Our education tells us a comp is similar and competitive. So how do we measure “comparability”?
If our job entails studying market data to get an answer … might it be important to know exactly how to describe a comp?
In researching the literature, including journal articles, I find little help. Only two or three articles address the issue at all. However, the Appraisal of Real Estate (14th. ed.) does provide the following:
- A good comparable sale is a competitive p.121
- The analysis … requires … competitive properties, such as information on comparable p.97
- The data used for comparison in the three approaches should come from properties that are similar to the property being appraised. 121
- …by comparing the subject property with similar (i.e., comparable) properties p. 45
So we know that comparable data is competitive is similar is able to be compared . . . . Are you dizzy yet?
This traditional education also intermixes the concept of similarity with the concept of adjustment. You adjust to make up for dissimilarity. In the 1930s, when appraisal was first theorized, this made sense. It appealed to the high challenge just to gather four or five “comparables”, then perhaps to explain your opinion of comparability. If it’s similar, you don’t need to adjust. If you need to adjust, it’s dissimilar. If you need to adjust “too much” it’s not a good comp. Simple. Yes or no.
Comp, or not a comp. Easy. Trust me.
So what’s the issue? Why should we care? I am a highly trained expert. I have a license. “Trust me. I know a good comp when I see one.”
Per the Dell Impossibility Theorem: “You can’t get objective output from subjective input.”
If we are to move toward an organized logical method of analysis – if we’re really charged with being “impartial, independent, and objective” – then we must start with data which is objectively selected. In today’s data science world, “trust me” is not enough.
In order to move valuation toward a science, a science of data, the first step is to recognize and admit that picking comps the old way assures the demise of the appraisal vocation, followed by the appraisal profession. The days of “trust me” are shortly over. Valueball is here.
There is a scientific basis for data selection. Evidence Based Valuation© (EVB) starts by reducing the valuation problem to its component parts. The reduction comprises The Five Dimensions of Competitive Similarity© which are separately analyzed and categorized. Adjustments are made using just three tools. In the data science approach to valuation, we call this initial data frame the Competitive Market Segment© (CMS). Once this ideal CMS© is delineated, an objective and more scientific valuation process is enabled. The data science approach also enables far greater service and results for client uses. Uses of value. Uses that can be monetized. For appraisers.
All this requires a new willingness and readiness to face the future now.
Are you willing?
Tom Terwilliger
March 21, 2018 @ 6:54 am
I’m am willing. I can’t wait until a class comes close enough to Iowa and we can get credit for the education. I loved Practical Regression Using Excel, which I know you helped write. I can’t wait to build on that foundation.
Steven DavisMRICS
March 21, 2018 @ 9:45 pm
non comp is meant us
Jeff Luzier
March 22, 2018 @ 6:04 am
Same here in PA, just waiting for a class not so far away. Ideally with CE credit.
Britt West
March 24, 2018 @ 5:15 pm
Ok George…synopsis of new class? CE credits
You got the best info and analysis out there.
Thanks
Mark Goodman
April 12, 2018 @ 1:35 pm
I came for the editorial, however this appears to end as an Advertisement.
The ‘old way’ of choosing comparable sales still requires logical analysis which I contend, is not equaled by technology. I work in one of the most technology advanced markets in the nation and have doubt that any current technology will choose correctly for a number of reasons, largely for differences which are not acknowledged or uniformly reported in MLS (quality, location factors, sale and marketability factors, obsolescence, physical or cosmetic condition). Many factors are not quantified, and factors noted in text are often exaggerated or omitted. Even Google Maps and similar products lack boundaries and have incorrectly named areas. These are factors which can make a big difference and should be considered: How are they gathered, graded and included? Are they included at all?
The measure for comparability is still whether a buyer for a presumed subject would accept a ‘similar’ property sale as a reasonable alternative. Differences in price are still determined by Buyers and Sellers’ Participation in each respective market segment. I am not saying that other methods, including regression are not useful; just that they are more accurate with uniform, similar and abundant data. Such uniformity is not observed in most market segments, and no ‘abundance’ for long-term undersupplied markets.
