I've never been married to paired sales. The absurd results you can get means it is often hopelessly wrong. By mixing both Regression and paired sales you can sometimes get a better result but again, if I actually created a formal "paired" sales data set, I bet I could plot the results against the dollar value of the house and it would assume some sort of expotential change with home value...
Thank you. Thank you. Thank you.
When I lived in Las Vegas it wasn't that unusual to find paired sales when appraising simple model match homes. That being said, for every pair I found, I could usually find three more that indicated something completely different. Regression has certain advantages, but it too can, (and does) yield misleading results. If regression was so great, AVMs would have put us out of business long ago. Paired sales, even under idyllic imaginary conditions, would only yield accurate results in an absolutely identical circumstance and in an absolutely static market. Personally I've never encountered either - even when appraising two simple, model match properties on the same street, within a few weeks of each other. Something is always different. When one part moves, it affects one or more of the other parts. Appraisals have lots of moving parts. Most of the time, in the appraisal of residential properties, the key to accuracy and reliability is experienced qualitative analysis - these qualities are not typically obtained through paired sale or qualitative analyses. The funny thing is, my files, and my reports, probably have more quantitative analysis in them than most, but these analyses are usually for those elements of the reports which lend themselves well to such analysis.
I just sent a residential report out this afternoon that involved three distinct market conditions analyses. Due to the overall paucity of data in the subject's area, I decided to approach the problem from a multitude of directions. Analysis A yielded a rate of depreciation of about one and a half percent per month. Analysis B yielded a rate of depreciation of about one percent per month. Analysis C yielded a rate of
appreciation of about one and a half percent per month. If you don't think I had to temper these statistics with some serious local knowledge and experience - think again. Conversely, as this report was prepared to assist the homeowner and a realtor with the establishment of likely sale and listing prices, I also performed a sale to list price ratio analysis. This gem was really helpful. I was able to derive a general pattern of list to sale price ratios, and, to temper it with a bit of common sense, as the subject is a high end property for the neighborhood, and the high end properties in the neighborhood tend to have list to sale ratios a few points lower than the median. The point here is, that although I prepared eight or ten tables to help me appraise this property, none of them were made part of the report, and, in the majority of the places where the rubber met the road, the calls were made qualitatively.
