Dale,
In a word, yes. We actually have 2 lists- one of our "generally" approved folks, and another for "special" assignments- the tough ones.
Sometimes, the AVM is quite accurate; other times not. Note also that I look at the data the AVM uses and not just the number it produces.
When the AVM and the appraisal conflict and I cannot tell which data is relevant (often), it goes out to one of our "experts" with localized knowledge. In those cases, I use "bump logic"- essentially allowing the appraiser to take the assignment as far as needed to produce credible results. Some could be done from the desk; others require a full blown new appraisal.
One very interesting exercise is found in using that database you cited. If I get an appraisal in what is clearly a well established subdivision, I do a radius search for sales within 1% of the subject size. Often revealing. If I find 5 sales of exactly the same size, they could well be model matches. More often than not, these sales are ignored by the first appraiser- as they do not support the value their client wants.
The expert finds them. If the first appraisal was egregious, we will and do report that to the state.
And to Randy, sure there are fraudulent appraisers out there in high demand. I never said they were not- only that there continues to be a good demand for the honest ones, and on that we totally agree, as well.
Brad Ellis, IFA, RAA