Over the years, I have argued that appraisers should not have to analyse the sales contract. To my mind, this should be the job of the underwriter: The appraiser’s job is to ascertain the home’s value, nothing more and nothing less. After all, appraisers are human beings, and every appraiser knows that if he produces a value that is below the contract price he is likely to take flak for it. No appraiser wants to spend days arguing back and forth because he “killed the sale.”
There is clear evidence that knowledge of the contract price influences an appraiser’s final opinion of value. A few years ago, I worked with my former colleagues Eric Rosenblatt and Hamilton Fout to produce these graphs:
As you can see, when the appraiser knows the contract price, he produces a value below that price far less often---in total, roughly 9% of the time---even though the split is close to fifty-fifty when the appraiser is blind to the contract.
The above demonstrates just how stark contract bias can be. Apparrently roughly 91% of buyers are getting a deal, if and only if the appraiser gets a chance to peak at the contract, and most sellers are getting taken for a ride---even though buyers are often selling their old homes just weeks apart from acquiring their new ones. It is hard to believe the exact same people are sharks one day and suckers the next.
While at Fannie, a number of us proposed asking the appraiser to value the home before being given access to the contract: In short, requiring the contract analysis to follow the delivery of the rest of the appraisal. Of course, this is logistically impossible; it would simply slow down the appraisal process too much. More importantly, the appraisal industry remains firm in its belief that appraisers must analyze the sales contract when forming their opinion of value even though it is perfectly valid to underwrite a loan with a pre-contract appraisal. If an analysis of the sales contract is absolutely necessary, why are pre-contract appraisals allowed? Indeed, why do the GSE’s offer appraisal waivers on sales at all? If no new appraisal was done, then it was obviously ok to skip the entire analysis of the sales contract. The policy suffers from a split-personality.
It is profoundly unlikely that appraisers will be freed from their obligation to analyze the sales contract: Despite all the arguments against it, the appraisal industry remains adamant. The industry argues that knowledge of contract concessions (Did the house come furnished?, Is the seller returning cash to the buyer at the closing table?, etc.) is necessary to properly value the home. However, the underwriter should be the one to analyze the contract, and the appraiser should focus on determining the home's value. If the underwriter has a fair and accurate value, he can observe that the property is selling above that value and judge the effect the concessions are having on the contract. It would be much easier to notice that concessions are inflating the home’s value if the appraiser, unaware of the contract, came to a value that was significantly below the contract price. As it is, many appraisers will just rubber stamp the contract price---ignoring all but the most egregious seller concessions. Indeed, at the end of the day, what does it matter if concessions are pushing the contract price or simple ignorance on the part of the buyer? The end effect on the lender is the same.
The next best option is to use an AVM to assess the validity of the appraised value. Our AVM can help protect stakeholders from gross misvaluations: The KAVM isn’t biased by knowledge of the sales contract, and is not scared of any loan officer. Once suspect values are flagged, an AMC’s reviewers can come up with their own opinion of value before looking at the sales contract. In this way, they can anticipate negative GSE feedback and help ensure that any loans their clients are holding in portfolio are valued correctly.
In a future post, I would like to discuss the relative accuracy of BPOs and traditional appraisals. What are your opinions about their relative accuracy?
Comments(4)