## _wp15184

## Source details

**Canonical URL:** [_wp15184](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15184.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15184.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15184.pdf.json)

---

### I. INTRODUCTION
- GCC corporate, financial, and public sectors are highly interconnected through ownership links among industrial, commercial, financial groups, and sovereigns.
- Large industrial and commercial groups hold ownership stakes in banks; GCC governments—through different government agencies—have stakes in several groups; financial groups are also shareholders in banks, industrial, and commercial groups.
- Consequence: GCC banks may have exposures to a group of counterparties with dependencies that imply they are all likely to fail simultaneously.
- Basel Committee on Banking Supervision standards (April 2014) — supervisory framework for measuring and controlling large exposures effective from 1 January 2019:
  - Large exposures defined as the sum of all exposures to a counterparty or to a group of connected counterparties equal to or above 10 percent of the capital base (Tier 1 capital).
  - Connected counterparties defined as a group of counterparties linked by a control relationship (either direct or indirect) or an economic interdependence; exposures to all connected counterparties should be considered as a large single exposure for reporting and compliance purposes.
  - Committee set a limit of 25 percent of the bank’s capital base for the largest exposures and recommended banks report exposures equal to or greater than 10 percent of the capital base, all exemptions granted, and the 20 largest exposures irrespective of the value of these exposures relative to the capital base.
- Network analysis proposed as a tool to visualize connected counterparties through direct and indirect ownership and control links.
- Objective of the paper:
  - Assess ownership and control links in GCC countries and identify connected counterparties and their consolidated debt using ownership data available in Bloomberg and GCC stock exchanges.
  - Analyze ownership concentration in GCC countries and identify key stakeholders.
  - Consider effects of different definitions of control on the distribution of consolidated debt.
- Key previewed findings:
  - Corporate ownership is strongly concentrated in the GCC countries.
  - Public sector institutions are central in GCC corporate ownership networks; holding companies, financial institutions, and family groups are also important.
  - Cross-shareholdings are not widespread; GCC ownership networks have hierarchical structures with multiple lines of command.
  - Debt concentration is highest when the wedge between ownership and control is largest; highlighted with Zingales (1994) definition where the largest shareholders hold at least 5 percent of total shares.

### II. INTEGRATED OWNERSHIP AND CONTROL — CONCEPTS AND METHODS
- Ownership analysis based on input-output theory to identify direct and indirect links among shareholders and corporates.
- Key matrices and definitions:
  - A: n-squared adjacency matrix where element aij represents the percentage of the equity that shareholder i holds in corporate j (direct ownership).
    - Ultimate shareholders: columns with zeros and rows with percentages of total shares held in different corporates.
  - Y: n-squared integrated ownership matrix with direct and indirect ownership links; yij is percentage of total shares held directly and indirectly by shareholder i in corporate j.
    - Series representation: Y = A + A^2 + ... + A^N.
    - Limit representation: Y = (I − A)^−1 A.
  - Adjusted integrated ownership matrix V to address double-counting from cross-ownership:
    - V = (I − A)^{-1} Diag(I − Ā) A where Ā = [ājj] and ājj = Σ_i aij.
  - C: matrix of direct effective control; cij equals aij under one-share-one-vote proportional normalization unless adjusted by specific control rules.
- Control rules implemented:
  - Zingales (1994, 1995):
    - Control exercised by the largest shareholder with at least 5 percent of total shares.
    - cij = 1 iff aij = max_i (aij,...,anj) and aij > 5 percent for corporate j; otherwise 0.
  - La Porta, Lopez-de-Silanes, and Sheleifer (1999):
    - Largest shareholder must have at least 20 percent of total shares to exert control; if largest < 20 percent, company considered widely held.
    - cij = 1 iff aij = max_i (...) and aij > 20 percent for corporate j; otherwise 0.
  - Chapelle and Szafarz (2002, 2005) 50 percent majority threshold:
    - cij = 1 iff aij ≥ 50 percent for corporate j; cij = 0 if ∃ k ≠ i such that akj ≥ 50 percent; otherwise cij = aij.
    - Variant: requires 50 percent majority threshold; control proportional to shares if largest shareholders do not have more than 50 percent: cij follows normalization Σ_i aij = 1 iff aij < 50 percent; 0 otherwise.
  - Adjusted effective control matrix T analogous to V:
    - T = (I − C)^{-1} Diag(I − Ā) C.
- Consolidated debt of connected counterparties:
  - D: n-dimension vector of shareholder and corporate debt.
  - G: n-squared conditional identity matrix with diagonal element gjj = 1 iff Σ_i tij = 0 (column j in T associated with an ultimate shareholder), gjj = 0 iff Σ_i tij > 0 (column j associated with a corporate).
  - W: connected counterparties’ consolidated debt obtained by W = (T + G) D.
    - T D adds corporate debt under direct and indirect control of shareholders; G D adds ultimate shareholder debt.

### III. OWNERSHIP IN THE GCC CORPORATE SECTOR — DATA AND APPROACH
- Data sources:
  - Ownership data from Bloomberg based on disclosures by GCC stock exchanges and private sources.
  - GCC stock exchanges provide information on domestic and foreign shareholders with more than 5 percent of total shares in publicly listed companies; list names and percentage of shares directly held by shareholders.
  - Some privately-owned corporates disclose information on their ownership structure.
- Analytical approach:
  - Construct direct ownership matrix A from disclosed shareholder percentages.
  - Apply integrated ownership and control frameworks (matrices Y, V, C, T) to assess ownership structure, interconnectedness, control links, and consolidated debt of connected counterparties.

