IFSB WORKING PAPER SERIES

WP-11/05/2019

INVESTIGATING INTERSECTORAL
LINKAGES IN ISLAMIC FINANCIAL
SERVICES INDUSTRY

May 2019
IFSB WORKING PAPER SERIES
Technical and Research Department

WP-11/05/2019
INVESTIGATING INTERSECTORAL
LINKAGES IN THE ISLAMIC FINANCIAL
SERVICES INDUSTRY

Abideen Adeyemi Adewale

Volker Nienhaus

May 2019

NOTE: IFSB Working Papers are published by the IFSB to seek comments and encourage
discussion on issues that are pertinent to the specificities of the Islamic financial services
industry. IFSB Working Papers present preliminary results of research in progress and
represent the views of the author(s); as such, they should not be reported as representing the
views of the IFSB.

Corresponding email: [email protected]

This working paper has benefited greatly from the feedback and guidance provided by a core team from
the Technical and Research Department of the IFSB Secretariat led by Dr. Jamshaid Anwar Chattha,
Assistant-Secretary-General, Technical and Research. The author would also like to acknowledge the
immense contribution of Prof. Volker Nienhaus, consultant to the IFSB, who provided valuable comments
and suggested numerous revisions to the draft paper. Many thanks to the members of the IFSB Technical
Committee, which reviewed and approved the paper. Thanks to the Financial Network Analytics (FNA)
for the access to and use of the trial version of the FNA Platform. Mr. Abdul Wahab Abdul Hamid (research
fellow) and Ms. Rosmawatie Abdul Halim, Member of the IFSB Secretariat provided support data collation
as well as in the formatting and publishing of the paper respectively. Finally, gratitude to the regulatory
and supervisory authorities, multilateral bodies, and institutions offering Islamic financial services that are
members of the IFSB for providing useful comments on the draft paper during members’ consultation.
Published by: Islamic Financial Services Board

Level 5, Sasana Kijang, Bank Negara Malaysia

2, Jalan Dato’ Onn, 50480 Kuala Lumpur, Malaysia Email: [email protected]; [email protected]

ISBN 978-967-5687-66-2

All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, or stored in
any retrieval system of any nature without prior written permission, except for permitted fair dealing under the Copyright,
Designs and Patents Act 1988, or in accordance with the terms of a license issued by the Copyright, Designs and Patents Act
1988, or in accordance with the terms of a license issued by the Copyright Licensing Agency in respect of photocopying and/or
reprographic reproduction.

Application for permission for other use of copyright material, including permission to reproduce extracts in other published
works, shall be made to the publisher(s). Full acknowledgement of the author, publisher(s) and source must be given.

© 2019 Islamic Financial Services Board

 ABOUT THE ISLAMIC FINANCIAL SERVICES
BOARD (IFSB)

The IFSB is an international standard-setting organisation, which was officially

inaugurated on 3 November 2002 and started operations on 10 March 2003. The
organisation promotes and enhances the soundness and stability of the Islamic
financial services industry by issuing global prudential standards and guiding
principles for the industry, broadly defined to include banking, capital markets and
insurance sectors. The standards prepared by the IFSB follow a lengthy due
process as outlined in its Guidelines and Procedures for the Preparation of
Standards/Guidelines, which involves, among others, the issuance of exposure
drafts, holding of workshops and, where necessary, public hearings. The IFSB
also conducts research and coordinates initiatives on industry-related issues, as
well as organises roundtables, seminars and conferences for regulators and
industry stakeholders. Towards this end, the IFSB works closely with relevant
international, regional and national organisations, research/educational
institutions and market players.
For more information about the IFSB, please visit www.ifsb.org.

i
ABBREVIATIONS
AAOFI Accounting and Auditing Organization for Islamic Financial Institutions

