Auto Lending Research Articles

Rise of the Machines: The Impact of Automated Underwriting

Using a randomized experiment in auto lending, we find that algorithmic underwriting outperforms the human underwriting process, resulting in 10.2% higher loan profits and 6.8% lower default rates. The human and machine underwriters show similar performance for low-risk, less complex loans. However, the performance of human underwritten loans largely declines for riskier and more complex loans, whereas the machine performance stays relatively stable across various risk dimensions and loan characteristics. The performance difference is more pronounced at underwriting thresholds with a high potential for agency conflict. These results are consistent with algorithmic underwriting mitigating agency conflicts and humans’ limited capacity for analyzing complex problems. This paper was accepted by Will Cong, finance. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2024.4986 .

The Economic Consequences of Risk-Absorption in B2B Relationships: Evidence from Indirect Auto Lending

The Economic Consequences of Risk-Absorption in B2B Relationships: Evidence from Indirect Auto Lending

Privacy-Preserving Personalized Revenue Management

This paper examines how data-driven personalized decisions can be made while preserving consumer privacy. Our setting is one in which the firm chooses a personalized price based on each new customer’s vector of individual features; the true set of individual demand-generating parameters is unknown to the firm and so must be estimated from historical data. We extend the existing personalized pricing framework by requiring also that the firm’s pricing policy preserve consumer privacy, or (formally) that it be differentially private: an industry standard for privacy preservation. We develop privacy-preserving personalized pricing algorithms and show that they achieve near-optimal revenue by deriving theoretical (upper and lower) performance bounds. Our analyses further suggest that, if the firm possesses a sufficient amount of historical data, then it can achieve a certain level of differential privacy almost “for free.” That is, the revenue loss due to privacy preservation is of smaller order than that due to estimation. We confirm our theoretical findings in a series of numerical experiments based on synthetically generated and online auto lending (CPRM-12-001) data sets. Finally, motivated by practical considerations, we also extend our algorithms and findings to a variety of alternative settings, including multiproduct pricing with substitution effect, discrete feasible price set, categorical sensitive features, and personalized assortment optimization. This paper was accepted by Vishal Gaur, operations management. Funding: R. Momot acknowledges financial support from the HEC Paris Foundation and the Agence Nationale de la Recherche (French National Research Agency) “Investissements d’Avenir” [Grant LabEx Ecodec/ANR-11-LABX-0047] during the initial stages of this project. Supplemental Material: The data files and online appendices are available at https://doi.org/10.1287/mnsc.2023.4925 .

Designing Dealer Compensation in the Auto-Loan Market: Implications from a Policy Change

We study dealer compensation in the indirect auto lending market that increases lender’s market share and protects consumers from potential discrimination.

Spillover Effects of the Opioid Epidemic on Consumer Finance

AbstractI examine the impact of the opioid epidemic on subprime auto lending. Using a difference-in-differences framework, I find that county-level increases in opioid abuse cause an increase in loan defaults. Moreover, I find that traditional credit scoring attributes (e.g., FICO score) fail to predict loan performance deterioration associated with opioid addiction. The weak predictive performance of traditional credit measures and the resulting higher default rates generate a negative externality for borrowers in opioid-afflicted areas, as evidenced by 5.7% higher loan costs for subprime borrowers.

Racial Disparities in the Auto Loan Market

Abstract We document racial disparities in auto lending. Combining credit bureau records with borrower characteristics, we find that Black and Hispanic applicants’ approval rates are 1.5 percentage points lower, even after controlling for creditworthiness. In aggregate, this effect crowds out 80,000 minority loans each year. Results are stronger where racial biases are more prevalent and lending competition is lower. Minority borrowers pay 70-basis-point higher interest rates, but default less ceteris paribus, consistent with racial bias rather than statistical discrimination. A major antidiscrimination enforcement policy initiated in 2013, but halted in 2018, reduced unexplained racial differences in interest rates by 60$\%$. Authors have furnished an Internet Appendix, which is available on the Oxford University Press Web site next to the link to the final published paper online.