### Network representation of shareholders and corporates
- Network graphs map rows and columns into nodes and links:
  - Nodes represent shareholders or corporates; directed arrows represent ownership relationships (tail = shareholder, head = corporate).
  - Arrow thickness proportional to percentage of shares held (example: a 29 percent ownership stake represented by a scaled arrow proportional to 29 percent).
  - Successive layers show direct and indirect ownership; ultimate ownership and control determined by direct and indirect links.
  - Shareholder portfolios: direct shares occupy the first layer in a pyramid; indirect ownership occupies lower layers.

### Data sources, gaps, and caveats
- Important gaps and caveats in analysis based on Bloomberg, GCC stock exchanges, and private sources:
  - Many privately owned corporates with no disclosed ownership information are unaccounted.
  - Ownership information on publicly listed companies might not contain details on ultimate shareholders (holding companies may hide ultimate shareholders if their shares are not publicly listed).
  - Inconsistent and imprecise registration of shareholder and corporate names adds uncertainty (example: the Arabic article Al can be part of a name in some records and absent in others).
  - Text mining required to reduce names to their roots and compare them.
- Ownership restrictions influencing emphasis on domestic shareholders (as stated):
  - Bahrain: Subject to government approval;
  - Kuwait: Foreign investors can own up to 49 percent of Kuwaiti companies;
  - Oman: Foreign ownership of shares in Omani companies is limited to 70 percent;
  - Qatar: Foreign investors cant own up to 49 percent of total shares;
  - Saudi Arabia: Direct ownership of shares by Qualified Foreign Investors up to 49 percent has been allowed since June 15, 2015;
  - United Arab Emirates: Foreign investors are limited to 49 percent of total shares.
- Note: Most countries do not impose ownership restrictions on GCC nationals; most countries allow exceptions in the foreign ownership restrictions for investments in specific sectors.

### Ownership concentration and key stakeholders
- Ownership concentrated in:
  - public sector institutions,
  - holding companies,
  - financial institutions, and
  - family groups.
- Public sector institutions central to ownership networks:
  - Social Insurance Organization (Bahrain);
  - Kuwait Investment Authority (Kuwait);
  - government and ministry of finance (Oman);
  - Qatar Investment Authority and Qatar Holdings (Qatar);
  - General Organization of Social Insurance, the General Retirement Organization, and the Public Investment Fund (Saudi Arabia);
  - Dubai government, Mubadala Development Company, and Sharjah Asset Management (United Arab Emirates).
- Examples of holding companies and groups:
  - Mumtalakat Holding (Bahrain);
  - Dholfar International Development and Investment Holding Company (Oman);
  - El Sharq Holdings, Gembal Holdings, and National Industries Group Holdings (Kuwait);
  - Masik Holding Company and Saudi Basic Industries Corporation (Saudi Arabia).
- Financial institutions and family groups noted:
  - National Bank of Bahrain;
  - Kuwait Finance House;
  - Abdul Qadir and Sons Company (Saudi Arabia).

### Network metrics and structure (selected statistics)
- Summary observations:
  - "The average number of links (the average degree) in GCC ownership networks is about one—implying that the average investor owns only one company (a low ownership concentration)—there are few nodes with a high number of links (degree)."
  - Public sector institutions are centers of networks with the highest number of links; holding companies, financial institutions, and family groups also have high degrees.
  - Figure 7 (Lorenz curve and implied Gini coefficient) portrays strong ownership concentration in GCC countries.
  - Table 2 indicates the median of shareholder blocks holds more than 26 percent of total shares in GCC countries.
- Selected exact numeric network statistics (from Table 1):
  - Bahrain: Number of Nodes 114, Number of Links 109, Average Degree 1.0, Average Distance (Avg. Path Length) 1.4, Krackhardt Hierarchy Score 1.0, Krackhardt Connectedness Score 0.4, Krackhardt Efficiency Score 1.0, Krackhardt LUBness Score 0.1
  - Kuwait: Number of Nodes 466, Number of Links 466, Average Degree 1.0, Average Distance (Avg. Path Length) 1.4, Krackhardt Hierarchy Score 1.0, Krackhardt Connectedness Score 0.4, Krackhardt Efficiency Score 1.0, Krackhardt LUBness Score 0.0
  - Oman: Number of Nodes 332, Number of Links 321, Average Degree 1.0, Average Distance (Avg. Path Length) 1.4, Krackhardt Hierarchy Score 1.0, Krackhardt Connectedness Score 0.5, Krackhardt Efficiency Score 1.0, Krackhardt LUBness Score 0.0
  - Qatar: Number of Nodes 312, Number of Links 220, Average Degree 0.7, Average Distance (Avg. Path Length) 1.0, Krackhardt Hierarchy Score 1.0, Krackhardt Connectedness Score 0.1, Krackhardt Efficiency Score 1.2, Krackhardt LUBness Score 0.3
  - Saudi Arabia: Number of Nodes 349, Number of Links 311, Average Degree 0.9, Average Distance (Avg. Path Length) 1.1, Krackhardt Hierarchy Score 1.0, Krackhardt Connectedness Score 0.1, Krackhardt Efficiency Score 1.0, Krackhardt LUBness Score 0.1
  - United Arab Emirates: Number of Nodes 347, Number of Links 312, Average Degree 0.9, Average Distance (Avg. Path Length) 1.1, Krackhardt Hierarchy Score 1.0, Krackhardt Connectedness Score 0.1, Krackhardt Efficiency Score NANA, Krackhardt LUBness Score (not provided)
- Structural interpretation:
  - Few cross-ownership links; networks resemble hierarchical structures with multiple lines of command.
  - Krackhardt measures:
    - Low (close to zero) Krackhardt connectedness scores imply weakly-connected structures overall.
    - High (close to unity) Krackhardt hierarchy scores imply cross-ownerships are non-existent.
    - Low (close to zero) Krackhardt least upper boundedness scores imply multiple lines of command.
  - Some shareholders have high eccentricity indicating long chains of consecutive ownership and control links.
  - A few corporates (example: Investment Corporation of Dubai) act as hubs with higher inner and outer degrees.

