Data sharing: Small, exclusive groups of peers you trust
Woody Fincham, Research Fellow
Since we finally went public with the labs site, the response has been rather good. We have received pats on the back, some good suggestions, and some rather "constructive" comments. I am humbled by the good things, but love the challenge of the folks who have been open enough to tell me how bad my ideas are.
You have to remember, folks: I am an appraiser as well, and I am used to being browbeaten and treated like my opinions are wrong, especially when they are not popular. So thank you to everyone who keeps me and the rest of the group in the know of what you folks like and dislike.
I have spent my last two months writing articles that talked about data sharing amongst appraisers. Some of you have responded with a resounding cry of negativity towards it. Most of the negative items have come from those of us that have a passionate opinion of automated valuation models, or AVMs. Let me be the first to say that I am one of appraisers out there that wholeheartedly believes that AVMs are bad tools to use for residential lending. I have seen too many people make bad decisions with them, and let's honest they can cut into my profits as an appraiser.
I realize that appraisers have been treated like the red headed stepchildren of the lending industry for many, many years. We have been lied to, stolen from and threatened with having our largest client base try and either completely replace us with computers or at least limit the influence we have. I get that, and like many of you it bothers me, but by the same token I understand that both industries must evolve to continue and co–exist.
I want to make it very clear that I would not be a part of this project if I ever got wind of things being put in place to harm my peers and industry as a whole. a la mode is not in the business of harvesting data to deliver it to AVMs. Dave Biggers, the Chairman and founder of a la mode , is very pro–appraiser and does not support the use of anything that is going to hurt our business. Why would he, as his company is the largest software provider to our industry, so much so that the next four competing companies don't have the installed user base as a la mode . The simple economics are enough to understand why a la mode wants us to stay around as long as we can, but this company actually believes in us and supports the reasons we are needed.
So with all that please understand that my data share project is a concept intended to use between appraisers, and only for appraisers. It is a tool for you and nothing else. There is not a motive to do anything with your data. So as long as I have been as clear as I can be about that let's talk about some things I've gleaned from reflection, hashing things out with colleagues and reading your feedback that can improve on my previous ideas.
The evolution of this idea the last couple of months has changed our thinking on it. Like I mentioned in the last article, I was leaning towards a large, location–specific database with a membership style system, very similar to how most MLSs work. I realized that there are two problems with this idea. Problem one is the fact that to simply generate enough revenue to maintain a system like this there would need to be many more appraisers than what is typical for most regions. Problem two comes from the number of folks that have expressed a negative opinion of this system. I can now see that in order to get this to work, it needs to be on a much smaller scale. To a degree that each appraiser involved has equal share, similar to a co–op. Let's look at this a little more closely. Like any problem, there are many ways to formulate a solution, and I am going to try and look at a couple of ways to do this.
Peer to Peer
Most of us are familiar with instant messenger programs that allow us to keep lists of our chat buddies on there. I think it would be pretty neat to be able to form alliances with shops in your area. Like most messenger programs you can share data back and forth as you choose. You could have a list of other appraisers that you could essentially share data with. The data exchanged would be similar to what is transferred through the current comps database interface, although I would like to see this application made a bit more user friendly.
This method works when you do not want to allow random access to your data. The access would be allowed only to who you choose to let in, but you still could have the option to only provide information as asked for (I refer to this method as "going fish" like the kid card game). It is very hands on and would require manual contact between peers to initiate the process. I see this as a drawback, considering what is possible between all the new Internet based applications available now.
Group of Peers
The second option is to allow for a group (say, three or more) of appraisers to share their data. Where peer to peer would be a desktop option, this design would lend itself better to a website. The reason for this is to allow automation for all the users, both exporters of the data and importers. This type of technology allows for secure transfer, at least as much as the individual networks for each user allows, and typical web security options for web protection.
Now, before I get e–mails stating otherwise, the only thing transferring is property data. Methods and conclusions would not be disclosed due to the possibility of USPAP non–compliance. Information identifying borrowers may need to be excluded, too due to financial privacy regulations. The information that transfers into the user database would only be accessible to allowed members, and to them only. a la mode would not be able to gain entry, nor anyone else as long as the security is set up properly.
This could run off of a site owned, set up and managed by the group, or through one appraiser providing access and storage through his/her individual site and network. It really wouldn't matter, but the option chosen would revolve around cost of the site and the degree of equipment it would be running on. Website cost should be split among the group, and of course if the person using it has old equipment then it will run slower than it would with newer stuff.
When an appraiser stores new comps, or updated comps in the Aurora database, the appraiser would have the option to include the new information into the shared database. Redundancies would be flagged and could be either explained or allowed as an alternative variation, so that no comp is overwritten or discounted improperly. This is important when looking at situations where some time has elapsed between two site visits of the same property. Small differences would not matter, and this could be determined by the group to allow variations within certain fields (square footage being the most obvious).
As in the peer to peer concept, the data would be pushed into a template, as uploaded, and kept in the database. When a user is looking for comps, the database in Aurora could be set up to search the shared system as well. User defined criteria would look through the information and return data as searched for. This would then be imported and the appraiser would then analyze, verify and use the data as needed.
Considerations for Both Concepts
The overall idea here is to work along side with people that you trust. I know that in my coverage area, there is no way that I can possibly cover every single appraisal needed by the public. So I don't feel like by doing this I will be helping my peers put me out of business. Like I said in my last article, I am confident in my abilities to out–service the majority of my competition, and the benefits of this type of tool outweighs those fears. If you have a less than positive view of someone's work, then you should not allow them to be in the group.
The simple fact: if the group has equal say then everyone coming in has to be approved by the existing members. If one person is providing the system/service of the data share, then that would be at his/her discretion.
Whatever you and your group/peer decide is really up to you and not anyone else. What you share exactly, how you decide to provide it, and anything else you can think about is really given to the users of the tool to choose. Of course, with any type of system similar to this I would think that user license agreements would be pertinent and understandable. You wouldn't want someone to be willing to illegally provide the information to AVMs, or otherwise steal data that is still your intellectual property.
Using this alongside your current MLS and tax systems could really make life easier for you. Adding other tools to analyze data could help you automate as much as you can leaving you with more time for more work, or some down time with the family. Either way you get to decide that for your self.
I have some other, non–database ideas to put out there in my next article. Until then, keep bringing us your opinions. Good or bad, long or short we are happy to hear them. That is one reason why we are here. If you feel really charged up over something here, let me know. I reply to every email I receive, and as you can probably tell from the evolution of my articles, your opinions have certainly made a difference.
