Spatial Automated Valuation Model (sAVM) – From the Notion of Space to the Design of an Evaluation Tool

Abstract

Assuming that it is not possible to detach a dwelling from its location, this article highlights the relevance of space in the context of housing market analysis and the challenge of capturing the key elements of spatial structure in an automated valuation model: location attributes, heterogeneity, dependence and scale. Thus, the aim is to present a spatial automated valuation model (sAVM) prototype, which uses spatial econometric models to determine the value of a residential property, based on identification of eight housing characteristics (seven are physical attributes of a dwelling, and one is its location; once this spatial data is known, dozens of new variables are automatically associated with the model, producing new and valuable information to estimate the price of a housing unit). This prototype was developed in a successful cooperation between an academic institution (University of Aveiro) and a business company (PrimeYield SA), resulting the Prime AVM & Analytics product/service. This collaboration has provided an opportunity to materialize some of fundamental knowledge and research produced in the field of spatial econometric models over the last 15 years into decision support tools.