The Assessment of a Building’s insurable Value using Multivariate Statistics: The Case of the Czech Republic

Abstract

Abstract When concluding a property insurance agreement, adjustment of the insured amount poses a certain risk. From the policyholder’s point of view, the risk measure translates into the chosen target amount, which should correspond to the insurable value. The aim of the research is to determine a statistical model for prediction of the insurable value with using current models in the Czech Republic. The model for insurable value prediction proposed in this paper accepts the risk of decision making under uncertainty suitably. The model’s foundation is a synthesis of four core models discussing the addressed issue. The methodology is based on a classification tree created by the CART method, and multivariate linear regression. After the classification tree is created, the input variables which contributed to the classification are used in the regression model. The database consists of 125 family houses which went through a detailed examination (they were documented, measured, and their technical state and legal status were determined), and described in experts’ reports. The obtained results showed a high degree of statistical association of selected predictors with the estimated insurable value of property, as well as with the acceptable risk, and subsequently, a relatively low percentage of misclassified objects. The proposed multiple regression model proved to be statistically significant and can be used for objective estimations of insurable values free of insurance companies’ strategy. The designed methodology may be applied in other areas as well, for example, in decision-making processes at the population level in crisis situations.