D
Deleted member 126639
Guest
The biggest obstacle/Fly in the ointment for Regression models is the lack of complete data for all residential houses in a particular area/market.
For example in my County on a yearly basis only about 30-40% of the Residential Sale went through MLS.. Our GIS is not the best on the planet. Its improving but that takes time and money. When you consider what a County tax assessor mission is then you understand that perfection is not necessary.
Multiply that one county Gaston x 99 other counties in NC and you begin to see the problem even for regression. Add in the other 49 States and it gets worse not better.
Not saying regression wont be useful, but its never going to be perfect nor IMO even close to being good, let alone Great. It will likely improve as more counties across the nation improve there syystems
40% of all sales is more than sufficient to produce a relevant dataset.
It doesn't need to be 100% accurate. 90% is sufficient, particularly when the credit worthiness of the borrower is orders of magnitude more important than the sale price when predicting defaults.
We as appraisers must remember that the #1 reason banks are trying to diminish our profession is large sampling of reports post-2008 effectively proved that the quality of the appraisal and most importantly the appraisal fee had no relevance in predicting default.
In any event, I only make this point to counter the historically strong libertarian presence here. Automation is affecting many jobs, but residential appraisers will be hit quite hard. Knocking on Starbucks workers for making half the average income of residential appraisers is not the solution to this intractable problem. The problem, ultimately, is the bourgeoisie is taking all of the country's wealth in any way that they can. We should be cheering the success of Starbucks workers and BoA employees. The reality is we as appraisers need similar reforms, although the solutions are a bit more complicated.