That said, I am sure that this technology works better, maybe even well in highly uniform areas. Given that I am responsible for any results I report, I am still obligated to analyze. As algorithms are either not reported, unintelligible or most likely secret, I cannot know for sure what is included or excluded in the process. Regression is not Market Approach Analysis. Big Data solutions work well with an abundance of data, and are observed to not work well with a scarcity of data, such as in undersupplied, and especially significantly undersupplied market segments. The science of data has been around for a long time. Where the science of data applies to appraisal, that is AVM and zillow. Collateral Underwriter is already taking away the low LTV and easy, tract appraisals by applying big data that they gather and control, from Our reported data. They forego the need for an appraisal for Market Value. That is the GSA’s prerogative.
Despite any of this, when You say that “picking comps the old way assures the demise of the appraisal vocation followed by the appraisal profession”, You are wrong. Choosing comparables still requires analysis of specific data from sales, direct comparison satisfactory to the Definition of Market Value. Any factor which added or detracted from marketability and mattered to the buyer should be considered, and if an appraisers unseen, automatic technology tool gives you ‘garbage in-garbage out for information or misses something, it is Appraisers responsibility. The only way Your quote is true is if the person choosing sales did not know enough about what they do.
If an appraiser really cared, one would Explain how comparables are chosen along with whether the data sampling was adequate in volume, and explain comparable’ difference in narrative. With this is Evidence. There is no ‘Trust Me’. If Appraiser is relying on technology, and especially solely on technology to perform a logical function as important as selection of comparables, it should be so reported. Further, if all Appraiser is doing is relying on technology, appraiser should be replaced by technology.
George, do You have a stake in EBV/CMS (C) ?
George Dell
April 15, 2018 @ 7:41 pm
Thank you all for your comments!
Mark, I appreciate the thought-out comments. You raise several of the critical issues about the contrast between human judgment and technology. My entire life as an appraiser and researcher has formed and evolved around this very topic.
The overall answer I favor, is that the best solution is the combination of each: appraiser savvy, and computer algorithms, modeled by the appraiser.
The primal human brain excels at generalization from anecdotal information. Once you have seen a friend eaten by a tiger, you avoid tigers, and tell your friends about the danger(s) of tigers. One friend remembers a story by another friend, (a credible friend) whose father got eaten by a tiger. You may even figure out a clever way to lure an enemy near a hungry tiger. If that works, you have evidence from three comparable cases of tiger danger. Then you adjust your behavior for times and locations and sizes and effective age of known nearby tigers. You may ask: “How can I keep tigers away at night when I am asleep?”
The human brain was not designed to handle more than a handful of cases. Why bother when you already have 3 to 5 cases? Easy to find. Easy to remember. Easy to understand. Easy to adjust for.
The human brain can implicitly handle multiple variables, and intuitively generalize them into life-saving decisions. Brain research reveals that humans may be an order of magnitude better at generalization than other species. It is a brain part, the neocortex. Also referred to as the association cortex, is specialized in learning the regularities in the environment – finding patterns with any two or more data.
Computers are terrible at generalization. Even pattern-matching algorithms find no patterns until there are at least 6 or 8 cases. And they are picky. And computers handle data well. They excel at manipulating data. They are good at calculating descriptives of data (like mean, median, variation, distribution). They are really quick and really good at following rules: algorithms. (Algorithms used by humans are called “models”). Computers are also really good at arranging data so that it is seen by the human. Visual methods include tables (like a grid), plots, maps, and interactive movement. Again, the human’s ability to see things instantly is accentuated.
The ability to scare the tiger at night using modern technology (sustainable fire) – makes humans even “smarter” and survive more. Another technology, the spear, adds to human ability to make good survival decisions.
But the nighttime fire alone will not protect you. The spear alone will not offend the tiger. The best survival uses both the technology, and the human ability to use the technology, at the right time and at the right place. Modeling.
The best valuation solution involves technology. The best observers and modelers are skilled humans. They must know how to throw a spear, how to keep a fire going all night, and tell the new truth convincingly to friends.
This is the essence of the data scientific method. It is called “Evidence Based Valuation”©.
©Please see next week’s blog (April 24, 2018) for explanation of why the copyright© reserves for re-use only under certain conditions. (It is primarily in keeping with the spirit of the open-source movement).
Best Appraisal Blogs of the Year (So Far) – DataMaster for Appraisers
March 28, 2023 @ 10:40 am
[…] What is a Comp? from George Dell […]