Could the very definition of market value hold bias?
Let’s look. First, we need to figure out what types of bias there might be. There are two major categories: 1) analytical bias; and 2) human bias. And there may be some overlap!
Analytical bias generally comes from two places:
- Error in data selection;
- Error in the predictive model.
Personal (human) bias may be of two types:
- Conscious bias, where the person is clear about attitudes and feelings. The National Center for Cultural Competence states that it may come out as physical or verbal means, or less obvious means such as exclusion. The person does not deny the prejudice and may take pride in it. In appraisal, conscious, declared bias is not likely, as it would likely be a self-revealing breach of ethical requirements.
- Unconscious, or implicit bias, is not in the person’s awareness. It may be in direct contradiction of stated beliefs and values. It is considered the most dangerous, because it can escape notice, while affect decision-making and personal relationships. Denial of bias is automatic and internally driven. In appraisal, implicit bias might enter through either data selection, or in choice of algorithm (model parameters).
Hidden human prejudice more comes from data selection than it does from choices of predictive ‘adjustments’. ‘Picking comps’ is fundamentally subjective. It also produces the greatest errors.
Thus, human bias transforms into analytic bias. This can occur in AVM code as well as in appraisals.
Human bias is a firm, preconceived notion of something, before we have full information. Preconceptions are not all bad. They are automatic responses to particular inputs. Neural shortcuts allow us to avoid paralysis when quick action is needed. Sometimes the decision for flight or fight or freeze must be immediate. There is no time to analyze what type of claws and teeth are coming. At other times, thinking, collecting data, and reasoning things out is required for good decisions.
So back to the definition of market value. First it is not really a measure of value as in “usefulness” or “benefit”. As defined in USPAP Advisory Opinion 22, market value is not value, it is exchange price. Exchange. Not benefit. Not fundamental value. Not intrinsic value. Not usefulness.
So here is the problem: “Most probable price” embeds bias, encloses it, and even may conceal it. The bias may come from buyers, sellers, sale agents, lenders, or law and regulations. It may take the form of exclusion, words, threats, intentional infliction of fear, and just plain groupthink — “that’s just the way it is”. Recall the days of “blockbusting”, “redlining”, and even deed restrictions. Today these seem disgusting. Back then they seemed normal, to be expected, and comfy – except to the people who were left out. It is a form of sanctioned or unconscious bullying by adults.
This is one outcome when the market (with all its “built-in” bias) sets the price. But what’s an alternative?” If we do not let the market set prices, then the government will. If we somehow start to “price” minority neighborhoods higher, then those who already have property there (the ‘haves’) gain more, but the propertyless ‘have-nots’ get more shut out of home ownership.
These issues can only be solved by getting to the root cause(s). Yes, minorities tend to occur in lower priced areas. But correlation does not equal causation. When two things happen together, they are usually caused by some other, third factor. We must find the real cause.
Simple solutions make us feel good (kinda like an unconscious bias) but do not solve the problem. Real solutions require addressing the past as well as the future.
Michael V. Sanders
April 21, 2021 @ 8:48 am
Scope of work (AO 22) and market value aside, the issue of bias is topical and concerning. I’d like to think that appraisers are professional enough to be above that, but events of the recent past have made me realize that we haven’t come nearly as far as I thought we had.
As far as “market value,” it is a CONCEPT that comes from neo-classical economics. It makes a number of assumptions about the perfectly competitive market, many of which are embedded as conditions in our definitions. The problem is that we have a multitude of definitions that vary in significant respects, particularly with regard to the value standard (“highest” v. “most probable” v. definitions with no qualification on price), and the normative conditions attached to the various definitions. Sometimes these conditions mirror the actual market, sometimes they don’t. I’m not a fan of the idea that a value conclusion can be different simply because of the definition that is used. Many appraisers who live mostly in the lending world are only aware of the standard market value definition from Title 12 of the Code of Federal Regulations that is part of FNMA/FHLMC appraisal forms. Those of us who live outside that world know (or should know) that there are many other definitions of market value and fair market value that are used for a variety of purposes – civil litigation, eminent domain, estate appraisal, casualty losses, marriage dissolution, etc. The existence of multiple definitions that apply to the same concept is a problem that needs to be addressed by the profession.
William M. (Bill) James MAI CCIM MBA
April 23, 2021 @ 8:49 am
George Dell’s comments are right on. If we used comparable sales or rents other than those that are the most comparable to the subject property, adjust them inconsistent with the behavior of actual buyers and sellers in that market, or analyze them different than actual buyers and sellers, we would be violating professional standards and predicting a price different than market value. Perhaps we need to make more clear that the number we deliver is not what it would be if the market participants behaved without bias, it is a number that necessarily results from whatever bias is inherent in the market participants.
Responding to Michael Sanders’ comments, yes, different intended uses of appraisals require different names and definitiions for the value appraised, but the intended resulting values for nearly all intended uses of appraisals have essentially the same attributes.