InfoBytes Blog
Filter
Subscribe to our InfoBytes Blog weekly newsletter and other publications for news affecting the financial services industry.
NYDFS offers guidance to insurers on AI models
On January 17, NYDFS issued a guidance letter on artificial intelligence (AI) intended to help licensed insurers understand NYDFS’s expectations for combating discrimination and bias when using AI in connection with underwriting. The guidance is aimed at all insurers authorized to write insurance in New York State and is intended to help insurers develop AI systems, data information systems, and predictive models while “mitigat[ing] potential harm to consumers.”
The guidance letter states that while the use of AI can potentially result in more accurate underwriting and pricing of insurance, AI technology can also “reinforce and exacerbate” systemic biases and inequality. As part of the letter’s fairness principles, NYDFS states that an insurer should not use underwriting or pricing technologies “unless the insurer can establish that the data source or model… is not biased in any way” with respect to any class protected pursuant to New York insurance law. Further, insurers are expected to demonstrate that technology-driven underwriting and pricing decisions are supported by generally accepted actuarial standards of practice and based on actual or reasonably anticipated experience. It was last noted that these rules build on New York Governor Hochul’s statewide policies governing AI.
7th Circuit: No causation in FCA claims against mortgage servicer
On June 14, the U.S. Court of Appeals for the Seventh Circuit affirmed a district court’s grant of summary judgment in favor of a defendant mortgage servicer, holding that while the plaintiff had sufficient proof of materiality with respect to alleged violations of the False Claims Act (FCA), plaintiff failed to meet her burden of proof on the element of causation. Plaintiff (formerly employed by the defendant as an underwriter) alleged the defendant made false representations to HUD in the course of certifying residential mortgage loans for federal insurance coverage. She maintained that HUD would not have endorsed the loans for federal insurance if it had known defendant was not satisfying the agency’s minimum underwriting guidelines. Defendant moved for summary judgment after the district court excluded the bulk of plaintiff’s “expert opinion,” arguing that plaintiff could not meet her evidentiary burden on the available record. The district court sided with defendant, ruling that as a matter of law, plaintiff could not prove either materiality (due to the lack of evidence that would allow “a reasonable factfinder to conclude that HUD viewed the alleged underwriting deficiencies as important”) or causation (the false statement caused the government’s loss).
On appeal, the 7th Circuit explained that to show proximate causation, plaintiff was required to identify evidence indicating that the alleged false certifications in reviewed loans were the foreseeable cause of later defaults, as defaults trigger HUD’s payment obligations. The appellate court noted that “it is not clear how a factfinder would even spot the alleged false statement in each loan file, let alone evaluate its seriousness and scope.” Without further evidence indicating how defendant’s alleged misrepresentations caused subsequent defaults, the plaintiff’s claims could not survive summary judgment.
However, the 7th Circuit disagreed with the district court’s reasoning with respect to materiality under the FCA. Although the district court held that plaintiff had failed to establish materiality, the appellate court determined that because HUD’s regulations “provide some guidance, in HUD’s own voice, about the false certifications that improperly induce the issuance of federal insurance, and those are precisely the false certifications present here” there was enough evidence to “clear the summary judgment hurdle” on this issue.
Freddie allows digital paystubs in underwriting
On May 22, Freddie Mac announced new capabilities allowing lenders to use a borrower’s digital paystub data when assessing income paid through direct deposit. Lenders will be able to access the enhancements to Freddie’s automated income assessment tool through the Loan Product Advisor (LPA) asset and income modeler (AIM). Freddie noted that in addition to providing access to direct deposit data, AIM is also able to “assess income from tax return data for self-employed borrowers as well as bank account data to identify a history of positive monthly cash flow activity” to help first-time homebuyers and borrowers in underserved communities who may not qualify through traditional methods of underwriting. AIM is also designed to notify lenders when submitting this type of account data may benefit a borrower. The new AIM capability will be available beginning June 7 to Freddie-approved sellers that use LPA.
