


3. The Fair Lending opportunity in auto finance

As mentioned in the last section: "Fair Lending accommodates economic discrimination potential existing in some market environments." Economic discrimination is often found in the car sales market environment. [xiv] In the companion article, Cutting through complexity: A confidence-building car buying approach, we provide processes and tools enabling individual buyers and borrowers to reduce or eliminate economic discrimination impact. As suggested earlier, economic discrimination may occur to anyone. Information and a disciplined decision process are foundational to negotiation success.

However, not everyone is able to follow a car-buying discipline that:

  1. Reduces the likelihood of a negative economic discrimination outcome and

  2. Enables the full expression of the value of one's preferences.

As such, the Fair Lending legal framework helps identify and effectively remediate the potential disparate impacts on protected-class auto loan borrowers. Next, we identify opportunities to advance Fair Lending's discrimination-reducing mandate.

At the time this article was published, the auto dealer portion of auto lending via indirect auto lending channels remained outside Fair Lending’s legal scope. The indirect auto channel is where 1) a bank or other lender provides auto financing and 2) the auto dealer participates as a loan origination agent in the indirect transaction. Within the auto dealership, the group typically charged with helping the car-buying consumer arrange financing is called the "Finance and Insurance ("F&I") Department." As part of the indirect auto lending process, the auto dealer receives multiple borrower loan options from multiple banks. These loan options provide different financial incentives for both:

  • the borrower => incentives based on the loan interest rate and terms as found on the promissory note.

  • the auto dealer => incentives based on dealer finance reserve or the difference between the bank’s interest rate (aka, the "buy rate") and borrower’s interest rate.

The auto dealer is aware of the incentives' value to both itself and the borrower. HOWEVER, the auto dealer is not obligated to disclose the auto dealer’s incentives and there is no obligation for the dealer to share all loan options with the borrower.

A lack of borrower transparency creates information asymmetry benefiting the auto dealer during sales negotiations.

Information asymmetry may lead to a negative impact and economic discrimination outcome.

It is possible a higher rate, higher auto dealer incentive loan would be presented to a car buyer and a lower rate, lower auto dealer incentive loan would not. In some cases, auto dealers may provide consumer disclosures or other materials containing at least a portion of the resources needed to make an informed borrowing decision. However, often, these information disclosures are presented in a dense "fine print" way. This disclosure design effectively discourages the understanding of that information. Behavioral economists call this "sludge." Sludge-like information disclosure may also lead to a negative impact and economic discrimination outcome. [xv]

In the eyes of Fair Lending regulation, because indirect auto loan originations are currently out of scope, an auto dealer may legally steer an unwitting car buyer toward higher-priced auto loans. The reason auto dealers are “out of scope” is a legal technicality, relating to the fact that an auto dealer is not a licensed lender. Consumer protection advocates make a poignant observation:

"From a consumer protection standpoint, auto dealers' lending organizations fail the traditional 'duck test' - 'If it looks like a duck, swims like a duck, and quacks like a duck, then it probably is a duck'"

Auto dealer F&I-related groups live in a technical gray area between the auto industry and the lending industry. As a result, it makes it difficult for traditional Fair Lending regulators to address Fair Lending challenges. As identified by U.S. federal regulators, the practice of auto loan product steering has proven to have a disparate impact on minority communities. [xvi] This sort of disparate impact is an example of a negative economic discrimination impact usually regulated via the Fair Lending regime.

In addition, today, the pandemic-impacted auto market environment creates upward pressure on car prices. Upward price pressure and supply scarcity likely provide more pricing power to the auto dealer. (possibly creating additional economic discrimination opportunities) As a case in point, in a recent article by Edmunds, the average vehicle transaction price as of January 2022 is above MSRP (list price). [xvii]

gray area between auto dealers and lenders

As of the writing of this article, the vast majority of auto loan originations occur via the indirect auto channel. This is despite the fact that an auto borrower is not required to arrange financing through the dealer. Auto borrowers may be better served to arrange financing directly with lenders to avoid the "gray area." [xviii] In 2013, the CFPB issued a rule to extend fair lending to auto dealers. This was rationalized since auto dealers often serve as loan-originating agents. (i.e., "Quacks like a duck.") In 2018, that rule was invalidated by congress. Since then, some states, like New York, have stepped in to implement indirect auto lending-based discrimination rules. Also, based on the "gray area" reference earlier, regulators may need to adapt. The FTC [xix], which has auto industry regulatory responsibility, could address auto dealer Fair Lending concerns.

