Paired Sales Analysis & Adjustments

[george foster]

Ron,

Just a little ancedotal story...I practice in a sparsely populated largely rural northern New England state. In 20 years of business, I've never been in 2 houses which were exactly the same. Those 1000 home cookie-cutter subdivisions you see in National Geographic in articles on sprawl might as well be on the moon.

I took an Appraisal Institute Paired Sales Analysis seminar put on by my local chapter. The instructor brought a set of local researched comp's to use in his demonstrations.

Well, to make a long story short-while diligently applying all the theorums discussed earlier in the class in the final demo, he simply could not get the numbers squared away to come up with a legitimate, decent looking array of final adjusted values. It was all quite embarrassing-

Comp's were too diverse-Paired Sales Analysis theory is one thing-practical application is another.

The support for my adjustments tend to be piece-meal abstractions put together from years of experience. You grab one here-you grab one there, and put it all in a file. Eventually it all comes together, and you have a credible system. There is reliance on logic and inference in this business.

But in rural areas, it sure "ain't" the black and white like they'll have you believe in Society Course 101.

Appraising, is "art" form, as much as it is numbers. You either have that innate aesthetic feel for something or you don't. Personally, I think good appraiser's have a certain gift. And this is also why one who have got "it" can see all the hacks for what they really are-"fast-buck" Charlies...

Just my 2 cents FWIW......

 

[Ariba]

MH,

So let’s say that I do not have the facts/data to back-up that statement due to the lack of sales and no paired sales in the market area. Should common sense prevail?

Another example, a subject and a paired sale in the same market area that is very similar in square footage (3,300), rooms, bedrooms, etc. The only difference is the year of construction, 1985 vs. 1995. The houses are in the $600,000 price range. The data seems to indicate a $100,000 adjustment for age. Common sense tells me that this is not the case in this market, but the data in this particular paired sales reflects a $100,000 adjustment. Do I go with common sense since there are no other paired sales available?

As has been posted in this thread, Appraising is an art and common sense has a role to play if the data seems flawed.

Ron <_<

 

[Pamela Crowley (Florida)]

Another example, a subject and a paired sale in the same market area that is very similar in square footage (3,300), rooms, bedrooms, etc. The only difference is the year of construction, 1985 vs. 1995. The houses are in the $600,000 price range. The data seems to indicate a $100,000 adjustment for age. Common sense tells me that this is not the case in this market, but the data in this particular paired sales reflects a $100,000 adjustment. Do I go with common sense since there are no other paired sales available?

"Paired sales", if it ever comes close to actually determing a value of a 'item', needs much more than 2 sales gridded.

In various sub-markets in my area, I regularly grid 9 - 12 sales to help determine adjustment values.

 

[Mike Garrett RAA]

Absolutely, under no circumstances, use common sense! :rainfro:

Paired sales analysis is the heart of determining adjustments. Most appraisers develop, over years and years of experience, adjustments that work and were proven up by sales analysis. Do they do paired sales for every item on every appraisal...uh huh, sure, of course, wink wink

I am curious what part of your market has no sales? I would also like to see proof that swimming pools have negative values?

How would you like to own "lake front" property on Lake Powell. Just read a study that says....."Powell will be gone by the year 2007". Find me some paired sales for that type of adjustment.

 

[Ariba]

Mike,

Since the house are similar could the proof be as easy as Day-On-Market?

Ron

 

[Alan Simmons]

Ron wrote, “Take your pool as an example. Here in Colorado we are having a drought, and for a period of time and you were not allowed to fill the pool. So a swimming pool was a negative in as far as determining value.”

My question is, did the pools actually loose value or were they just suffering from a short-term external obsolescence? Did the drought actually shift the paradigm or is the historical trend line still valid but just in a lull?

Seems to me that the smart thing to do would be to buy a house with a pool at a discount during the drought. If the buyer’s expectation is to own the house for any length of time, as soon as it starts to rain the buyer could expect to capture some of seller’s unrealized value.

How much of a discount are houses with pools currently realizing? What are the expected maintenance costs to mothball a pool? When is the drought expected to end?

 

[Ariba]

Alan wrote

When is the drought expected to end?

If I had the answer to that question I would not be appraising. I would be trading in futures and be rich! No LO, what a thought!

Ron

 

[Fred]

I don't get it. How do you make adjustments then? Please don't come down on me just a newbie dumb dumb asking? I often wondered how often you experts review your adjustments

As they say in Star Wars, to me “adjustments” are the Dark Side of The Force. “Matched” pairs would work, hypothetically. If you could flap your arms fast enough, you could fly, hypothetically. Matched pairs from workbooks can be good for trainees the way push-ups are good for football players.

Paired sales analysis is the heart of determining adjustments. Most appraisers develop, over years and years of experience,

Maybe. But who has years and years? The answers are there every day. Finding paired sales and adjustments are like feeling your way around in the dark with a stick. I don’t know why so many appraisers are so completely dedicated the most difficult possible way of doing this. The sales have too many characteristics to check and they mixed around in the sales that one pair gets in the way of isloating the other.

