Legacy appraisal practice “the valuation process” is well established.  Therein lies the problem.  It is not designed to use modern methods toward objective, unbiased results.

First, lets distinguish the different kinds of bias.  We need to separate personal bias from analytic bias.

Personal bias can be intentional, or unintentional (unconscious).  Analytic bias itself can come from two parts of the valuation model:  1) bias in the data selection; and 2) bias in the model/algorithm selection.

Here we concern ourselves only with data selection.  (However, note that data selection bias can be of the two types above:  personal and model.  Also, we look at only the traditional legacy appraisal model.  Not AVMs or other hybrids or “non-appraiser appraisals.”

This traditional appraisal model follows a simple universal valuation path:  1) identify the problem; 2) select the comparable sales (the data); 3) apply adjustments; and 4) explain/reconcile inconsistencies.

In selecting three to six comps, we are told that they should be competitive, similar, and ‘able to be compared.’  To the best of my knowledge, no ‘similarity’ algorithms are to be found anywhere in the appraisal education requirements for licensing, nor in more ‘advanced’ text.

The selection of comparable sales is based on the appraiser’s knowledge of the location, the type of property, and features of the property (such as living area, site size, floor plan, or income potential).  Traditionally, appraisers used three to six sales because:

  • Sale data was incomplete and measurements seldom uniform from one property to another.
  • Just gathering and verifying the data was a lot of work, perhaps 80% of the overall work.
  • The human brain does not do well in understanding beyond six or seven data points.

Data bias comes from two sources:  randomness of convenience — the ease of getting a sale data; and personal judgment bias of what is ‘truly’ comparable or not.

The motivation, or ‘source’ of the bias usually comes from the source of the work:  the client.  To keep a client, whether a lender, or management company (AMC) becomes the overriding motivation (conscious or unconscious).  The unconscious part itself has several sources.  The most obvious of these is called “anchoring.”  Essentially, the human brain simply comes near the first number seen. (Like the sale price.)

As it turns out, this itself is justified by current subjective appraisal standards (USPAP), which require the appraiser to meet or exceed “the expectations of parties who are regularly intended users for similar assignments.  This is followed by what I call Billie’s mom’s rule:  “But mom! Billie’s mom lets him do it!”

Much of the current debate about how to modernize appraisal goes back to pervasive, root causes.  Our expectations, habits, education, groupthink, and organizational policies and laws – all line up squarely to avoid looking at the two real issues: 1) lack of objectivity in data selection; and, 2) misuse of the very definition of ‘market’ value.  (Market price does not equal value – see “loan conditions” adjustments.)

The FHFA and other governmental regulatory agencies must look squarely at the societal burden lined up against innovation and improvement of collateral evaluation systems.  The burden of 55 state/territory/DC + Appraisal Foundation — fees and approvals inhibit progressive appraiser education.  And prohibit education of technology-optimized valuation and risk assessment methods.

For appraisers, “getting CE hours for license renewal” has become detached from getting real progressive modern analytics education, which leads to unbiased results.