### Ownership blocks and international comparison (selected exact values)
- Median, Min, Max, and Average Block by Firm (selected exact figures):
  - Australia: Median Block 7.2, Min Block 5.4, Max Block 50.2, Average Block by Firm 9.9
  - Canada: Median Block 67.9, Min Block 5.0, Max Block 78.0, Average Block by Firm 21.7
  - France: Median Block 22.6, Min Block 5.0, Max Block 87.3, Average Block by Firm 21.5
  - Germany: Median Block 17.7, Min Block 5.8, Max Block 99.1, Average Block by Firm 23.6
  - Hong Kong: Median Block 30.0, Min Block 5.0, Max Block 84.6, Average Block by Firm 56.9
  - India: Median Block 15.6, Min Block 5.1, Max Block 89.5, Average Block by Firm 45.8
  - Indonesia: Median Block 50.5, Min Block 5.1, Max Block 99.1, Average Block by Firm 62.9
  - Malaysia: Median Block 14.8, Min Block 5.2, Max Block 64.6, Average Block by Firm 57.6
  - New Zealand: Median Block 12.7, Min Block 5.0, Max Block 83.5, Average Block by Firm 48.4
  - Norway: Median Block 16.6, Min Block 5.0, Max Block 79.7, Average Block by Firm 47.2
  - Singapore: Median Block 15.2, Min Block 5.0, Max Block 82.3, Average Block by Firm 41.1
  - South Africa: Median Block 12.7, Min Block 5.3, Max Block 74.5, Average Block by Firm 38.1
  - South Korea: Median Block 8.7, Min Block 5.0, Max Block 78.0, Average Block by Firm 38.2
  - Turkey: Median Block 24.6, Min Block 5.0, Max Block 93.3, Average Block by Firm 67.9
  - UK: Median Block 17.6, Min Block 5.6, Max Block 33.1, Average Block by Firm 7.3
  - USA: Median Block 26.9, Min Block 10.6, Max Block 40.7, Average Block by Firm 7.2
  - GCC countries (selected):
    - Bahrain: Median Block 43.9, Min Block 5.7, Max Block 90.2, Average Block by Firm 41.7
    - Kuwait: Median Block 44.3, Min Block 5.6, Max Block 95.8, Average Block by Firm 46.0
    - Oman: Median Block 55.0, Min Block 10.2, Max Block 98.1, Average Block by Firm 53.0
    - Qatar: Median Block 41.0, Min Block 8.3, Max Block 64.2, Average Block by Firm 40.0
    - Saudi Arabia: Median Block 26.2, Min Block 5.0, Max Block 95.0, Average Block by Firm 31.7
    - United Arab Emirates: Median Block 47.1, Min Block 5.0, Max Block 99.9, Average Block by Firm 47.4
- Interpretation: "The median of shareholder blocks holds more than 26 percent of total shares in GCC countries." Ownership in GCC countries is generally more concentrated than most selected countries.

### Control, corporate governance, and related-party links
- One-share-one-vote rule in GCC countries implies large ownership stakes associate with rights to appoint board members and control firms.
- Mechanisms creating a wedge between ownership and control include dual-class stock, voting agreements, pyramids, and disproportional board representation; effective control can exceed direct ownership when pyramids and indirect links exist.
- Corporate governance observations:
  - "54 percent of MENA banks do not have independent directors," implying limited use of independent non-executive directors.
  - "Only 20 percent of the companies in the S&P/Hawkamah Pan Arab ESG index...conducted their related party transactions on market terms."
- Board interlocks and related parties:
  - GCC Board of Directors Institute (2011) reported that nine percent of GCC board members sat on more than 4 boards.
  - Average board size in the MENA region is 8.9 directors.
  - Related parties defined to include entities exerting control, major shareholders, board members, senior management, close family members, and corresponding persons in affiliated companies (per Basel core principles).

### Debt concentration, connected counterparties, and consolidation (exact Herfindahl indices)
- Basel supervisory guidance:
  - Connected counterparties defined broadly; exposures to connected counterparties should be consolidated as a single exposure.
  - Banks recommended to report large exposures equal or above 10 percent of the capital base (Tier 1 capital), all exemptions granted, and the 20 largest exposures irrespective of value relative to capital base.
- Herfindahl indices for connected counterparties’ consolidated debt (End-2013) under five consolidation rules:
  - Bahrain: No Consolidation 0.1725, Zingales (1994) 0.1297, La Porta et all (2002) 0.1911, Modified Chapelle and Szafarz (2002) 0.1888, Integrated Ownership 0.1297
  - Kuwait: No Consolidation 0.0769, Zingales (1994) 0.1006, La Porta et all (2002) 0.1235, Modified Chapelle and Szafarz (2002) 0.1084, Integrated Ownership 0.1019
  - Oman: No Consolidation 0.1302, Zingales (1994) 0.1485, La Porta et all (2002) 0.1552, Modified Chapelle and Szafarz (2002) 0.1351, Integrated Ownership 0.1486
  - Qatar: No Consolidation 0.1789, Zingales (1994) 0.2027, La Porta et all (2002) 0.2089, Modified Chapelle and Szafarz (2002) 0.1981, Integrated Ownership 0.2054
  - Saudi Arabia: No Consolidation 0.0800, Zingales (1994) 0.1447, La Porta et all (2002) 0.1804, Modified Chapelle and Szafarz (2002) 0.1646, Integrated Ownership 0.1620
  - UAE: No Consolidation 0.1126, Zingales (1994) 0.1905, La Porta et all (2002) 0.2007, Modified Chapelle and Szafarz (2002) 0.1866, Integrated Ownership 0.1905
- Interpretation:
  - Herfindahl values 0.15 to 0.25 indicate moderate concentration; values above 0.25 indicate high concentration.
  - Debt amounts based on long- and short-term debt disclosed in 2013 annual reports and syndicated loans and bonds tracked by Bloomberg.
  - Debt concentration is largest when control is associated with low ownership thresholds (e.g., Zingales (1994) at 5 percent).
  - Herfindahl index is larger for connected counterparties’ consolidated debt than for non-consolidated corporate debt under any definition of control, underscoring importance of consolidation.
  - Largest Herfindahl index often obtained under Zingales (1994) definition (lower thresholds imply more debt consolidation).
  - Under Zingales (1994) the debt distribution shifts to the right relative to individual corporate debt distribution — a lower peak but a fatter tail — indicating a larger number of highly indebted borrowers.

