Appraisal Statistics

[RCA]

Do you have MARS? I found out last week that you can no longer get it at the old rate - unless you were already subscribed to it. They want to you to pay $15,000 upfront for all of their packages. If you don't have it, you could try begging that you are a poor appraiser. But they sounded pretty strict to me that the only way out of the fee is to teach a class in MARS or be a student, in which case, all the students get a free license for one year. I guess just go to a MARS class every year .... But give them a call.

I would iterate: Why is MARS important? Because it gives you the highest R2. The R2 indicates what percentage of the price variance is explained by the model. If the R2 is 50 and includes GLA, LotSize and BathCount, then you know those 3 adjustments will take care of 50% of your problem. You have to figure out the other 50% - which is almost always some combination of appeal and condition (Quality, Condition, View, ...). Most of your work will be in trying to calculate the residuals from the model, and then modeling the residuals against a bunch of comps. If you can estimate the C-n and Q-n and throw in a view numerical (e.g. % of views better than the compable), then rerun your MARS, you will then get a new model with Quality, Condition and View. I would bet you that you would likely be well over 90% of your price variance by then. Throw everything in the grid and fine tune based on your ground on the feet observations, phone calls and so on.

Last note - I've always said for many years that Salford Systems MARS is really the only regression tool that every worked for me on complex areas like Northern California (I've tried 3 other versions of MARS without much success). People turned a blind ear and eye and just went for linear regression or multi-linear regression. They could have maybe swung a deal with Salford-Systems (now owned by MiniTab) to provide a specially price package for appraisers only, but it's probably too late for that.

 

[Charles West]

No I don't have it but would like to play with it for awhile. If that is correct that's a little steep to put it mildly. Which package is suitable for appraisal use?

 

[RCA]

No I don't have it but would like to play with it for awhile. If that is correct that's a little steep to put it mildly. Which package is suitable for appraisal use?

The product you have to purchase now is "SPM 8.2" which includes

MARS is what I have always used. I think it is pretty easy to understand what this does with some explanation. This is the best tool around for extracting adjustments. Maybe the only tool for complex properties..
CART (Classification and Regression Trees) - It is specifically designed to group your data into chunks and could be used for a variety of purposes; MARS is based on CART. Most appraisers could probably understand the concepts
Random Forests - Looks very interesting and I wish I could afford it. More difficult to explain what it does.
TreeNet - Also looks interesting but again, a bit more difficult to explain what it does to a layman.

I'm sure you could find use for CART (which can be used to create groupings and is used as a basis for MARS).

 

[Deleted member 130081]

It's just that sometimes, even with plenty of data, due to the complex relationships, it just about takes an advanced degree in statistics to extract the adjustments. There is a rhyme and reason to about everything. But, everything, as complex as it is, is made even messier by a bunch of assumptions that have been around for ages - like there is one and only one correct market area or neighborhood for a subject property. In reality it is often far more complex. People look far and wide to find what they want - which very often places a higher priority on features other than location. If you are narrowly focused on the traditional features of importance, you will often believe there is no way to extract credible adjustments. But who can argue with you without some details? - Some example cases. I like a 1.5 story home, where I can lock off the upstairs for my wife and myself and rent out part of the downstairs to help pay the mortgage, sharing the kitchen. I kind of know exactly the kind of house I want. My wife would prefer a condo to my large house and small farm concept - because she'd rather do other things than clean house all day long. We both together have to decide on something - somehow. Well, you'd be surprised how many people have the same kinds of problems. Yes, they have a preference for a certain location, for this or that, but they aren't necessarily going to find what they want. They settle on the closest thing - as they perceive closeness - which a talented sales agent can alter to his/her benefit, an appraiser can stomp on, a reviewer or underwriter can mitigate and so on. And don't forget the house inspector.

So, even though you agree with me, you don't agree with me - lol.

