Leveraging AI for Equity in Mortgage Credit Access: A Data Science and Mathematical Perspective

This study delves into the transformative potential of Artificial Intelligence (AI) within the mortgage industry to address long-standing barriers hindering homeownership for marginalized communities, specifically Black, Brown, and lower-income groups. We introduce a multidimensional framework encompassing societal, ethical, legal, and practical considerations essential for the development and deployment of AI models. Employing this framework, we analyze various AI applications reshaping the mortgage sector, including digital marketing, incorporation of non-traditional "big data" in credit scoring algorithms, AI-driven property valuation, and loan underwriting models. Our assessment reveals that while existing AI models may perpetuate historical biases in mortgage lending, opportunities exist for proactive AI model development aimed at dismantling systemic credit access barriers.

Introduction

The emergence of AI technologies in mortgage origination and servicing offers a promising pathway to revolutionize the housing sector and address the historically underserved segments of the population. Enhanced technologies coupled with non-traditional data acquisition hold the potential to streamline processes, enhance consumer experience, and reduce costs. However, without meticulous design, AI implementation risks entrenching prevailing inequalities in the housing market, conflicting with efforts to boost homeownership rates among marginalized groups.

Ethical Considerations in AI Model Development

The pivotal challenge pertains to the judicious selection of data for creating foundational models, particularly in evaluating credit risk. Mere predictive power of a variable, such as educational background or spending patterns, does not suffice for its ethical inclusion. The SCALE typology, proposed herein, offers a structured approach for stakeholders, including industry, regulators, and policymakers, to evaluate the merits and demerits of deploying diverse data in mortgage origination or servicing. Such analysis could inform regulatory enhancements, fair lending expansions, and privacy legislation prohibiting the use of specific data types for mortgage credit evaluation.

Addressing Biases and Socio-Political Priorities

AI models have raised concerns regarding perpetuating biases observed in human decision-making. Instances across various sectors, from Microsoft's AI chatbot to discriminatory outcomes in facial recognition programs, underscore the risks of embedding biases in AI systems. In mortgage contexts, biased feedback loops perpetuate historical credit barriers, potentially leading to systematic errors. Adoption of the SCALE framework by stakeholders is essential for the responsible deployment, evaluation, and monitoring of digital tools in mortgage operations.

Regulatory and Policy Recommendations

Proposed legislative measures, like the Algorithmic Accountability Act, aim to regulate AI's use in critical decision-making processes. However, challenges persist in adequately auditing the dynamic nature of model development. Industry self-regulation, complemented by collaborative efforts as demonstrated by initiatives like FinRegLab, presents a pragmatic alternative to conventional regulatory oversight. Revisiting existing HUD rules and adopting a nuanced approach to the disparate impact legal standard are recommended for more effective enforcement in addressing discriminatory practices.

AI's Role in Advancing Equity and Overcoming Bias

AI strategies in mortgage markets hold the potential to redress systemic discrimination. Thoughtfully designed tools, leveraging empirical research, can mitigate biases and enhance predictive accuracy. However, the inclusion of race at the design stage, contrary to existing policies, may be pivotal in offsetting discriminatory effects in credit assessment models, thereby advancing equity and homeownership opportunities.

Future Directions

Critical questions persist regarding the goals and outcomes of AI/ML adoption in lending decisions. Defining success criteria for digitalization projects, particularly in expanding homeownership for underserved communities, necessitates a broad and contextual understanding. Moreover, employing AI/ML for fair and equitable treatment verification in decision-making processes offers a responsible use case, advocating algorithmic reparation principles.

Conclusion

AI's integration in mortgage markets presents a dual opportunity of enhancing efficiency while addressing historical inequities. By adhering to responsible design principles, leveraging diverse data ethically, and monitoring for fairness, AI has the potential to transform mortgage credit access while advancing social and political goals of equity and inclusion.