COVID-19 and Auto Loan Origination Trends

We study the impact of the COVID-19 crisis on auto loan origination activity during 2020. We focus on the dynamic impact of the crisis across lending channels, Equifax Risk Score (Risk Score) segments, and relevant geographic characteristics such as urbanization rate. We measure a significant drop in auto loan originations in March‒April followed by a near rebound in May‒June. Originations remain slightly depressed until October and fall again in November‒December. We document the largest drop and the smallest rebound in the subprime segment. We do not find any suggestive evidence that used car loan originations exhibited patterns significantly different from the rest of the market. We also document a more pronounced impact in the Northeast and the Pacific, seemingly influenced by the higher urbanization rate in these regions. Bank-financed originations experienced the largest drop and the smallest rebound, thus resulting in a loss of market share and continuing a 10-year trend of bank share loss in auto lending. We find that the drop in auto loans originated by banks was particularly significant among subprime borrowers. The impact of the COVID-19 crisis across origination channels contrasts with the experience during the Great Recession when banks contributed the largest support to the auto loan origination segment during periods of stress and finance company-originated auto loans were depressed.

Ensemble learning from model based trees with application to differential price sensitivity assessment

The assessment of price sensitivity is a relevant issue with important implications in decision making for revenue management. The issue has attracted interest among companies evolving towards the data-driven culture through the exploitation of their data sources. Thus, the design of pricing strategies that rely on analytics to identify groups of customers that exhibit differential price sensitivity has a great potential for revenue managers. This work proposes a data-driven approach, using ensemble learning from model based trees, to assess differential price sensitivity in a similar way as random forests algorithm does to assess variable importance. A differential price sensitivity score is defined and a ranking is obtained as a result so that the top ranked variables can be selected as candidate inputs for segmentation and differential price sensitivity group finding. Then optimal price allocation is carried out on the derived groups in order to compute the expected revenues which are compared with the revenues given by un-optimized prices and by optimal price allocation derived from the logit estimation of the bid response function. The proposed approach is validated in synthetic experiments and by application to the real business case of an auto lending company; the resulting revenues show its benefit.

Bursting the Auto Loan Bubble in the Wake of COVID-19

Before the COVID-19 pandemic, auto loans outstanding in the United States had soared to record highs. The boom in lending spanned new and used cars and traditional and subprime loans. With loan delinquencies also hitting new highs almost every quarter, predictions that the auto lending market could burst soon abounded. When the economy came to a grinding halt and unemployment skyrocketed in the wake of the pandemic, auto lenders knew they were facing a crisis. Throughout 2020, auto lenders granted more payment forbearances to consumers, while slashing interest rates on new loans. Auto manufacturers similarly made promises to buyers, such as the ability to return new cars for up to a year upon job loss. Combined with the CARES Act’s relief rebates and moratoria, the bottom did not fall out of the auto loan market.These measures, however, are temporary. The pandemic alone will not reduce people’s need for cars, but it will burst the auto loan bubble. The economic fallout will require interventions in the auto sale and loan markets, which presents a moment to transform America’s car economy. This symposium essay details a range of financial and related measures that can be implemented in the near future to shift auto financing away from promoting economically unequal and environmentally unfriendly use and access to automobiles, and, more broadly, to shift the United States away from prioritizing automobiles as the primary means of personal transportation.

Driven to Debt: Social Reproduction and (Auto)Mobility in Los Angeles

Since the financial crisis of 2008, subprime lending in the United States has flourished in auto loan markets. This article charts, for the first time, some of the contours of this underresearched part of the subprime landscape. In so doing, it makes two contributions. First, it widens and resituates debates about subprime lending by building on a suite of feminist political economic scholarship to argue that the sites, practices, and agents of social reproduction provide an essential—and largely neglected—perspective on the endurance and deepening of subprime markets. Second, the article leverages the intrinsically geographical, place-bound nature of social reproduction to provide a more holistic treatment of the financialization of everyday life. The article uses (auto)mobility as a novel vantage point from which to connect hitherto disparate social science literatures on financialization, transportation planning, welfare, and urban form. Drawing on secondary data and qualitative fieldwork in Los Angeles, the research explores the proliferation of subprime auto lending and the gendered, raced, and classed inequalities in mobility that mediate everyday life—the daily commute, the school run, the grocery shopping trip—and the demand for subprime debt.