The regulatory environment remains dynamic. [xx] Since the 2018 congressional rollback, the federal regulatory posture is generally more permissive for enabling auto dealers to capture consumer surplus and attendant economic discrimination. Owing to the less clear "gray area"-based auto lending regulatory environment, this places a premium on consumer car buyer information gathering and decision-making abilities. In our companion article, we provide processes and tools to help all people make better car-buying decisions. These solutions are helpful regardless of the regulatory posture: Cutting through complexity: An auto buying approach

good decisions to reduce disparate impact

4. Conclusion

It is a long game. The definition of discrimination is nuanced and intertwined with economics and cultural history. American laws and social attentions have momentum for reducing discrimination in the long run. Congress' 2018 CFPB auto-based fair lending rule invalidation action may prove to be a speedbump on the longer-term road to reducing discrimination. Rules extending Fair Lending to all lending activities, including auto dealer-based loan sales, appear consistent with the Civil Rights Act and the Fair Lending regime’s intent.

From a business standpoint, the current movement toward implementing state-based Fair Lending-like rules may create more problems for auto dealers and lenders. Historically, laws inconsistently implemented across states have proven to create operational complexity, business inequity, and additional business costs. [xxi] At some point, impacted auto businesses may wish for the consistent federal Fair Lending regime to again be applied to auto dealers. Also, because car costs are rapidly increasing, people are more willing than ever to travel further to purchase their next car. [xxii] This suggests state laws creating a more buyer and consumer-friendly sales environment will attract buyers more willing than ever to travel to purchase their next car.

Finally, the tools of social justice are evolving. This is occurring via:

  • The application of behavioral economics,

  • The availability of tools and processes helps to improve decision-making and to reduce the impact of economic discrimination,

  • The availability of data and RCT-like testing, and

  • The availability of research like that provided in The Why Axis.

Given today's dynamic and not always consumer-friendly regulatory environment, it is wise to follow the timeless, self-directed consumer protection advice:

"Caveat Emptor" - Buyer Beware!

In today's data-abundant world, the information is available to successfully buy a car. The car-buyers' challenge is to know which information to utilize and how to utilize it to make the best car decision. Next, car decision support tools are suggested to help car buyers reduce their exposure to economic discrimination and make the best car-buying decisions.


5. Resources

Economic discrimination may occur in many contexts, not just when buying a car. We suggest the key to making the best decisions is to be the captain of your own decision process. The following are resources to help you own your decision.

Definitive Choice: For individual or small organization groups - This smartphone app provides a convenient way to enter and weigh your preference criteria, then, enter your potential decision alternatives and their costs. Behind the scenes, it uses decision science to apply your tailored preferences and preference weights to score each of your alternatives. Ultimately, it renders a rank-ordered report to help you understand which alternatives will give you the biggest bang for your buck. Using a decision support app will 1) save you time, 2) optimize your economic value achieved, and 3) increase your decision-making confidence!

Next are a few real-life examples of using choice architecture solutions to make the best decision:


6. Appendix

A1) Theories of discrimination:

Discrimination causing disparate impact generally falls into 3 theory categories:

  • Taste-based discrimination

  • Statistical discrimination

  • Structural discrimination

Disparate impact manifests as 1) discouraging members of impacted groups from attempting an opportunity. or 2) Successfully achieving that opportunity even after engaging it.

Harvard economist and discrimination scholar Roland Fryer [xxiii] said:

"Not every disparity is discrimination." and "It's just really hard to pinpoint discrimination {disparate impact} without really good data and really great techniques."

