A timely recap of Time:
The FHFA working paper 24-07 provided four strategies used by appraisers to satisfy the prior Fannie Mae and Freddie Mac Selling Guide requirement for time adjustments. The strategies comprise:
- Grouped data
- Paired sales
- Indexed similar-sales
- Avoidance
Editor’s Note: Read the entire series (so far) here.
The Appraisal Institute News Release of Dec 13, 2024 simply states the above-referenced FHFA report “Contains Multiple Flaws,” without factual/analytic support for this opinion. Thus, the AI response seems to provide no positive help towards resolving market conditions and bias issues.
An underlying comment here is that both GSEs always required a time adjustment. The requirement simply was not enforced upon lenders, and therefore also not enforced upon appraisers. Avoided.
The Fannie Mae updates to their lender Selling Guide, early in 2025, noted five “acceptable” methods to “support” time adjustments. These included two actual methods, and three broad generalizations. Please note that the generalized methods are repeated from The Appraisal of Real Estate, 15th ed.
- HPI (Housing Price Index),
- “Statistical,”
- “Modeling,”
- Paired sales, and
- “Other commonly accepted methods.”
Of the five GSE “methods” only paired sales and HPI can be considered definitive to produce a number.
Combining the above, we are left with four options: 1) grouped data; 2) paired sales; 3) Similar-sales index; and 4) HPI (compiled/aggregated Housing Price Indexes).
Of these four, we have learned:
- Grouped data (grouping time segments) discards information, and cannot ‘bracket’ the closing (current) date of the appraisal. (Simpson’s Paradox proves inherent math error in this method.) The bar-plot and histogram are visually appealing but oversimplify and misrepresent reality.
- Paired sales seldom match the timespan needed for each comparable, and seldom produce an identical or similar sale at the closing (current) date of the appraisal. This method also requires subjective reconciliation of the several non-parallel results.
- Indexed similar sales have none of the above modeling issues, and apply a repeatable algorithm with clear analyst decision points, (including trend-change identification). Universally applied in econometrics, engineering, and social sciences. It is viable … except where there is lack of data.
- Housing Price Indexes (HPI) are published by several commercial enterprises, as well as government agencies (including FHFA, and GSE). HPIs cover broader property features and locational elements, including compositional elements of personal preferences, financing, and motivation – not paralleling the features and similarities of the specific subject market segment.
HPI may be the only fallback where data is sparse and/or highly differentiated.
So, which way is the best way to calculate market conditions, time adjustments? We will answer this question in the next three posts of this Analogue Blog. This will include:
- Overview of strategies, “acceptable” methods and the recent Appraisal Institute News Release.
- Best method conclusion, and reliability strategies when there are genuine data deficiencies.
- A checklist, designed for lender compliance both for underwriter/reviews and appraisers.