The announcements this summer that the automated underwriting systems Desktop Underwriter and Loan Prospector are now free are the latest cause for pause in a tsunami of industry developments. But for industry veterans, this announcement should bring up so much more than the cost of DU and LP.
Making something free, especially when there is no apparent upside to doing so, is highly questionable. The government-sponsored enterprises' move has been called a "desperate" move by some colleagues, and viewed as an attempt to simply see more loan data. Free or not, it's time to re-think agency AUS.
DU and LP exist only to determine the eligibility of the loan for sale to Fannie Mae and Freddie Mac. Lost in the excitement of the announcements is the fact that a salable loan is not always a well-underwritten loan.
Before DU and LP, the loan originator understood the borrowers' circumstances and assessed their willingness and ability to repay the mortgage in a high-touch transaction. Once the loan originator served the borrower's financial need for a loan, the underwriter served the investor's interests by referencing the Seller-Servicer Guide as they made the final credit decision.
For decades, default rates on U.S. mortgages were lower than 3%, demonstrating that these practices kept mortgage assets pristine and housing values generally stable.
The first generation of agency AUS products was impressive. Facilitating standardization and liquidity in the U.S. housing industry with a convenient tech tool is a technology adoption success story. The industry went from zero to virtually 100% adoption in 10 years, an impressive feat for any industry. Beyond that, the agencies charged their seller-servicers up to $35 per loan to be their data entry staff. It was brilliant.
As a former executive vice president of Keystroke.com, an early pioneer of online mortgage application and automated underwriting technology, I was fascinated by lenders that would use approve/eligible AUS decisions to empower a loan originator to go "direct to docs." I saw human underwriters devolving into "documentation engineers," helping aggressive loan originators game the AUS system. (Full Disclosure: I have also previously worked as a consultant to Calyx Software and LoanScoreCard, its AUS technology division.)
Fannie's decision to purchase a $250,000 loan by a lender to Jane Borrower is based on a 0.000008% impact on its $3.140 trillion portfolio. It's pretty easy for the GSEs to approve a $250,000 loan with a 100% total debt ratio at that rate of marginal risk. But for Jane Borrower, the impact is 100%. Qualified Mortgage and Ability-to-Repay are in place for a reason, and looking at this development from the borrower's perspective helps see that point.
From the economy's perspective, regardless of politics, the concentration of risk in the agencies' portfolios needs resolution. Progress is being made with credit risk transfer securities and higher guarantee fees. How is making DU and LP free contributing to that goal?
Perhaps making the AUS technology free was a concession to the dramatic increase in the cost of G-fees. Fannie and Freddie have surely noticed that newly-originated loans being held in bank portfolios have increased from 15% to 25% in the last few years.
But in making the AUS free, Fannie and Freddie only left about $250 million in annual revenue on the table. Divide that by the number of agency seller-servicers, and barely a dent is made in improving access to mortgage capital.
Lenders must further consider that if a loan will not be sold to Fannie or Freddie, if it will be held in portfolio, or sold to a private investor, then DU and LP are not only irrelevant for their portfolio or investor, they're also risky.
From a credit risk management perspective, Jane's loan will have a dramatically different impact on the portfolio of a bank or credit union than on Fannie or Freddie, and needs different risk management tools. It will have a .0625% impact on a $400 million portfolio, a .025% impact on a $1 billion portfolio, and a .000625% impact on a $40 billion portfolio.
Lenders must ask themselves why the agencies' risk parameters are acceptable for any portfolio. There are excellent AUS and credit risk models available that can be calibrated to unique credit risk metrics and appetite. Reminder: DU and LP are not for lenders; they exist only for Fannie and Freddie.
Most importantly, this raises the question of data privacy: should borrowers' private loan data be shared with the agencies on loans that won't be sold to them? Neither of the agencies will have financial interest in these loans, but they'll receive thousands of bits of highly personal borrower data. Lenders should proceed with utmost caution.
Many in the industry owe their careers to Fannie and Freddie. They provided a roadmap for mortgage lending for decades, and remain the standard for the mortgage secondary market. Housing would have collapsed during the crisis, along with the economy, without them.
However, it's time to consider whether DU and LP should remain the duopoly they are today, even if they are free. Lenders should consider shaking off pre-crisis habits for their portfolio and nonagency loans and consider using AUS tools that serve borrowers, protect borrower data privacy, and are appropriate for their portfolio or investors' risk tolerance.
In doing this, lenders will sustain the industry, increase private, nonagency investor confidence, help prevent the next calamity in U.S. housing, and leave a great legacy. An agency salable loan is not always a well-underwritten loan, but a well-underwritten loan can always be sold.
Eileen O'Grady is the founder of Elliott Bay Associates, a consulting firm based in Seattle.