### Conclusions and supervisory relevance
- Main conclusions:
  - (i) Corporate ownership is strongly concentrated in the GCC countries.
  - (ii) Public sector institutions are at the center of GCC corporate ownership networks; holding companies, financial institutions, and family groups are also important shareholders.
  - (iii) Cross-shareholdings are not widespread; ownership networks have hierarchical structures with multiple lines of command.
- Integrated ownership by ultimate shareholders determined by both direct and indirect ownership links; effective ownership can exceed direct ownership and creates a wedge between ownership and control.
- Debt concentration is maximized when the wedge between ownership and control is largest (example: Zingales (1994) with at least 5 percent).
- Supervisory implication tied to Basel framework:
  - Bank exposures to connected counterparties arising from ownership and control links should be consolidated at the level of the controlling shareholders for reporting and compliance under the Basel supervisory framework.

*Source: _wp15184 - Staff calculations, Bloomberg, GCC stock exchanges, and private sources (extracted content from the provided PDF).*

### References .............................................................................................................

### _wp15184 - References .............................................................................................................

### I. INTRODUCTION
- The GCC corporate, financial, and public sectors are characterized by a high degree of interconnectedness through ownership links among industrial, commercial, financial groups, and sovereigns.
- Large industrial and commercial groups have ownership stakes in banks; GCC governments—through different government agencies—have stakes in several groups; financial groups are also shareholders in banks, industrial, and commercial groups.
- Consequence: GCC banks may have exposures to a group of counterparties with dependencies that imply they are all likely to fail simultaneously.
- Basel Committee on Banking Supervision standards (April 2014) set a supervisory framework for measuring and controlling large exposures effective from 1 January 2019:
  - Large exposures are defined as the sum of all exposures to a counterparty or to a group of connected counterparties equal to or above 10 percent of the capital base (Tier 1 capital).
  - The framework further defines connected counterparties as a group of counterparties linked by a control relationship (either direct or indirect) or an economic interdependence; exposures to all connected counterparties should be considered as a large single exposure for reporting and compliance purposes.
  - The Basel committee has also set a limit of 25 percent of the bank’s capital base for the largest exposures and has recommended that banks should report to the supervisor exposures equal to or greater than 10 percent of the capital base, all the exemptions granted, and the 20 largest exposures irrespective of the value of these exposures relative to the capital base.
- Network analysis is presented as a useful way to visualize connected counterparties through direct and indirect ownership and control links.
- Objective of the paper:
  - Assess ownership and control links in GCC countries and identify connected counterparties and their consolidated debt.
  - Focus on integrated ownership and network arising from ownership data available in Bloomberg and GCC stock exchanges.
  - Analyze ownership concentration in GCC countries and identify key stakeholders.
  - Consider the effect of different definitions of control on the distribution of consolidated debt.
- Key findings previewed:
  - Corporate ownership is strongly concentrated in the GCC countries.
  - Public sector institutions are at the center of GCC corporate ownership networks; holding companies, financial institutions, and family groups are also important.
  - Cross-shareholdings are not widespread; GCC ownership networks have hierarchical structures with multiple lines of command.
  - Debt concentration is highest when the wedge between ownership and control is largest; highlighted with Zingales (1994) definition where the largest shareholders hold at least 5 percent of total shares.