I use bracketing for simple problems and couple of types of regression. I just can’t imagine appraising without some help from regression. It’s the Jedi’s light saber. Neither makes “adjustments,” but you can use regression output to make adjustment grids. The basic solution is to just put the sales on a table and let the computer figure out what correlates. A few macros can solve the regression equation and make an “adjustment” grid in a blink.

This is a hypothetical set of 7 sales from a “perfect” market. There was high correlation on year built and size, so dumping everything else leaves a simple table. There are no “matched” pairs. There is an exact and simple solution, but I don’t see how fumbling through the dark with a stick (pencil) will find it.
The data columns are year built, size (sf), and price.

1 2002 2,100 414,000
2 2000 2,200 418,000
3 1999 1,750 380,000
4 1996 1,800 378,000
5 1995 1,900 384,000
6 1994 2,050 394,000
7 1992 2,000 386,000

With a subject built in 1997 containing 1,950 sf, simple “bracketing,” tells me it’s in the middle of the age range and the middle of the size range. The middle of the price range (average, whatever) in the lower half of the $390,000’s somewhere. That’s a sound solution as far as I’m concerned. Done. No “adjustments.” Or I can run the regression equation, which comes to exactly $392,000.

Because people expect to see that adjustment grid, here it is – except this one is sideways instead of up-down. This is “perfect” market data and all sales adjust to the same indicated value.
The data columns are price, size adj, age adj, and adjusted value.

1 414,000 -12,000 -10,000 392,000
2 418,000 -20,000 -6,000 392,000
3 380,000 +16,000 -4,000 392,000
4 378,000 12,000 +2,000 392,000
5 384,000 +4,000 +4,000 392,000
6 394,000 -8,000 +6,000 392,000
7 386,000 -4,000 +10,000 392,000

 

[Terrel L. Shields]

At least you guys and gals are asking the right question. However, I suspect a lot of "paired sales data analysis" is, in fact, allocation more than analysis. And I believe allocation, in fact, to be a more sensible approach to estimating an adjustment than I do paired sales data analysis in a slow or rural market. I can slice, dice, and mull over rural property attempting to divine the contribution of CHA, Fireplaces, etc. etc., but the reality is I need to estimate a land value as if vacant, estimate the contribution of the non-complex outbuildings, and, eventually, get back to a per SF contribution for the dwelling by itself. In my opinion, I first need to identify any functional or external market influences. So if I, for example, have found a 2,000 SF 20 year old home contributes $80,000 ($40/SF) and it appears the RCN is $60, then I don't think it unreasonable to attribute most, if not all, the depreciation to normal wear and tear.

If I do this same analysis to the comparables available, perhaps something will "jump" out at me. i.e.- comps without att. garages are $2-3 / SF more as an example, thus 2,000 SF x $2 = $4000 adjustment. That's more allocation than paired sales in my opinion. And bracketing this adjustment reinforces that "sense" a good appraiser should develop in estimating adjustments

For age/depreciation adjustment, I would be more inclined to follow a series of resales in a regression analysis (which is just a quantified form of allocation), kicking out any where remodeling has clearly been done. By 20 or so years old, most have been remodeled.

When I feel compelled for some reason to do paired sales, then I go to a newer subdivision, typically one of 30-60 dwellings, with consistent style, extract the sales data for a 3 yr period or so, and pair for whatever will fall out....and few things usually do except per SF, maybe a fireplace or garage. The only reason to do such analysis is to keep Fannie mae off your fannie. I do not like the new form certification (I supposed I could live with the form if that were the only change) that basically tells you that the report thus created is a "limited" report, but does not say how it is limited. Limiting the reporting does not limit the appraisal. Nevertheless insists you use all sales "locationally, physically, and functionally" closest to the subject. And, if you cannot find data on a known nearby sale? You can't throw it out? Use it with the caveat that you assume the conditions were arm's length? The list of "unacceptable appraisal practices" is an insult to professionals, so I assume Fannie mae does not consider us professionals, only cannon fodder to feed congress when the next market crash occurs. The statement regarding the "use of adjustments...that do not reflect the market's reaction" implies to me that fannie mae is still married to paired sales and woe to any appraiser who uses an alternative method of determining an adjustment.

Finding paired sales and adjustments are like feeling your way around in the dark with a stick.

Or a blind man trying to find a black cat that isn't there in a dark room, and complaining that if he could only find the light switch he could find the cat.

 

[moe cohen]

Steven
I am sure you are aware of multicollinearity, your variables are ok but even age and size are not independent, let alone bedrooms and bathrooms. I think that the casual user of regression analysis should not use it unless they study it well.
there are methods to try and cure it such as stepwise regression and principal components but that may be beyond the scope of our assignment.
Moe