Goal of research
A comprehensive analysis of approaches to aggregation of individual credit assessments using methods of expert information analysis and computational social choice.
Computational social choice theory, convex analysis, memetic algorithms and data visualization based on cutting-edge research papers. Numerical computations were carried out using the Matlab software.
Empirical base of research
Developed approach to aggregation is applied to long-term credit ratings assigned by seven credit rating agencies (CRAs) to Russian banks in national scales over the period from 2010 to 2015. Data on defaults of Russian banks is used in order to measure the discriminatory power of aggregated rating
Results of research
We characterize credit assessment of entities as a weak partial order of objects and determine aggregated ratings as a ranking which is most consistent with all individual credit assessments, i.e. aggregated rating is Kemeny median. As Kemeny median is generally multiple for weak partial orders we introduce a supplementary criterion and set lexicographic ordering optimization problem in order to obtain a unique solution. Solving this problem we obtain a Kemeny median which is optimal according to the supplementary criterion. As this problem is computationally hard we reformulate it into regularization problem and adopt memetic algorithm to find solution approximately with practical precision over reasonable time. Prior to applying our approach to real data, we carry out a simulation study in order to make sure that numerical solution is robust. The result of an aggregation of ratings assigned by seven credit rating agencies (CRAs) to Russian banks is highly consistent with individual ratings and demonstrates practical discriminatory power, therefore aggregated rating can be used for the purposes of credit analysis, mapping of credit ratings and benchmarking.