Is USPAP biased? (Appraisal Foundation, Uniform Standards of Professional Appraisal Practice)
Is this possible? It can’t be!
This part 8 of a multipart look at issues in valuation standards.
Editor’s Note: This is Standards, part 3.8 of George Dell’s series on How Do I Move to EBV? Links to the earlier posts are here.
Let’s look at the relevant USPAP sections in the order they appear.
Per USPAP definitions, an appraisal is “an opinion of value.” And bias “precludes an appraiser’s impartiality, independence, or objectivity.” And every appraisal must be credible “worthy of belief.”
Our goal here is to see how USPAP rules may discourage, allow, or even encourage bias.
The Ethics Rule, Conduct section, states that bias is not allowed!
The Scope of Work Rule states that credibility is measured in the context of the client’s intended use.
It additionally says that the client communicates most of the information necessary. Although the appraiser is responsible for identification of subject characteristics. The Acceptability section states clearly that scope of work should meet the “expectations of parties [clients] who are regularly intended users for similar assignments.”
We should expect that clients will regularly desire an appraiser to meet their expectations, biased or not. (Or should we expect that users will not have any bias toward their $ goal?)
Standards Rule 1-3(v), requires analysis of “market area trends.” Unfortunately, “market area” is not defined. Market area trend is not relevant! (An area, or neighborhood will contain a wide variety of property types sizes, ages, and locations, with varying price trends! What matters is the trend of competitive property sales. Wrong data in, produces biased results out.
Legacy appraisal methods prescribe that an appraiser is to select some “comparable sale data.” Unfortunately, what is a “comp” is not described nor defined in USPAP. (The Appraisal of Real Estate defines a comp is “similar, competitive, and “able to be compared.”) In turn, these terse terms also are not defined.
“Picking comps” is a doubly-biased enterprise. The selection is not random, not complete, nor defined by the market itself. Any selection which does not include the complete competitive market segment (CMS)© is statistically biased. It does not analyze the market. It analyzes the hand-picked sales data.
The bias is analytic in nature (as taught in statistics classes). AND, it is biased by the mind-set of the appraiser at that moment. Note: the mind-set might be convenience, client expectations, unconscious desire to keep a client (and future fees) happy. Bias: analytic and personal. Darn.
USPAP does take the first step toward unbiased data selection. It suggests the use of all comparable sales data as are available and as are necessary. (Not as picked.)
USPAP is founded on believability, not reliability. Unfortunately, subjective “reviewer” believability becomes a USPAP “violation.” .
Such costs are not imposed on competing valuation enterprises. In fact, the history of regulation and quasi-governmental mega lenders has been to chip away at appraiser USPAP principles using exceptions, exclusions, hybrids, automateds, bifurcateds, “price opinions,” “evaluations”, “wavers”, and “acceptances.”
USPAP does not explicitly promote bias. Unfortunately, it enables and obscures analytic bias as well as personal intentional or unconscious bias under the cover of the goal of subjective believability-credibility rather than measurable reliability.
Tom Stowe
July 26, 2023 @ 8:15 am
Does USPAP regulate how to appraise, or does it regulate the appraiser’s behavior?
How “Is USPAP biased?” depends on that answer.
Has TAF morphed from its original purpose, due to political pressure, and is regulating how to appraise in order to regulate behavior?
Since the early 2000s, as a member of various MLS’s, I have worked with large data sets, including historic sales, pending, active, and expired or cancelled listings. The issue has always been selection of the physical area to be used for analysis. I was taught to know the subject as well as a prospective buyer would know it, then analyze the substitutes for that property, and that would determine the market area. The limitation being that real property is location specific and cannot be moved. Is this analysis biased?
George
July 26, 2023 @ 8:56 am
You are correct. The difference is that you are selecting the competitive market segment, while USPAP only requires location. Location alone does not define similarity.
John Pratt
July 28, 2023 @ 10:37 am
You use the word “Bias” a lot in your comments. What is YOUR definition of “Bais”? My definition must be different than yours. Bad data in does not necessarily result in bias, it might result in incorrect results however that doesn’t mean it is bias.
George
July 28, 2023 @ 11:21 am
John, thank you for your comment and questions.
Not sure if you are asking about human bias, (conscious or unconscious). Or analytic bias.
In stats/econometrics/data science we have the “bias-variance trade-off”. Too little information creates bias. Too much creates increased uncertainty.
I don’t have any “personal” definitions which might differ from dictionary definitions.
George
July 28, 2023 @ 11:23 am
BTW, we cover this very topic in the valuemetrics curriculum. It is an important foundational concept in the data science approach.
George
July 28, 2023 @ 12:54 pm
Oops, one more thing.
You might find my blog around the end of June, 2022, which went into more detail on the types of bias, and their inter-relationships. (Causing us to be vulnerable to such accusations.
Glen Kemp
August 1, 2023 @ 1:04 pm
Thank you, George, for taking the time to call me and direct me to where I can learn more about data science, especially as it applies to real estate appraisal. I would hazard to say that appraisal bias is the subjectivity that is inherent in “comp selection.” Since the goal is to allow the market to speak for itself, it is imperative that we avoid applying subjective judgment (no matter how experienced/astute) to the comparable selection process. Instead, as you have clearly stated, utilize all sales that meet objective criteria without regard to our perceived individual sale fitness. By so doing, we allow the market to speak for itself without subject filters that may bias the results. No more worries about sample sizes and the other various artificial data set restrictions/measures that define traditional statistics. My next stop is to acquire a reasonably well-regarded certification or degree in data science. I will keep you posted about my progress. Best Regards, GK