Can a Hybrid Automated Valuation Model Outperform Individually Assessed Capital and Site Values.

In recent years the use of Hybrid Automated Valuation Models has been widely discussed in the property taxation literature. Such models are now recognised in the IAAO's standard for AVM's. Given recent court decisions in Australia that seem to require valuers to consider sales with improvements when assessing site value in "thin markets", such models may prove to be a useful tool in mass appraisal. This is particularly relevant following problems with the delivery of acceptable valuations for rating and taxation in several states in Australia. In 2004 a variety of AVM's were developed for a study area in Adelaide and presented at the 2005 PRRES conference. That research showed that a single model could produce capital and site values of a similar average accuracy to the actual assessed values that were created using multiple computer assisted valuation (CAV) models at the small sub-market level. This paper extends the research, by improving the models using the outcomes from the previous research but also by applying a wide range of tests. Assessments based on a large number of small models may lead to assessment bias and an abnormal distribution of assessment ratios typically evidenced by low relative valuations for high priced properties. In this research various measures from the IAAO standards on ratio studies are adopted as the tools to assess the performance. In particular the A/S ratios are tested for the level of assessment using mean, median, weighted mean and geometric mean and for variability using the coefficient of dispersion (COD),coefficient of variation (COD) and quartile ranges whilst reliability is tested using confidence intervals and vertical inequities with the price related differential (PRD) and normality with the Shapiro- Wilk test. The Mann-Whitney test is used to check for sales chasing and reliability between the model and hold-out data.