Valuation accuracy and variation: a meta analysis

Abstract

This paper provides a first order approximation of the accuracy of commercial property valuations for comparison with the ±5-10% threshold of tradition, convention and judicial acceptance. The nature of ranges is considered in relation to the uniform and normal probability density functions and the effects of bias considered. Summary statistics are examined for gross differences (differences between property valuations and subsequently realised transaction prices) recorded in the Investment Property Databank (IPD) database and significant yearly changes noted in both the means and standard deviations. A meta analysis of previous work is presented which shows all other results involving gross differences (GDs) to be reasonably consistent with statistics yielded by the IPD database. The results for the two main studies of net differences (GDs adjusted for the lag in time between valuation and transaction dates) also suggest yearly trends of a similar nature to the GDs. The role of intravaluer variability is also examined. The variability associated with the time lag is then estimated and applied to the IPD figures to give the approximation sought, suggesting ±5-30% to be a realistic range in place of ±5-10%.