Automated Hedonic Valuation Models: A Variety of Projects in Research and Industry

Real estate valuation is often cost- and time-intensive. This is particularly the case for mass appraisal in portfolio management or time critical valuations for acquisition or risk management purposes. Therefore, investors, operators and service providers are looking for solutions to automate real estate valuation. From the institutional perspective, it should be considered that banks, insurance companies, larger housing companies and their service providers, are potential users."Automatic Valuation Models" (AVMs) are often based on and tested by regression models, in particular hedonic models. Hedonic models are fast and cost-efficient as the calculation as well as the data management related to internal and external parameters can get highly automated. The definition and calibration of these models requires large data sets with appropriate parameters.In a first step, a valuation model can be developed based on the data collection and its statistical analysis. On the next level model validations, adjustments and optimizations are carried out. The results of the developed valuation models and their underlying regression equations should be manually examined, interpreted and documented, in order to assure optimal results for various valuation scenarios. The model automation is considered as a crucial step for the efficiency of mass appraisals. As a result of this automation, manual inputs, valuations, controls and other processing steps are mainly avoided. This increases the speed of the processing and at the same time decreases resource consumption. AVM has been established as the main term for this group of models and processes. The goal of efficient valuations with simultaneous quality assurance has been discussed for several years. The current status of the model development and its application on the valuation process are subjects of current research. To identify the gap between the demand of excellent AVM and achievements to date the study examines 1) past research in scientific theory, 2) the past development projects in the valuation practice, and 3) the perspective of associations, political systems and legislations on AVMs.The evaluation of the three research components assesses the achieved interim level and indicates further research requirements. Analyzing these three research components across several countries, similarities and differences can be identified. In particular, this refers to a certain acceptance and intensity of use of AVMs.