References

1. Gartner (n.d.). Digitalization. Available online: https://www.gartner.com/en/information-technology/glossary/digitalization (accessed on 7 February 2023).
2. IBM Cloud Education (n.d.). What Is Artificial Intelligence (AI)? Available online: https://www.ibm.com/cloud/learn/what-is-artificial-intelligence?lnk=fle (accessed on 7 February 2023).
3. Brown, S. Machine Learning, Explained. 2021. Available online: https://mitsloan.mit.edu/ideas-made-to-matter/machine-learning-explained#:~:text=What%20is%20machine%20learning%3F,to%20how%20humans%20solve%20problems (accessed on 7 February 2023).
4. Akinwumi, M.; Merrill, J.; Rice, L.; Saleh, K.; Yap, M. An AI Fair Lending Policy Agenda for the Federal Financial Regulators. Policy Brief. 2021. Brookings Center on Regulation and Markets, Brookings Institution. Available online: https://www-brookings-edu.cdn.ampproject.org/c/s/www.brookings.edu/research/an-ai-fair-lending-policy-agenda-for-the-federal-financial-regulators/?amp (accessed on 7 February 2023).
5. FinRegLab. The Use of Machine Learning for Credit Underwriting. The Milken Institute. 2021. Available online: https://finreglab.org/wp-content/uploads/2021/09/the-Use-of-ML-for-Credit-Underwriting-Market-and-Data-Science-Context_09-16-2021.pdf (accessed on 7 February 2023).
6. Arnold, D.; Dobbie, W.; Hull, P. Measuring Racial Discrimination in Algorithms. AEA Pap. Proc. 2021, 111, 49–54. [Google Scholar ] [CrossRef ]
7. Martin, K. Designing Ethical Algorithms. MISQ Exec. 2019, 18, 129–142. [Google Scholar ] [CrossRef ]
8. Kroll, J.A. The Fallacy of inscrutability. Philos. Trans. R. Soc. A 2018, 376, 2018008420180084. [Google Scholar ] [CrossRef ]
9. Johnson, K.N.; Pasquale, F.A.; Chapman, J.E. Artificial intelligence, Machine Learning, and Bias in Finance: Toward Responsible innovation. Fordham Law Rev. 2019, 88, 499. [Google Scholar ]
10. Perry, V.G.; Schnare, A.B. Tipping the SCALE: Will Alternative Data in Credit Scoring Promote Or Impede Fair Lending Goals? Presentation at the National Association of Realtors Public Policy Forum. 2021. Available online: https://www.nar.realtor/events/public-policy-forum/tipping-the-scale-will-alternative-data-in-credit-scoring-promote-or-impede-fair-lending-goals (accessed on 7 February 2023).
11. Martin, K. Ethical Implications and Accountability of Algorithms. J. Bus. Ethics 2019, 160, 835–850. [Google Scholar ] [CrossRef ]
12. Davis, J.L.; Williams, A.; Yang, M.W. Algorithmic Reparation. Big Data Soc. 2021, 8. [Google Scholar ] [CrossRef ]
13. Martin, K.; Nissenbaum, H. What Is It About Location? Berkeley Technol. Law J. 2020, 35, 251–326. [Google Scholar ] [CrossRef ]
14. Nissenbaum, H. Privacy in Context: Technology, Policy, and the Integrity of Social Life; Stanford Law Books; Stanford University Press: Stanford, CA, USA, 2009. [Google Scholar ]
15. Walzer, M. Spheres of Justice: A Defense of Pluralism; Basic Books: New York, NY, USA, 1984. [Google Scholar ]
16. Blattner, L.; Nelson, S. How Costly Is Noise? Data and Disparities in Consumer Credit [Working Paper]. 2021. Available online: https://arxiv.org/abs/2105.07554v1 (accessed on 7 February 2023).
17. Consumer Financial Protection Bureau(CFPB). Using Publicly Available Information to Proxy for Unidentified Race and Ethnicity. Consumer Financial Protection Bureau; 2014. Available online: https://files.consumerfinance.gov/f/201409_cfpb_report_proxy-methodology.pdf (accessed on 7 February 2023).
18. Heaven, W.D. Bias Isn’t the Only Problem With Credit Scores—And No, AI Can’t Help. MIT Technology Review, 17 June 2021. Available online: https://www.technologyreview.com/2021/06/17/1026519/racial-bias-noisy-data-credit-scores-mortgage-loans-fairness-machine-learning/ (accessed on 7 February 2023).
