$ per Square Foot Significance

[Austin]

In my experience, the best method of learning this stuff would be online in some type of forum. The thing that makes it difficult as I have said many times before, ignoring the nay sayer-tinfoil-hat with copper grounding strap crowd, (incidentally Lee Ann; there is a forum for these people called the Watercooler), is that you have to have the capability to post graphs. Second and maybe more significantly, you have to do a lot a of math iterations. For benefit of the Watercooler Peter Pan anti-elitists groupies that means you can work circular reference problems by experimenting with different numbers until you zero in on the correct answer. This is important because I have devised a system of doing multiple regression on a marketing grid using this method, that converts complex mutliple regression into a simple regression graphical picture. Maybe Wayne could allow us to post graphical files that individuals could down load and view if that is possible.

Lee Ann raised a legitimate concern in her last posts. I have written two articles on regression and spent many hours discussing the subject on this and other forums. Let me tell you what it is like from my perspective. I offer the benefit of my research free so you can take it or leave it. I have had to listen to crap from people like slacker as evidenced from his above post. When my article was on e-mail Appraisal forum that I can’t recall the name of, some guy from California sent me an e-mail and stated that he was a graduate of the University of California with an advanced degree and had spent 26 years researching regression to do appraisals and if he couldn’t do it neither could I, therefore I was a quack. Besides he stated, it was obvious I was totally ignorant of statistics. Then he demanded to see my algorithm. I told him to go to hell.
Then a few days later he e-mails me asking some question. Then about two weeks later he e-mails me again and says: “Upon further review, maybe you know what you are talking about. Lets make peace and talk.” I told him to go to hell again. This is a type one person.
A type two person is like slacker. Here is their line: “The elitists are taking over the World with their complex methods that the rest of us can’t understand and we have got to stop them or we will be flipping hamburgers for 5 bucks an hour.” You probably know them as the anti-AVM groupies. They have no idea what an AVM is but they know it won’t work just because it won’t. There use to be some guy on this site that attack every thread on this subject but I haven’t seen him for a while. Typical line of crap like slacker posted above.
The 1st time I heard regression mentioned in a CE class was some idiot from NC, (there seems to be a plethora of them down there) told the crow story. “Seems a man went hunting and a crow flew by. His 1st shot was one foot in front of the crow and his second shot was one foot behind the crow. Statistically, he killed the crow.” The class really thought that was a hoot. I replied that there is another way of looking at his crow story: “The 1st two shots just bracketed the crow. With this information in hand, the shooter knew exactly where to aim on the 3rd shot. His shooter just never heard of statistical significance.”
I relate these experiences so that you will appreciate what I and others have to deal with when sharing this “free information.” Now you know the rest of the story.

 

[Pamela Crowley (Florida)]

Austin,

Liked the crow story. I can only imagine what you've gone through is probably similar to my trying to explain the significance of $/SF and how to use it to the Realtors.

Count me in if we can come up with a way to learn this from you. In my feeble little brain, it makes a lot sense.

 

[Steve Owen]

I haven't had time to read the latest responses, so please excuse me if this has already been covered. But, the original question was:

I've looked in some of my appraisal books and the $ per SF is not really discussed except as an adjustment which doesn't really relate to the total $ per SF.

Since the question involves total price per square foot, which is listed as price per gross living area in most of the forms we use, it seems to me like all of Austin's MRA info is a little bit off subject. Austin, are you not talking about the "Gross Living Area" adjustment on the sales comparison grid?

Austin said:

you can’t deal with size adjust or $/sf adjustments in the same way that you do with other properties.

both this statement and everything else Austin is talking about relates to the per square foot adjustment, an entirely different subject from the total $ per sq. ft., which Pam originally asked about. Correct me if I'm wrong, Pam.