BCBS Basel Committee on Banking Supervision

BGD Bangladesh

BIS Bank for International Settlements

CRD Capital Requirements Directive

FNA Financial Network Analytics

GFC Global Financial Crisis

GOV Government

G-SIFI Global systemically important financial institution

G-SII Global systemically important institution

H-H Household

HLA Higher loss absorbency

HQLA High-quality liquid assets

IADI International Association of Deposit Insurers

IAH Investment account holder

IAIS International Association of Insurance Supervisors

ICM Islamic capital market

IDN Indonesia

IFSB Islamic Financial Services Board

IFSI Islamic financial services industry

IIFS Institutions offering Islamic financial services

IMF International Monetary Fund

IOSCO International Organization of Securities Commissions

ISIC International Standard Industrial Classification of All Economic

Activities

JOR Jordan

LCR Liquidity coverage ratio

MYS Malaysia

NAV Net asset value

NFC Non-financial corporation

NCR Net capital rule

NPISH Non-profit institutions serving households

NSFR Net stable funding ratio

ii
OFI Other financial institutions

OMN Oman

OTC Over-the-counter

PAK Pakistan

PSIFIs Prudential and Structural Islamic Financial Indicators

ROW Rest of the world

RSA Regulatory and supervisory authority

RWA risk-weighted asset

SAU Saudi Arabia

SDG Sustainable development goal

SDN Sudan

SIFI Systemically important financial institution

SME Small or medium-size enterprise

SNA System of National Accounts

UAE United Arab Emirates

iii
GLOSSARY

Sharīʻah The practical divine law deduced from its legitimate sources: the Qurʼān,
Sunnah, consensus (Ijmāʻ), analogy (Qiyās) and other approved
sources of the Sharīʻah.
Ṣukūk Certificates that represent a proportional undivided ownership right in
tangible assets, or a pool of tangible assets and other types of assets.
These assets could be in a specific project or a specific investment
activity that is Sharīʻah-compliant.
Takāful A mutual guarantee in return for the commitment to donate an amount
in the form of a specified contribution to the participants’ risk fund,
whereby a group of participants agree among themselves to support
one another jointly for the losses arising from specified risks.

iv
Abstract

Research carried out since the Global Financial Crisis (GFC) of 2007–8 to in an
attempt to understand the systemic risk origination, propagation and mitigation relating
to sectoral linkages has rarely focused on the specificities of the Islamic financial
services industry (IFSI). The main aim of this IFSB working paper is to fill this gap by
exploring intersectoral financial linkages within the IFSI, as well as between the IFSI
and the real economy, via a financial network analysis. The preliminary indicative
outcome is envisaged to prepare the ground for future work by the IFSB on
macroprudential guidelines as it applies to the specificities of the IFSI. The Islamic
banking balance sheet data from 4Q13 to 4Q17 extracted from the IFSB Prudential
and Structural Islamic Financial Indicators database is used to generate a bilateral
exposure adjacency matrix that indicates assets and liabilities across sectors of the
IFSI and the real economy. The Financial Network Analytics software is used for
missing data generation based on maximum entropy as well as for both the financial
network measures and visualisation. The results obtained did not show the anticipated
significant commonality across countries that would enable a strong (albeit descriptive)
conclusion to be drawn. The results did, however, indicate what analysis could be done
if the requisite data were available. While some inexplicable results were noticed, in
general, all seven sectors studied across the four selected jurisdictions1 have
increased in interconnectivity over time. Both the other financial institutions and
household sectors recorded increased bilateral exposure in terms of obligations to
other sectors. Notwithstanding, the results indicated a reduced likelihood of the Islamic
banking sector creating a devastating effect on the financial network in the event that
a systemic risk originated from it.

1
The countries selected feature a systemically significant Islamic banking sector or membership of the G20 economies.
These countries are by no means representative of the entire global IFSI. As such, financial network analysis conducted
is country-specific with an assumed possibility of finding cross-country common trends. The results obtained are
assumed indicative, rather than inferential, due to data constraints; as such, they should be viewed with caution.
Notably, the peculiarities of each country in terms of economic diversification and the size of the Islamic financial sector
were not considered. Similarly, the possibility of interconnectedness of the IFSI with the conventional financial
institutions and likely cross-border claims and obligations were also discounted due to data limitations.