### II. INTEGRATED OWNERSHIP AND CONTROL — CONCEPTS AND METHODS
- Ownership analysis relies on input-output theory to identify direct and indirect links among shareholders and corporates (Ellerman (1991) and extensions).
- Notation and matrices:
  - A: n-squared matrix where element aij represents the percentage of the equity that shareholder i holds in corporate j (direct ownership / adjacency matrix).
    - Ultimate shareholders: columns with zeros and rows with percentages of total shares held in different corporates.
    - Corporates: columns with percentages of total shares held by different shareholders; rows with zeros if they have no investments in other corporates; in cross-ownership, corporates have rows with percentages held in other corporates.
  - Y: n-squared integrated ownership matrix with direct and indirect ownership links; element yij is percentage of total shares held directly and indirectly by shareholder i in corporate j.
    - Series representation: Y = A + A^2 + ... + A^N (equation form shown).
    - Limit as N → ∞ gives Y = (I − A)^−1 A (equation 2’ form).
  - Double-counting from cross-ownership addressed by adjusted integrated ownership matrix V:
    - V = (I − A)^{-1} Diag(I − Ā) A where Ā = [ājj] and ājj = Σ_i aij (sum of shares in corporate j held by shareholders and corporates in the network).
    - Diag(I − Ā) includes shares held by outsiders not in direct ownership matrix A.
- Control definitions and matrices:
  - C: matrix of direct effective control; cij under one-share-one-vote proportional to aij subject to normalization conditions.
  - Zingales (1994, 1995) control rule:
    - Control exercised by the largest shareholder with at least 5 percent of total shares.
    - cij = 1 iff aij = max_i (aij,...,anj) and aij > 5 percent for corporate j; otherwise 0.
  - La Porta, Lopez-de-Silanes, and Sheleifer (1999) control rule:
    - Largest shareholder should have at least 20 percent of total shares to exert control; if largest < 20 percent, company considered widely held.
    - cij = 1 iff aij = max_i (...) and aij > 20 percent for corporate j; otherwise 0.
  - Chapelle and Szafarz (2002, 2005) 50 percent majority threshold:
    - cij = 1 iff aij ≥ 50 percent for corporate j; cij = 0 if ∃ k ≠ i such that akj ≥ 50 percent; otherwise cij = aij.
    - Variant: requires 50 percent majority threshold; control proportional to shares if largest shareholders do not have more than 50 percent: cijfollows normalization Σ_i aij = 1 iff aij < 50 percent; 0 otherwise.
  - Adjusted effective control matrix T analogous to V: T = (I − C)^{-1} Diag(I − Ā) C (equation 11 form).
- Consolidated debt of connected counterparties:
  - D: n-dimension vector of shareholder and corporate debt.
  - G: n-squared conditional identity matrix with diagonal element gjj = 1 iff Σ_i tij = 0 (column j in T associated with an ultimate shareholder), gjj = 0 iff Σ_i tij > 0 (column j associated with a corporate).
  - W: n-dimension vector of connected counterparties’ consolidated debt obtained by W = (T + G) D (equation 13).
    - First term T D adds corporate debt under direct and indirect control of shareholders; second term G D adds ultimate shareholder debt.

### III. OWNERSHIP IN THE GCC CORPORATE SECTOR — DATA AND APPROACH
- Data sources:
  - Ownership data from Bloomberg based on disclosures by GCC stock exchanges and private sources.
  - GCC stock exchanges provide information on domestic and foreign shareholders with more than 5 percent of total shares in publicly listed companies; list names and percentage of shares directly held by shareholders.
  - Some privately-owned corporates disclose information on their ownership structure.
- Analytical approach:
  - Use direct ownership matrix A (adjacency matrix) constructed from disclosed shareholder percentages.
  - Apply integrated ownership and control frameworks (matrices Y, V, C, T) to assess ownership structure, interconnectedness, control links, and consolidated debt of connected counterparties.
- Paper organization (as structured in source):
  - Section II discusses integrated ownership and corporate control (presented above).
  - Section III analyzes ownership networks in GCC countries.
  - Section IV discusses impact of different definitions of control on consolidated debt of connected counterparties.
  - Final section discusses main findings.

*Source: _wp15184 - References (extracted content from provided PDF).*

### section II.B contains information on shareholders and corporates displayed in rows and

### _wp15184 - section II.B contains information on shareholders and corporates displayed in rows and

### Network representation of shareholders and corporates
- Network graphs represent the same information contained in rows and columns by using nodes and links.
- Nodes represent either shareholders or corporates; arrows (links or edges) represent ownership relationships.
- Shareholders are located on the tail side of the arrow; corporates are placed on the head side.
- Arrow thickness is proportional to the percentage of shares held by shareholders.
  - Example: "the 29 percent ownership stake of the Dubai government in Emaar Properties would be represented by a scaled arrow—proportional to the 29 percent share—starting at the node named Dubai government and pointing to the node named Emaar Properties."
- Successive layers link shareholders and corporates to show direct and indirect ownership; ultimate ownership and control are determined by direct and indirect links in the ownership network.
- Shareholder portfolios: direct shares in corporates occupy the first layer in a pyramid; indirect ownership is given by lower layers.

### Data sources, gaps, and caveats
- Analysis based on information available in Bloomberg, GCC stock exchanges, and private sources has important gaps:
  - A large number of privately owned corporates with no disclosed ownership information is left unaccounted.
  - Ownership information on publicly listed companies might not contain details on the ultimate shareholder (holding companies may hide ultimate shareholders if their shares are not publicly listed).
  - Inconsistent and imprecise registration of shareholder and corporate names adds uncertainty to the network (example: the Arabic article Al can be part of a name in some records and absent in others).
  - Text mining is required to reduce names to their roots and compare them.
- Ownership restrictions in GCC countries influence emphasis on domestic shareholders:
  - Bahrain: Subject to government approval;
  - Kuwait: Foreign investors can own up to 49 percent of Kuwaiti companies;
  - Oman: Foreign ownership of shares in Omani companies is limited to 70 percent;
  - Qatar: Foreign investors cant own up to 49 percent of total shares;
  - Saudi Arabia: Direct ownership of shares by Qualified Foreign Investors up to 49 percent has been allowed since June 15, 2015;
  - United Arab Emirates: Foreign investors are limited to 49 percent of total shares.
- Note: "Most countries do not impose ownership restrictions on GCC nationals. Most countries also allow exceptions in the foreign ownership restrictions for investments in specific sectors."