19. Schnare, A.B. Alternative Credit Scores and the Mortgage Market: Opportunities and Limitations. Progressive Policy Institute. 2017. Available online: https://www.progressivepolicy.org/wp-content/uploads/2017/12/UpdatedCreditScoring_2017.pdf (accessed on 7 February 2023).
20. Engler, A. The E.U and U.S. Diverge on AI Regulation: A Transatlantic Comparison and Steps to Alignment. Brookings Institution. 2023. Available online: https://www.brookings.edu/articles/the-eu-and-us-diverge-on-ai-regulation-a-transatlantic-comparison-and-steps-to-alignment/ (accessed on 26 August 2023).
21. Akinwumi, M.; Rice, L.; Sharma, S. Purpose, Process, and Monitoring: A New Framework for Auditing Algorithmic Bias in Housing and Lending. National Fair Housing Alliance. 2022. Available online: https://nationalfairhousing.org/wp-content/uploads/2022/02/PPM_Framework_02_17_2022.pdf (accessed on 26 August 2023).
22. Federal Reserve. The 2019 Survey of Consumer Finances. 2020. Available online: https://www.federalreserve.gov/econres/scfindex.htm (accessed on 28 August 2023).
23. Fannie Mae. COVID-19, Mortgage Digitization, and Borrower Satisfaction. 2021. Available online: https://www.fanniemae.com/media/40491/display (accessed on 7 February 2023).
24. Atske, S.; Perrin, A. Home Broadband Adoption, Computer Ownership Vary by Race, Ethnicity in the U.S. Pew Research Center. 2021. Available online: https://www.pewresearch.org/fact-tank/2021/07/16/home-broadband-adoption-computer-ownership-vary-by-race-ethnicity-in-the-u-s/ (accessed on 7 February 2023).
25. Vogels, E.A. Digital Divide Persists Even as Americans with Lower Incomes Make Gains in Tech Adoption. Pew Research Center. 2021. Available online: https://www.pewresearch.org/fact-tank/2021/06/22/digital-divide-persists-even-as-americans-with-lower-incomes-make-gains-in-tech-adoption/ (accessed on 7 February 2023).
26. Ehrentraud, J.; Garcia Ocampo, D.; Quevedo Vega, C. Regulating Fintech Financing: Digital Banks and Fintech Platforms. FSI Insights on Policy Implementation, 2020, No. 27, Financial Stability Institute, BIS. Available online: https://www.bis.org/fsi/publ/insights27.pdf (accessed on 7 February 2023).
27. nLIFT. Fulfilling the Promise of Fintech: The Case for A Nonprofit Vision and Leadership. The Aspen institute. 2018. Available online: https://www.aspeninstitute.org/wp-content/uploads/2018/09/nLIFT-Manifesto-FINAL-1.pdf?_ga=2.176913637.1579357171.1541431738-807926508.1541431738 (accessed on 7 February 2023).
28. Friedline, T.; Chen, Z. Digital Redlining and the Fintech Marketplace: Evidence From US Zip Codes. J. Consum. Aff. 2021, 55, 366–388. [Google Scholar ] [CrossRef ]
29. Haupert, T. The Racial Landscape of Fintech Mortgage Lending. Hous. Policy Debate 2022, 32, 337–368. [Google Scholar ] [CrossRef ]
30. Matz, S.C.; Menges, J.I.; Stillwell, D.J.; Schwartz, H.A. Predicting individual-Level income From Facebook Profiles. PLoS ONE 2019, 14, e0214369. [Google Scholar ] [CrossRef ]
31. Matz, S.C.; Netzer, O. Using Big Data as a Window into Consumers’ Psychology. Curr. Opin. Behav. Sci. 2017, 18, 7–12. [Google Scholar ] [CrossRef ]
32. Libai, B.; Bart, Y.; Gensler, S.; Hofacker, C.; Kaplan, A.; K?tterheinrich, K.; Kroll, E.B. Brave New World? On AI and the Management of Customer Relationships. J. Interact. Mark. 2020, 51, 44–56. [Google Scholar ] [CrossRef ]
33. Ali, M.; Sapiezynski, P.; Bogen, M.; Korolova, A.; Mislove, A.; Rieke, A. Discrimination Through Optimization: How Facebook’s Ad Delivery Can Lead To Biased Outcomes. Proc. ACM Hum.-Comput. Interact. 2019, 3, 199. [Google Scholar ] [CrossRef ]
