Fannie’s Loan Buyback Sophistry Relies on Modifying Analysts’ Behavior
At the heart of Fannie Mae’s valuation system is a tool that shunts a subset of the millions of loans purchased by the mortgage giant each year into a “high risk” pool. Fannie then pressures and incentivizes a corps of in-house analysts to create further doubt about the loans in order to send repurchase demands to lenders for up to 50 percent of the cases. So says a Fannie Mae whistleblower who spoke with appraiser-author Jeremy Bagott on the condition of anonymity.
According to the source, when Fannie’s collateral analysts review appraisals of properties categorized as high-risk, they can issue one of two outcomes: a “finding” or a “defect.” A “finding” is equivalent to a slap on the wrist. A “defect” triggers a demand letter to the lender to buy back the mortgage it sold to Fannie. Loans with higher loan-to-value ratios draw greater scrutiny.
Originally, in-house human appraisers were employed by Fannie to create a check on its automated tool. But this evolved, said the whistleblower, into the mortgage giant using a system of behavior modification to both pressure and encourage analysts to agree with the results of the algorithmic tool, known as Collateral Underwriter. The check and balance to the automated system has become a coerced validation of it.
“Fannie has a large investment in the technology with much prestige riding on it,” said the source. “The values and adjustments created by [the tool] need to be right for the sake of Fannie’s technocratic vision.” Fannie’s analysts complain they are now pushed to validate the automated system’s findings, said the insider.
“Careers ride on Collateral Underwriter,” said Brian Jarrard, a former Fannie Mae subject-matter expert. “When human opinion differs from the algorithmic findings, trouble follows for the human. You’d better start agreeing with the tool’s output or you will find yourself in a professional development program.” Jarrard corroborated much the whistleblower said about the internal workings of Fannie. He left the mortgage giant before it began incentivizing analysts to issue the buyback letters and penalizing those who don’t find a sufficient number of so-called “defects.”
“The goal with [appraisal] reports in the high-risk pool is to provide alternate data and discredit the outside appraisers’ comparable sales,” said the insider. “We always win, but if we had to go in front of an independent arbitrator, we would mostly lose,” said the insider.
In the spring of 2022, Fannie’s scoring system for evaluating the “production” of its in-house collateral analysts changed. The system now awards analysts more points if they are able to find overvaluations leading to lender repurchase demands. Having a high number of production points means you’re a high-flyer at the mortgage giant. High production points come with perks, prestige and job security. Staffers accumulate more production credits when they conclude appraisals are overvaluations.
“Many of the repurchase letters make no sense,” said the insider. If analysts don’t write enough defects on those files, they get hammered. If analysts fall short of their peers in “production points,” it’s considered their fault, said the insider.
“The stepped-up repurchase demands,” said the source, “are based on analysts at Fannie Mae trying to hit their numbers. It’s led to people pushing files into the repurchase bin that are extremely weak. If an [analyst] wants, or needs, to write a repurchase letter, he will find a way to do it, regardless of the quality of the appraisal.”
An offshoot of this stepped-up “productivity” in mortgage repurchase letters to lenders has been a flood of unsigned complaints to state appraiser boards, which seem to be connected. For at least a year, Fannie has been sending at times bizarre complaints to state agencies that respond to them in different ways. In some states, like Ohio and Texas, it has caused many hapless licensees to expend considerable time and money to clear their names.
Fannie Mae emphasizes the unsigned complaints are not automated. “Our expert analysts validate the appraisal results by asking questions like, Do the comparable sale selections make sense? Is the data accurate? Did the appraiser make appropriate adjustments? Are we getting the most probable value?”
Others, like the insider, cite lack of accountability.
“Because these [complaints] are not signed and the reviewer is not identified, there is no accountability to the reviewer,” said New York-based appraiser and blogger Jonathan Miller.
In 2022, Fannie Mae provided $684 billion in liquidity to the U.S. mortgage market, which enabled the financing of approximately 2.6 million home purchases, refinancings and rental units. That year, Freddie Mac, which uses a similar algorithmic tool, provided $614 billion in liquidity to the market, which enabled 2.5 million home transactions. In all, the two mortgage giants provided $1.3 trillion in liquidity to the market for over 5 million home transactions last year.
The median price of an existing home was just $278,200 in August 2019, according to the National Association of Realtors. But that figure surged to $407,100 as of August 2023.
As derivative data is repeatedly recycled through algorithms at Fannie and Freddie – something Cincinnati-based appraiser and podcaster Phil Crawford calls “data cancer” – expect similar rapid run-ups in median home prices as upside bias insulates markets from true price discovery.
- Fannie’s Loan Buyback Sophistry Relies on Modifying Analysts’ Behavior - October 16, 2023
- Finding of Bias in Home Valuations Fails by Own Measure - August 14, 2023
- The Nightmarish End of Home Appraisals - July 31, 2023