$ per Square Foot Significance

[Bill_FL]

Dee Dee

That senarios happens alot in the mountain area of NC. You can have a million dollar home on top of the mountain and a single wide next door. Homeowners are quick to point out all of the nice homes around their home and expect you to use those sales as comps, when infact they are not comaprales, they are simply sales.

Austin,

I think you are correct in saying that many appraisers over adjust, or shall I say, make to many adjustments. I dont think that in most cases buyers of sf homes consider as much as some appraisers. You stated that quality was the number one issue. I agree with this, however this comes down to comaprable selection. If one selects the correct comaprables, the quality issues should not be relevant. I think this can come with only experience and knowing your market area as Dee Dee has implied. However, I think that one can over analyze also. You will never get your graph flat no matter how much you adjust, it cant be done. No two buyers and sellers have the same motivations. By way of your reasoning, I would assume you have taken a class from McKissock and his multiple linear regression software that he used to sell. He called this the purple factor. There are some things that take place in the transaction that can not be accounted for. Why would two new construction homes next door to each other and identical quality/floor plans by the same builder close at two different prices? Buyer seller motivations and you wont be able to graph/adjust this out.

 

[EDWARD BERRY]

AUSTIN,

YOUR method is very interesting, you are sold on it and I agree in some respects.

However, being an old Arkansas chicken farmer(but have done over 7,500 appraisals) it is not too clear to of the procedure. BUT I would like to try your method.

Several months ago you furnished some info but I cannot find it.

Would it be possible for you to screen dump some of process and paste it into MS Word and e-mail it?

I will try it, and give feed back.

It sounds simple enough and with the vast amount of MLS data today would be an interesting project.

Thanking you in advance-"Ed Berry"<berry@IPA.net>

 

[Farm Gal]

Bill: I quite agree with your:

There are some things that take place in the transaction that can not be accounted for. {....}Buyer seller motivations and you wont be able to graph/adjust this out.

Actually with enough data you CAN approximate adjusting individual motivation out and approximate the 'typical' buyer! Multilinear regression IF there is reasonably clean data going in will perform that very task, in a manner we cannot perform on our own through mathematical manipulation but with enough experience by gosh we do it just fine through intuition and 'experience!! Problem with MLR is it takes a good appraiser and personal knowlege of the individual market to make the software do what it is supposed to do!

If one selects the correct comparables, the quality issues should not be relevant. I think this can come with only experience and knowing your market area as Dee Dee has implied.

Well yes, but sometimes on that mountaintop, you only GET three 'fair' comparables instead of even one 'good one'. So whats a poor appraiser to do? I think the regression programs now available on your home confuser are part of the answer, but it would sure be nice if there was a bit more of a primer on how to use them.

Personally I think about 20 of us ought to pool our money and have Austin teach us a class: as near as I can tell I am not the only one who wants to KNOW a lot more about how to massage the stuff like he does, and I DON'T have the time to learn all the ropes on my own. I have tried McKissock, Appriasal Institue and reading on my own, but I want to sit in front of somebody who does this and take some notes about what they have learned about residential and how to SPOT those areas where I need to look deeper cause I am missing something! The commercial guys are starting to be sold, but usun's are lagging way behind...

I used MLR in a past life in another discipline, I don't have to come up to speed on the thery or buttons (tho its a whole bunch faster than the punch cards we used to use... :wink: .

Any one else game? (Austin stop shakeing in your boots, don't throw the locks, the disciples have arivved! Most of these folks are not beleivers (yet).

 

[Frederick R. Ruffell]

Lee Ann,
I am up for a class from Austin, I started doing some of his basic regression analysis and found I wanted more. Specifically how to set up the Excell program once so all I would have to do is grab the template and enter the data. Where are you located Austin?

 

[Pamela Crowley (Florida)]

You know I'm ready!

The Florida Keys - or Las Vegas - or ________

Trip time with a write off!!!

 

[Frederick R. Ruffell]

Vegas!!!!

 

[Fred]

Austin posted,
"The price per square foot does not always decrease as the GLA increases"

Austin,
It may sound funny but what you refer to there is price per square-foot, per square-foot. Unit price can increase, decrease or stay the same within any sample or part of a sample.

The contributory value of additional square feet will not consistently drop off until the scale of improvements reach a point (of dismissing returns) that exceeds (over-improvement) the highest and best use of the land.

