BankThink

The Federal Home Loan banks and the disappearing American dream

Homeownership gap
Due in part to deficient consumer data and biased risk prediction technology, the homeownership gap between white and non-white Americans has widened since the Fair Housing Act was passed 55 years ago.
Monkey Business Images/Monkey Business - stock.adobe.co

The American dream has long been associated with homeownership, but many Americans have been shut out of home buying opportunities due to an antiquated housing finance system that could be described as redlining for the 21st century. The Fair Housing Act (FHA) was passed 55 years ago to protect Americans from such discrimination. However, due in part to deficient consumer data and biased risk prediction technology, the homeownership gap between white and non-white Americans has widened since the FHA was passed. The Federal Home Loan banks, which were created to democratize homeownership nearly 100 years ago, have been hamstrung by decades-old, automated loan underwriting technology. Mechanisms for enabling homeownership like the Home Loan banks' Mortgage Partner Finance program, which was designed to help lenders reduce their risk exposure, rely on technology crafted more than 30 years ago. While this technology can reasonably predict the delinquency risk of prime borrowers that have deep credit histories and high credit scores, it is woefully inaccurate for all others. Tens of millions of Americans have been unfairly denied homeownership due to fixable inaccuracies in the very technology the banks impose on their ecosystem of lenders.

The Home Loan banks have come under scrutiny by Congress and by Federal Housing Finance Administration Secretary Sandra Thompson. Many are questioning the decisions of Home Loan banks that provided financing of high-risk balance sheets at Silvergate and Silicon Valley Bank and criticizing the Home Loan Bank System for falling short of its mission to support fair and responsible mortgage lending, which is necessary to strengthen our communities. The banks continue to rely on obsolete technology to guide their Mortgage Partner Finance approvals, which also guides $600 billion of mortgage underwriting decisions executed by nearly 7,000 lenders that are members of the Home Loan bank ecosystem. Lisa Rice, president and CEO of the National Fair Housing Alliance, recently noted, "Technology is the new frontier of civil rights. While we have passed civil rights statutes designed to stop discrimination; we have not designed laws to dismantle the systems of inequality that are still producing biased impacts."

The automated underwriting technology developed in the 1990s that is still imposed today by the Home Loan banks on their member banks has become a major culprit in undermining the civil rights of Americans. The racial homeownership gap is wider today than it was one generation after the abolition of slavery. This gap has been an indelible stain on the fabric of American economic opportunity. It endures in spite of regulators' and legislators' best efforts because of how deeply entrenched the monopoly of automated underwriting systems has been within U.S. housing finance.

Indeed, the Home Loan banks dictate to their member banks that Mortgage Partner Finance will not be provided unless borrowers are approved by these old systems. These systems rely on even older credit scoring methods developed by the private Fair Isaac Corporation, which produces the ubiquitous FICO score. The automated underwriting systems may not have a hard minimum FICO score — but generally speaking, anyone with a score below 620 is highly unlikely to receive loan approval for a mortgage. An individual may be perfectly responsible with their financial obligations, such as apartment rent payments, utility bills, childcare expenses and so forth, yet have a thin history of debt that causes their FICO score to flounder.

Herein lies the problem. The automated underwriting systems impose a hard minimum FICO wall, not necessarily because borrowers below that threshold are riskier, but because the technology cannot accurately predict their risk. This is a byproduct of a self-reinforcing loop where mortgages were approved for applicants who already established a prior record of multiple debts, which leads to a gross underrepresentation of younger people (Gen Z and millennials) and people of color in the historical data. This imbalanced data is used to calibrate risk prediction models. The models learn from individuals with higher FICO scores, becoming ever more accurate for such borrowers, which motivates further lending to other similar people. This cycle of data concentration repeats year over year, locking out low-scoring FICO households from the opportunity to climb the economic ladder through homeownership.

Sen. Elizabeth Warren, D-Mass., in a letter to the Consumer Financial Protection Bureau, challenged the idea that credit card late fees serve as a deterrent to delinquency, instead saying issuers told her office that some of them earn tens of millions of dollars collecting late fees.

June 30
Elizabeth Warren

Even if a person with a low FICO score has the financial strength to repay a mortgage, they are deemed high-risk due to the underwriting system's own inaccuracy in predicting delinquency for lower-scoring FICO borrowers. This is not a failure of the individual, but a failure of the algorithms embedded in the automated underwriting technology that pervades the Home Loan bank ecosystem. The victims of this pervasive failure number in the tens of millions.

The good news is that powerful new streams of consumer data, modern credit scoring algorithms, innovative loan underwriting technology and the development of novel algorithmic debiasing techniques from fintech and civil rights organizations have progressed dramatically in recent years. Those advances have been deliberately formulated to underpin a more equitable, safe and sound housing finance system. It is up to the Home Loan banks to modernize by harnessing these advances. At a minimum, it means removing the mandate that their 6,800 member banks must rely solely on underwriting technology created decades ago.

The loan repayment risk of even low-scoring FICO individuals can now be predicted with far greater accuracy than was imaginable when those automated underwriting systems were originally conceived in the 1990s. This leap in predictive accuracy for low-scoring FICO borrowers will empower the Home Loan banks to finally make good on their founding mission: to responsibly increase homeownership opportunity across all communities.

We call upon the Home Loan banks, as well as all U.S. housing and financial services agencies and regulators, to embrace their full obligation under our nation's fair housing and fair lending laws to use fair, modern technology in their advertising, underwriting, pricing and other systems to safely expand homeownership opportunities for millions of Americans — especially younger generations and people of color, who have fallen into a homeownership gap kept open by obsolete technology. Now is the time to let go of archaic automated underwriting systems that limit upward mobility toward the American dream.

For reprint and licensing requests for this article, click here.
Community banking Redlining Underwriting
MORE FROM AMERICAN BANKER