AVMs and UAD

[Pittsburgh Pete]

That's your response? Get a clue about appraising? Good one Pete. But what does appraising have to do with algorithms and database analysis? What makes you think that you can process relevant and salient property features better than a computer? Maybe you need a tour of the Google compound. And like any board would have any opinion on the topic of AVM's or the future of the industry for that matter.

I don't get it Pete...my point in my original post was to not grant property rights to appraisal management companies now that the tools available give them unprecedented ability to process and analyze data. Are you making the counterpoint?

I have no problem with anyone getting the data. One can give boards, nails, etc. to anyone but it doesn't mean they can build a house. That takes talents that are honed over time.

What I do with the data in the way of interpretation and analysis is what makes the appraisal. While I believe that the quantity of data available for residential valuation makes some computer-assisted valuations possible, such computer-assisted valuations will never compete with a walking, talking, competent appraiser.

Ask anyone on this forum what they have to offer that a computer can't provide? Most, I'm sure, will find your analysis flawed.

 

[lefty2cox]

Well Pete, then you have nothing to worry about. Sounds to me like 90% of your business comes from not residential, non mortgage related clients. I envy you. All that complex property you're appraising... I apologize. I didn't realize that you don't appraisal any ranches, capes or colonials in suburban neighborhoods made up of ranches, capes and colonials. If you really think that I can't pump that data you gather into a computer program and have a more credible value pop out the other end, then you're delusional. I'll even write my own code, use my own program...I won't rely on some multi-million dollar platform built by MIT grads. We're talking about market share gained over time, not replacing you overnight.

And Michigan, is that sarcasm? I've seen your posts...another genius.

Maybe you guys should check out the some studies that have been done showing antiquated AVM technology outperforming appraiser's work in lending portfolios.

But you keep signing away those rights. Your wifes job covers your heath insurance, you got nothing to worry about. Hell, you're not even entrepreneurs...you got lucky. Right place at the right time. Poof, you're a business owner. Start acting like one.

 

[J Grant]

This is just craziness. Why do banks need "our" data? They can get all the data they want from public records, or from public views of MLS, they have been using their own data from their own soruces for years in AVM's.

I doubt the GES'e would have gone to the trouble of revamping the forms if they wanted to eliminate appraisals so soon.

If all they wanted was data, couldn't they pay realtors to enter data for them on properties? It makes no sense. The data will trickle in on appraisals on certain properties and not others, and by the time a bank needs it again in 3 years, some of it will be obsolete. I suppose they will store the data and make some use of it, but I don't think it is an evil conspiracy to mine data and then replace us AVM's using the mined data.

Data is so freely available now with Zillow, Google earth, County property appraiser records online, public views of MLS listings, plus banks that use AVM's subsribe to their own data bases.

If one day they lenders truly don't need us to understand market forces and pendings and analyize the sales data and inspect the property, then they won't.

 

[Fritz in NY]

Whats the point in this thread other than Pete being beat up by lefty????

 

[J Grant]

It is not "our" data, the public records data is free, if we add any input from MLS not on public recrord, if they want data they can get that too. Data is free, cheap, online, data is everywhere now. I am suprised they even bother gettting it from us.

We offer our conclusions from analysis, our market knoweldge, our inspection of the property. If one day this input is not valued, so be it.

I have not seen any studies of AVM's outperforming appraisals in lender porfiolios...out performing by what standard? Lower foreclosure rates? More loans made?

 

[lefty2cox]

Fritz,
I don't mean to beat up Pete but he's an employee. He's trying to justify to himself why he continues to walk blindly down the path that his competition is leading him and assuming they're benevolent. Or he's a mole. Have the admin check his credentials.

Listen, there's no one out there who recognizes more the difference between an appraisal and an AVM. I agree that currently, an AVM is no competition for a qualified appraiser. But I'm not talking about right now.

J Grant. You are 100% correct about the data and farming of the data. But the point being missed is that until now, the important, missing data would have to be entered. Entering data is not feasible. The data must present itself for integration. Otherwise, the cost of moving the data would exceed the threshold for profitability.

Besides, brokers aren't allowed to utilize MLS data in that way, typically. And brokers know better. They're obviously smarter than appraisers. They have BPO's to sell. They're not going to give there data over to someone who is going to utilize it to create a model to compete with them. In addition, they know that once this model is mastered, it will be marketed to the everyman to minimize the need for listing services. Entry only listings will become more and more common. Besides, MLS data, as everyone knows, is biased. Its not presented impartially because it's a sales tool, not an analytical tool.

This is my point folks. The public records available for importation lack certain detail and facts only available when analyzed and gathered by a person, an appraiser. We have all talked about them with clients when discrediting a value obtained from zillow or some other AVM type product. We tell our clients that many factors such as quality, condition, GLA verification, locational factors and so on, are not properly considered in these models. We are usually correct. This doesn't mean that this information isn't available, as stated earlier by more than me, lenders and management companies could technically copy this data manually into a property database from appraisals they have obtained. However, having a coded system that can interface directly with their model would solve the efficiency problem.

