How Many Have Figured Out

[Randolph Kinney]

 

[Elliott]

The behavior of buyers and sellers do not understand that their actions are being evaluated using a complex series of equations. They don't communicate with one another except on a gross sales price basis. They make decisions based on their knowledge of the market, their finances, the available inventory, their particular preferences, and their interpretation of the future. I seriously doubt if interviewed they would agree with the results massaged by an appraiser to meet a 0.95 R2.

Using the most sophisticated mathmatics, the aggregate of millions of dollars of polling, Hillary Clinton is now our President.

 

[George Hatch]

I have to keep repeating, that what I am concerned with are the difficult appraisals.

But let's just suppose what I am really concerned with is perfection. Appraisal for the sake of appraisal. Appraisal like pure mathematics, without regard to sales. Just as a discipline in itself to work towards a future that doesn't exist yet.

The lenders are a different story. You all should know that. They don't want just the cheapest appraisal they can get that satisfies regulatory requirements. They want a rubber stamp that does that.

So, "pure appraisal" like "pure mathematics" is without concern for application in the real world, as in, making money by selling a service that someone wants.

But studying "pure appraisal" helps you develop a better understanding of the issues involved, rather than just pushing them to the side.

The lack of understanding keeps building up, the markets become more complex at an accelerated pace with the advance in technology. No one is doing much about this. The Tower of Babel eventually collapses on itself. What does that mean: An appraiser is brought into court or before some board or before executives or wherever to explain just how it was he came to his value conclusion. And "they" really want to understand the logic. He explains. They drill him on the where and why and In the end, all he can do is say, "based on my experience". The whatever board is not satisfied and they go out and try to find someone who can prove it's all BS, only they can't find anyone. Everyone goes home with "based on my experience" ringing in their ears.

"Meaningful" is an interesting term that is distinct from truthful, relevant, logical and supported. You need to consider all issues here. If your audience lack competence, you can probably deceive them into thinking your arguments are meaningful, even if they are built on sand.

We are not going to get at the truth here any time soon. Looks like a slow painful process.

Read that book: https://www-bcf.usc.edu/~gareth/ISL/ISLR Seventh Printing.pdf. Section 8.2.3 and 8.3.1 are interesting and have to do with building a regression model, getting the residuals and then analyzing the residuals further. This is the BEST book by far I have come across on modern statistics for appraisers. It gets into all the latest techniques used by companies like Salford Systems (Minitab), bagging, random forests, boosting, ..... If there is one book to try to understand, this is your best bet. AND it has a lot of R examples/scripts using the good old Boston Housing Data Set.

The paranoia about supporting adjustments is grossly overblown, IMO. Nobody is going to get yanked by a state board because they used $80/sf for the GLA adjustment vs $92.77/sf. In fact, when I review appraisal reports I don't even look at the charts and graphs and I don't fret the adjustments or how they were developed unless something looks obviously unreasonable or inconsistent. Most of the valuation occurs when the appraiser picks their comps, and what amounts to "most similar" can be readily measured on the objective basis. If there's a lie or a gross incompetency occurring in the analysis, that's where it's going to occur, not in the adjustments to those comparables. All the adjustments do after that is to refine the range of value indicators in which the appraiser was going to rank the subject. Buyers and sellers do their thing without using any adjustments, and they literally bet the farm when they act on those decisions.

If you started out with an unadjusted range with a 22% spread and you ended up with an adjusted range under 10% that fits within the original unadjusted range then I genuinely don't care how you did it. Do. Not. Care. I'm looking for the reasonable result within the context of the intended use, not someone's unarticulated vision of the perfect result. That's because the appraisal standard a reviewer is supposed to be operating on is *reasonable*, not perfect.

And that includes the state boards. If the typical buyers and sellers in a market segment are operating on the qualitative-only basis (and almost all of them do in the SFR markets), then there is no logic involved to argue that approach as being an unreasonable or not accepted appraisal technique when an appraiser does it with a less-than-optimally adjusted set of value indicators in a Sales Comparison.

IRL, none of the valuation models I mentioned: all-qualitative, the legacy "adjustment list", sensitivity analysis, RA or hueretics or Johnny Mnemonic - no valuation model will work unless the appraiser actually works them. THAT'S the problem that the lenders have been complaining about - appraisers putting zero thought or effort into what they're doing. The quants and the Ivy League B-school grads have stepped in and sold the residential lenders on the idea that the EASY button widget is the only solution when the widget was never the only solution, and sometimes isn't even a functionally viable solution.

Now as to your other point - sure, progress is good and somebody has to work toward that in order for it to happen. But as an appraiser, the applications I use work for me - I do not work for them. Meaning, I'll let the people who enjoy being the early adopters struggle with the various processes to see which will and won't get adopted by the users before I invest any of my own time in learning them.

 

[George Hatch]

BTW, every SFR appraiser should take note of the point that the logical conclusion of the ongoing quest to build the better EASY button is to reduce the remaining economic life of the bottom 60% of their group. Any button you push can just as easily be pushed by the $13.97/hr drone in cubicle 247-b at the Corelogic Matrix.

You want the process to work for you - not the other way around.

