Appraisals are psychology

[sputnam]

The current push for objectivity in appraisals is missing a part of reality. As appraisers, our job is to gather and analyze available data in order to form an opinion about the value, usually market value, of the the subject property. Value is what sellers will accept and buyers will offer in exchange for the property. It's psychology. The piece that is missing in the sometimes absurd push for objectivity is... Buyers and Sellers are not 100% rational when making Buy/Sell decisions. It's also emotional. The market isn't some thing that's independent. It's the collective actions of people. Every good appraiser knows that... if just intuitively.

 

[J Grant]

The current push for objectivity in appraisals is missing a part of reality. As appraisers, our job is to gather and analyze available data in order to form an opinion about the value, usually market value, of the the subject property. Value is what sellers will accept and buyers will offer in exchange for the property. It's psychology. The piece that is missing in the sometimes absurd push for objectivity is... Buyers and Sellers are not 100% rational when making Buy/Sell decisions. It's also emotional. The market isn't some thing that's independent. It's the collective actions of people. Every good appraiser knows that... if just intuitively.

Agree.
But the push for objectivity had another objective - automating or outsourcing appraisals . The psychology, or understanding buyer/seller motivations and how much it influences prices, and sometimes an entire market is central to appraisal. But by saying it is subjective and therefore suspect, they paved to rid appraisals of it.

The AMC time crush and price crush also favored appraisers, leaving out teh crucial time it takes to analyze and understand markets and instead, data dump, use computer programs to spit out result and sign. GREED GREED GREED.

I'll go one further, that by not understanding enough to identify the buyer, or why they pay or choose X, the appraisal itself is not credibly developed .

 

[CGinMN]

Appraisers utilize subjective reasoning for decisions that are immeasurable. For example, how much did the seller motivations of Comp 1 impact its transaction price, and therefore how much weight should it be given in the reconciliation?

Appraisers also utilize subjective reasoning when making data driven decisions. For example, should depreciated cost, paired data, or regression be the basis of this adjustment, or should all methods should be used and reconciled?

In either cause, subjective reasoning is not inherently unfair or leaning to a specific desired or implicit result. There are subjective processes when developing an appraisal, and that is not the same thing as bias. Further, no big data quantitative analyses can account for the subjective processes in appraisal. This is the case for AVMs and it is also the case for appraisal gap/appraisal bias studies.

Data scientists should know these limitations. Can they build an AVM which gets you close enough to make a decision without an appraisal? Yes. Can they use big data to determine whether prohibited bias exists? No, because big data can’t measure the subjective parts of appraisal.

We also have a language problem. The closest opposite of bias is objective. The closest opposite of subjective is objective. Bias and subjectivity are not the same thing, and the ambiguity can be exploited to cast unfair doubt on appraisals. Just because our process isn’t 100% objective doesn’t mean it’s biased. It can also be said that a more objective process isn’t always stronger. AVMs might be more objective than appraisal, but only because they ignore subjective processing.

 

[Tom D]

Why do we keep going to the cemetery and looking at our graves.
When i was a broker it was all emotional, the hot button. But way back here the 80's, in this big city, we were told what words, and/or actions not to do, or you would be in trouble.
This is called race baiters, who have found a way to make money for themselves.

 

[alebrewer]

Fully agreed. However, as the ability to quantitatively analyze data sets continues to improve, so should the reliance on subjective analyses continue to decrease. For instance - the selection of neighborhood boundaries is a subjective task - hopefully rooted in well supported opinions, but subjective nonetheless. Selection of neighborhood boundaries can reflect appraiser bias or not - depending on the appraiser's selection of said boundaries. However, what if we had the ability to quantitatively determine neighborhood boundaries by analyzing the common elements of comparison and thus eliminating different neighborhoods based on the difference in elements of comparison. Homes on the west side of Main street were constructed primarily in the 1940s through 1960's and homes on the east side of Main St were constructed primarily in the 1990's and 2000's. We would then quantitatively/objectively select Main St as a boundary.

That's just one example. To your point about which technique to use, I don't think it will be long before we'll be able to build models that tell us which technique most accurately reflects market behavior.

I'm not poo-pooing the subjective components of our analyses - I'm just stating that the natural progression is to reduce subjectivity and increase objectivity. The prior generation of appraisers (for the most part) had no idea how to quantify adjustments. Now we have a plethora of tools.

