Who Sits Where? — Automated Detection of Director Interlocks in Indian Companies

Prateek Sancheti, Kamalakar Karlapalem, Kavita Vemuri · 2026

Read Who Sits Where? — Automated Detection of Director Interlocks in Indian Companies by Prateek Sancheti, Kamalakar Karlapalem, Kavita Vemuri (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-22860.txt.

Introduction

Sancheti, Karlapalem, and Vemuri build a system to detect shared directors across Indian companies. Such interlocks can spread expertise, but they also raise conflict and compliance risks. Manual checks fail at this scale, with family-run groups adding complexity. Their graph framework uses breadth-first search over a large curated dataset. It covers tens of thousands of directors and companies and hundreds of thousands of links. Language models help describe personal and work ties. About three in five directors sit on two or more boards, they find.

Catalog record

CREATORPrateek Sancheti, Kamalakar Karlapalem, Kavita Vemuri
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE45642
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.22860

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Who Sits Where? — Automated Detection of Director Interlocks in Indian Companies Prateek Sancheti, Kamalakar Karlapalem, Kavita Vemuri Abstract Interlocking directorships—where individuals simultaneously serve on the boards of multiple corporations—can facilitate the exchange of expertise and strategic alignment but also present risks including conflicts of interest, economic ‘oligarchy’, and regulatory non-compliance. In contexts such as large, family-controlled corporate conglomerates in India, the manual detection of interlocks is hindered by the high volume of corporate entities and the complex involvement of extended familial networks. This study introduces a scalable, graph-theoretic framework for the systematic identification and analysis of interlocking directorships. Using Breadth-First Search (BFS) traversal, we examined a curated dataset comprising over 50,000 directors, 85,000 companies, and 300,000 director–company affiliations, yielding a comprehensive representation of corporate network structures. Large Language Models (LLMs) were integrated into the analytical pipeline to characterize both personal and professional linkages among directors. Empirical results indicate that 17% of directors hold positions in exactly two companies, while 58.6% maintain directorships in two or more companies. The combined BFS–LLM methodology enables the detection of recurrent director–company clusters, indicative of strong network cohesion, and provides qualitative insights into potential underlying drivers of these interlocks. The proposed approach enhances the capacity for automated, data-driven detection of complex intercorporate relationships, offering actionable implications for corporate governance, regulatory monitoring, and systemic risk assessment. Keywords Interlocking Directorships \cdot Indian Corporations \cdot LLMs \cdot BFS graph traversal \cdot Cliques 1 Introduction A corporation is a legal entity separate from its owners, capable of conducting business, owning assets, and being liable for its debts. These corporations are run by directors who are elected by shareholders to hold crucial responsibilities concerning the governance of the company, ensuring its operations align with shareholder interests. A corporation can have multiple individuals serving as directors, and similarly, an individual can hold directorial positions at multiple corporations, creating a many-to-many network structure. In India as per the Companies Act 2013, a minimum of 2 and max of 15 directors are allowed. 18 testing for corporate governance by the agency and resource dependency theories, found that larger board size resulted in positive performance. This leads to the question of the composition of the board of directors. In the complex and modern structure of corporate governance, the phenomenon of interlocking directorships (IDs) is a common yet nuanced aspect. An interlocking directorship occurs when an individual holds directorial positions across multiple corporations, creating a network of shared influence ( 24 ) . Such interlocks not only connect individuals with diverse expertise and vision across corporate boards, but also foster an intricate web of professional relationships that can have profound implications for business practices, policy formation, and strategic alliances. According to 7 , the five key factors behind the formation of board interlocks are: (1) corporation size, (2) degree of managerial control, (3) financial ties, (4) interaction with rivals, and (5) existence of local economic interests. Some studies also suggest that resource-dependence theories provide another explanation for interlocking ( 29 ; 28 ) . Early research on interlocking directorates and corporate networks dates back to the formative period of corporate capitalism in the early twentieth century ( 32 ; 7 ) . These studies focused on the reasons for the creation and maintenance of interlocks ( 28 ; 27 ; 35 ) , whereas more recent research has shifted attention to the consequences of such interlocks on business behavior and corporate ideology ( 24 ; 5 ; 13 ) . While more recent research has focused on impact on specific sectors like banking ( 3 ) , policies like green innovation ( 39 ) , gender inclusivity ( 31 ) and family involvement and IPO pricing ( 34 ) . The dual-edged nature of director interlocking raises significant concerns. On one hand, engaging visionary leaders and implementing strong corporate governance can promote the dissemination of innovative policies and allow for the flow of valuable information and resources. Interlocks can also help corporations gain a competitive edge by monitoring business ties and gathering confidential information, which leads to improved management practices ( 21 ; 26 ; 20 ; 23 ) and innovation ( 12 ) . On the other hand, the independence of board decisions might be compromised by shared directors ( 1 ) , harbor potential risks related to corporate malpractices or quid pro quo situations, including, but not limited to, corruption, collusion, the formation of shell companies, facilitation of hostile takeovers, and unfair business practices ( 14 ; 16 ) and can form an economic oligarchy. Transparency and awareness among investors, stakeholders, and regulatory bodies regarding such interconnected networks are crucial for maintaining the integrity of corporate governance. Motivated by the critical need for clearer insights into the structure of director interlocking, particularly within the context of Indian corporations, we present a pipeline to navigate through the complexities of these networks. We identify weakly and strongly connected components using Graph Cliques and Frequent Itemsets of directors and companies within these corporate networks, highlighting how a small set of directors can control many companies and how a small set of companies can be controlled by a large number of directors. Additionally, we identify relationships among connected directors by extracting director profiles from the web. These relationships are categorized as either personal, including family ties such as Daughter–Father or Wife–Husband, or professional, encompassing shared work experiences, professional memberships, or educational backgrounds. The objective is to provide nuanced information and model for an informed analysis of corporate structures. The contribution of the paper is the method to mine, retrieve and analyse large web data of corporates and extract the interlocks. Towards this we leverage a BFS graph traversal-based approach. We extracted a data corpus of over 50,000 directors, 85,000 companies, and 300,000 company-director connections. We analyze the corporate networks and present results revealing insights into the complex corporate structures of Indian corporations. 1.1 Cliques, Maximal Cliques, and Maximum Cliques In the context of graph theory, a Clique represents a specific subset of vertices within an undirected graph, characterized by complete adjacency among its members. A Maximal Clique is a clique that cannot be expanded by including any additional adjacent vertices. This definition implies that a maximal clique is not wholly contained within the vertex set of a larger clique. Essentially, it represents a subset of vertices that forms a complete graph and is as large as possible without being subsumed by a greater clique. This characteristic makes maximal cliques particularly significant in network analysis, as they help identify the largest connected groups. Mathematically and computationally, identifying all cliques of all maximal is a NP-Complete Problem ( 17 ) . To extract all the cliques in a graph G with vertices V , we need to identify all subsets and check if all the vertices in the subgraph are connected to each other. For a subgraph with n nodes, to be a clique needs \binom{n}{2} edges and therefore the time for extracting all maximal cliques in the graph G can be exponential time of the vertex number V ( 25 ) . Various approaches such as Pivot Bron-Kerbosch Algorithm and Tomita Algorithm have been studied to identify number of Cliques in a given graph. 1.2 Frequent Itemsets

Read the complete text