I said,

FNMA wants price per gross living area bracketed.

and that seems to have resulted in a lecture from Austin, including the quote above. But total $ per Sq. Ft. is not at all complicated. It is simply the division of the sale price by the total gross living area. Please explain to me how the amount of adustment per square foot difference in GLA has anything to do with total price per gross living area?

Now, I think that MRA can offer a lot of valuable insight to an appraiser. But, I don't think it is the end all be all. We are paid to think, not just to do math. I've seen quite a few appraisals that had very complex and sophisticated math be way off because the appraiser forgot to ask this simple question. "Is the result reasonable?"

Our county assessor recently used a new AVM program with MRA and raised property values an average of about 40% in one year. The real increase was about 2%!!!! Garbage in Garbage out. After the new appraiser took office the incompetent who had been in charge of the computer quit and most of the values that had been increased beyond reason were lowered again.

If you are not usuing your head and being a real appraiser, I don't give a hoot how complex your math is.

 

[Austin]

Steve Owen wrote:

“Please explain to me how the amount of adustment per square foot difference in GLA has anything to do with total price per gross living area? “

Reply: When you finish adjusting for all factors including size difference and then graph GLA vs. adjusted $/sf, the test to determine if everything has been accounted for will show up as variance of each sale from the trend line. The trend line is total price per square foot for the data set after all adjustments have been made, one data point for each sale. You have to understand my algorithm to understand where I am coming from. There is a force at play in appraisal known as multicollinearity of variables. This means that there are value factors that cannot be separated, they must be dealt with in the aggregate. For example, bigger houses have better quality with more amenities that are of better quality and the lots are bigger with more and better site improvements. Size and quality of construction, the two most significant price influencing forces are covariant meaning you can’t separate one from the other. You can’t deal with multicollinear value factors the way they are presently handled on a marketing grid simply because they are collinear. An analogy is sugar dissolved in water. It is not water and it is not sugar. It is syrup. You can’t drink it and you can’t sprinkle it on your corn flakes. The estimated price is not equal to the cost of a quart of water plus the cost of a pound of sugar, like a Pepsi for example.
Total dollars per square foot is similar. Total dollars per square foot is a vicarious variable meaning everything contributing to price is included in that number. Breaking this $/sf number into its component parts is the whole purpose of appraising, that is to determine what is affecting price. The only possible method of doing this is with MRA because all value influencing factors are covariant to some degree.

Steve Owen wrote:

“Now, I think that MRA can offer a lot of valuable insight to an appraiser. But, I don't think it is the end all be all. We are paid to think, not just to do math.”

Reply: If appraisers are such a good thinkers, then why do they use match pairs when match pairs are not statistically significant and completely ignor normal random variance which can be plus or minus 5%? How do thinking appraisers explain the fact that something as significant as normal random variance is not even address in appraisal literature? How can thinking appraisers make adjustments when they don't even know the proper sequence of adjustments because the sequence of adjustments comes from the market and appraisal theory has no system to extract the correct sequence from the market? Why do thinking appraisers try to answer the question posed in the definition of market value “most probable price” by using three comparable sales when is it a mathematical fact that you can’t answer that question with less than 30 sales? How do thinking appraisers think they can solve multicollinear problems when these problems are not solvable using existing methods? How do thinking appraisers know when a data set is perfectly adjusted when they don’t even know the definition of a perfectly matched set? How can thinking appraisers argue about the meaning of total $/sf when they don’t know what the number represents? Name one thing in the existing appraisal theory that is an “end all be all? Existing appraisal methods in light of the above sure as hell is not an en all be all. Who said MRA was an “end all be all? Would it be reasonable to pay an appraiser to think when he/she does any or all of the above? I think not. It would appear that thinking appraisers in general have not been doing much thinking in recent history if the existing system is the best they can come up with.

