Evidence-Based Valuation Enforcement is the topic of the Appraisal Institute’s June 22 News Release.

Read the original AI news release here.

This recognition is immensely gratifying to me.  As many of you know, I have been an advocate of  EBV (Evidence Based Valuation)^© for some 20 years, since I was on the Appraisal Institute’s curriculum and technology committees, as well as development teams for the “advanced” education.

We are finally adapting to today’s data technology instead of sticking to “judgment-based” valuation and regulation.

My extensive graduate education in econometrics led me to a belief in this inevitable result.

It is good to be clear on the differences between legacy judgment-based valuation and modern Evidence Based Valuation, EBV^©.  EBV is the “Market Analysis” approach, where objective data, logic and empirical  market information is applied to every estimate, adjustment, and prediction (rather than personal intuition/experience, habit, or “rules of thumb”).  All information is verifiable and replicable.  EBV work is reproducible, leading to measurable estimates of risk and reliability – the true needs of valuation users.

Traditional practice relies on careful selection of a few comparables (usually 3 to 6), and adjusting features which differ from the subject property.  This applies to income comparison and cost/land comparisons, as well as direct sales comparison.  Where the “three approaches” do not align, the appraiser is to justify and explain the differences, and weight conflicting and “inbred” data and analysis.

The subjective nature of traditional data selection and adjustment practices preclude much of today’s advantages of using artificial intelligence under expert direction.

Legacy standards and coerced practices require an expert opinion of a single point value, with narrative reconciliation of differing point numbers from differing legacy “Approaches To Value.” Quantifying of risk/reliability is not even attempted — it’s not a “peers-actions,” “user-expectations” practice.

Evidence based practice relies on measured market information, using similarity parameters and similarity algorithms, each derived from the market segment itself.  The quantity of data used is also objectively, econometrically determined, based on the data available.  The “three traditional approaches” to value are incorporated in a single but comprehensive “market analysis” approach.  This integration manages the inherent inbreeding of the “three approaches.”

The data science focus is on data gathering and selection.  The process emphasizes market analysis prior to identifying of illustrative report comparables.  (The traditional process requires picking comparables first, then doing market analysis – thereby skewing judgment to the precluded comp selection.)

Data Science principles govern EBV:

• Consider and analyze complete data sets.
• Reduce to the CMS (Competitive Market Segment)^©
• Expand to indirectly competitive information, if necessary.
• Rely on visuals, particularly graphs (histograms and scatterplots).
• Always apply market-specific price indexing, as required by clients and standards
• Consider the level of competence of the intended user, and their intended uses.
• Apply the consistent, repeatable path of analysis which enables artificial intelligence

Next week’s blog will ask the question:  If HUD (and other agencies’ regulators) should evaluate bias evidence on an appraisal, shouldn’t appraisers simply provide that evidence in the first place?

More information is available at Valuemetrics.info and here at GeorgeDell.com