This document regression_appraisal_parameters was derived from running a multiple regression analysis on many different neighborhoods using parameters standard to appraisal type valuations of homes. These are characteristics such as the square feet of the home, number of bedrooms, number of bathrooms, garage spaces, lot size, basement and basement finish, etc. Each page of the document is a different neighborhood. So each page demonstrates how erratic the correlation between each physical characteristic is to value based on different neighborhoods and price ranges.
You can quickly flip through the pages of this document to understand that the correlations of these physical characteristics to value (sale price) are not stable. In other words sometimes size in square feet is highly correlated to value; sometimes it has little or almost no correlation to value. Sometimes the number of bedrooms or bathrooms has a high correlation to value, other times the number of bedrooms or bathrooms has little or nearly no correlation to value. You’ll find that multiple regression analysis shows these erratic results for all physical characteristics of homes and value.
What these results really say is that you cannot value a home based solely on physical attributes. And even when you utilize the physical attributes as part of a hybrid valuation model each valuation must be custom made for each house. The marketplace of home buyers place different values on physical attributes based on neighborhoods and price ranges, and the AVM must adjust for this every time it values a home.
Why is this so? Simply because the people who buy homes do not purchase them based solely on physical characteristics. Thus no automated valuation model can accurately assess a property’s value using just the typical characteristics found in appraisals. In fact neither could appraisers if it weren’t for a lot of subjective evaluation on their part.
The other reason the correlation between all physical characteristics and value is not steady is that every neighborhood is different. The buyers within different neighborhoods desire different attributes. As an easy example, just consider two different neighborhoods, one with an average price of $150,000 and another with an average price of $950,000. You can place the exact same attributes into homes in these neighborhoods and have completely different value impact. Visualize nice cherry cabinets and granite counters. In the $150,000 neighborhood these items would take the emotional desires of the buyer off the charts–thus they value them very highly; they would show a high correlation to value in multiple regression analysis. In the $950,000 neighborhood these attributes would go hardly noticed–they were expected; their impact on value is negligible. So the correlation of each home’s attributes to value is very much specific to location, and even more particularly the desires of the buyers for a specific area.
AVMs fail because 1) they derive their value based on easily measurable physical characteristics, and 2) they don’t even take into consideration the emotional impact of the kinds of attributes the market really uses to determine which homes they value more, and how much more they will pay to meet those desires.
QValue™ overcomes all these things and much more. Each valuation is customized, based on what the market desires the most, and how much they are willing to pay to obtain those desires.