Formula - to Adjust GLA ?

[incognito]

Just read Rattermans pg. 97-101. Read it before, but an interesting refresher. As he mentions, his method does involve backing into an adjustment figure, but it appears basically sound and well supported.

However, even Ratterman in this book speaks of a possible method for deriving adjustment factors for living area (other than the obvious paired sale analysis) is:

some appraisers use a percentage of the average dollar-price-per-square-foot of gross building area including land as a baseline adjustment. For example, many appraisers would use 30% to 40% of the average of these sale price ratios...

Sound familiar?

Additionally, Ratterman emphasizes by example the necessity of bracketing the subject by living area with the comparables, in his sensitivity analysis discussion.

So Mr. Bone, convinced? I could go on...

Thanks Metamorphic for the reference. It is a must read for real appraisers!

 

[Michigan CG]

........Jerry, for a more specific answer to your question, I would recommend Residential 101...

OK, let's be nice. You have an opportunity to enlighten, let's not chase folks away.

There are many people who use wrong approaches to adjust for GLA and that is the way they were taught. There is a large skippy mill south of me where they are taught $25/SF no matter what.

Many folks were taught to take the sold $/SF and apply a percentage, it was even taught in skippy schools.

If we were to to to USPAP and cite the "what our peers do" I would guess that the people doing the GLA extraction correctly would be in the minority.

The Ratterman book starting on page 97 details the process well and I highly recommend it to anyone who has been doing this wrong. I provided a link to the book earlier in this thread.

 

[CANative]

Isolate the feature, bracket it, then apply sensitivity analysis (play with the adjustments) until the high and low start going in the wrong direction.

Easy stuff.

 

[incognito]

Isolate the feature, bracket it, then apply sensitivity analysis (play with the adjustments) until the high and low start going in the wrong direction.

Easy stuff.

Exactly. Just back into it!


just having fun!

 

[David Beasley]

Here's a little spread sheet that does basically what Ratterman is talking about.

I've put some sample formulas in there for people that aren't really excel savvy.

That works pretty good. I setup one based on your posted image and like it. It backs up my method on the 2 cases I plugged into it after I entered/copied/pasted all the formulas to the correct cells.

I set mine up for 6 comps, and for the last 3 comps, I used the following formula to force it to return a blank cell so it didn't skew the results *if* I only had 3 comps. This way, I can use any number between 3-6 comps without skewing the MIN/MAX. Obviously, it can be blown up as far as you want it to go. I setup mine from $25 psf to $125 psf.

=IF((B5<=0),"",$C5+($B8-$B5)*F1) Where =IF((logical_test),result if true,result it false)

(And Where B5 is the GLA cell for comp 4 based on my formatting). If the GLA figure for a comp is < or = 0 (zero), then the next "" forces it to return a blank instead of a zero. This is imperative for accurate MIN/MAX calculations. If it is > zero, then it processes the cell with the indicated PPSF and the difference between the subject and that comp's GLA.

For the first real-world case I tried it on today, my normal method told me about $38 psf for the GLA adjustments ($37.68). When I applied those comps to the above spreasheet, the indication was "somewhere" between 35-40 as 40 was the lowest MIN/MAX spread. Not bad.

 

[CANative]

Exactly. Just back into it!


just having fun!

I think the reason this doesn't occur to many appraisers is because it IS so easy and SEEMS like backing in to a solution.

If the data set was huge this wouldn't work so well. But residential assignments usually rely on very limited data sets. This works well for small samplings.

 

[incognito]

For the first real-world case I tried it on today, my normal method told me about $38 psf for the GLA adjustments ($37.68). When I applied those comps to the above spreadsheet, the indication was "somewhere" between 35-40 as 40 was the lowest MIN/MAX spread. Not bad.

Great stuff David. Interestingly enough, in your example the raw $/sq.ft. average is $155/sq.ft. Your conclusion of $38/sq.ft. adjustment factor is 25% of the raw factor. Ratterman threw out 30-40% as examples, but certainly this supports his general idea that the raw $/sq.ft. factor must be significantly discounted...

NOW, I believe you could in future appraisal cases, having completed this proof, utilize 25% of whatever raw $/sq.ft. adjustment factor you may derive on your sales grid from the comparables to derive a supported $/sq.ft. adjustment factor for living area (of course for similar homes in similar market areas).

When I have time I want to play with the Excel idea. For my area I would be inclined to start much lower, say at $10 psf, and work up to about $90 psf or so. I rarely see myself out of those ranges in real world situations in MY market.

Nice job!

 

[CANative]

Interestingly, $40± is about the same as on all those "inherited" adjustment lists.

 

[Metamorphic]

The other thing you have to think about is covariance.

Basically, anything you've already adjusted for that *is* GLA will reduce the size of your GLA adjustment.

The obvious example is beds and baths....both of those are included in GLA, so if you subtract $5000 per bed and bath more than your subject, you've already gone some (maybe all) the way to adjusting for GLA.

In the most strict sense, your Bed/Bath adjustment should only reflect your adjustment for market reaction to the utility of the additional space. The adjustment for the actually SFs should be in in GLA adjustment. As a practical matter, I've gotten around to where on most jobs I include bed/bath adjustments in GLA. Seems like unless you're in a lower end entry level market, people dont react as strongly to bed count as GLA, and bath count seems to be as big a function of where the baths are located rather than how many [With bath it seems like the places that have lots of extra baths need lots of extra baths because there would be a bath in all the necessary places if there were any fewer. IOW its a ** issue rather than a count issue.]

 

[Metamorphic]

Exactly. Just back into it!


just having fun!

Interestingly, this is EXACTLY the theory behind regression analysis; trying to find a a set of adjustments which reduce the variability in the data set. Only regression does it to a whole raft of properties and multiple variables at the same time in a fashion that's less easily observed.

I've always kind of bridled at the "backing into it" suggestion though.

The whole point of the SCA is that the market is rational and that the difference in prices between two or 20 properties is a result of the collective value of the differences in amenities. All the grid based SCA is is an attempt to put dollar ammounts to the differences. There is no right answer, only logical answers, and answers that successfully reduce the variability between the adjusted sales prices. The whole method is "backing into it."