Check This Out (AVM)

[bradellis]

Steve,

I do know a bit about what is happening in the AI from many friends and colleagues. I was a member for about 10 years (candidate/associate).

I spoke at length with Ann Spitzey about the AIRD as well and have formally invited her to make a presentation (agree to by our President) to the NAIFA board.

Frankly, it is my hope that sooner, rather than later, we all will come together to foster the betterment of the profession. The umbrella organization is a start; NAIFA's reduced fee offer will hopefully not only increase numbers, but will get some folks in who will be out telling everyone that we are a quality organization and that it might erase some of the perceptions that members of one organization are somehow better or more knowledgeable than members of another organization.

I personally have great respect for the other organizations, and perhaps the egos will finally calm down and see the big picture. Till then, I may have to tilt at windmills.

I'm off to the NAIFA instructor clinic 3/18. Danny Wiley, SRA, will be putting on the USPAP instructor certification there. I'd like to say that he will be pleasantly surprised by the quality of our folks, but I think he already knows this and will not be surprised at all.

Brad Ellis, IFA, RAA

 

[slacker]

Are the AVM products an inevitable solution to many valuation situations? Of course they are. Are they the Silver Bullet that’s going to have all the answers? Of course not.

The bottom line comes when the loan enters a bankruptcy or bad debt situation. The holder of that mortgage needs to settle that amount through sale, liquidation, bad debt write off, or what ever. Over time the banks will be able to determine if the bad debt write off’s increase, decrease, or remain the same. If they increase we’re in good shape. If they can maintain or decrease their level of risk, why not use them? Only time will tell.

In the meantime, the AVM features available through my local MLS are great time saving tools for my appraisal research. So I say keep the technology coming.

I’m putting my faith in the Federal Government to make the right decision. They gave us the “Susan B. Anthony” dollar coin didn’t they?

I’m not in panic mode yet. I’d rather worry about the ice melting on Ruth’s lake.

 

[Tom Hildebrandt]

Bradellis

I just got back from the instructors course in Atlanta. Wiley is a pretty good instructor, you will like the material, lots of concepts, mentally challenging and one h*** of a test. All in all, a top rate class.

Current gouge from the students who have attended is to spend a lot more than the 10 hours the ASB recommended in studying the materials up front. Some say 20 hours is more like it. Also, get a copy of the new FAQ as it give the most current guidance on practical solutions to the questions posed in the course material.

Who else is going to be in your class from this forum? We had several posters, Bob I. and Red B.

Post me back on what your impressions are when you finish.

Regards

Tom Hildebrandt GAA

 

[Austin]

I have been debating regression and AVM for a number of years. I keep hearing the same statement: “AVM will work in cookie cutter subdivisions but they will give wrong answers outside cookie cutter subdivisions.” I totally agree. The problem I have with this argument is that the same thing can be said of the present sales comparison-marketing grid. I doubt if there are many articles or books on regression or AVM that I have not read and they all condemn regression and AVM but refuse to apply the same standards to the existing appraisal theory. The AI’s appraisal journal in the last edition had an article by a statistics professor and the article critiqued an industrial property appraisal done by an appraiser using regression analysis. I reworked the problem in about two minutes, and in my opinion both the appraiser and professor missed the boat for different reasons. The appraiser used four independent variables, three of which did not correlate. He could have used one variable and had an adequate appraisal. His mistake was that he did not know how to do stepwise regression and how to test variables to see if they affected price. He just did not understand the algorithm. The professor knew the math and went off on a tangent about test statistics. My question is: “What test statistics do we use on the present marketing grid. The answer is none, we are just asked to take the appraisal word for it. Actually I found that the test statistics are not relevant because we are not working with random variables. For example, if I do a regression on properties from across the property spectrum and come up with a standard deviation of predicated prices of 10% and an R^2 value of 65%, that does not look so good because the price range is pretty big. But, if I select the 30 most similar sales to the subject from a narrow market segment, know which variables affect price, and come up with the same test statistic results, they mean something entirely different. . For example, with a 10% standard deviation on a $100,000 property is a range of $90,000 to $110,000. But on a property worth $10,000 the range is $9,000 to $11,000. Same test statistic but a totally different picture. Test statistics are not the criteria for measuring success. In my regression model I predict the prices of the comparables and the measure of success is the variance between the actual and predicted prices. You can come up with a perfect equation to explain the correlation between dependent and independent variables and have terrible test statistics. The entire objective of the appraisal process is to find the perfect regression equation that explains 100% of the independent variables. Regression analysis is the perfect tool for doing just that. The poblem is not the math, it is the appraisal theory and algorithm.

 

[Joe Birrell (NY)]

The IRS has no concept of what an appraisal is and no concept of market value. How do I know this ? I spent an entire day at a deposition (hired by propety owner) to agrue why the IRS "appraisal" (so-called appraisal) was not valid. Their idea of an appraisal was a public records print out of all the sales in the condo project (since it was built in the late 80's). They took the average price per square foot of all the units, and multiplied it times the square footage of the subject unit. Unfortunately for the owner, he had a roof top penthouee unit in a project with no current demand or appeal, located in an area where the project and the subject have no market acceptance. Subject unit was over-improvement to say the least.
Their value estimate was $300,000 higher than the best comp from competing project (they had no comps). They even sent an attorney from Washington DC office to drill me all day long. I told him why don't you hire appraisers ? After all, your the same government that makes us jump through hoops like trained seals to become appraisers. He gave no response.

 

[Pamela Crowley (Florida)]

Joe,

I do like your Attitude!

 

[Tom Hildebrandt]

Austin and others

I like your discussion on this topic better than the last time we discussed it :lol:

I particularly like the idea that in appraising we do not have to follow the rigid dogma of the empirical academic regression gurus. In most instances, as appraisers, we should not be involved with the entire data set, or even do random sampling, we should be using judgement sampling.

The 12th edition of Real Estate Appraising recognizes this concept (and your opinion Austin )as valid on page 653 stating "While certain statistical tests used with random samples can not be applied to judgment samples, in many circumstances judgement samples can provide superior results."

We appraisers know this instinctively, it is the basis for the historic acceptance of pair sales as a valid technique. We have evolved past paired sales, as Austin has often suggested, multiple linear regression analysis is the best of the tools available to appraisers today for our daily work.

Regards

Tonm Hildebrandt GAA