Automated valuation models (AVM) according to the RICS AVM Standards working group are systems that use one or more mathematical techniques to provide an estimate of the value of a specified property at a specified date, accompanied by a measure of confidence in the accuracy of the result, without human intervention post-initiation. They combine property sales data, property attributes data as well as local market information (RICS 2013, Corelogic (n.d)); these form the variables that are fed into the model. Models typically comprise one dependent variable which is the estimated property value and several independent variables (property attributes data) which take turns in explaining the dependent variable (RICS, 2013). AVMs vary depending on the modelling technique adopted, the methodology and independent variables adopted. Choice is solely down to the provider’s specification (RICS, 2013). Examples of the different models include; multiple regression model, indexation, sales comparison models and automated comparable selection and artificial neural networks. AVMs have been around for a while. However, market acceptance has been slow, tentative and somewhat phased.
There has been a significant increase in the use of AVMs in valuing properties for taxation purposes (RCIS 2013, Adair and McGeal 1988). This is primarily due to the somewhat cumbersome nature of the traditional single parcel appraisal method adopted by valuers, which is also very cost intensive (O’Neil, 2004).