 

[ucbruin]

Sales people sell the sizzle....
Appraisers aren't in the sales end of the industry....

 

[J Grant]

Appraisers utilize subjective reasoning for decisions that are immeasurable. For example, how much did the seller motivations of Comp 1 impact its transaction price, and therefore how much weight should it be given in the reconciliation?

Appraisers also utilize subjective reasoning when making data driven decisions. For example, should depreciated cost, paired data, or regression be the basis of this adjustment, or should all methods should be used and reconciled?

In either cause, subjective reasoning is not inherently unfair or leaning to a specific desired or implicit result. There are subjective processes when developing an appraisal, and that is not the same thing as bias. Further, no big data quantitative analyses can account for the subjective processes in appraisal. This is the case for AVMs and it is also the case for appraisal gap/appraisal bias studies.

Data scientists should know these limitations. Can they build an AVM which gets you close enough to make a decision without an appraisal? Yes. Can they use big data to determine whether prohibited bias exists? No, because big data can’t measure the subjective parts of appraisal.

We also have a language problem. The closest opposite of bias is objective. The closest opposite of subjective is objective. Bias and subjectivity are not the same thing, and the ambiguity can be exploited to cast unfair doubt on appraisals. Just because our process isn’t 100% objective doesn’t mean it’s biased. It can also be said that a more objective process isn’t always stronger. AVMs might be more objective than appraisal, but only because they ignore subjective processing.

Agree. So-called "subjective" reasoning is how the human mind works. It allows us to form conclusions, judgments, and analyses. In real estate markets, buyers and sellers think subjectively, so appraisers need to think like they do, and then we go further test the results against the data, prices, and trends.

Pairing subjective with basis, or with the negative connotation of it is about personal feelings, was used to denigrate it.

True thought uses subjective reasoning and is more time-consuming than clicking and seeing a computer spit out a result. The priority of speed and efficiency is applied to a loan that will last 15-30 years and a financial decision that can impact people and markets for a very long time.

 

[J Grant]

Sales people sell the sizzle....
Appraisers aren't in the sales end of the industry....

But we need to understand why buyers bought the sizzle and how it affects prices.

Market value is what the property is worth when the sizzle is gone.

 

[CGinMN]

Fully agreed. However, as the ability to quantitatively analyze data sets continues to improve, so should the reliance on subjective analyses continue to decrease. For instance - the selection of neighborhood boundaries is a subjective task - hopefully rooted in well supported opinions, but subjective nonetheless. Selection of neighborhood boundaries can reflect appraiser bias or not - depending on the appraiser's selection of said boundaries. However, what if we had the ability to quantitatively determine neighborhood boundaries by analyzing the common elements of comparison and thus eliminating different neighborhoods based on the difference in elements of comparison. Homes on the west side of Main street were constructed primarily in the 1940s through 1960's and homes on the east side of Main St were constructed primarily in the 1990's and 2000's. We would then quantitatively/objectively select Main St as a boundary.

I understand what you are saying, and I think it makes sense to a degree. WRT to above, I assume you mean neighborhood in the sense that the GSEs do (e.g. market area). If so, what if the subject was on the west side of Main but it was constructed in 1995? The question in that scenrio becomes, will buyers consider both areas part of the same market area? Properties that are directly adjacent to each other can have different market areas. So I have no doubt AVMs will be able to automatically draw boundaries based on data, but whether that data reflects behavior of current market participants, which is constantly changing due to redevelopment, supply and demand, economic factors, etc. is another question. I know this is just an example to illustrate a point, which i do understand, but boundaries based on historical data (age of home) rather than the behavior of current market participants could actually be construed as redlining. This is how AVMs perpetuate structural discrimination.

 

[J Grant]

We don't need AI to "objectively" define neighborhood boundaries, and perhaps this question should not be asked, given the current climate of bias.

If an appraiser can not think it through enough or be area faimilar enough to define the applicable neighborhood boundaries, they are not competent for the job. All this will mean is that lazy, incompetent, or out-of-geo area appraisers ( or tech folks outsourced in India) can plug in AI and have it spit out "objective " boundaries, which may or may not be relevant.