Designing Dealer Compensation in the Auto Loan Market: Implications from a Policy Experiment

We study the design of dealer compensation policy in the indirect auto lending market, where most lenders give dealers the discretion to mark up interest rates. To protect consumers from potential discrimination by the dealer discretion, several banks switched to a new compensation scheme by fixing the markup as a percentage of the loan amount. We document that the market share of these banks responded positively (negatively) in the consumer segment where the policy increased (decreased) the interest rate — a reversal of the usual demand curve — which highlights the influence of dealers on the bank choice for financing loans. Accordingly, we develop and estimate an empirical model that allows for dealer–consumer bargaining, which depends on both the dealers' and the consumers' utility. Based on the estimation results, we explore alternative compensation policies that also eliminate dealers' discretion. We show that a lump-sum compensation that pays dealers a fixed dollar amount per loan dominates the current policy for the banks in terms of gaining market share. This is because dealers' equivalent markup rates would better align with their bargaining power. Our study highlights the importance of accounting for the interests and bargaining power of middlemen in designing a compensation scheme.

Privacy-Preserving Personalized Revenue Management

This paper examines how data-driven personalized decisions can be made while preserving consumer privacy. Our setting is one in which the firm chooses a personalized price based on each new customer's vector of individual features; the true set of individual demand-generating parameters is unknown to the firm and so must be estimated from historical data. We extend this classical framework of personalized pricing by requiring also that the firm's pricing policy preserve consumer privacy, or (formally) that it be differentially private -- an industry standard privacy preservation. We develop privacy-preserving personalized pricing algorithms and show that they achieve near-optimal revenue by deriving theoretical (upper and lower) performance bounds. Our analyses further suggest that, if the firm possesses a sufficient amount of historical data, then it can achieve certain level of differential privacy almost for free. That is, the revenue loss due to privacy preservation is of smaller order than that due to estimation. We confirm our theoretical findings in a series of numerical experiments based on synthetically generated and On-line Auto Lending (CPRM-12-001) data sets. Finally, we also investigate the problem of privacy-preserving personalized assortment optimization and derive results parallel to those in the pricing setting.

Human vs. Machine: Underwriting Decisions in Finance

Using a randomized experiment in auto lending, we show that algorithmic underwriting outperforms the human underwriting process, resulting in 10.2% higher loan profits and 6.8% lower default rates. The machine performance is more stable across various risk dimensions and loan characteristics, whereas the performance of human underwritten loans largely declines for riskier and more complex loans. Moreover, the performance difference is more pronounced at underwriting thresholds with a high potential for agency conflict. These results are consistent with algorithmic underwriting mitigating agency conflicts and humans' limited capacity for analyzing complex problems.

The Fast and the Usurious: Putting the Brakes on Auto Lending Abuses

The car loan market is rife with consumer abuses: inflated pricing, discriminatory lending, and a variety of deceptions and scams. These abuses all stem from the dealer-centric nature of the auto finance market that ties the vehicle purchase to the vehicle financing. The overwhelming majority of consumers finance their purchases through the car dealer, but consumers cannot learn dealer financing terms in advance. They learn the offered financing terms only after spending substantial time and energy negotiating a car price, a trade-in price, warranties, insurance, and vehicle add-ons. At this point, because most consumers lack alternative financing options, they face a take-it-or-leave-it choice that leaves them especially vulnerable to dealer abuses. Not only does the lack of alternative financing options deprive consumers of the protection of competition in auto loans, but competition in the dealer-based lending market also actually works against consumers. Dealers auction off loans to financial institutions based largely on which institution allows the dealer the greatest compensation in the form of a markup on the loan. These ultimate lenders compete for the dealers’ business, not the consumers’, which results in consumers paying supracompetitive rates because of the dealer markup. The discretionary nature of the markups also enables discriminatory lending, with minorities often charged more for car loans, as well as a number of outright frauds and scams that cannot occur with third-party financing. This Article proposes to fix these auto lending abuses by requiring a three-business-day waiting period before delivery of the vehicle for con- sumers who do not have a bona fide third-party financing offer, as well as a prominently disclosed, penalty-free prepayment right for the loan during this period, and a system of mandatory data collection on auto loans to enable regulatory oversight. A penalty default waiting period would incentivize consumers to shop for financing separately from the vehicle purchase transaction, which will create positive competitive forces lowering dealers’ supracompetitive markups of financing, reduce opportunities for discriminatory lending, and protect consumers from other deceptive practices.