Taste-based discrimination is an economic model of labor market discrimination arguing that employers' prejudice or dislikes in an organizational culture rooted in prohibited grounds can have negative results in hiring minority workers, meaning that they can be said to have a taste for discrimination. The model further posits that employers discriminate against minority applicants to avoid interacting with them, regardless of the applicant's productivity, and that employers are willing to pay a financial penalty to do so.

The taste-based discrimination model was first proposed by Gary Becker in 1957 in his book The Economics of Discrimination.  The taste-based approach may be thought of broadly in society, beyond employers.


Statistical discrimination is a behavioral model in which group inequality arises when economic agents, like consumers, workers, employers, etc., have imperfect information about individuals with whom they interact. According to this model, inequality may exist and persist between demographic groups even when those economic agents are behaving rationally. This is distinguished from taste-based discrimination which emphasizes the role of prejudice (sexism, racism, etc.) to explain disparities in labor market outcomes between demographic groups. 

Statistical discrimination occurs when imperfect information causes a decision-maker to associate past affiliated judgments of a group of people with a decision impacting another person or group.  For example, Let us say someone judged an ethnic group as being “not good workers” based on some past experience and a prospective new hire is a member of that ethnic group.  Then statistical discrimination is where the hiring decision-maker would not hire them because of their association with the ethnic group, even though they will turn out to be a good worker.  Because the hiring decision-maker has imperfect information, they fall back upon past judgments to represent needed information to fill the information gap.  In behavioral psychology, statistical discrimination is known as a cognitive bias called “representativeness bias.”

The theory of statistical discrimination was pioneered by Kenneth Arrow and his 1972 article The Theory of Discrimination.  Statistical discrimination is also known as information discrimination.


Structural discrimination is a form of institutional discrimination impacting economic agents, such as race or gender, which has the effect of restricting their opportunities. It may be either intentional or unintentional, and it may involve either public or private institutional policies.  Such discrimination occurs when policies or procedures of these organizations have disproportionately negative effects on the opportunities of certain social groups.

The theory of structural discrimination is generally researched by sociologists.  Structural discrimination is also known as systemic discrimination.

Specific to lending, home appraisal bias is cited as an example of structural discrimination.


A2) A Deeper dive - Economic mechanics:

Economic Discrimination Framework: Market environment v Impact

To further understand why economic discrimination is not always "bad," it is essential to consider some of the economic mechanics. Think of economic discrimination as the outcome of the supply-demand negotiation for consumer surplus. [v] As an outcome, economic discrimination is measured by impact. It can be neutral, low, or high, favoring the buyer or seller. At the bottom of section 1 is a typical supply/demand curve. The equilibrium, or the point where the 2 curves intersect, is the market-clearing price. The seller attempts to extract consumer surplus by negotiating a higher sales price. The buyer attempts to keep consumer surplus or extract producer surplus by negotiating a lower sales price. The actual market-clearing price is a measured result of market operations. The market-clearing price is unknown to participants during the market-making process.

Shown in the economic discrimination graphic at the bottom of section 1 is a neutral impact of economic discrimination. This occurs because the sales price is at the market clearing price. The impact of a buyer's economic discrimination is measured as the consumer surplus area over the dotted price line. What if the final sales price was higher than the market clearing price? In this case, the green area shrinks, and the impact of the buyer's economic discrimination increases.

As an essential idea: it is possible a buyer with a complete understanding of their own preferences (aka, utility) would be willing to buy a product above the market clearing price. The point is, a person with a lower understanding of their own utility is more likely to pay a higher price and to be subject to a higher economic discrimination impact.

There is a difference between the market environment creating the potential for economic discrimination versus the economic discrimination impact.

First, it is important to appreciate that the "discrimination environment" is associated with the market-making "process." Whereas "discrimination impact" is a measure of the market "outcome." In the graphic introducing the appendix, we show this visually in the lower, left-hand corner box.

