- Joined
- Jan 15, 2002
- Professional Status
- Certified General Appraiser
- State
- California
By the time we're done we invariably quantify our qualitative analyses and we qualify our quantitative analyses.
I don't either but they can indicate a quality issue. I did a lot of regression analysis when I first tried it just to see what the drivers of value were, excluding land. Regression fails with items like fireplaces were common presenting either absurd adjustment or low Rsquare values. Eventually I concluded that size, age-condition (effective age), garages, and quality (oft reflected in brick, part brick, or all brick exterior as a proxy) were driving the bulk of general adjustments. The rest of adjustments should come from cost related adjustment or pairing if possible.don't adjust for small features like a fireplace
I had Realstat and using the example (which was commercial property), it really needed a couple dozen sales or even 40 or so, which is problematic with rural and small towns- impossible with commercial or farm properties basically...and town properties typically are somewhat more predictable than rural. So I went to a cheap add on and whittled it down to 4 or 5 factors. The first thing I did was take the land out of the equation and value based on land prices...which we had far more of those than large fully developed areas would. This allowed me to work with as few as 8 or 9 sales, with 10 or more being desirable. By doing both paired sales and sensitivity along with the RA, it often were very close results which, in my mind, meant the adjustments were credible.Austin spent a lot of time advocating RA on this forum and he always said (and I agree, in principle) that the data involving typical SFRs usually stops responding significantly once you increase the variables beyond about 4 factors.
I like sensitivity analysis and use it all the time because it has the benefit of being internally consistent with the dataset I'm adjusting, it's simple to perform, and its easy to demonstrate and explain.I had Realstat and using the example (which was commercial property), it really needed a couple dozen sales or even 40 or so, which is problematic with rural and small towns- impossible with commercial or farm properties basically...and town properties typically are somewhat more predictable than rural. So I went to a cheap add on and whittled it down to 4 or 5 factors. The first thing I did was take the land out of the equation and value based on land prices...which we had far more of those than large fully developed areas would. This allowed me to work with as few as 8 or 9 sales, with 10 or more being desirable. By doing both paired sales and sensitivity along with the RA, it often were very close results which, in my mind, meant the adjustments were credible.
To @ucbruin 's point about the "Book" - really I think in a given area, you could develop such adjustments periodically and apply them favorably within price groups. OTOH, with the spreadsheet sensitivity analysis, it's so quick to simply punch in the comps and see what they say, and an added benefit is if one of your sales is "far out", then perhaps you need a better comp.
It's curious to me that so many folks get disjointed about the idea of the Book of Adjustments....
Yet have no issue with the idea that one can apply common sense to a reach a conclusion....