Spillover Effects of the Opioid Epidemic on Consumer Finance

Appendix available at https://ssrn.com/abstract=3324720. I examine the impact of the opioid epidemic on subprime auto lending. Using a difference-in-differences framework, I find that county-level increases in opioid abuse cause an increase in loan defaults. Moreover, I find that traditional credit scoring attributes (e.g., FICO score) fail to predict loan performance deterioration associated with opioid addiction. The resulting higher default rates and weaker predictive performance of traditional credit measures generate a negative externality for borrowers in opioid-afflicted areas, as evidenced by 4.4% higher loan costs for subprime borrowers.

Discrimination in the Auto Loan Market

We provide evidence of discrimination in auto lending. Combining credit bureau records with borrower characteristics, we find that Black and Hispanic applicants’ loan approval rates are 1.5 percentage points lower, even controlling for creditworthiness. In aggregate, discrimination crowds out 80,000 minority loans each year. Results are stronger where racial biases are more prevalent and banking competition is lower. Minority borrowers pay 70 basis point higher interest rates, but default less ceteris paribus, consistent with racial bias rather than statistical discrimination. A major anti-discrimination enforcement policy initiated in 2013, but halted in 2018, reduced discrimination in interest rates by nearly 60%.

Consumer Protection Laws in Auto Lending

We examine the impact of consumer protection laws on loan terms and outcomes in auto financing. We find that dealers strategically adjust contracts such that similar borrowers face the same monthly payments in states with and without interest rate limits. However, monthly payments are higher in states that prohibit post-default collections through wage garnishment. Default rates do not vary with usury law when lenders can seek deficiency payments; however, default rates are higher where wage garnishment is prohibited. We find that default rates are lower among borrowers who cannot discharge their debts due to a recent Chapter 7 bankruptcy, suggesting moral hazard as a plausible explanation for the higher default rates in states that prohibit wage garnishment. Although wage garnishment laws benefit borrowers who default, borrowers who pay their debt in full face higher total observable loan costs where wage garnishment is prohibited.

Credit Where Credit is Due: Drivers of Subprime Credit

We use individual credit histories to study how creditor-friendly repossession rights in auto lending affect borrowers. Results from a quasi-experimental setting show that bankruptcy rates among subprime auto borrowers increase twice as much following natural disasters in states with strong creditors’ rights compared to states with more borrower protection. Further tests show that auto repossessions increase the likelihood of bankruptcy, and reduce borrowers’ future access to both uncollateralized and collateralized credit, including home mortgage loans. Our findings suggest that creditors’ rights can have broad, negative effects on borrowers that extend beyond merely losing a collateralized asset.

Product Design in Selection Markets *

Abstract In selection markets, where the cost of serving consumers is heterogeneous and noncontractible, nonprice product features allow a firm to sort profitable from unprofitable consumers. An example of this “sorting by quality” is the use of down payments to dissuade borrowers who are unlikely to repay. We study a model in which consumers have multidimensional types and a firm offers a single product of endogenous quality, as in Spence (1975) . These two ingredients generate a novel sorting incentive in a firm’s first-order condition for quality, which is a simple ratio. The denominator is marginal consumer surplus, a measure of market power. The numerator is the covariance, among marginal consumers, between marginal willingness to pay for quality and cost to the firm. We provide conditions under which this term is signed and contrast the sorting incentives of a profit-maximizer and a social planner. We then use this characterization to quantify the importance of sorting empirically in subprime auto lending, analytically sign its impact in a model of add-on pricing, and calibrate optimal competition policy in health insurance markets.

The Effectiveness of Field Price Discretion: Empirical Evidence from Auto Lending

In many markets, it is common for headquarters to create a price list but grant local salespeople discretion to negotiate prices for individual transactions. How much (if any) pricing discretion headquarters should grant is a topic of debate within many firms. We investigate this issue using a unique data set from an indirect lender with local pricing discretion. We estimate that the local sales force adjusted prices in a way that improved profits by approximately 11% on average. A counterfactual analysis shows that using a centralized, data-driven pricing optimization system could improve profits even further, up to 20% over those actually realized. This suggests that centralized pricing—if appropriately optimized—can be more effective than field price discretion. We discuss the implications of these findings for auto lending and other industries with similar pricing processes. This paper was accepted by Serguei Netessine, operations management.