### Ownership concentration and key stakeholders
- Ownership in GCC countries is concentrated in:
  - public sector institutions,
  - holding companies,
  - financial institutions, and
  - family groups.
- Examples of public sector institutions central to ownership networks:
  - Social Insurance Organization (Bahrain);
  - Kuwait Investment Authority (Kuwait);
  - government and ministry of finance (Oman);
  - Qatar Investment Authority and Qatar Holdings (Qatar);
  - General Organization of Social Insurance, the General Retirement Organization, and the Public Investment Fund (Saudi Arabia);
  - Dubai government, Mubadala Development Company, and Sharjah Asset Management (United Arab Emirates).
- Examples of holding companies and groups mentioned:
  - Mumtalakat Holding (Bahrain);
  - Dholfar International Development and Investment Holding Company (Oman);
  - El Sharq Holdings, Gembal Holdings, and National Industries Group Holdings (Kuwait);
  - Masik Holding Company and Saudi Basic Industries Corporation (Saudi Arabia).
- Financial institutions and family groups noted as important:
  - National Bank of Bahrain;
  - Kuwait Finance House;
  - Abdul Qadir and Sons Company (Saudi Arabia).

### Network metrics and structure (summary of Table 1 and related findings)
- Table 1 summary observations:
  - "the average number of links (the average degree) in GCC ownership networks is about one—implying that the average investor owns only one company (a low ownership concentration)—there are few nodes with a high number of links (degree)."
  - Public sector institutions are at the center of networks with the highest number of links; holding companies, financial institutions, and family groups also have high degrees.
  - Figure 7 (Lorenz curve and implied Gini coefficient) portrays strong ownership concentration in GCC countries.
  - Table 2 indicates that the median of shareholder blocks holds more than 26 percent of total shares in GCC countries.
- Selected numeric network statistics (from Table 1):
  - Bahrain: Number of Nodes 114, Number of Links 109, Average Degree 1.0, Average Distance (Avg. Path Length) 1.4, Krackhardt Hierarchy Score 1.0, Krackhardt Connectedness Score 0.4, Krackhardt Efficiency Score 1.0, Krackhardt LUBness Score 0.1
  - Kuwait: Number of Nodes 466, Number of Links 466, Average Degree 1.0, Average Distance (Avg. Path Length) 1.4, Krackhardt Hierarchy Score 1.0, Krackhardt Connectedness Score 0.4, Krackhardt Efficiency Score 1.0, Krackhardt LUBness Score 0.0
  - Oman: Number of Nodes 332, Number of Links 321, Average Degree 1.0, Average Distance (Avg. Path Length) 1.4, Krackhardt Hierarchy Score 1.0, Krackhardt Connectedness Score 0.5, Krackhardt Efficiency Score 1.0, Krackhardt LUBness Score 0.0
  - Qatar: Number of Nodes 312, Number of Links 220, Average Degree 0.7, Average Distance (Avg. Path Length) 1.0, Krackhardt Hierarchy Score 1.0, Krackhardt Connectedness Score 0.1, Krackhardt Efficiency Score 1.2, Krackhardt LUBness Score 0.3
  - Saudi Arabia: Number of Nodes 349, Number of Links 311, Average Degree 0.9, Average Distance (Avg. Path Length) 1.1, Krackhardt Hierarchy Score 1.0, Krackhardt Connectedness Score 0.1, Krackhardt Efficiency Score 1.0, Krackhardt LUBness Score 0.1
  - United Arab Emirates: Number of Nodes 347, Number of Links 312, Average Degree 0.9, Average Distance (Avg. Path Length) 1.1, Krackhardt Hierarchy Score 1.0, Krackhardt Connectedness Score 0.1, Krackhardt Efficiency Score NANA, Krackhardt LUBness Score (not provided)
- Structural interpretation:
  - Few cross-ownership links; networks are similar to hierarchical structures with multiple lines of command.
  - Krackhardt measures indicate:
    - Low (close to zero) Krackhardt connected scores implying weakly-connected structures overall.
    - High (close to unity) Krackhardt hierarchy scores implying cross-ownerships are non-existent.
    - Low (close to zero) Krackhardt least upper boundedness scores implying multiple lines of command.
  - Some shareholders have high eccentricity indicating long chains of consecutive ownership and control links.
  - A few corporates (example: Investment Corporation of Dubai) act as hubs with higher inner and outer degrees.

### Ownership blocks and international comparison (Table 2 excerpts)
- Median, Min, Max, and Average Block by Firm statistics (selected exact values preserved as presented):
  - Australia: Median Block 7.2, Min Block 5.4, Max Block 50.2, Average Block by Firm 9.9
  - Canada: Median Block 67.9, Min Block 5.0, Max Block 78.0, Average Block by Firm 21.7
  - France: Median Block 22.6, Min Block 5.0, Max Block 87.3, Average Block by Firm 21.5
  - Germany: Median Block 17.7, Min Block 5.8, Max Block 99.1, Average Block by Firm 23.6
  - Hong Kong: Median Block 30.0, Min Block 5.0, Max Block 84.6, Average Block by Firm 56.9
  - India: Median Block 15.6, Min Block 5.1, Max Block 89.5, Average Block by Firm 45.8
  - Indonesia: Median Block 50.5, Min Block 5.1, Max Block 99.1, Average Block by Firm 62.9
  - Malaysia: Median Block 14.8, Min Block 5.2, Max Block 64.6, Average Block by Firm 57.6
  - New Zealand: Median Block 12.7, Min Block 5.0, Max Block 83.5, Average Block by Firm 48.4
  - Norway: Median Block 16.6, Min Block 5.0, Max Block 79.7, Average Block by Firm 47.2
  - Singapore: Median Block 15.2, Min Block 5.0, Max Block 82.3, Average Block by Firm 41.1
  - South Africa: Median Block 12.7, Min Block 5.3, Max Block 74.5, Average Block by Firm 38.1
  - South Korea: Median Block 8.7, Min Block 5.0, Max Block 78.0, Average Block by Firm 38.2
  - Turkey: Median Block 24.6, Min Block 5.0, Max Block 93.3, Average Block by Firm 67.9
  - UK: Median Block 17.6, Min Block 5.6, Max Block 33.1, Average Block by Firm 7.3
  - USA: Median Block 26.9, Min Block 10.6, Max Block 40.7, Average Block by Firm 7.2
  - GCC countries (selected):
    - Bahrain: Median Block 43.9, Min Block 5.7, Max Block 90.2, Average Block by Firm 41.7
    - Kuwait: Median Block 44.3, Min Block 5.6, Max Block 95.8, Average Block by Firm 46.0
    - Oman: Median Block 55.0, Min Block 10.2, Max Block 98.1, Average Block by Firm 53.0
    - Qatar: Median Block 41.0, Min Block 8.3, Max Block 64.2, Average Block by Firm 40.0
    - Saudi Arabia: Median Block 26.2, Min Block 5.0, Max Block 95.0, Average Block by Firm 31.7
    - United Arab Emirates: Median Block 47.1, Min Block 5.0, Max Block 99.9, Average Block by Firm 47.4
- Interpretation: "The median of shareholder blocks holds more than 26 percent of total shares in GCC countries." "Generally, ownership in GCC countries is more concentrated than most selected countries."