 

[RCA]

The paranoia about supporting adjustments is grossly overblown, IMO. Nobody is going to get yanked by a state board because they used $80/sf for the GLA adjustment vs $92.77/sf. In fact, when I review appraisal reports I don't even look at the charts and graphs and I don't fret the adjustments or how they were developed unless something looks obviously unreasonable or inconsistent. Most of the valuation occurs when the appraiser picks their comps, and what amounts to "most similar" can be readily measured on the objective basis. If there's a lie or a gross incompetency occurring in the analysis, that's where it's going to occur, not in the adjustments to those comparables. All the adjustments do after that is to refine the range of value indicators in which the appraiser was going to rank the subject. Buyers and sellers do their thing without using any adjustments, and they literally bet the farm when they act on those decisions.

If you started out with an unadjusted range with a 22% spread and you ended up with an adjusted range under 10% that fits within the original unadjusted range then I genuinely don't care how you did it. Do. Not. Care. I'm looking for the reasonable result within the context of the intended use, not someone's unarticulated vision of the perfect result. That's because the appraisal standard a reviewer is supposed to be operating on is *reasonable*, not perfect.

And that includes the state boards. If the typical buyers and sellers in a market segment are operating on the qualitative-only basis (and almost all of them do in the SFR markets), then there is no logic involved to argue that approach as being an unreasonable or not accepted appraisal technique when an appraiser does it with a less-than-optimally adjusted set of value indicators in a Sales Comparison.

IRL, none of the valuation models I mentioned: all-qualitative, the legacy "adjustment list", sensitivity analysis, RA or hueretics or Johnny Mnemonic - no valuation model will work unless the appraiser actually works them. THAT'S the problem that the lenders have been complaining about - appraisers putting zero thought or effort into what they're doing. The quants and the Ivy League B-school grads have stepped in and sold the residential lenders on the idea that the EASY button widget is the only solution when the widget was never the only solution, and sometimes isn't even a functionally viable solution.

Now as to your other point - sure, progress is good and somebody has to work toward that in order for it to happen. But as an appraiser, the applications I use work for me - I do not work for them. Meaning, I'll let the people who enjoy being the early adopters struggle with the various processes to see which will and won't get adopted by the users before I invest any of my own time in learning them.

But the real problem is NOT adjusting things like GLA. They are pretty easy to get at with regression. Not that you won't see BIG differences on these adjustments between appraisers; you may very well if one appraiser measures by the ANSI standard and the other is mostly concerned about creating the illusion that the GLA is what is in the assessor's database (because he/she is absolutely unsure that of their own measurements.) -- These are what I would call objective variables which are invariably tangible as well. The difficult ones are the ones that require subjective judgement, such as condition (and while we may both walk into a house and agree it is C4, we would probably differ on the adjustments), quality, style, aesthetics, appeal, functional utility and so on. Some intangibles are partially tangible, because their value comes from their aggregation and configuration in say a kitchen or bathroom. The latter becoming more of an issue with high end homes.

 

[RCA]

View attachment 37171

Lets see if you do this by hand (calculator), how long will that take?

Most use R-Language which is free. See www.georgedell.com. Bruce Hahn on this forum, pushes for using R as well. And this book I mentioned has a number of examples using the Boston Housing Data Set. I think this book is THE book for appraisers. George Dell has written a few articles in the AI journal, including one last year. They are kind of philosophical and not of much practical use. Bruce does some nice things with graphics. What I have seen from George Dell and others is mostly simple linear or possibly curvilinear regression - the sort of the thing that may be useful for describing trends in pricing, but doesn't work very well in the SF Bay Area for extracting good adjustments with high R2 values.


A lot of appraisers don't like what they see coming down the road. But that is where we are headed. Is there money in this? Well, maybe not. Lenders give money to people to buy houses. They buy them. If they don't pay back the loan, the lenders sell the homes and go after the owners for the difference. I'm not sure that works very well over the long term; you know, with baby boomers getting old, retiring and all. I guess we will see. Reverse mortgages? Interesting. Now they want two appraisals. What's coming down the Road?

 

[Randolph Kinney]

I think this book is THE book for appraisers.

This book isn't something appraisers can use. Unless you comprehend the math used in it and how to work it, you might as well use Zillow's Zestimate.

Why isn't CompCruncher more acceptable and more in use today?

Published on Mar 18, 2011

 

[Randolph Kinney]

What I have seen from George Dell and others is mostly simple linear or possibly curvilinear regression - the sort of the thing that may be useful for describing trends in pricing, but doesn't work very well in the SF Bay Area for extracting good adjustments with high R2 values.

How similar is the Bay Area to the rest of the US? You have a method for the Bay Area which isn't needed in the vast majority of areas across the US.

 

[Elliott]

It makes more sense to hold a few variables fixed, smaller value adjustments, and develop a rational basis for the remainder of the adjustments, trying to discern what the market is doing. It is probably how most appraisers do a report so it is a reasonable, somewhat conservative analysis, which is appropriater so that it is acceptable to lenders. I won't deny a non-meat-and-potatoes appraiser from using esoteric models to influence the next level of appraiser users, but its not what most appraisers do. And that doesn't mean there is a better way to do appraising because it is not intuitive. If your not doing more than 150 reports a year, you may not be a residential appraiser. : )

 

[RCA]

How similar is the Bay Area to the rest of the US? You have a method for the Bay Area which isn't needed in the vast majority of areas across the US.

Well, it depends on what you mean by similar. I am looking right now at the San Diego MLS in the La Jolla Alta North. You just don't find that here. It's different. But in terms of difficulty appraising, it could certainly be similar. But it also depends on market conditions; whether you have a lot of sales comparables or not, pressure on prices, etc., then whether your subject is in the ballpark or an outcast.

WRT to using regression, I almost always see it used with tract homes. …