 

[jtrotta]

Austin

curious, of all the people in the Appraisal Field (80,000 +/-) how many would you think use your method :?:

over the past 30 years, how many have used your (or similar) methodology :?:

perhaps, there is some theory or use of what you speak, but on occassion you have mention a fact, that 30 sales are needed - what happens in your graph when you cannot locate those numbers, under current requirements by FNMA & USPAP :?:

also, noticed that you have not mentioned anything in regards to "emotion" - how do you graph - emotion :?:

you have also mentioned in this thread, that the better quality of construction, the higher the per SF sold cost would be. I don't think that fact needs to be graphed; similarily, a lower cost house will reflect that fact in the SF sold cost. the real question that would raise my interest some, would be if you could graph the value of "AIR" - as in condo. sales
and how that affects that market.

8)

 

[Austin]

Jtrotta: What is your hourly rate for thinking? If 80,000 appraisers want to do it wrong, that is their problem. Get out your copy of USPAP and read the comments under section 1-1A then come back to the forum and tell me about what USPAP compliant appraisal methods and techniques are. FNMA cannot dictate appraisal theory. If they want guidelines that are impossible to conform to, that is their problem not mine. If you can’t locate 30 sales, then you can’t report a “most probable price.” If you are reporting a “most probable price” using less than 30 sales you are performing a fraudulent and misleading appraisal. Read standard 1-1B & 1C. Using existing appraisal methods is nothing short of a series of errors on top of errors than cannot but significantly affect the results of an appraisal in the aggregate and affect the credibility of the results.
Can you give me an appraisal reference on dealing with emotion? Can you use matched pairs to demonstrate how the buyer and seller were feeling on the day of sale? How do you make an adjustment for PMS or what the buyer’s favorite color or flower is? That emotion crap is nothing but a ruse to mask voodoo and astrological prognostications. Do you include an addendum describing your emotional state as of the date of appraisal? How do you feel about that?

 

[Tim Hicks (Texas)]

Funny, Pam, I am getting the same response from a appraisal I performed this week. Sales price-$207,500. My value-$200,000. I threw six comps on there just to prove my point. So, the listing agent has the whole problem solved with a pending sale that sold for $72.00 a foot, but is 300 SF smaller. That pending sale is the only sale (without a pool, like mine) anywhere near that price per foot. However, mine is on the smallest lot, has no lawn sprinkler system (appears to be the only one wihtout one in the addition), needs carpets cleaned or replaced, no alarm (once again the only one), standard upgrades (nothing special) and has been on the market twice as long as everything else. When I comp out the pendind sale what do I get ($200,000), but the listing agent thinks I am just impossible to work with. There is not one sale in this addition with over 40 sales in the last year that supports their $207,500.

On a side note, isn't it amazing how much they think a pool adds to the value of property, but when it does not have a pool, they try to argue that pools add no value on the comps. This same agent doesn't remember our discussion last year on why I would not give them a $30,000 adjustment for their $30,000 pool on my one comp that had no pool. They want to talk addition, but never subtraction on their properties.

 

[Terrel L. Shields]

Many credible people have written learned articles published in the likes of the Appraisal Journal only to have someone else disect the evidence and conclude that there were flaws in the way they operated the model. Typically flaws are masked by an abundance of data. The problem with linear regression is that by the time we have sufficient sales, the model is OBE (overtaken by events). The market changed, 9/11 came, etc. 30 sales would be a bare minimum. 70 is better, 1000 even more so.

Then we can go into the predictive model, with Monte Carlo simulation. Nothing is more misleading that doing MCS with insufficient data.

Linear regression in a court room setting sets the appraiser up for a great fall unless they are very well versed in same. The advantage is that no one else in the room, certainly not the judge, will have a clue what you are talking about. It is difficult to refute that you cannot understand. But if the lawyer is smart enough to hire an actuary or other statistically knowledgeable person, your goose is likely cooked. That expert can refute your testimony with B. S. and nobody will understand him either, so they will discount your testimony even if you are right.

One of the great fields I think for the future will be legal counseling. Appraisers sitting at the table with the lawyer, not as expert witness, but interpreting what an appraiser is saying and preparing questions to support or refute that testimony.