Your response is full of assumptions, but also some truths, about how people shop for property. In the end, all you have said only adds to my point. I bolded one massive assumption for argument sake. You say people look far and wide, which equates to more emphasis on features than location. Do you have a stat on how "often" that is? I'll bet you don't. So here's a story from my life if I can trade one back. My wife and I decided we wanted to live in a very specific part of our village. Probably a couple hundred homes total (might not even be that many). The homes in this area are a total mix - everything from small shacks to some pretty large places, numerous ages and physical conditions (no mansions). Because there are only so many homes, and only so many go up for sale every year, we knew we would settle for whatever became available, and that we would not have much choice in the matter, and considering everyone who buys here pays a premium for the location, the dynamic is reflected in all sales in the area. That's pretty much the exact opposite of what you said. LOL - I wont do a mortgage appraisal in my own neighborhood, but I do all other assignment types.

 

[RCA]

So, even though you agree with me, you don't agree with me - lol.

Yep. But, enough said, otherwise I will get off subject.

 

[Rlong]

So, even though you agree with me, you don't agree with me - lol.

Your response is full of assumptions, but also some truths, about how people shop for property. In the end, all you have said only adds to my point. I bolded one massive assumption for argument sake. You say people look far and wide, which equates to more emphasis on features than location. Do you have a stat on how "often" that is? I'll bet you don't. So here's a story from my life if I can trade one back. My wife and I decided we wanted to live in a very specific part of our village. Probably a couple hundred homes total (might not even be that many). The homes in this area are a total mix - everything from small shacks to some pretty large places, numerous ages and physical conditions (no mansions). Because there are only so many homes, and only so many go up for sale every year, we knew we would settle for whatever became available, and that we would not have much choice in the matter, and considering everyone who buys here pays a premium for the location, the dynamic is reflected in all sales in the area. That's pretty much the exact opposite of what you said. LOL - I wont do a mortgage appraisal in my own neighborhood, but I do all other assignment types.

While I'm sure this is true for YOU, I think it is a stretch to extrapolate how a certified general appraiser, who has full access to the MLS shops, to the typical buyer in the market. Most appraisers are (or should be) actual experts in the market, AND/OR analyzing properties in any market they are considering. While many point out on these forums that the "typical" buyer in the market is not generally well educated in real estate market dynamics. But, I DO likely shop more like you; laser focus in on what are MY priorities; of course I can run as many searches as I like, and change as I see fit. In the end, I don't think comparing statistical analytics of what is historically happening in a market, to how an individual appraiser shops is a realistic assessment of property analytics.

 

[Deleted member 130081]

While I'm sure this is true for YOU, I think it is a stretch to extrapolate how a certified general appraiser, who has full access to the MLS shops, to the typical buyer in the market. Most appraisers are (or should be) actual experts in the market, AND/OR analyzing properties in any market they are considering. While many point out on these forums that the "typical" buyer in the market is not generally well educated in real estate market dynamics. But, I DO likely shop more like you; laser focus in on what are MY priorities; of course I can run as many searches as I like, and change as I see fit. In the end, I don't think comparing statistical analytics of what is historically happening in a market, to how an individual appraiser shops is a realistic assessment of property analytics.

? Well exactly. That's my point. This would be a case where you need to put the posts in context of the other posts.

My point is that there is little that is truly typical. The precision analysis we are asked to do becomes too precise for the data/housing stock/human behaviors/intangibles available, and when we push it through anyways, simply to satisfy the ignorant whim of our mortgage clients, the assignment results are actually less credible, and misleading, in most cases.

There is a significant need for mortgage appraisals to look the same from everywhere. This in itself is not bad, and in fact is a very good thing, from a certain point of view. The first word of URAR is Uniform - that is not an accident. The first word of USPAP is also Uniform - also not an accident. The difference however, is that USPAP is mostly written by appraisers, and the URAR is mostly written by mortgage clients. The resulting texts reflect these influences. URAR has stuff in there appraisers would do very different, or not at all, and is a rigid format. However USPAP is written intentionally vague in many places, to allow flexibility. Last and not least, the URAR was written before the internet, back in the days when appraisers had to sort through deeds and assessor records at the courthouse, and were pretty darn lucky to find 3 decent comps. I wonder what the appraisers in those days would have done if the clients came crashing down to "support" their adjustments back then? They could pretty much say anything and get away with it, true or false, and no one could really dispute it. Today, the data is out in the open for all to see.