v
Table of Contents

ABOUT THE ISLAMIC FINANCIAL SERVICES BOARD (IFSB) ............................................i

ABBREVIATIONS......................................................................................................................ii
GLOSSARY .............................................................................................................................. iv
Abstract ................................................................................................................................. v
SECTION 1: INTRODUCTION ......................................................................................... 1
1.1 Background .................................................................................................................... 1
1.2 Objectives....................................................................................................................... 4
1.3 Scope of the Paper ........................................................................................................ 4
1.4 Methodological Assumptions and Limitations ........................................................... 4
1.5 Structure of the Paper ................................................................................................... 5
SECTION 2: SYSTEMIC RISKS, INTERCONNECTEDNESS AND HIDDEN
VULNERABILITIES IN THE IFSI ..................................................................................... 6
SECTION 4: SECTORAL FINANCIAL NETWORK ANALYSIS IN IFSI ................ 12
4.1 A Network Analysis of Sectoral Linkages in IFSI ..................................................... 12
4.2 Network Construction ................................................................................................. 13
4.3 Network Measures, Visualisation and Analysis ....................................................... 15
4.3.1 Network Measures .................................................................................................. 15
4.3.2 Network Visualisation ............................................................................................. 15
4.3.3 Network Centrality Algorithms ................................................................................ 19
SECTION 5: CONCLUSION AND RECOMMENDATIONS ...................................... 21
References .......................................................................................................................... 25
Appendix 1 ......................................................................................................................... 29
Various Network Centrality and Connectedness Measures Based on Maximum
Entropy ............................................................................................................................... 29
Appendix 2 ......................................................................................................................... 31
PSIFIs as a Bridge to Flow of Funds ............................................................................. 31
Appendix 3 ......................................................................................................................... 34
Data Classification from PSIFIs Database.................................................................... 34

vi
SECTION 1: INTRODUCTION

1.1 Background
The Global Financial Crisis (GFC) of 2007–8 offered a sharp reflection of the systemic
linkages that exist across the various sectors and jurisdictions of the financial
ecosystem. The consequences of the GFC on households, corporations and
governments manifested in many guises and resulted in arguably the deepest peace-
time contraction of the world economy since the 1930s.2

The vulnerability of the existing financial, fiscal and monetary policies at that time was
also exposed to an extent that the prominent global banks then had excess leverage,
and insufficient stable funding to see them through the crisis without a government
bail-out.3 In addition, the speed at which initial losses in a financial institution or sector
can be magnified and transmitted across other sectors through risk propagation
mechanisms became more discernible.4 The GFC thus provided a number of lessons
about the potential grave consequences of global financial instability.

Post-GFC, it became apparent that forces within the finance industry and interactions
among financial institutions and particular segments of the real economy (such as the
construction and real estate sector) played a major role in the propagation of the crisis. 5
Existing macroeconomic models had to be extended, and new models developed, to
capture the internal dynamics of the finance industry and mitigate against systemic
crisis in the financial sector and massive losses to the real economy.

While there is speculation about whether another financial crisis may be looming, a
unanimous view holds that any next financial crisis may be fundamentally different
from the previous crises, both in terms of origination and propagation.6 Today,
innovations in the financial sector arising from advancement in financial and
technological developments differ in many regards from what obtained during the last
GFC. For instance, the Bank for International Settlements (BIS) has noted the
implication of the regulatory challenges posed by crypto-assets, digital currencies,
shadow banking, etc., for financial system stability, especially under stress situations.7

A myriad of policy responses within and across the financial ecosystem has been put
in place post the GFC.8 This is with the ultimate aim of forestalling its impacts and
contagious spread on the one hand, and reoccurrence across sectors and jurisdictions

2
Financial institutions with large credit exposures to the property sector were greatly affected, while the consequent
bail-out by the government with taxpayers’ money resulted in a risk transfer from the banking sector to the government
sector. This transfer of risk had severe negative implications for the sovereign bond yields in the capital market sector
(Cussen, 2017, p. 2). Based on data covering 40 years and 187 different episodes of banking crisis in 126 countries,
van Dijk (2013) found that a number of socio-economic indicators were significantly affected by the financial crises. For
instance, life expectancy fell by nine months, school enrolment dropped by 3.5%, while fertility fell by 5.5%. In addition,
there was an increase in outbound refugees and in inbound foreign aid.
3
BIS (2018), Annual Economic Report, p. 80.
4
Cortes, Lindner, Malik and Segoviano (2018), p. 5.
5
The GFC led to the realisation that the interaction between the financial and the real economy, as reflected in the
interlinkages in the asset prices and macroeconomic outcomes across the various financial subsectors, was more
complex than hitherto thought.
6
Depository Trust and Clearing Corporation (2018).
7
BIS (2018), Annual Economic Report.
8
Claessens and Kose (2018), p. 5.

