What adjustment is the “right” one?

The world of appraisal adjustments is problematic.  We get a wide range of promises and incantations.

Proprietary sellers of appraisal software tend to make claims like:  “Use our software to prove your adjustments!”  Education providers tend to use vague promises verbally describing “techniques” which are inherently mathematical in nature.  Some client/users would like to see “rigorous” (but impossible) methods to support adjustments.

None of these work.  Darn.

These pronouncements tend to go back to the original, outdated sources of “established” appraisal practice: the Appraisal of Real Estate (“ARE”), by the Appraisal Institute.

The methods proffered by the ARE are definitive in language, but scant in technique or real-life example.  These include:  1) Paired sales; 2) Grouped data; and 3) Secondary data.  Also:  “statistical,” “graphical,” and “scenario” analysis.  Finally, reliance on the other two “approaches”:  “Cost-related” and capitalizing income differences.

Finally, there is an admonition that “the appraiser must be careful” and that the math reflects “the reactions of market participants.”

However, we are now in the age of instantaneous, complete data, analytic computation power, and visual interface to human expertise.  Appraisers do not have to hand-operate algorithms.  Appraisers do need an in-depth competence in models of micro-economic relationships, and  human behavior.

I find appraiser education (in this respect) to be woefully out of date.  A lot of generalizations and admonitions – where clear specification and evidence are needed instead.

The appraisal “process” is simple:  Scope, Data, Prediction, Delivery.

The “scope” or hypothesis of an appraisal must interface with the client need:  Risk to loan, investment return, or justice-fairness.

Data should be complete and optimal (all relevant data, and no more).  Errors and missingness matter.

Prediction must reflect conformity (variation), including a reliability/risk score.  The analysis must consider the differences in the multidimensional nature of real estate data.  It involves adjustments of:    1) property type and rights; 2) contract/transaction; 3) indexed pricing; 4) location; and 5) personal preferences (“within-group” differences).

Delivery should include explanation of the reliability, and its possible impact on the user’s purpose. This reliability measure must consider game-theoretic aspects, as well as the “subjective” elements.

In the practice of EBV (Evidence Based Valuation)^© we find that the principles of data science (as above) eliminate or dramatically reduce the need to consider subjective adjustments.  Objective adjustments, such as for time (and most concessions), are deterministic once the data is set.

We find that traditional adjustments are barely necessary!  In fact, they may reduce the understandability and even the usefulness of valuation results.

Traditional appraisers continue being replaced by analytic substitutes for old “trust me” opinions!

The future is asset analysts armed with analytic technology, data science, and econometric theory.