Fannie expands underwriting eligibility to help "credit invisible" borrowers
On December 6, Fannie Mae announced enhancements to its Desktop Underwriter to create more homeownership opportunities for “credit invisible” borrowers by changing its automated underwriting system to expand eligibility and further simplify the borrowing process for loans where borrowers do not have a credit score. Fannie noted that close to 15 percent of Black and Latino/Hispanic people are credit invisible (as compared to nine percent of their white and Asian counterparts), explaining that these imbalances lead to racial disparities in access to credit and quality affordable housing. “We believe consumers should benefit from their responsible money management habits and a steady stream of income when buying a home, even if they don’t have an established credit history,” Mallory Evans, Executive Vice President and Head of Single-Family Business at Fannie Mae, said in the announcement. “Traditional lending practices make it hard for borrowers with no credit score to access credit, so we’ve taken steps that may help them responsibly qualify for a home loan using data that provides a more holistic view of how they manage their money.”
Beginning December 10, enhancements made to the Desktop Underwriter will (i) update borrower eligibility criteria for those with no credit score to align with Fannie’s standard selling guide requirements; (ii) enable the system to evaluate “a borrower’s monthly cash flow over a 12-month period to potentially enhance their credit risk assessment”; and (iii) simplify the mortgage process by automating the current selling guide requirement for documenting nontraditional sources of credit.
Freddie to consider bank account data in automated underwriting
On October 17, Freddie Mac announced that beginning November 6, borrowers’ bank account data will be included as part of its loan purchase eligibility assessments. This “industry-first capability” will be made available to lenders and brokers through Freddie’s automated Loan Product Advisor (LPA) underwriting system. “With the addition of positive monthly cash flow data, our underwriting system can help with more accurately predicting a borrower’s ability to pay their mortgage because it uses a comprehensive view of how personal finances are managed over time,” Freddie said in its announcement. “Our latest innovation levels the playing field and helps make homes more accessible to borrowers whose lenders might not have qualified them with traditional methods of underwriting. This should particularly help first-time homebuyers and underserved communities.”
Lenders and brokers must obtain borrowers’ permission in order to submit financial data showing 12 or more months of cash flow activity. Data may be obtained from checking, savings, and investment accounts, including those used for direct deposit of income and monthly bill payments, such as rent, utilities, and auto loans, Freddie said, stressing that “account data submitted can only positively affect the borrower’s credit risk assessment.” Lenders and brokers will also be able to obtain financial account data from designated third-party service providers through LPA’s asset and income modeler—the same automated process used to verify assets, income, employment, and on-time rent payments, Freddie explained. Additionally, LPA will advise lenders when a borrower may benefit from the submission of additional account data.
The announcement follows Freddie’s decision to start considering on-time rent payments as part of its loan purchase decisions to increase homeownership opportunities for first-time homebuyers. (Covered by InfoBytes here.)
WA Superior Court: Insurance commissioner overstepped in banning credit scoring in underwriting
On August 29, the Washington State Superior Court entered a final order declaring that the Washington Insurance Commissioner exceeded his authority when he issued an emergency rule earlier this year banning the use of credit-based insurance scores in the rating and underwriting of insurance for a three-year period. As previously covered by InfoBytes, several industry groups led by the American Property Casualty Insurance Association (APCIA) sued to stop the rule from taking effect. The rule was intended to prevent discriminatory pricing in private auto, renters, and homeowners insurance in anticipation of the end of the CARES Act, and specifically prohibited insurers from “us[ing] credit history to place insurance coverage with a particular affiliated insurer or insurer within an overall group of affiliated insurance companies.” The rule applied to all new policies effective, and existing policies processed for renewal, on or after June 20, 2021. Industry groups countered that the rule would harm insured consumers in the state who pay less for auto, homeowners, and renters insurance because of the use of credit-based insurance scores to predict risk and set rates.
According to a press release issued by APCIA, earlier this year the superior court issued a bench decision granting the trade group’s petition for a declaratory judgment and invalidating the rule. The superior court “held that the Commissioner could not rely on the more general rating standard statute that prohibited “excessive, inadequate, or unfairly discriminatory” rates to “eliminate all meaning from the more specific credit history statutes by which the legislature had authorized its use.” Calling the final order “an important victory for Washington consumers, particularly lower risk senior policyholders who were forced to pay more to subsidize higher risk policyholders because the rule eliminated the use of credit,” the trade groups said they were pleased that the court agreed with their position that the Commissioner “exceeded his authority when he acted contrary to the longstanding statute that authorized the use of credit in the property and casualty insurance space.”