34. Evans, C.; Miller, W. From Catalogs To Clicks: The Fair Lending Implications of Targeted internet Marketing. Consum. Compliance Outlook 2019, 3. Available online: https://consumercomplianceoutlook.org/2019/third-issue/from-catalogs-to-clicks-the-fair-lending-implications-of-targeted-internet-marketing/ (accessed on 7 February 2023).
35. Hermann, E. Leveraging Artificial intelligence in Marketing for Social Good—An Ethical Perspective. J. Bus. Ethics 2022, 179, 43–46. [Google Scholar ] [CrossRef ]
36. Martin, K.D.; Murphy, P.E. the Role of Data Privacy in Marketing. J. Acad. Mark. Sci. 2017, 45, 135–155. [Google Scholar ] [CrossRef ]
37. Martin, K.D.; Palmatier, R.W. Data Privacy in Retail: Navigating Tensions and Directing Future Research. J. Retail. 2020, 96, 449–457. [Google Scholar ] [CrossRef ]
38. Stefano, P.; Reczek, R.M.; Giesler, M.; Botti, S. Consumer Experiences With Marketing Technology: Solving the Tensions Between Benefits and Costs. NIM Mark. Intell. Rev. 2022, 14, 25–29. [Google Scholar ]
39. National Fair Housing Alliance. Fair Housing Groups Settle Lawsuit with Facebook: Transforms Facebook’s Ad Platform Impacting Millions of Users. 2019. Available online: https://nationalfairhousing.org/national-fair-housing-alliance-settles-lawsuit-with-facebook-transforms-facebooks-ad-platform-impacting-millions-of-users/ (accessed on 7 February 2023).
40. Jan, T.; Dwoskin, E. Facebook Agrees to Overhaul Targeted Advertising System for Job, Housing and Loan Ads After Discrimination Complaints. The Washington Post, 19 March 2019. Available online: https://www.washingtonpost.com/business/economy/facebook-agrees-to-dismantle-targeted-advertising-system-for-job-housing-and-loan-ads-after-discrimination-complaints/2019/03/19/7dc9b5fa-4983-11e9-b79a-961983b7e0cd_story.html (accessed on 7 February 2023).
41. Brown, A.K. Fair Lending—Digital Marketing and HMDA 2018. Compliance Session at the Marquis User’s Conference, Skadden Foundation. 2019. Available online: https://gomarquis.com/wp-content/uploads/2019/10/Austin-Brown-Fair-Lending-Digital-Marketing-and-HMDA-2018.pdf (accessed on 7 February 2023).
42. Ballard Spahr. DOJ/CFPB/OCC Settle Redlining Lawsuit Against Mississippi-Based National Bank. Consumer Finance Monitor. 2021. Available online: https://www.jdsupra.com/legalnews/doj-announces-major-new-initiative-2017442/ (accessed on 7 February 2023).
43. Humber, N.J.; Matthews, J. Fair Housing Enforcement in the Age of Digital Advertising: A Closer Look at Facebook’s Marketing; Working Paper; Roger Williams University: Bristol, RI, USA, 2020; Available online: https://docs.rwu.edu/cgi/viewcontent.cgi?article=1308&context=law_fac_fs (accessed on 7 February 2023).
44. Aronowitz, M.; Golding, E.; Choi, J. The Unequal Costs of Homeownership. MIT Golub Center for Finance and Policy. 2020. Available online: https://gcfp.mit.edu/wp-content/uploads/2020/10/Mortgage-Cost-for-Black-Homeowners-10.1.pdf (accessed on 7 February 2023).
45. Ramirez, E.; Brill, J.; Ohlhausen, M.K.; McSweeny, T. Big Data: A Tool for Inclusion or Exclusion? Understanding the Issues, FTC Report, Federal Trade Commission; 2016. Available online: https://www.ftc.gov/system/files/documents/reports/big-data-tool-inclusion-or-exclusion-understanding-issues/160106big-data-rpt.pdf (accessed on 7 February 2023).
46. Kreiswirth, B.; Schoenrock, P.; Singh, P. Using Alternative Data to Evaluate Creditworthiness. Consumer Financial Protection Bureau; 2017. Available online: https://www.consumerfinance.gov/about-us/blog/using-alternative-data-evaluate-creditworthiness/ (accessed on 7 February 2023).
