Probability of Default Models Development Approaches for Insurance Companies

The master’s work explored various models that predict the default likelihood of Russian insurance companies, using the logistic model of the binary choice and decision tree. Considered a wide range of models for calculating the probability of default for insurers in relation to the Russian reality on the basis of annual financial statements and summary reporting data on insurance activities for the period from 2012 to 2016. Also, binary factors of the type of insurance activity, the regional factor and the year are added to the features of research. In addition, the models were validated on 2017 data and defaults for the entire 2018. The relevance of the work is that the above models will help predict defaults of Russian insurance companies and, therefore, will be in demand both for the insurance sector and the regulator, and for consumers of insurance services .