1
on the other.9 For example, based on the Basel III accords, a two-stage approach was
adopted to mitigate the contagious spread of the financial crisis in the banking sector.
In order to combat pro-cyclicality,10 a higher overall equity capital requirement for
banks was established, including a counter-cyclical buffer of between 0% and 2.5%
and a capital conservation buffer of 2.5%.

Financial institutions may also be required to maintain a higher loss absorbency (HLA),
depending on their relative importance in terms of size or position within a system, as
a possible origin of a systemic financial crisis if they have attained “too big to fail”11 or
“globally systemically important financial institution”12 status. Furthermore,
macroprudential policy tools such as capital and liquidity surcharges have been
suggested to lessen the effects of systemic risk.13 Arguably, these various measures
that were meant to provide a buffer and build resilience especially in the banking sector
have proven to be useful complements of other financial, fiscal and monetary policies,
albeit at the price of an unintended slow credit growth and restraints on a financial
boom.14

For the Islamic financial services industry (IFSI), relevant Islamic Financial Services
Board (IFSB) standards already contain requisite modifications relating to pertinent
macroprudential issues.15 For instance, provisions relating to capital buffer, leverage
ratio, dynamic provisioning, sectoral risk weights, etc. are been revisited by the IFSB
with a view to ensuring the stability of the IFSI.16 In addition, the IFSB has produced a
guidance note which provides key parameter settings that address the concerns
relating to liquidity coverage ratio (LCR), net stable funding ratio (NSFR) and high-
quality liquid assets (HQLA).17

Ensuring global financial stability entails that the focus of the intervening policies
should be complementary and related to how they affect the real economy18. This is
particularly relevant to the IFSI whose activities, considering its essential underlying
philosophy, are strongly tied to the real economy.19 Arguably, the effect of the GFC on

9
BIS (2018), Annual Economic Report, p. 80.
10
The tendency of financial variables to fluctuate around a trend during the economic cycle (Landau, 2009).
11
Loepfe, Cabrales and Sánchez (2013), p. 1.
12
Global systemically important financial institutions (G-SIFIs) are defined by the Financial Stability Board as institutions
of such size, market importance and global interconnectedness that their distress or failure would cause significant
dislocation in the global financial system and adverse economic consequences across a range of countries.
13
In the insurance sector, the International Association of Insurance Supervisors (IAIS) states that from 2019,
regardless of all group activities including non-insurance subsidiaries, global systemically important institutions (G-SIIs)
will be expected to hold regulatory capital that is not less than the total required by the sum of the BCR and HLA
requirements. In the securities sector, the International Organization of Securities Commissions (IOSCO) has updated
its 1989 Report on Capital Adequacy Standards for Securities Firms (Capital Standards Report) with a focus on the net
capital rule (NCR) approach and the capital requirements directive (CRD) following the Basel Committee approach
although with recognition of relevant national variations.
14
BIS (2018), Annual Economic Report, p. 63.
15
The IFSB has also published its flagship Islamic Financial Services Industry Stability Report on an annual basis since
2013. The report tracks the stability and resilience-related developments in the IFSI across the Islamic banking, takāful
and Islamic capital market sectors by providing in-depth trend analysis of, and the likely future outlook for, the IFSI as
part of its core mandate to ensure the industry’s financial soundness and resilience.
16
The IFSB is presently revising a number of its standards, especially IFSB-15: Revised Capital Adequacy Standard
for Institutions offering Islamic Financial Services (Excluding Islamic Insurance (Takāful) Institutions and Islamic
Collective Investment Schemes].
17
See GN-6: Guidance Note on Quantitative Measures for Liquidity Risk Management in Institutions offering Islamic
Financial Services [Excluding Islamic Insurance (Takāful) Institutions and Islamic Collective Investment Schemes ].
18
Silva, da Silva, and Tabak (2017) p. 4.
19
Aziz (2008) p. 3.

2
the Islamic financial sector was generally considered to be relatively less than that on
the conventional financial sector. Plausible reasons in addition to the IFSI’s linking its
transactions to real assets include the inherent foundational principles that promote
transparency, market discipline, and the avoidance of interest and uncertainty in
contractual obligations.20 For instance, Islamic banks, whose risk-sharing business
model is essentially equity-based, were protected from the consequence of exposure
to the high leverage and risk taking that underlay the GFC.21

There is, however, evidence that suggests the Islamic financial services sector was
not absolutely immune from the effects of the GFC, given its link to the real sector of
the economy. The vulnerability of institutions offering Islamic financial services (IIFS)
to financial crisis is reflected in their link to the real estate sector – for instance, in the
Gulf countries – as a second-round effect of the GFC.22 This view is based on the
notion that while an institution – or a financial sector, for that matter – may be sound
in isolation, its susceptibility to, for instance, the real sector’s pro-cyclicality can also
be very costly23.