47. Cochran, K.T.; Stegman, M.; Foos, C. Utility, Telecommunications, and Rental Data in Underwriting Credit. Urban Institute Research Report. 2021. Available online: https://www.urban.org/research/publication/utility-telecommunications-and-rental-data-underwriting-credit/view/full_report . (accessed on 7 February 2023).
48. Lee, A.S.; Schnare, A.; Turner, M.A.; Walker, P.D.; Varghese, R. Give Credit Where Credit Is Due: Increasing Access To Affordable Mainstream Credit Using Alternative Data. Urban Markets Initiative Report. Policy and Economic Research Council and the Brookings Institution. 2006. Available online: https://www.brookings.edu/research/give-credit-where-credit-is-due-increasing-access-to-affordable-mainstream-credit-using-alternative-data/ (accessed on 7 February 2023).
49. CFPB. Report on the Use of Remittance Histories in Credit Scoring. Consumer Financial Protection Bureau; 2014. Available online: https://www.consumerfinance.gov/data-research/research-reports/report-on-the-use-of-remittance-histories-in-credit-scoring/ (accessed on 7 February 2023).
50. Drehobl, A.; Ross, L. Lifting the High Energy Burden in America’s Largest Cities: How Energy Efficiency Can Improve Low Income and Underserved Communities. ACEEE. 2016. Available online: https://www.aceee.org/sites/default/files/publications/researchreports/u1602.pdf (accessed on 7 February 2023).
51. Byrne, J.; Portanger, C. Climate Change, Energy Policy, and Justice: A Systematic Review. Anal. Krit. 2014, 36, 315–343. [Google Scholar ] [CrossRef ]
52. Carley, S.; Konisky, D.M. the Justice and Equity Implications of the Clean Energy Transition. Nat. Energy 2020, 5, 569–577. [Google Scholar ] [CrossRef ]
53. National Consumer Law Center. the Credit Score Pandemic Paradox. 2022. Available online: https://www.nclc.org/images/pdf/special_projects/covid-19/IB_Pandemic_Paradox_Credit_invisibility.pdf (accessed on 7 February 2023).
54. Choi, J.H.; Young, C. Owners and Renters of 6.2 Million Units in Small Buildings Are Particularly Vulnerable During the Pandemic. 2020. Available online: https://www.urban.org/urban-wire/owners-and-renters-62-million-units-small-buildings-are-particularly-vulnerable-during-pandemic (accessed on 7 February 2023).
55. FinRegLab. The Use of Cash-Flow Data in Underwriting Credit: Empirical Research Findings. The Milken Institute. 2019. Available online: https://finreglab.org/wp-content/uploads/2019/07/FRL_Research-Report_Final.pdf (accessed on 7 February 2023).
56. Friedman, B.; Nissenbaum, H. Bias in Computer Systems. ACM Trans. Inf. Syst. 1996, 14, 330–347. [Google Scholar ] [CrossRef ]
57. Odinet, C.K. Predatory Fintech and the Politics of Banking. Iowa Law Rev. 2021, 10, 1739–1800. [Google Scholar ]
58. Federal Housing Administration. Underwriting Manual: Underwriting and Valuation Procedure Under Title II of the National Housing Act; Federal Housing Administration: Washington, DC, USA, 1938. Available online: https://www.huduser.gov/portal/sites/default/files/pdf/Federal-Housing-Administration-Underwriting-Manual.pdf (accessed on 7 February 2023).
59. Rothwell, J.; Perry, A.M. Biased Appraisals and the Devaluation of Housing in Black Neighborhoods. Brookings Institute. 2021. Available online: https://www.brookings.edu/research/biased-appraisals-and-the-devaluation-of-housing-in-black-neighborhoods/ (accessed on 7 February 2023).
60. Folk, J.; Chen, K. Avoiding Overvaluation Risk and Appraisal Bias in Today’s Uniquely Challenging Market Session. Clear Capital. 2021. Available online: https://www.clearcapital.com/avoiding-overvaluation-risk-and-appraisal-bias-in-todays-uniquely-challenging-market-session (accessed on 7 February 2023).
