I’ve written about appraisal bias, specifically the importance of the valuation community to participate in the narrative. Regardless of how appraisers feel about the issue, it doesn’t matter. What matters is for us to seize the opportunity to be part of the conversation.
Financial Institution Valuation Advisors
At FIVA (Financial Institution Valuation Advisors) we have had two webinars on the appraisal bias issue. The speakers included Peter Christensen of the Christensen Law Firm (Part 1) and Ed Pinto, senior fellow and director of the American Enterprise Institute (AEI) Housing Center (Part 2), and the following.
AEI Resources:
Additional speakers included Tom Munizzo, IFA, CMAR, DAR, ASA Controlling partner in Accurity Midwest and TJ McCarthy, SRA, ASA, IFA of T.J. McCarthy & Associates.
Since those webinars, Freddie Mac came out with a report that suggests bias: appraisal values are more likely to fall below the contracted sale price of a home in areas with a higher share of minority households. Some of their research highlights:
“A large portion of appraisers who completed appraisals in both minority and non-minority areas generated statistically significant gaps.”
“The patterns observed based on the aggregate national data mostly persist; thus, the appraisal gaps seem pervasive.”
More questions than answers
Why does the 1.7 million loan sample where the race or ethnicity is known (Exhibit 2) had a much lower Black and Latino gap vs. White, while the 12.7 million appraisal sample (Exhibit 1) found much wider appraisal gaps for minority vs. White tracts?
Why does the results for Exhibit 2 differ substantially from Exhibit 1? This raises many issues that were unaddressed, including the possibility that there are omitted variables.
That said, having no preconceived notion of the final outcome of the research, the goal is to make sure the research is completed in a robust and complete fashion.
Chief appraiser perspective
Questions to consider from a chief appraiser perspective.
1. While there may be correlation with race and the valuation gap, is it the cause/driver?
2. If these markets are in areas with lower education levels than other areas (especially with respect to financial literacy) are buyers more prone to pay higher prices as a result? If that is true, one would expect the appraisals to come in lower.
3. What percentage of the transactions that were undervalued ultimately proceeded to closing? What savings did the buyer’s realize as a result of the undervaluation? What are the default rates for those properties that were undervalued?
4. Does this suggest that the Realtors and mortgage brokers/lenders are not properly advising their clients when making offers, thereby running the risk of overpayment?
Other interpretations
Some “read into” the Freddie Mac report that it’s possibly inconclusive. If the research is based on 12 million appraisals in a five-year period with the researcher saying, “Our research marks the beginning of a comprehensive effort to better understand the key drivers contributing to the appraisal gap.” Sounds like they don’t really know but suspect its bias.
The research may be due to other influences such as census tract demographic inconsistencies, socio-economic variations, or minorities don’t align with the market value definition of an informed buyer.
Is Sales Comparison Approach the problem?
From the Freddie Mac press release, “We’re uniquely positioned to investigate potential gaps and provide data-driven research like this to advance solutions that promote equity across the valuation process. The company is also testing whether alternatives to traditional appraisals offer a more objective analysis of property value.”
Is Credit the problem?
Getting a loan starts with lending and credit. Freddie Mac, “Our goal is to develop solutions to this persistent problem, including appraisal best practices, uniform standards for automated valuation models, enhanced consumer disclosures, improved value processes, and revised fair lending exam procedures and risk assessments.”
Are AI and algorithms the problem?
Do mortgage algorithms perpetuate lender racial bias? 5 questions to consider.
- Does AI or algorithms eliminate the potential for human bias?
- Does algorithmic bias potential have a disparate impact on minority borrowers?
- Can AI exacerbate bias potential, reinforcing biased credit?
- Is there an assumption that technology is bias-free?
- Can the reliance on tech potentially hides discrimination in lending, making it harder to find?
Let’s be professional
Regardless of how it plays out, if we aren’t part of the conversation, we’ll be in reaction mode. Creating more work for all of us. Speak up in positive productive ways to tell the truth/facts. Take an honest but thoughtful look at the data. If our industry needs correction, it should be us doing it not external players. Let’s be professional.