Federal government directs agencies to eliminate medical debt as an indicator of creditworthiness
On August 25, the Director of the Office of Management and Budget (OMB) issued a memo directing “agencies with direct loan and loan guarantee programs that focus on consumer loans or small and medium businesses where a consumer’s credit history is a factor, to whenever possible and consistent with the law take actions to reduce the impact of medical debt in the underwriting of Federal credit programs.” Although OMB recognized that some agencies such as the Department Veterans Affairs and the CFPB have already taken some steps to lessen medical debt burdens, it found that these prior efforts have been insufficient. Instead, the memo stresses that “[t]he collective efforts of the Federal Government, working with the private sector” are necessary to “remedy the impact of the issue of medical debt as an indicator for creditworthiness.” The memo outlines guidance for agencies to develop a plan to eliminate medical debt as a factor for underwriting in credit programs. These steps include (i) “[i]dentifying any statutory, regulatory, or administrative changes that would be required to modify criteria and consideration factors, exclude medical debt, or otherwise lessen the impact of medical debt consideration or underwriting in Federal lending programs”; (ii) conducting an “[i]nitial qualitative assessment and cost-benefit analysis of any statutory or regulatory changes” or any anticipated changes; (iii) conducting an “[a]ssessment of whether model updates are required for FCRA cost estimation, especially if the exclusion of medical debt would explicitly or implicitly affect particular underwriting requirements such as debt-to-income ratios, etc.”; and (iv) incorporating stakeholder input and assessing known risks that may impact an agency’s goal of achieving its plan.
Freddie to consider rent payments in automated underwriting
On June 29, Freddie Mac announced that it will begin considering on-time rent payments as part of its loan purchase decisions to increase homeownership opportunities for first-time homebuyers. Starting July 10, with a borrower’s permission, mortgage lenders and brokers will be able to submit bank account data showing 12-months of on-time rent payments through Freddie’s automated underwriting system. According to Freddie, bank account data will be “obtained from designated third-party service providers using the same automated process used to verify assets, income and employment” using its asset and income modeler. Freddie explained that eligible rent payment data includes checks, electronic transactions, or digital payments made through specific payment apps. “These automated capabilities provide greater efficiencies to lenders and allows them to deliver a better borrower experience while continuing to meet Freddie Mac’s strong credit underwriting standards,” the announcement said. Additional requirements for submitting rent payment data to Freddie’s underwriting system will be announced in an upcoming July Single-Family Seller/Servicer Guide Bulletin.
CFPB grants financial services company’s request to end no action letter
On June 8, the CFPB announced an order to end a “no-action letter” (NAL) issued to a consumer lending platform in 2020 in response to the NAL recipient’s request to shorten the term of the letter, which the company noted was necessary in order to be able to keep its underwriting models accurate and up-to-date during a period of significant economic change. As previously covered by InfoBytes, the Bureau first issued a NAL to the company in 2017, as part of the CFPB’s Project Catalyst initiative, relating to the Equal Credit Opportunity Act and the company’s loan underwriting and pricing model. Under the NAL, the company was immunized “from being charged with fair lending law violations with respect to its underwriting algorithm, while the ‘no-action letter’ remained in force.” In November 2020, the Bureau issued a second NAL after the first one expired (covered by InfoBytes here).
Freddie’s automated underwriting now includes bank account data
On May 26, Freddie Mac announced new automated underwriting capabilities that will allow mortgage lenders to verify assets, income, and employment using borrower-approved bank account data. The new functionality is available starting June 1, through Freddie’s asset and income modeler (AIM) within the Freddie Mac Loan Product Advisor. According to Freddie’s announcement, the automated underwriting capability “provides the borrower’s current employment status using borrower-approved bank account (direct deposit) or payroll data obtained from designated third-party service providers” in order to give “lenders a more efficient option than obtaining oral or written verification of employment prior to closing.” Freddie cited a recent study, which found that adopting offerings like AIM helps lenders “significantly boost efficiency and shorten cycle times by as much as 15 days.” These efficiencies, Freddie said, “translate into a 30 percent reduction in loan origination costs, greater customer satisfaction, and an increase in applications being completed and closed.” The announcement also noted that Freddie recently released the industry’s first automated-assessment of direct deposit income, which enables AIM to access additional fixed income or alternative income sources such as retirement, Social Security, Veteran Affairs benefits, alimony, and child support, as well as income from an applicant’s tax return for self-employed individuals.