Low potential discrimination environment / Low discrimination impact example: The car retailer CarMax® has a "no-haggle" policy. They have one price, and the buyer may accept it or walk away. This is a low-potential economic discrimination environment. In effect, they are setting their own market clearing price. They have a pricing process that closely mirrors the broader market for the particular car. In this example, this results in a very low or neutral economic discrimination impact because of the low potential economic discrimination environment.

High potential discrimination environment / High discrimination impact example: Take a more typical car dealer sales environment that prices every car individually for each customer. This is a potentially higher economic discrimination environment. This is the prototypical environment where an uninformed buyer may accept a higher price. But it could work in favor of either the buyer or the seller. Take a seller that is trying to hit its numbers at the end of the year. They may be willing to make price concessions to meet a sales volume target. Thus, there is high economic discrimination working to the advantage of the buyer. Economic discrimination does not always work against the buyer. This is the segment regulators typically consider for Fair Lending violations based on the disparate impact of protected classes. This is discussed in section 2.

High potential discrimination environment / Low discrimination impact example: Similar to the last example, this is the typical environment that prices every car individually for each customer. However, as in the "Agents of discrimination" example earlier, both the buyer and the seller are fully capable and willing to represent their economic interests. In this case, it is more likely little or no economic discrimination impact will occur.

Low potential discrimination environment / High discrimination impact example: We consider this a unique fraud environment where the seller creates the illusion of a low discrimination environment, but turns out to have a more extreme economic discrimination impact. The famous Bernie Madoff Ponzi scheme scandal [xxiv] is a notorious example where unwitting investors were dupped out of their savings. However, more subtle and legally questionable activities may occur in our day-to-day economic life. For example, a car seller "fudges" some of the car data published on the internet as a means to lure customers into their sales process.


7. Notes

[i] We explore the "Big 4" decision-making cognitive biases in the following article:

[ii] See Chapter 6: Gneezy, List, The Why Axis, 2013

The Randomized Control Trial or "RCT" is generally considered the gold standard for determining causality. This includes developing test and control groups by varying a single "control" aspect of a hypothesis. That is a question where one is trying to determine "why" that hypothesis is occurring. Upon completing the test and by using statistical hypothesis testing, a researcher may determine whether the "not different" null hypothesis has been validated or not.

Gary Becker is widely considered the father of modern economics-based discrimination theory:

[iii] A. Thernstrom, S. Thernstrom, Black Progress: How far we’ve come, and how far we have to go, The Brookings Institution, 1998

[iv] Increasingly - systemic bias, statistical model bias, and other factors are calling into question the traditional “loan delinquency is a choice” assumption. There is evidence some people may not have as much choice in making payments as we may traditionally believe. In the book Scarcity, the authors argue that scarcity, whether from money, time, or other resource constraints, causes a short-sightedness ("tunneling") that creates a sort of self-fulfilling prophecy in terms of negative feedback loops. These negative feedback loops may cause a failure to meet scarcity-based commitments. Based on the book’s thesis, think about how you feel when:

  1. Your work is so busy you feel like you will never get your work done. You may feel tired and short-term focused. You may just want to get to the end of today.

  2. You have been working for 14 hours and you have one more item on your list. If it is not life-threatening, you may want to put it off until tomorrow. You may feel tired and irritable.

That feeling is similar to someone without enough money to make a payment. Except, for the poor, this feeling never goes away.

In our article Resolving Lending Bias - a proposal to improve credit decisions with more accurate credit data we discuss systemic bias that may reverse the payment delinquency causality arrow. We provide an actionable approach to both 1) test for protected class systemic bias found in the FICO score and 2) resolve systemic model bias by addressing the data utilized to develop the FICO score.

[v] Consumer Surplus and Producer Surplus, Corporate Finance Institute (

In the context of consumer surplus, a buyer’s surplus is defined as the difference between what they actually pay and the value of the utility the good provides. In the case of a car, understanding your “value of utility” is difficult. A car is used to drive to work, to get groceries, to drive the kids to their events. But there is more to it, an owner may get value from the car’s performance, how fast it accelerates, the music system, how good it looks to their neighbors, etc.