### Control, corporate governance, and related-party links
- One-share-one-vote rule in GCC countries implies large ownership stakes associate with rights to appoint board members and control firms.
- Mechanisms that create a wedge between ownership and control include dual-class stock, voting agreements, pyramids, and disproportional board representation.
- Effective control can be larger than direct ownership when pyramids and indirect links exist.
- Corporate governance implications:
  - "54 percent of MENA banks do not have independent directors," implying limited use of independent non-executive directors (OECD et al. (2009)).
  - Minority shareholders can be left unprotected with few independent non-executive directors to monitor conflicts of interest and related-party transactions.
  - "Only 20 percent of the companies in the S&P/Hawkamah Pan Arab ESG index...conducted their related party transactions on market terms." (Aaltonen (2013))
- Board interlocks and related parties:
  - GCC Board of Directors Institute (2011) reported that nine percent of GCC board members sat on more than 4 boards.
  - Average size of board members in the MENA region is 8.9 directors (Aaltonen (2013)); close inter-links can arise between corporates, banks, and public institutions via related-party connections.
  - Definition of related parties incorporates entities exerting control, major shareholders, board members, senior management, close family members, and corresponding persons in affiliated companies (per Basel core principles).

### Debt concentration, connected counterparties, and consolidation (Table 3 and Figure 8)
- Basel definitions and supervisory guidance:
  - Connected counterparties defined as a group of counterparties linked by a control relationship (direct or indirect) or economic interdependence.
  - For connected counterparties, exposures to all connected counterparties should be considered as a single large exposure.
  - Basel committee recommends banks report large exposures equal or above 10 percent of the capital base (Tier 1 capital), all exemptions granted, and the 20 largest exposures irrespective of value relative to capital base.
- Consolidation approach: Table 3 contains Herfindahl indices for connected counterparties’ consolidated debt obtained under four definitions of control detailed in section II.B:
  - No Consolidation,
  - Zingales (1994) — largest controlling shareholder with at least 5 percent of total shares,
  - La Porta, Lopez-de-Silanes, and Sheleifer (1999) — at least 20 percent,
  - Modified Chapelle and Szafarz (2002) — at least 50 percent,
  - Integrated Ownership (proportional to the shares in the integrated ownership for the largest shareholders, equation 6 in section II.B).
- Interpretation of Herfindahl index in this context:
  - Values ranging from 0.15 to 0.25 indicate moderate concentration; values above 0.25 indicate high concentration.
  - Debt amounts based on long- and short-term debt disclosed in 2013 annual reports and syndicated loans and bonds tracked by Bloomberg.
- Table 3 exact Herfindahl index values (End-2013):
  - Bahrain: No Consolidation 0.1725, Zingales (1994) 0.1297, La Porta et all (2002) 0.1911, Modified Chapelle and Szafarz (2002) 0.1888, Integrated Ownership 0.1297
  - Kuwait: No Consolidation 0.0769, Zingales (1994) 0.1006, La Porta et all (2002) 0.1235, Modified Chapelle and Szafarz (2002) 0.1084, Integrated Ownership 0.1019
  - Oman: No Consolidation 0.1302, Zingales (1994) 0.1485, La Porta et all (2002) 0.1552, Modified Chapelle and Szafarz (2002) 0.1351, Integrated Ownership 0.1486
  - Qatar: No Consolidation 0.1789, Zingales (1994) 0.2027, La Porta et all (2002) 0.2089, Modified Chapelle and Szafarz (2002) 0.1981, Integrated Ownership 0.2054
  - Saudi Arabia: No Consolidation 0.0800, Zingales (1994) 0.1447, La Porta et all (2002) 0.1804, Modified Chapelle and Szafarz (2002) 0.1646, Integrated Ownership 0.1620
  - UAE: No Consolidation 0.1126, Zingales (1994) 0.1905, La Porta et all (2002) 0.2007, Modified Chapelle and Szafarz (2002) 0.1866, Integrated Ownership 0.1905
- Key conclusions from Table 3 and Figure 8:
  - Debt concentration is largest when control can be associated with low levels of ownership (e.g., Zingales (1994) at 5 percent).
  - Herfindahl index is larger for connected counterparties’ consolidated debt than for non-consolidated corporate debt under any definition of control, emphasizing the importance of consolidation.
  - The largest Herfindahl index is obtained under the Zingales (1994) definition of control (lower thresholds imply more debt consolidation).
  - Figure 8 shows that under the Zingales (1994) definition the debt distribution shifts to the right, implying a lower peak but a fatter tail than individual corporate debt distribution — a larger number of highly indebted borrowers under the Zingales (1994) definition.