Excell and other spread sheets offer us no magic bullets, but certainly the industry would be well served if someone offered a good linear regression template for Excell for reasonable price and teach a supporting class to apply it.

Without a doubt the hard part is not finding and inputting data, but it is in determining the reliablity of that data and finding any flaws in the assumptions made.

ter

 

[Steve Owen]

curious, of all the people in the Appraisal Field (80,000 +/-) how many would you think use your method

I don't know the answer to that question. However, in my various careers (college student, marketing, community planning, appraising) I have used linear regression, multiple linear regression, number of standard deviations from the mean, chi sq tests, and many other economic statistical analyses. I have no problem with appraisers using tools; any tools that will help lead the appraiser to a conclusion that can aid the appraiser in rendering a value opinion. I do have a problem with some appraisers and non-certified people in the appraisal market trying to assert that the method is the end and the tool can replace the thinking human.

Your post asks some very interesting questions, JTrotta. Because the market is not perfect and buying decisions are made by imperfect people rather than perfect machines, the machine (AVM) will not give a perfect answer most of the time.

I am interested in what Austin might have to teach us about mathematical tools. (I am always interested in learning something new.) However, some of his assumptions appear to possibly be contrary to economic laws. His assertion that the three best sales (always, is implied) will not produce as accurate a conclusion as 30 sales flies in the face of reason; it would be true if there were 30 nearby, similar, and recent sales with which a regression model could be applied.

It might take him 30 sales to get market value, but I do it with three over and over again; the proof is that people I have appraised for a listing price tell their friends after the property sells very near appraised value and their friends call me to appraise their own property. The proof is in a restaurant that recently sold for more than my appraised value (in numeric dollars). "I knew it was worth more than that," said the seller. "Well let's see," I responded, pulling out my calculator "I appraised it for you at $x and properties on that part of z road (busy commercial street) have appreciated at about 3% per year over the last few years." While doing the calculations: "You sold it for y after holding it for three years. Hmm. X plus 3% plus 3%, plus 3%, is gee, that's within a thousand dollars of my appraised value."

Yes, I would be interested in learning more about math anytime, after all, I don't claim to be a math wiz. However, I'm not interested in learning about appraisal from anyone who has such apparent disdain for the profession.

There was nothing inherently wrong in what Slacker said in his post. (All of you people who are appraisers, go back and re-read it, please. And, shame on you for not calling Austin on it; this forum is supposed to be a place for sharing of information and opinions, not for belittling anyone who expresses an opinion contrary to your own.) The kind of personal attack that was leveled following that, well I've seen it before.... It's been my experience that when someone doesn't respond to the logic of his detractors, but simply says, "you'll have to trust me, I know more about it than you do," often has some hidden agenda. (Forgive me, Austin, if that is not the case with you, I am simply reacting to things you have said in this thread.) In this thread, Austin seems to be playing the part of a guru or evangelist. I've learned to hold on to my pocket book when faced with either.

_________

All propaganda has to be popular and has to adapt its spiritual level to the perception of the least intelligent of those towards who it intends to direct itself.

Adolf Hitler

 

[Terrel L. Shields]

"You sold it for y after holding it for three years. Hmm. X plus 3% plus 3%, plus 3%, is gee, that's within a thousand dollars of my appraised value."

Don't that just irk you? Or, likewise, you appraise a property but they sold the business and equipment, too then smugly announce you were way off or the guy is carrying 30% at 5% interest.

Recall one of those when a Realtor informed me I was way too low and he could get zzz for a place a borrower was putting up for sale and for collateral to buy another small farm. Borrower mad cause it was about to kill deal. Another appraiser came in with the cash, so deal went thru for new place. Old place (I appraised) was on market for 400+ days and sold below my appraisal, let alone the other guy's. Few months later the guy loses the property he bought and filed bankruptcy....could have prevented all that if they had simply said NO.
ter