How Many Have Figured Out

[George Hatch]

“Yours is the superior intellect”
Joachim

That dude was triggered when he said that.

 

[RCA]

Nope nobody has figured it out yet. I'll give it another week or two. Can't be that difficult. Well, I guess it is.

Another Hint: You will be able to value a house without comps, - if you know the subject market area very well and can therefore rank the subject percent better/worse for the area.

A function will then tell you the value of the home. But to fill out the URAR, to support your value with comps, you will need some comps and adjustments - and it will give you all adjustments; although only a total adjustment for intangibles [and that is all you need to arrive at an opinion of value]. So, you will need to break down the total adjustment for intangibles and explain. But this method will give you adjustments for any comps, no matter how dissimilar they are.

Will this replace the appraiser? No. A computer cannot view a home inside and out, take in the hundreds of issues and rank it for a complex neighborhood. People are much better that, especially if experienced.

What impact will this have on hybrids? They will be less of an advantage. Critical to this method is the ability to rank the subject with respect to the criteria laid out by the regression model. Appraisers doing hybrids can do that, but they will be less accurate and incur more risk - wrt to the subject. The other question is how well they know the market area, if they can't see the subject and don't know the market area from recent experience, then they are in a pretty weak position.

What impact will this have on the appraiser?
1. Can more easily deal with complex market areas.
2. Can more easily deal with subject properties without good comps.
3. Will not be guessing at the value of the comps intangible features, such as condition, quality and view - it will be a given.
4. Thus faster turn-around and better accuracy for complex properties.
5. Little or no impact on tract homes with good comps.
6. Will need to understand the process.

 

[Skeletor]

Your ability to split hairs is truly masterful.

 

[DWiley]

Your ability to split hairs is truly masterful.

I am trying to understand the point.

Are we being asked to guess/surmise the recognized methods and techniques the OP is using? I have no interest in such a quest/game

Or, is the OP purporting to have developed a new methodology? Again, I have little interest until it is peer reviewed and becomes a recognized method -because USPAP does not allow use until that happens.

 

[Meandering]

statistical analysis is a recognized method of developing an opinion of value.

https://www.appraisalinstitute.org/assets/1/7/N300GR_TOC-OVER.pdf

 

[Meandering]

A Guide to Appraisal Valuation Modeling
A Guide to Appraisal Valuation Modeling will introduce skeptics to the mathematical modeling of market behavior. The handbook provides historical perspectives, statistical fundamentals, support for assertions of causality in appraisal reports, and the basics of regression analysis and model construction. Many of these topics are brought together in a case study on the valuation of lots in an actual residential subdivision.

 

[Michigan CG]

I am trying to understand the point.

He lost me at 300 comparable sales.

 

[RCA]

It could be regarded as an extension to multivariate non-linear regression.

We do the regression on the tangible features such as GLA, Lot Size, room counts, stories, and Y/N values on whether or not it has features like certain heating types, certain kinds of views (without regard to quality), certain kinds of roofs and so on. We leave out those things that require subjective judgment, such as the quality of the view, the condition, quality of construction, aesthetics, etc.. We then predict the value of the home based on these tangilble features. It will fall short of the sale price by an amount that reflects the value of the intangible features we did not model for. The actual sale price minus our estimate is a measure of the total contribution to the value of the house of those intangible features. This is how we get a FIRM (relatively speaking) value estimation for the subjective intangible features.

Note that minor tangible features may be left out of the primary regression model, simply because they are minor and provide no significant contribution to value on their own. However, a large number of such minor features can become a major contributor to value - and thus must be thrown into the pool of "intangibles" that are not measured in the first regression. A kitchen with high quality granite counter tops, the best sinks, two Sub-Zero refrigerators, a $25,000 oven from France, and so on, might be treated in its entirety as an intangible. ... And it is precisely this sort of thing that requires a certain degree of finesse by the appraiser in scoring the subject property.

The sale price is in effect a "grade" for the subjective quality/appeal of a home.