### Conclusions and supervisory relevance
- Main conclusions summarized:
  - (i) Corporate ownership is strongly concentrated in the GCC countries.
  - (ii) Public sector institutions are at the center of GCC corporate ownership networks; holding companies, financial institutions, and family groups are also important shareholders.
  - (iii) Cross-shareholdings are not widespread; ownership networks have hierarchical structures with multiple lines of command.
- Integrated ownership by ultimate shareholders is determined by both direct and indirect ownership links; effective ownership can be larger than direct ownership and creates a wedge between ownership and control.
- Debt concentration is maximized when the wedge between ownership and control is largest (e.g., Zingales (1994) with at least 5 percent).
- Relevance to Basel supervisory framework:
  - The new Basel supervisory framework for measuring and controlling large exposures defines connected counterparties broadly and treats exposures to connected counterparties as a single exposure for reporting and compliance.
  - From a supervisory perspective, bank exposures to connected counterparties arising from ownership and control links should be consolidated at the level of the controlling shareholders.

*Source: Staff calculations, Bloomberg, GCC stock exchanges, and private sources as presented in the supplied content.*

### REFERENCES

### _wp15184 - REFERENCES

### Empirical studies on corporate ownership and control
- La Porta, R., F. Lopez-de-Silanes, and A. Sheleifer, 1999, “Corporate Ownership around the World,” Journal of Finance, Vo. LIV(2), pp. 471-517.
- Brioschi, F., L. Buzzacchi, and M. Colombo, 1989. “Risk Capital Financing and the Separation of Ownership and Control in Business Groups,” Journal of Banking and Finance, Vol. 13, pp. 747-772.
- Villalonga, B. and R. Amit, 2008. “How are U.S. Family Firms Controlled?” Review of Financial Studies, Vol. 22(8), pp. 3047-91.
- Rotundo, G. and A. M. D’Arcangelis, 2010. “Ownership and Control in Shareholding Netwoks, Journal of Economic Interaction and Coordination, Vol. 5, pp. 191-219.
- Ellerman, D., 1991. “Cross Owership of Corporations: A new Application of Input-Output Theory,” Metroeconomics, Vol. 42(1), pp. 33-46.
- Alves, P., 2008. “Who Owns the Largest Firms around the World?” MPRA Paper no. 52113. MPRA Paper no. 52355.
- Alves, P., 2010. “Corporate Ownership: Some International Evidence,” MPRA Paper no. 51380.
- Alves, P., 2012. “The Puzzle of Corporate Control,” MPRA Paper no. 52113.
- Chappelle, A. and A. Szafar, 2002. “Ownership and Control: Dissecting the Pyramid,” Centre Emile Bernheim Working Paper no. 03/002 (Brussels: Unveristé Libre de Bruxelles).
- Chappelle, A. and A. Szafar, 2005. “Controlling Firms through the Majority Voting Rule,” DULBEA Working Paper no. 05-05.RS (Brussels: Unveristé Libre de Bruxelles).
- Gutiérrez, L., C. Pombo, and R. Taborda, 2005. “Ownership and Control in Colombian Corporations,” IDB Research Network Working Paper no. R-520 (Washington: Inter-American Development Bank).
- Gutiérrez, L., C. Pombo, 2007. “Corporate Governance and Firm Valuation in Colombia,” IDB Working Paper no. 568 (Washington: Inter-American Development Bank).

### Regional corporate governance and banking (MENA / GCC)
- Aaltonen, A., 2013. Environmental, Social and Corporate Governance Practices in the MENA Region 2007–2012, (Dubai: Hawkamah The Institute for Corporate Governance).
- GCC Board of Directors Institute, 2011, Embarking on a Journey: A Review of Board Effectiveness in the Gulf, (Dubai: GCC Board of Directors Institute).
- OECD, Hawkamah-The Global Corporate Governance Forum, and Union of Arab Banks, 2009. Policy Brief on Improving Corporate Governance of Banks in the Middle East and North Africa Region, (Paris: OECD).
- International Monetary Fund, 2014. “Assessing Concentration in GCC Banks,” GCC Background Papers (Washington: International Monetary Fund).
- Kingdom of Saudi Arabia Capital Market Authority, 2015. Rule for Qualified Foreign Financial Institutions Investment in Listed Shares, (Riyadh: Kingdom of Saudi Arabia Capital Market Authority).

### Network analysis, graph theory, and software tools
- Bastian, M., Heymann, S., and M. Jacomy, 2009. Gephi: An Open Source Software for Exploring and Manipulating Networks, International AAAI Conference on Weblogs and Social Media.
- Glattfelder, J. B., 2013. Decoding Complexity, chapter 3 (Berling: Springer-Verlag).
- Krackhardt, D, 1994. “Graph Theoretical Dimension of Informal Organizations” in Carley, K. and M. Prietula (eds), Computational Organizational Theory (Hillsdale, NJ: Lawrence Erbaum Associates), pp. 89-111.
- Kogut, B. and G. Walker, 2001. “The Small World of Germany and the Durability of National Networks,” American Sociological Review, Vol. 66, pp. 317-35.
- Kunegis, J. and Julia Preusse, 2012. “Fairness on the Web: Alternatives to the Power Law,” WebSci, June.

### Policy, supervisory, and international institutional reports
- Basel Committee on Banking Supervision, 2014, Supervisory Framework for Measuring and Controlling Large Exposures, (Basel: Bank of International Settlements).
- International Monetary Fund, 2014. Annual Report on Exchange Arrangements and Exchange Restrictions, (Washington: International Monetary Fund).

*References list as provided in the source PDF.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15184.pdf_