Utility is intensely personal. Also, good salespeople attempt to “build” consumer surplus-based utility value perception by highlighting many of the cars' “producer surplus” enabling benefits. Sometimes the buyer may get confused and not fully understand their own “value of utility.” Finally, auto finance is often part of an auto dealer’s strategy to build the buyer’s consumer surplus value perception. An increase in the cost of $10,000 sounds a lot worse than an increase in payment of $100 a month. Salespeople may attempt to focus the buyer's attention on the “lower salience” car payment than the “higher salience” car price.

As we recommend in Cutting through complexity: An auto buying approach, it is important to perform a personal “value assessment” before car shopping. It is important to clarify the car buying objective. When in doubt, a car buyer may start with a more utilitarian-based “value of utility” statement like “For safe transportation.” When comparing alternatives, a utilitarian objective value statement will help the buyer keep costs down. Also, we suggest an approach to integrating auto finance into the car buying process that 1) minimizes its impact on the car buying objective and 2) enhances the buyer’s ability to capture consumer surplus.

[vi] While information is "king," today, the ability to use information is even more important when evaluating and weighing preferences. People often struggle with "drowning" in too much information. In today's world, information may be so abundant as to be overwhelming and noisy. Also, some sellers or sellers agents utilize uncurated information (e.g., "fake news") as a means to confuse and create a form of censorship.

[vii] "Self-interest" is easy to confuse with "Selfishness." According to the Merriam-Webster dictionary, selfishness is "a concern for one's own welfare or advantage at the expense of or in disregard of others."

Self-interest is different. Self-interest is an economic concept concerning the expression of utility. While self-interest may have an aspect of selfishness for some people, in general, it is the expression of individual utility across various factors. For example, a person could be a philanthropist that gets great pleasure by helping others. In this case, their highest weighted self-interest would be giving to others. As is often the case, self-interest is complex for most people, including weighted elements of selfishness and concern for others.

[viii] BATNAs, No Deal Option, and other negotiation standards are explored in the article:

Hulett, Negotiating success and building your BATNA, The Curiosity Vine, 2021

[ix] Protected classes include: Race, Color, Religion or Creed, Sex, Age, Disability, and Family Status

For many people, determining an individual's protected class membership is a box-checking exercise. Generally, a higher number of boxes checked increases the strength of the protected class affiliation. Many have more than one class box checked, plus, the number of boxes likely changes throughout our life.

For example, for me, when I was under 40 years old, I had no protected class boxes checked. Today, I have one age-based protected class box checked.

[x] The Federal Reserve Compliance Handbook, Federal Fair Lending Regulations and Statutes Overview

[xi] We explore hard-to-see challenges of systemic bias as found in credit bureau data:

[xii] Traditionally, FICO Score and related credit assessment predictive statistical models have been accepted as a necessary form of economic discrimination. In the last decade, a growing base of research suggests systemic bias found in the data being used to train these statistical models may be responsible for inappropriate credit decisions. In fact, very recently, the GSEs (the mortgage giants Fannie Mae and Freddie Mac) are being compelled by their regulator and conservator the Federal Housing Finance Agency to change FICO score models to include training data better representing protected classes.

[xiii] We describe the natural challenge for people to express their own utility. At its core, the challenge relates to how our brain naturally processes information and attendant naturally occurring cognitive biases.

Hulett, Assessing value like Warren Buffett, The Curiosity Vine, 2022

[xiv] John List is an economics professor at the University of Chicago. Uri Gneezy is an economics professor at the University of California, San Diego. These economists suggest several reasons why economic discrimination may be found in car sales:

“Car sales are some of the most common and important transactions in the economy for most individuals, as roughly sixteen million cars are sold annually in the United States. In addition, the stakes are high, but the transactions are relatively short….”

Gneezy, List, The Why Axis, 2013

Also, auto transactions are infrequent and complex. Oxford University fellow, business strategist, and choice architecture researcher Olivier Sibony suggests infrequent decisions where we do not receive regular feedback are particularly difficult to build intuitive decision-making ability. Auto purchases certainly fall in this category.