61. Williamson, J.; Palim, M. Appraising the Appraisal: A Closer Look at Divergent Appraisal Values for Black and White Borrowers Financing Their Home. Fannie Mae. 2022. Available online: https://www.fanniemae.com/media/42541/display (accessed on 7 February 2023).
62. House Canary. Reducing Racial Bias in Home Appraisals Using Automated Valuation Technology. 2021. Available online: https://www.housecanary.com/wp-content/uploads/2021/12/Reducing-Racial-Bias-in-Home-Appraisals-Using-Automated-Valuation-Technology-December-2021.pdf (accessed on 7 February 2023).
63. Perry, V.G.; Aronowitz, M.; Choi, J.H.; Golding, E.; Green, M.; Green, R.K.; Jourdain-Earl, M.; Aiko Nelson, A.; Rhue, L.; Rice, L. 2020 State of Homeownership in Black America. 2020. Available online: https://www.shiba2020.com/ (accessed on 7 February 2023).
64. FinRegLab; Blattner, L.; Spiess, J. Machine Learning Explainability Fairness: Insights From Consumer Lending, Working Paper. 2022. Available online: https://finreglab.org/wp-content/uploads/2022/04/FinRegLab_Stanford_ML-Explainability-and-Fairness_insights-from-Consumer-Lending-April-2022.pdf (accessed on 7 February 2023).
65. Kluttz, D.N.; Kohli, N.; Mulligan, D.K. Shaping Our Tools: Contestability as a Means To Promote Responsible Algorithmic Decision Making in the Professions. In After the Digital Tornado: Networks, Algorithms, Humanity; Werbach, K., Ed.; Cambridge University Press: Cambridge, UK, 2020; pp. 137–152. [Google Scholar ]
66. Huard, F. How Machine Learning Is Making It Easier to Spot Fraud and Mitigate Risk for Underwriters. Core Logic. 2023. Available online: https://www.corelogic.com/culture-stories/how-machine-learning-is-making-it-easier-to-spot-fraud-and-mitigate-risk-for-underwriters/ (accessed on 26 August 2023).
67. Schwartz, O. In 2016, Microsoft’s Racist Chatbot Revealed the Dangers of Online Conversation. IEEE Spectr. 2019. Available online: https://spectrum.ieee.org/in-2016-microsofts-racist-chatbot-revealed-the-dangers-of-online-conversation (accessed on 7 February 2023).
68. Collier, K. Twitter’s Racist Algorithm Is Also Ageist, Ableist and Islamaphobic, Researchers Find. NBC News 2021. Available online: https://www.nbcnews.com/tech/tech-news/twitters-racist-algorithm-also-ageist-ableist-islamaphobic-researchers-rcna1632 (accessed on 7 February 2023).
69. Guynn, J. Google Photos Labeled Black People “Gorillas”. USA Today. 2015. Available online: https://www.usatoday.com/story/tech/2015/07/01/google-apologizes-after-photos-identify-black-people-as-gorillas/29567465/ (accessed on 7 February 2023).
70. Rosen, E.; Garboden, P.M.E.; Cossyleon, J.E. Racial Discrimination in Housing: How Landlords Use Algorithms and Home Visits to Screen Tenants. Am. Sociol. Rev. 2021, 86, 787–822. [Google Scholar ] [CrossRef ]
71. Kaye, K. This Senate Bill Would Force Companies to Audit AI Used for Housing and Loans. Protocol. 8 February 2022. Available online: https://www.protocol.com/enterprise/revised-algorithmic-accountability-bill-ai (accessed on 7 February 2023).
72. Myers, S.; Samuel, M., Jr. On the Surprise of the Minnesota Paradox and Large Racial Disparities. Video. Humphrey School of Public Affairs, University of Minnesota. 2020. Available online: https://www.youtube.com/watch?v=6TsLSv1KCWM (accessed on 7 February 2023).
73. Ajunwa, I. The Paradox of Automation as Anti-Bias intervention. Cardozo Law Rev. 2019, 41, 1671–1742. [Google Scholar ]
74. Thomas, R.; Uminsky, D. The Problem with Metrics Is a Fundamental Problem for AI. arXiv 2020, arXiv:2002.08512. [Google Scholar ]