The most widely used and perhaps most accurate method is the comparison of comparable properties, commonly referred to as comps. Typically, this involves three properties that possess a degree of likeness to the subject property, and that were either sold or evaluated within a reasonably recent period of time. Appraisers apply a number of methods to quantify the comparison, some, simple averages of major features, others, detailed assessments of minute components. Still, the factors used in the evaluation remains fairly arbitrary. Moreover, there is an inherent level of bias toward meeting client expectations, especially as matters such as loan approval often hang in the balance.

All that being said, a variety of statistical methods are included in the process, including simple linear regression, cluster analysis, and neural networks. Major lending institutions clearly bare the burden of risk in making loans and assessing real estate portfolios, and recent studies by both corporate and government lending agencies have shown that the statistical evaluation can improve the accuracy of commercial appraisals by as much as 100%, especially when coupled with third-party information such as demographic data. Simple linear regression can define the basic characteristics that give a property quantifiable worth, such as gross square footage and surrounding land, age, construction type, and location. For its part, cluster analysis provides an optimum method for selecting comparable properties, while given the high levels of correlation among demographic variables, factor analysis can also be implemented to further bolster accuracy. At the same time, neural networks have shown themselves to be excellent for generating both micro and macro models.

In spite of this increased accuracy, however, as regulations now stand, no structure can be evaluated using these techniques without the approval of a certified appraiser unless the loan issued is less than $1 million and the property is used as secondary collateral. A further dilemma to a wider acceptance of statistical analysis is the need for clean, comparable data - a problem for most industries just discovering the world of analytics. Additionally, the uniqueness associated with commercial properties poses another obstacle. Unlike residential properties, which exist in large developments with high turnover, commercial structures tend to be far more distinctive and turn over far less often. Still, many commercial appraisers feel a need to quantify unique attributes such as the layout of a lobby, or a building’s overall aesthetics.

The question remains, however, are such attributes in need of special handling? While many appraisers continue to feel that such things have a real impact on a property’s value, statistically, many of these features turn out to be insignificant. One example of this has to do with the amount of land on which a structure sits. While the size of a structure clearly has a major impact on its value, because of the high degree of correlation between the size of a building and the amount of land that surrounds it, the amount of land is not in itself statistically significant, a concept appraisers find difficult to accept. In fact, it still seems likely to be some time before analytics are truly accepted by the industry as a whole. Until then, they will continued to be questioned, as in some cases, perhaps they should.

More Stats in Industry