Continuity of the valuation of property portfolios with stratified sampling: a case study

The purpose of this paper is to establish whether the estimation accuracy of the sample-based valuation of a fairly homogenous real estate portfolio with stratification based on principal component and cluster analysis is robust over multiple valuation dates. We use a model portfolio of 2400 rental apartment buildings in Germany to extrapolate the portfolio value from fairly small samples and calculate bootstrap confidence intervals to estimate the precision. The samples are based on cluster allocation using a theoretical statistical process. The continuity of the sample-based valuation model is analysed by comparing cluster allocation and confidence interval accuracy and precision over multiple valuation dates. The results confirm that the value of a fairly homogenous real estate portfolio can be estimated sufficiently well using small samples and that the performance of the approach is reasonably robust regarding its temporal aspect. Our model is an efficient alternative to valuing real estate portfolios of significant size under tight temporal and financial restrictions. This paper extends previous research on sample-based valuation with regard to its temporal dimension.