GLA Adjustments

[Terrel L. Shields]

What happens to the sensitivity analysis if all (best) comps are larger or smaller than the subjects GLA?
Do you place a "not best" comp into the grid just to bracket?

Should not matter, I think

 

[Mr Rex]

Biased.

 

[RCA]

I am a little nonplussed that recognition of fraud potential is apparently exclusive to subjective opinion Nobody has EVER been able to manipulate a statistical result (heavy sarcasm included).

Then it comes down to the credibility of the subjective analyses vs. the credibility of the regression model. Done right, a judge is going to understand the logic behind, and give credence to, a subjective opinion. If the opposition has created a good regression model...they should support each other. A subjective support explanation should be easy to follow and will not make anyone's eyes glaze over tho. Maybe I'm weird but I really don't take issue with opposition providing good information so a judge can formulate an informed opinion. Never went into it with an I win/you lose attitude.

Both methodologies are predictive models. In one, sample data was observed and numerous market participants were interviewed for reaction to establish a likely range of feature value. In the other, sample data is analyzed statistically, produces a likely feature reaction result and is given a confidence score. If the adjustment is that tricky to begin with, both methods are are going to suffer in the confidence department. Personally, I think both would be best.

Consider that stats can be manipulated to produce a desired result and it is not always readily apparent how or at what stage this was done. That type of skew is a little more difficult to achieve when using simple logic because if the conclusion doesn't follow the logic it won't make much sense.

I am not a fan of, nor want to be positioned as, advocating for one methodology over another. They are tools in an appraiser's toolbox. My Daddy taught me to use the proper tool for a job...also didn't let me use a dangerous tool until he was confident I understood how to properly operate it. You may be able to explain a regression methodology in great detail on the stand. I am not so comfortable attempting that when opposing counsel is looking for ways to make me look bad and I would speculate that a good percentage of practitioners are in the same position.

I totally advocate greater exposure to and education in statistical methodologies. Very powerful tools and sometimes the best tool for a problem...but not the only one. You've got me and numerous members, I am sure, taking a closer look at these methods and perhaps considering classes. Thank you for that...haven't had too many threads discussing it for a while and it make this a better place.

When you provide statistical support, you should also be ready to provide the dataset you used, the input parameters so that a reviewer could replicate your results and check you methodology. So, no, in the context of USPAP, you couldn't get away with fraud by using statistics. However, it is difficult in court to discount the subjective opinion of an experienced appraisers, if subjective judgement is allowed. Thus, the "system" is not as willing as in the past to automatically give credit to subjective opinion - especially when non-subjective evidence to the contrary exists.

 

[Mr Rex]

Figures often beguile me, particularly when I have the arranging of them myself; in which case the remark attributed to Disraeli would often apply with justice and force: "There are three kinds of lies: lies, damned lies and statistics."
- Mark Twain

 

[Joyce Potts]

Hey joyce,

If I came to your service area and used the same regression analysis you were using, what would make your report better than mine? Your years of training and education and experience in the neighborhood, much more than me, has to be evident some where in that opinion of value. Or is the real secret of all that training and experience supposed to be found in the explanation of the same regression model, like, it wouldn't be boiler plate? Maybe we could argue over the data used to produce that model was better selected because of training and experience, but when you are using all sales, the model picks the best.

Becareful of what you advocate for, as it's easy to create software to come up with numbers and sure as heck no experience in real estate appraising is even needed to download all the sales into a program and copy and paste boilerplate.

.

Marion, are you suggesting that I'm advocating MLR for support adjustments? If anything, I'm endorsing the total opposite.

 

[Joyce Potts]

It still amazes me that you can't get a room full of appraisers to agree on the most basic concepts like support for GLA. My fear-mongering comments referred to people selling their courses who grandstand that FNMA will go after you if you don't have solid mathematical or other solid market support for every adjustment. Sorry, but most markets simply don't work like that. Erratic and inconsistent being the keywords.

 

[RCA]

Marion, are you suggesting that I'm advocating MLR for support adjustments? If anything, I'm endorsing the total opposite.