What I am saying is that it is a matter of time before services, perhaps from CoreLogic, will provide up-to-date quality/appeal grades for home sales shortly after they occur. For example, you could go out to a site such as Realist and provided you have paid for the service, see an extra field for properties called something like CQA (Condition/Quality/Appeal) that would range say from 0.0-10.0 or possibly 0-100%. AS SOON AS A HOME SELLS, it can be given this score. BUT, you will also get a detailed regression model for the subject market area, a definition of the subject market area, and a breakdown of the features contributing to the value of the sale. The sum of the value contributions will of course add up to the sale price. From this, you will be able to use the same model for the subject and quickly calculate all adjustments. In fact, you can calculate a value for the subject without even having to look at comps.

But central to your valuation is YOUR score for the subject. If that is off, your value is off. However, note, your inaccuracy is limited to the percent of the total value of the intangible features, as the tangible features are already valued by the model. So if the model has an R2 of 0.80, your score for the subject only accounts for 20% of the value. Thus a 10% error in your scoring is effectively only a 2% error.

Your score for the subject is in relation to the breadth of the properties in the subject market area. If you don't have a good sense of the character of the market area, from worst to best homes is, or of the subject, then you will not do a good job of valuation.

If someone wants a value of +/-5%, they will need to have the subject inspected inside and out by an appraiser experienced in the neighborhood.

If they need a value of +/-10% to +/-15%, they can probably get by with a hybrid.

If they need a value +/-20% they can probably get by with an AVM, - and that is assuming that the home is not at the extremes of quality, low or high. AVMs are of course good at estimating the value of average homes, average that is to say, in relation to recent sales. If you are in a market area where most homes are upgraded before sale, those AVM prices can be far different than the value of a home being refinanced.

And, I'm talking about complex market areas, not tract homes, not relatively homogeneous subdivisions. For example, the SF Bay Area is a network of mostly complex market areas.

 

[J Grant]

What you present is a way to model prices but not necessarily values, or should I say not value in the way that is accepted for appraising. You should try promoting the idea to non appraisers, I personally don't see it as viable though. There are a plethora of AVM's and instant valuations including RPR from our own MLS service that work fairly well , or not well, depending... any of these programs are subject to the same issues.

Appraisal as a methodology does not need to be "fixed". Its a time tested method of establishing a market value model. The issues around training appraisers , client ordering and selection of appraisers overly influenced by turn time and fees do need to be fixed. THAT is what hurts appraising, not the methodology. Everyone wants to make $ with ways to make appraising "more efficient" or to "support adjustments", but nobody has a way to fix the real problems at hand.

 

[RCA]

What you present is a way to model prices but not necessarily values, or should I say not value in the way that is accepted for appraising. You should try promoting the idea to non appraisers, I personally don't see it as viable though. There are a plethora of AVM's and instant valuations including RPR from our own MLS service that work fairly well , or not well, depending... any of these programs are subject to the same issues.

Appraisal as a methodology does not need to be "fixed". Its a time tested method of establishing a market value model. The issues around training appraisers , client ordering and selection of appraisers overly influenced by turn time and fees do need to be fixed. THAT is what hurts appraising, not the methodology. Everyone wants to make $ with ways to make appraising "more efficient" or to "support adjustments", but nobody has a way to fix the real problems at hand.

I do believe that you are in a different market area. In California, in the higher income areas, appraisers are really thrown into situations where they are flying by the seat of their pants, dancing around hundreds of thousands of dollars in adjustments without support, except to say something like, "... based on historical records." If ALL of your uncertainty is reduced to a single score - the appraiser's scoring of the subject, that is actually very nice. That really simplifies things.

Why are my market areas different than yours? People here, average people, earn 6 figure salaries. a couple earning over $200K/year is fairly common. They often have inherited money and/or gift money from parents, often from overseas so that gifts from parents are totally tax free. If they want, they often do, they can spend wads of money on upgrades to their homes. Some homes have, some don't have upgrades. It is not always that evident from the outside. Houses up for sale in affluent neighborhoods are few and far between. Supply/demand. In this market it is often difficult to understand why homes sell for what they do. It may be nothing more than an additional 20 minutes commute each way to San Francisco. It may be the temperature gradient from being next to the ocean as opposed to being 10 minutes inland. It's nearness to hospitals, being in the right school district. On and on. Like I said in a previous post, in our MLS, if you expand the lookups, you have about 5000 features to consider in the MLS database.