Applying Cluster Analysis Methods to Assess the Heterogeneity of Russian Households by the Level and Structure of Wealth

Abstract. This paper is devoted to the application of cluster analysis to economic statistics. In particular, the heterogeneity of economic agents by the level and structure of wealth (i.e., their financial situation) is studied. The advantages and disadvantages of different clustering methods and the possibilities of their combined use are identified. The clustering methods applied include iterative k-means and Affinity Propagation methods, hierarchical Ward’s and complete link methods, and Density-Based Spatial Clustering of Applications with Noise (DBSCAN). Numerical modeling is carried out using Python (the scikit-learn library) based on available data from the Russian Household Survey on Consumer Finances, conducted by the Central Bank of the Russian Federation in 2022. (The survey was coordinated and implemented by Demoscope LLC, including field work.) The objects of this study––households––are divided into six relatively homogeneous groups (clusters) based on their financial situation. The clusters are characterized, and the potential impact of monetary policy on different household groups is analyzed.

Keywords: heterogeneity, household financial situation, unsupervised learning, clustering, k-means, Ward’s method, statistical surveys.