You will now be telling a client and any go between that the quality of the subject is Q3.
Q3- Dwellings with this quality rating are residences of higher quality built from individual or readily available designer plans in above-standard residential tract developments or on an individual property owners site. The design includes significant exterior ornamentation and interiors that are well finished. The workmanship exceeds acceptable standards and many materials and finishes throughout the dwelling have been upgraded from "stock" standards.

C2: Kitchen-updated less than one year ago; Bathrooms remodeled one to five years ago.

These codes are strait from FNMA's new system and now give the user of the report (keep in mind i'm not talking about your well meaning lender. I'm talking about your go-between who has asked for the property rights to your work and is a company in business to profit not give you an income) a field and code for mapping to a database.

These and other characteristics of a property are now quantified and served up in computer language. This format IS NOT necessary for the mere underwriting of an appraisal but is yummy to a computer.

Step back. Look at this objectively.

 

[NC Appraising]

This is just craziness. Why do banks need "our" data? They can get all the data they want from public records, or from public views of MLS, they have been using their own data from their own sources for years in AVM's.

I doubt the GES'e would have gone to the trouble of revamping the forms if they wanted to eliminate appraisals so soon.

If all they wanted was data, couldn't they pay realtors to enter data for them on properties? It makes no sense. The data will trickle in on appraisals on certain properties and not others, and by the time a bank needs it again in 3 years, some of it will be obsolete. I suppose they will store the data and make some use of it, but I don't think it is an evil conspiracy to mine data and then replace us AVM's using the mined data.

Data is so freely available now with Zillow, Google earth, County property appraiser records online, public views of MLS listings, plus banks that use AVM's subscribe to their own data bases.

If one day they lenders truly don't need us to understand market forces and pendings and analyize the sales data and inspect the property, then they won't.

They care. money, MONEY mONEY......
http://www.fncinc.com/Products/ncd.aspx

http://www.fncinc.com/Products/epo.aspx

http://www.fncinc.com/Products/data_express.aspx


A excellent biz model. They charge you to upload your appraisal, and then charge you again to use the data. :blush:



ACI and appraisal dot com are heavy players in the UAD. The other big AVMS, banks, and AMCs are in the MISMO program and have their hands in it.


One cannot rule out that the above companies will not find a way to data mine our reports. Given all the factual data and how we will now will have to rate things like view, quality, location, and condition, things that a AVM has problems with, appraisal data will be a gold mine for AVM companies.

I do not see it destroying us or putting us of business tomorrow; although we will make the AVM companies a lot of money by them mining our data. No more template average comments for everything.

It will affect our business and the amount of business we get. Currently many AMCs offer free AVMs or the broker will order a AVM for a Comp check. For the most part, they are accurate enough to give the broker a yAh or a nay on whether or not to go forward with the lending process. EVER WONDER WHY YOU ARE ONLY GETTING THE HARD ONES? When I speak with other appraisers, that is all they say. Yes, depending on the loan type and LTV, AVMs are being used under 250K.

As the AVMs improve, the better the COMP checks will become, and will be used more and more under the 250K.

 

[J Grant]

But the data still gets dated. If we enter a code for kitchen updated 1 year ago, then what happens 3 years from now? The evil company stealing our data updates the kitchen to 4 years ago remodeled, if they need to do an evaluaion in 3 years?

I guess of all the things to worry about, I can't worry about it, at least not that much!

 

[Michigan CG]

And Michigan, is that sarcasm? I've seen your posts...another genius.

I will take that as a compliment.

The truth is things change, and with change one must adapt. I don't see residential appraising coming to an end anytime soon but it will develop and change like every other business.

If I owned a video store or a book store I would cry the sky is falling but our industry will be around for quite awhile.

 

[lefty2cox]

Thank you NC my brother. And their probably just utilizing the usual average, good and so on that we provide them. Also ganking GLA, room count and other choice morsels that usually require verification.


JGrant. Yes. You hit the nail on the head. And what makes that such a far fetched idea? And remember, I'm not talking about never having to appraise that property again. I'm talking about having a meaningful basis for use of that property as a comparable. What's wrong with depreciating that with an algorithm? And remember, the 12 month sales guideline is a FNMA guideline. It doesn't mean that sales data over 12 months old can't be reconciled to a current value with statistical modeling. And who's to say their not sprinkling a little of our value opinion into the mix? I may be wrong on this but I don't know of any federal regulation on how an automated valuation is derived. They look like trade secrets to me. Please correct me so I don't have to worry about them utilizing value opinions from refinance appraisals to enhance the dataset. What we're talking about is a dramatic increase in the quality of the data fed into a model which will result in a dramatic increase in the reliability of said model.