Ownership Concentration and Wealth Inequality in Russia

Kirill Polovnikov, Nikita Pospelov, Dmitriy SkougarevskiyIJCAI 2022 · DOI

TL;DR

This paper introduces a novel network-based algorithm to uncover hidden ultimate owners in complex corporate ownership structures, even those with cyclic ownership. Applied to Russia, it reveals significant wealth inequality, with the state holding the largest share of net assets, followed by individuals and foreign entities.

What Problem Does This Paper Address?

Understanding who ultimately owns companies is crucial for monitoring wealth inequality, combating corruption, and enforcing anti-money laundering regulations. However, corporate ownership structures are often intentionally complex, using intricate networks of intermediaries and cyclic ownership (where entities own each other in a loop) to obscure beneficial ownership. Traditional graph-based methods struggle with these cycles, leading to an incomplete picture of control. This paper tackles the challenge of accurately identifying ultimate owners and quantifying their asset holdings in large-scale, complex ownership networks, specifically focusing on Russia, a jurisdiction where cyclic ownership is prevalent.

Methods

The authors propose a network-based algorithm that leverages Katz centrality to overcome the limitations of cyclic ownership. Instead of directly calculating transitive shares (which can diverge in cyclic networks), they introduce a regularization parameter to ensure convergence. The core of their approach involves a multi-step process:

  1. 1,1-core decomposition: This step identifies the "core" of the network – the subgraph containing all cyclic ownership structures. By isolating this core, the computational complexity is drastically reduced, as the rest of the network (a forest of trees) can be analyzed more simply.
  2. Core-shell-forest decomposition: The core is further broken down into "in-shells," "out-shells," and the core itself, reflecting a "bow-tie" structure. Tree components (DAGs not connected to the core) are also identified.
  3. Transitive shares calculation: Transitive ownership matrices are computed for each component (shells, core, trees). For the core, the regularized Katz centrality formula is used. For shells and trees, finite series are sufficient due to their acyclic nature.
  4. Network-wide transitivity: The final transitive ownership matrix for the entire network is constructed by combining the matrices from the decomposed components, accounting for interactions between them.

This decomposition allows for an efficient calculation of ultimate ownership shares, even for millions of nodes.

Datasets

The primary dataset used is administrative data from Russia's Uniform State Register of Legal Entities (EGRUL), specifically a January 1, 2021 snapshot. This register contains legally binding information on equity ownership, including taxpayer identifiers, names, and countries of origin for individuals and organizations.

Additional data sources include:

  • The Forbes-200 list of wealthy individuals in Russia (2021 version) to identify ultra-high-net-worth individuals.
  • Authoritative country lists of tax havens (OECD, Hines Jr, Tørsløv et al.) to identify foreign entities operating from such jurisdictions.
  • Net asset data for non-financial firms in Russia as of 2018, sourced from the Russian Statistical Service.

Key Insights and Ideas

  • Katz centrality as a solution for cyclic ownership: The paper demonstrates how a modified Katz centrality measure can effectively handle the divergence issues arising from cyclic ownership in network analysis, providing a robust way to calculate transitive influence.
  • Decomposition for scalability: The core-shell-forest decomposition is a clever way to manage the computational complexity of analyzing massive ownership networks by breaking them into manageable, topologically distinct parts.
  • Russia as a case study for complex ownership: The choice of Russia, with its history of privatization and prevalent cyclic ownership, makes it an ideal and challenging testbed for their algorithm.
  • State dominance in Russian asset ownership: The findings highlight the significant role of the state in the Russian economy, holding the largest portion of net assets.
  • Hidden wealth in tax havens and foreign entities: The study quantifies the substantial ownership held by foreign entities, including those in tax havens, which is often obscured.

Key Results

  • The algorithm successfully identified ultimate owners in a network of 5,967,800 organizations and participants.
  • The distribution of asset ownership in Russia follows a power law, indicating strong wealth inequality with a Gini index of 0.93.
  • 51.7% of net assets of non-financial companies are ultimately held by the state and state-owned enterprises.
  • 25.0% of net assets are held by individuals (including 3.4% by Forbes-200 listed individuals).
  • 11.3% of net assets are owned by foreign entities, with 5.7% originating from tax havens.
  • The study provides Lorenz curves and Gini indices for different owner categories, illustrating the extent of inequality.

Limitations and Future Work

  • Data completeness: The authors acknowledge that some organizations may fail to submit their equity ownership data, or that financial data might be erroneous. They also note that shareholder equity in publicly traded companies is outside the scope of the register, meaning their estimates represent a lower bound of asset ownership.
  • Foreign ownership: The paper suggests that wealthy individuals might exercise control through foreign entities not fully captured in the data, which could explain some individuals appearing in lower wealth percentiles.
  • Future work: The authors' method is presented as applicable to any corporate network, suggesting its potential for analyzing ownership structures in other countries. Further exploration of the drivers and implications of the observed wealth distribution is also implied.

Summary updated

BibTeX
@inproceedings{polovnikov2022ownership,
  author = {Kirill Polovnikov and Nikita Pospelov and Dmitriy Skougarevskiy},
  title = {{Ownership Concentration and Wealth Inequality in Russia}},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year = {2022},
  doi = {10.24963/ijcai.2022/713},
}
Figure