Wednesday, May 23, 2012

Statically supported Appraisals in Hardin and Jefferson Counties

The appraisal industry has had three main approaches to value for quite a while: cost, sales, and income approach.  However, I’m hearing more and more about statically supported appraisals that include correlation and regression analysis. 

A correlation (r) is a relationship between two variables, and it expressed numerically between -1 and +1.  A correlation can be extracted from a population or sample data set; it shows zero, negative, and positive relationships.  A regression analysis takes the correlation and predicts further results.  For example, Insurance companies have shown that a positive relationship exists between the age of drivers and the amount of accidents that they have; through regression, they are able to predict the amount of accidents someone could potentially have based off their age and adjust your insurance accordingly (Johnson,R & Kuby,P, 2010, p.60). 

This is all fine and dandy, but how does this apply to appraising real estate?  In today’s market of tighter lending practices and fluctuating markets, appraisals need to be more accurate than ever.  In the past, the final opinion of value may have been based off one or two approaches to value, but the value can be manipulated by the comparable sales that are chosen.  A regression analysis can add statically supported data to the approaches to value, thus giving the final opinion of value greater weight and lowering the potential risk for investors. 
I’m not going to get into the nuts and bolts of regression, but I will say this: the accuracy of the data is based off the size of your sample or population.  In other words, select your sample without bias or subjectivity and get a large sample if at all possible. 

A regression analysis is not intended to replace an appraiser's ability to use his or hers training to determine the value of a piece of property, just to support their conclusion.  It is possible that we will be seeing a greater use of regression, with the expansion of on-line access to MLS and PVA data and increased demands from lenders.  I have been using regression to support my appraisals for over a year now, and I wish I could use it in all the areas that I cover.  Hopefully as more of our rural stats become available on-line, we will be able to expand our use of regression analysis into these areas.

Do you use regression in your appraisals?  What has been your experience with this approach to value?  Tell us your stories, and don’t forget to follow us on Facebook. 

1.    Johnson, R. & Kuby, P. (2010) Stat 2: Student Edition
    Brooks/Cole Cengage Learning, Boston, MA

No comments:

Post a Comment