In this part, we look at modeling, the third “acceptable method.” What kind of a method is “modeling”? That’s pretty broad.
Editor’s Note: Read the entire series (so far) here.
A common definition of “model” is “a small representation of something.” More formally, it’s “an abstract/logical depiction to improve understanding of prediction.”
The Fannie Mae GSE Selling Guide provided five “acceptable methods” for market conditions adjustment for use by licensed appraisers. (Non-appraiser valuations are exempt, of course.)
These “methods” are:
- HPI (Housing Price Index),
- “Statistical,”
- “Modeling,”
- Paired sales, and
- “Other commonly accepted methods.”
Can an appraiser simply use “modeling” to get a good specific dollar amount time adjustment, and apply it to a comparable sale which took place say 93 days ago?
NO.
Appraisers apply specific models to specific algorithms for results (to provide opinion “support”).
It appears this “method” is not a method. It simply says an appraiser can use methods to get a result. Darn. No real help yet. Deeper we go.
Summarized, an analytic model is mathematical, using variables and factors. A modeling decision is required to decide which model to apply to an appraisal problem. Like time adjustment, or price indexing, or other adjustment.
The guideline guides . . . A model is acceptable to solve a modeling problem. So go model.
Unfortunately, this “method” is not. This solution is an assertion that an appraiser should use some method from the list — to solve the problem. Crisp and clear guidance. A model method.
Hand-picked comparable sales, as always. Carefully selected, and “able to be compared.” Acceptable.
So how do we put “modeling” to use in our context? How does it relate to the other three “acceptable methods”? Only “paired sales” falls in as a model.
The Fannie Mae selling guide requires “market-derived,” market conditions, market analysis.
HPI (Housing Price Indexes) are not “market-derived” from your relevant market segment. An aggregated HPI is not a model. It’s an average. It’s not the subject market. It’s not USPAP relevant data.
“Statistical” is not a model. Maybe some algorithm, decided by a modeling method.
“Paired sales” is a model. Paired sales has been thoroughly debunked in academic and practice literature, although it continues to be a false dogma in required appraiser licensing education.
However, there is an appraisal standards requirement that appraisers do what their peers do in similar situations.
So it appears that following an erroneous model is okay, so long as everyone follows the same erroneous model. (Legislated Groupthink?)
The next blog considers whether “Other commonly accepted methods” are acceptable. Yes.
You can’t get objective results from subjective data.
March 17, 2025 @ 7:35 am
The client, makes the rules. Repeat: the client, makes the rules. “I will pay, for this; this, is what I want” Do what you deem best, while following the rules. Get to the client, to change the rules. “Should”, has nothing to do with it. Determine your time adjustments as you believe they ought to be, and if so inspired, seek to influence the “client” to make rules supporting your correct beliefs. It’s apparent, they don’t know. Aside from that, it’s all just an academic exercise.