[xv] We discuss the definition and impact of sludge in the following articles:

Hulett, Cutting through complexity: An auto buying approach, The Curiosity Vine, 2020

[xvi] The authors analyzed data from before and after the implementation and repeal of the 2013 CFPB rule extending the reach of Fair Lending to auto dealers. The rule was invalidated in 2018. They find a significant reduction in race-based discrimination when Fair Lending is extended to auto dealers.

Butler, Mayer, Weston, “Racial Discrimination in the Auto Loan Market,” Consumer Financial Protection Bureau, 2021

As an example, the CFPB regularly brings lawsuits against auto lenders that violate Fair Lending laws. In the cited example of Credit Acceptance Corporation, the lawsuit suggests the lender is "misrepresenting the cost of credit and tricking its customers into high-cost loans on used cars." Please notice, as a result of the limits of Fair Lending's scope, the CFPB is not permitted to seek damages from the auto dealers that enabled Credit Acceptance's bad behavior. Unfortunately, the CFPB is limited to resolving the lender as a symptom of the problem, without addressing the auto dealer and related root causes of the problem.

Newsroom, "CFPB and New York Attorney General Sue Credit Acceptance for Hiding Auto Loan Costs, Setting Borrowers Up to Fail," Consumer Financial Protection Bureau, 2023

[xviii] "You don’t have to finance your auto loan through the dealership, and you should find out if you can obtain better interest rates through a bank, credit union, or another lender that you contact separately from the dealer."

Editors, "What is a Finance and Insurance (F&I) department?," Consumer Finance Protection Bureau, 2016

[xix] The Federal Trade Commission - The FTC asserts regulatory authority over the Automobile industry.

[xx] From the author's experience, there are 2 distinct levels to the banking regulatory environment.

  1. The law - is slow to change and requires significant constituent prodding of lawmakers to make changes. Lawmakers' incentives are to "party-first" mandates. It is only if the voters "scream loud enough" will lawmakers come together in a bi-partisan way to craft law changes. The passing of the Dodd-Frank Act is an example of "voters screaming loud enough" from voter pain created by the financial crisis.

  2. The enforcement of the law - this is more likely to be influenced by the current executive branch. The president may be more or less in favor of certain legal enforcement activities. For example, the Obama administration was more in favor of active legal enforcement of the Dodd-Frank Act than the Trump administration. In an article addressing the presidential administration's regulatory impact, Shiran Weitzman mentioned: "Donald Trump's administration rolled back some aspects of the Dodd-Frank Wall Street Reform and Consumer Protection Act." Most laws have wiggle room for interpretation and the amount of regulatory funding will impact the resources available for enforcement activities.

This regulatory environment leveling is the nature of the dynamic regulatory environment. This is also why it is suggested that the individual consumer be the captain of their own decision process. Good decisions will result in less economic discrimination impact. A good decision process makes the regulatory enforcement environment less relevant.

[xxi] In general, national bank subsidiaries cannot preempt state laws with federal law. Law provides some preemption exceptions. As such, when the Federal government reduces its consumer protection focus, some states are more likely to enact state-specific laws as a substitute for federal law. The author experienced this in the early 2000s when leading a mortgage and consumer lending division for a national bank. At the time, state and local predatory lending laws were being enacted after several high-profile examples of predatory lending were reported in the popular press. These bespoke local and state laws created significant operational complexity for national lenders. Over time, Federal law has become stronger regarding predatory lending. An example is the Unfair, Deceptive, and Abusive Acts and Practices (UDAAP) implemented via the Dodd-Frank Wall Street Reform and Consumer Protections Act of 2010.

Da-Wai, Huntington, Mi, Frey, Bergman, Hirsh, Weiss, Summary of Dodd-Frank Financial Regulation Legislation, Harvard Law School, 2010

[xxiii] Roberts (Host), Roland Fryer on Race, Diversity, and Affirmative Action, EconTalk, The Library of Economics and Liberty, 2023

[xxiv] CFI Team, Bernie Madoff, The Corporate Finance Institute, 2022


bottom of page