 

[Bill_FL]

Lee Ann

Actually with enough data you CAN approximate adjusting individual motivation out and approximate the 'typical' buyer! Multilinear regression IF there is reasonably clean data going in will perform that very task, in a manner we cannot perform on our own through mathematical manipulation but with enough experience by gosh we do it just fine through intuition and 'experience!! Problem with MLR is it takes a good appraiser and personal knowlege of the individual market to make the software do what it is supposed to do!

Actually, my point is that the machine can not do it all. This is why appriaser judgement is such a key issue. If the machines could do it all, we wouldnt be having this discussion, they would all ready have put us out of jobs. If you have 30 sales in your MLR (30 being the least number of statistically valid sample may have) and each of those sales have the unknown of buyer seller motivation, than how can you be assured that the MLR is producing the results you want? My example before of the two indentical homes in the new subdivision that sell for $5k difference. Now multiply that by the 30 sales used in your MLR. How is that accounted for? I do not belive it can be done. I think the MLR would be flawed as the differences in sales prices it finds it would attempt to associate with something that we as appraiser would consider not a significant difference inthe properties. Another example was posted elsewhere about a reviewer taking to task an appraisr for not making an adjustment for 1200 sf difference in a lot. The lot size was around 18,000sf. I would not have made the lot size adjustment either. However, most of the MLR's I have seen would have found this as the only difference and associated the difference in sales price with the lot size, when in fact it was due to negotiating skills of the buyers/sellers.

Well yes, but sometimes on that mountaintop, you only GET three 'fair' comparables instead of even one 'good one'. So whats a poor appraiser to do? I think the regression programs now available on your home confuser are part of the answer, but it would sure be nice if there was a bit more of a primer on how to use them

I empathize with you Lee Ann, I too appraise in a mountain market and know of what you speak. However, location is a starting point for these type properties. What I have found, in my market, is that it is best to start to look in the immediate area, same mountain, etc for comaprable QUALITY homes. If none are there, do not select homes that are not comaprable, rather look for homes that the buyer might have considered as equal to the subejct. This may mean going some distance, however I would rather compare apples to apples in terms of quality and make needed adjustments for locations and views. In my opinion, the comparable home that is on the next mountain, or the mountain 15 miles away, with a similar view, etc, is a better comparable than the home on the same mountain that requires significant quality adjustments. I say this because I dont think the buyer, the market participant, would consider a home of much lesser quality, just because it was on the same mountain. the same goes for the home on the other end of the spectrum.

Just my observations, though I know there are probably a million ways to approach this. I guess as long as the adustments can be supported and explained, you could approach it any way you wanted, my point was that I try to replicate the market. What homes would the buyer have considered.

I also know that some will repsond that in certain cases, the buyers wouldnt have considered another location. They must me in that one subdivision, on that one mountain. Those are few and far between, but they do exist. In those cases, I agree with the approach to use only sales from that location and adjust for what needs adjsuting. Some times those are in places such as homes inside speific ski resorts, golf communities and even some town limits.

I hope this has explained what I meant more clearly than before.

 

[Fred]

Bill,
You are right about judgement, but this statement:
"30 sales in your MLR (30 being the least number of statistically valid sample may have),"
is another one like the inverse relationship between size and price that has grown to urban legend.

The number of sales needed depends on several things including how many variables you are regressing (degrees of freedom, etc.). The appraiser's judgement goes into to which variables to use and how to interpret the output.

Keep in mind that Austin is only referring to single-variable regression.

 

[EDWARD BERRY]

The IAAO has used multi regression for years, similar to the work done by Mckissock.

Mr. Gene Dillmore has written several articles also.

As stated you need about 30 good sales and since much of his work was going in the early 1980's and lack of good sales prevented many from expanding the work.

Also, a Mr. Carbone developed a MRA system for IAAO, where by as a new sale was entered, it recalculated the formulas again. Thus the Asserrors could stay current on the assessments.(either up or down)

I obtained a copy of his work from a place in Ann Arbor that has copies of most PHD diissertations.(sic)

It was at a time that computers were VERY slow--I wish I had the knowledge to rework that project since we have so much MLS data and fast computers.

If anyone has info on any other work please post.

ed in arkansas