Good grief. A good appraiser, or should I say a top-notch appraiser, or should I rather say the best appraiser with complex properties and non-homogeneous neighborhoods, difficult appraisal assignments and time to do a good appraisal, will be a master at a wide range of techniques: He will be able to measure accurately (if you can't measure, give up), he will be able to draw accurate floor plans (if needed), he will have large databases of property data at his disposal, he will know non-parametric and parametric statistical methods, in particular multivariate NON-LINEAR regression, he will be a master at Excel , SQL and some programming language, he will be an investigator, in fact almost akin to a CIA field agent (I think in fact pay is about the same (not considering federal benefits), and I'm still leaving out a lot.

Of course there are those who are in environments where they can thrive on trac homes, and whose pay check is invariably based on how well they know regulations, guidelines and can do the same thing over and over again at speed.

 

[Joyce Potts]

Marion, are you suggesting that I'm advocating MLR for support adjustments? If anything, I'm endorsing the total opposite.

I'm here to tell you that

Good grief. A good appraiser, or should I say a top-notch appraiser, or should I rather say the best appraiser with complex properties and non-homogeneous neighborhoods, difficult appraisal assignments and time to do a good appraisal, will be a master at a wide range of techniques: He will be able to measure accurately (if you can't measure, give up), he will be able to draw accurate floor plans (if needed), he will have large databases of property data at his disposal, he will know non-parametric and parametric statistical methods, in particular multivariate NON-LINEAR regression, he will be a master at Excel , SQL and some programming language, he will be an investigator, in fact almost akin to a CIA field agent (I think in fact pay is about the same (not considering federal benefits), and I'm still leaving out a lot.

Of course there are those who are in environments where they can thrive on trac homes, and whose pay check is invariably based on how well they know regulations, guidelines and can do the same thing over and over again at speed.

Are you suggesting there is no one size fits all answer?

 

[RCA]

I'm here to tell you that


Are you suggesting there is no one size fits all answer?

Of course. But I calculated 10 years ago, that doing the kind of appraisals I do didn't make sense unless I could get a minimum of $700-$800 on an SFR appraisal, when I could easily switch over to a tech job like software engineering and make more money working 40 hours/week (working at a large company rather than a start-up), not having to deal with liability issues and all the other crap appraisers have to deal with). So, then, such appraisers need to do a good job of advertising just what their skills are, set a minimum fee that is about double the others and live with it. That is a tall order. They may not get any business. So, they have to have another source of income as back-up. That is a tall order again. You just have to decide what kind of an appraiser you want to be: High-volume/easy assignments, Low-Volume/difficult-assignments, set your fee accordingly and stick to your guns.

 

[Joyce Potts]

Of course. But I calculated 10 years ago, that doing the kind of appraisals I do didn't make sense unless I could get a minimum of $700-$800 on an SFR appraisal, when I could easily switch over to a tech job like software engineering and make more money working 40 hours/week (working at a large company rather than a start-up), not having to deal with liability issues and all the other crap appraisers have to deal with). So, then, such appraisers need to do a good job of advertising just what their skills are, set a minimum fee that is about double the others and live with it. That is a tall order. They may not get any business. So, they have to have another source of income as back-up. That is a tall order again. You just have to decide what kind of an appraiser you want to be: High-volume/easy assignments, Low-Volume/difficult-assignments, set your fee accordingly and stick to your guns.

Of course. But I calculated 10 years ago, that doing the kind of appraisals I do didn't make sense unless I could get a minimum of $700-$800 on an SFR appraisal, when I could easily switch over to a tech job like software engineering and make more money working 40 hours/week (working at a large company rather than a start-up), not having to deal with liability issues and all the other crap appraisers have to deal with). So, then, such appraisers need to do a good job of advertising just what their skills are, set a minimum fee that is about double the others and live with it. That is a tall order. They may not get any business. So, they have to have another source of income as back-up. That is a tall order again. You just have to decide what kind of an appraiser you want to be: High-volume/easy assignments, Low-Volume/difficult-assignments, set your fee accordingly and stick to your guns.

So what did you do the first ten (10) years?