Correlation Network of the Russian Stock Market in 2014–2021 and 2022–2026: Comparison and Stability of Centrality Estimates

This paper examines the transformation of the correlation-network structure of the Russian equity market after 2022 and evaluates the ro bustness of network centrality measures to data incompleteness in the presence of a structural shift. The study proposes an approach based on the separate analysis of two sub-periods using thresholded correlation networks, random-matrix spectral analysis, formal tests for scale-free behavior and core–periphery structure, and simulation experiments involving the removal of edges and nodes. The empirical analysis relies on daily stock-price data from the Moscow Exchange (MOEX ISS API) for two separately constructed samples: a baseline period from June 2014 to December 2021 (1,908 observations and 86 stocks) and a post-2022 period from April 2022 to June 2026 (1,177 observations and 83 stocks). The study aims to determine whether the structure of the correlation network changed after 2022 and to identify critical levels of data loss beyond which centrality estimates cease to be reliable. The contribution of the paper lies in the systematic comparison of random and targeted data removal in an actual financial market across two distinct sub-periods, together with formal tests of hypotheses concerning the existence of a dominant node and sectoral differentiation in the post-2022 period. The practical relevance of the findings lies in establishing quantitative benchmarks for statistical monitoring. The results show that rankings based on degree centrality remain robust when up to 40–60% of network edges are lost, whereas the identification of a unique node with the highest degree centrality becomes unreliable with losses of only 5–15%. These findings suggest that monitoring procedures should rely on group-level network characteristics and the composition of bridge nodes should be reassessed when the structure of the market changes.