September 17, 2021

MP Now News

Mortgage News

Is there racial bias baked into GSE underwriting?

One recently syndicated article on racial prejudice from The Markup on Racial Bias, published by ABC News, AP, Market Watch, and others, with support from some and attacks from others, has attracted widespread attention from trade groups, policymakers, and housing advocates.

The story “The Secret Bias Hidden in Mortgage Approval Algorithms” included testimony from fair housing activists who concluded that there was “systemic racism” in the mortgage process. But the same story was told by the MBA and the ABA, as well as CHLA, on the grounds that the data were selective and led to a biased conclusion.

The MBA was clearly so disturbed by history that he was a public comment posted on their website and posted to Newslink subscribers, in part with the statement, “From the outset we have told The Markup that its analysis of HMDA data and its pre-determined conclusions regarding mortgage lending do not take into account several”. Key components that form the backbone of credit decisions, including a borrower’s creditworthiness and credit history. “

I was also quoted in the article which said, “This is a relatively new world of automated underwriting engines that are deliberately not discriminatory, but likely by their effects.

In an effort to defuse this debate and at least give an additional perspective that takes both points of view in different ways, I want to highlight why I believe there is some level of racial bias and discrimination in the way we do The GSEs are pricing and drawing mortgages, which is likely to attract more minority applicants to one FHA Mortgage or maybe end up without a mortgage at all.

First, I have a problem with the item for several reasons. HMDA data shows results for a variety of variables, including but not race I AM Information. Obviously, this is a big deal for anyone in the industry since FICO is a threshold data point that determines eligibility for a mortgage.

second, the reporters only looked at conventional loans, that is, loans only from Fannie Mae and Freddie Mac, and did not include loans from FHA/will/USDA. This is important because FHA in particular has a much higher mix of successful mortgages to minority borrowers. On these two points alone, the MBA and ABA agreed that the conclusions were skewed because they missed these key points.

The authors argued that the MBA was one of the organizations working to ban the publication of FICO results in HMDA reporting, but I agree with the reasoning of the MBA here. With the items reported to HMDA, the privacy risk would dramatically increase if FICO had been included in public reporting. However, this does not release the authors from at least recognizing why this data element is so important.

Now, after all that has been said, let me formulate why I believe, as I said in my quote, “This is a relatively new world of automated underwriting engines being driven by intention must not discriminate, but through effect probably do. “

Risk-based pricing

The GSEs risk the base price on their mortgages, while the FHAs lump their mortgages. The bottom line here is that borrowers with FICO scores below, say 720, with a high loan-to-value (LTV) often find a better deal by going to the FHA. What I am complaining about with the GSEs’ pricing methodology is how they combine price adjustments at the credit level (NSP.As) in addition to the g-base fees for low down payment borrowers, which some may find excessive.

The GSEs have private mortgage insurance to cover their losses even with a minimal down payment from the borrower up to an LTV of around 65%. If private capital insures losses to an almost catastrophic level of risk in the event of a failure, why are they charging such exorbitant premiums on top of that coverage?

For example, a 95% LTV purchase with a 640 FICO adds an additional 275bp fee on top of the base G fees and MI expenses. The reason for this is simple, the GSEs believe, as stated in a public statement by FHFAthat MI companies have insufficient counterparty risk to hedge this initial loss in a deep recession as MI companies can fail and claims may fail to pay.

But over the past few years the GSEs have put new capital requirements on MI companies to protect themselves against a major event, which makes me wonder why they still haven’t changed the aggressive LLPA structure to this deep one First loss coverage to be taken into account. Simply put, many low down payment borrowers are hit by additional costs from MI and LLPAs that lead to higher interest rates or potentially force them to FHA. And low down payment borrowers have a higher concentration of minority applicants, which in part contributes to rejection rates.


The use of credit scores in underwriting became the standard for assessing willingness to repay in the mid to late 1990s. FICO is the standard used for this process, and while the GSEs argue that they use their own credit scoring method in their underwriting systems, the lender still uses FICO to set floors, set prices, and more.

There are two problems here. First, the GSEs haven’t updated their AUS models with updated FICO models, let alone the fact that they refuse to test even alternative models like the Vantage Score, which claims to be able to rate thousands of more borrowers to the ones today access to credit is denied.

Second, the FICO model does not attach value to timely payment of rent, utilities, cell phone bills, and other consistent payments from people that are not scored. While Fannie Mae for her last announcement that they can incorporate rental history into a credit rating that is too recent an announcement to judge effectiveness.

The challenge here is that Americans with underserved Americans who make timely payments but have no credit, or perhaps less than the multiple lines of credit that make a high FICO score, will have more difficulty getting a mortgage. This particularly affects Hispanic and African Americans, whose scores may be lower just because their proven repayment history doesn’t fit into a scoring model.

2 of 3 rule

In January of this year, just before Mark Calabria has left office, the FHFA, the Treasury and the GSE a change completed to the Preferred Stock Purchase Agreement (PSPA), which covered a number of things involving second homes, 2-4 units, cash sales, and more. In addition, the rule limits purchases from each lender to a maximum of 6% of the purchase money mortgages and a maximum of 3% of the refinancing mortgages during the last 52 weeks, which may have two or more higher risk characteristics at the time of issuing: combined loan-to-value ratio (LTV) greater than 90% ; Debt-to-income ratio of more than 45%; and FICO (or equivalent creditworthiness) less than 680. ”

This rule still applies today. This 2 out of 3 factor rule will clearly have a greater impact on first-time home buyers and minority groups, who are more likely to hit at least two of these caps at higher interest rates than white non-Hispanic borrowers.

Many argue that all of this is fine, that the GSEs are quasi-private companies and have set standards based on their risk assessment, and that the FHA, VA, and USDA can help offer options in the event a borrower does not qualify for a conventional mortgage. I am not here to argue for or against this argument. But none of this changes the result. The effect of some of these measures leads to a selection bias with regard to the results.

Earlier this year, a Publication of the Harvard Joint Center For Housing Studies concluded that black borrowers across all income levels are charged higher interest rates. In the research article, they admitted that “many other factors affect interest rates, including assets, debt, creditworthiness, down payment, mortgage amount, and term”. However, the study also concluded that “black homeowners have experienced systemic barriers to home ownership and wealth-accumulation that have limited their ability to access credit, which is a key component of maintaining low mortgage rates.”

When I read the MBA response to The Markup’s article, I thought this was a classic response from the trade association representing the industry, and when I read their comments, I agreed with the concerns I raised at the beginning of this comment would have. But I would argue that the divide in home ownership rates by race, credit scoring methods, and pricing structures should be an area where industry leadership is joining forces with consumer advocates to argue for change.

When I started my career as a loan officer, there was no credit rating. The philosophy was that no credit is not necessarily bad credit and insurers could make judgments to approve a loan. The problem is that the old process also led to racial prejudice. Technology and scoring take color from the decision-making process. But that doesn’t mean it isn’t without its flaws. As much effort should be made to highlight these shortcomings as to defend the industry against the points made in this story.

David Stevens has held a variety of real estate finance positions including Senior Vice President of Single Family at Freddie Mac, Executive Vice President at Wells Fargo Home Mortgage, Assistant Secretary to the Housing and FHA Commissioner, and CEO of the Mortgage Bankers Association.

This column does not necessarily reflect the views of the editors of HousingWire or its owners.

To contact the author of this story:
Dave Stevens at [email protected]

To contact the editor responsible for this story:
Sarah Wheeler at [email protected]