, 0

Why is modernizing necessary?

This Modernizing Appraisal series is about the analysis process, not the reporting revisions –(New URAR, UAD3.6).  We step through:  scope, data, and ”adjustments.”  It sets today’s path to competence, using data science with artificial intelligence.                                           Read the earlier parts of this series here.

In this blog series, we see the past reality, today’s endemic “appraisal” problems, and what changes might truly help society.  This includes proper prompting when using artificial intelligence in your appraisal practice.

This part three explores the valuation challenge in fundamental ways.  This includes noting some unique elements of real property valuation, ongoing legacy practices, and the evolution to modern data and analytics technologies.

The econometric view of the problem is fundamental.  Econometrics comprises math, statistics, and economic theory.  For asset economics, the valuation challenge is threefold: 1) mathematically, there are more “unknowns” than there are inputs; 2) every property sale is unique in some way; and 3) the problem itself is multidimensional.

It is multidimensional in terms of the property type; contract/motivational aspects; time; location, and the several intertwined features such as size, quality, age, and income potential.  (In the Evidence Based Valuation – EBV© approach this is called the “five dimensions of similarity.”)  And each of these dimensions has different data/statistical qualities, with differing mathematical solutions.

Another element is the explicit uncertainty around so much of the input (model or algorithm).  The uncertainty may arrive in the original problem description, including assumptions.  Uncertainty can also be variation in the data source, or even its completeness and missingness.

While often understated, valuation is and must be market analysis.  This forms a basis to compare the current legacy “approaches” to what is possible given today’s process technology.

Traditional appraisal attempts to represent (sample) “the market” in a very simple manner:  through “careful” selection of comparable sales, then “adjusting” the “elements of comparison.”  Note that no random sample is drawn, nor claimed to be drawn.  This disqualifies the entire body of inferential statistics, which is based on equal-probability sampling.  (Not judgment nor convenience sampling.)

Adjustments theoretically should be “supported” by “paired comparison” or other indirect methods.  These are questionable, for the same reasons as above:  More unknowns than data points, and concomitant uncertainties.

Analytic bias is inherent from the subjective (although “careful”) selection of comps.  Please note that analytic bias is different from any potential personal bias — a different topic entirely.

Modern methods are built on the basis of the science of data.  Data science is different from statistics in that it explicitly recognizes the concurrent need for a field related expert – in our case, the appraiser.

Modernization makes it possible to improve the appraisal product in several ways:

  • Where data is available, it is possible to optimize for better sureness of the result.
  • More comprehensive models can be instructed, as computation algorithms are instantaneous.
  • Expert analyst and client-user interface is readily done, visually in graphs, maps, and table. This includes interactive analysis/brain connection.  It now becomes possible for dynamic-decision visualization, beyond just dynamic data entry.

The coming parts of this blog will wrap up the general scenario of valuation, and step through each of the four fundamental parts of the process:  Formulation, Data reduction, prediction, communication.

This is about modernizing the analysis, not the report:  URAR, form, narrative, or interactive.