Notwithstanding the fact that the value of the IFSI is relatively small and Islamic
financial institutions do not have a strong multinational presence in comparison to the
conventional financial service industry in most jurisdictions,24 the former has assumed
systemic importance in a number of jurisdiction.25 In addition, most countries have
recognised the potential of Islamic finance as being crucial to their financial inclusion
agenda, while the United Nations notes its potential contribution to achieving the
financing goal aspect of sustainable development goals (SDGs).26

As the IFSI continues to gain traction across the world, its ascendance to being an
integral part of the financial ecosystem will also grow in parallel due to increased
intrasectoral linkages among the Islamic banking and non-banking financial sectors on
the one hand, and intersectoral linkages with the real economy on the other. What is
not clear is whether such interlinkages either impede or enhance the financial stability
of the IFSI.

Due consideration of the implications of a financial crisis for the IFSI therefore seems
pertinent now, more than ever. As such, an understanding of the multidimensional and
evolving complex nature of systemic sectoral linkages requires in-depth research,
especially in terms of systemic risk amplification mechanisms and consequential
contagion effects peculiar to the IFSI, in order to come up with appropriate policy
responses. Moreover, and quite specifically, the various macroprudential policy
responses tailored to conventional finance, while applicable in many regards to Islamic

20
Alqahtani, Mayes and Brown (2015); and Hashem and Giudici (2016), p. 2.
21
Aziz (2008, p. 3.
22
Kammer et al. (2015).
23
For instance, due to external shock, the real sector’s assets may be jeopardised triggering defaults on its financing
obligations to the banking sector. The latter’s need to comply with regulatory capital requirements may lead to a credit
crunch to the former putting it at a risk of further default thus amplifying the systemic shock(Silva, da Silva, and Tabak,
2017, p. 7).
24
Only Sudan and Iran have a full-fledged Islamic financial system.
25
Koster (2012).
26
IMF, Core Principles for Islamic Finance Regulation (2018).

3
finance, may also inadvertently discountenance the specificities of the IFSI given the
latter’s structural fragmentation and relatively smaller size in most jurisdictions.27

1.2 Objectives
This working paper is generally aimed at preparing the ground for the future work of
the IFSB on developing macroprudential guidelines. As a prelude to another working
paper on the effectiveness of macroprudential tools for financial stability in the IFSI,
this preliminary working paper specifically aims to investigate the systemic linkages
among the various IFSI sectors and their interconnectedness to the real sector in the
selected countries.28

1.3 Scope of the Paper

This preliminary working paper, like most studies on sectoral linkages in the financial
services industry, also faces an issue relating to requisite data availability on a granular
scale29. As such, the scope of the working paper is mainly conceptual with a modest
fractional empirical backing as much as the available data can permit to conduct
financial network analysis and visualisation with a specific focus on four jurisdictions
selected for this study.30

1.4 Methodological Assumptions and Limitations

A systemic analysis of the interlinkages between the IFSI and the real sector as well
as among the various sectors of the IFSI requires data on a granular basis, which is
presently not available. Specificities of Islamic finance were long ignored, both in
national and international statistics. The IFSB’s Prudential and Structural Islamic
Financial Indicators (PSIFIs)31 should be able to solve this dilemma in the future;
however, for now, the issue of data limitation remains pertinent to conducting a
granular data-driven analysis of the prevalence, transmission and effect of
intersectoral systemic linkages in the IFSI. As such, some methodological assumptions
are required, especially in relation to estimating the intersectoral bilateral exposure
data needed for conducting a financial network analysis.

Based on the PSIFIs data, while it may be possible to know both the aggregate of the
funding received (liabilities) from and financing advanced (assets) to the other sectors
by the Islamic banks in a country, no data are available on what other sectors provided
in funding to or received in financing from each other as counterparties. Consequently,

27
Apart from Iran and Sudan, which have an Islamic financial system, only a few countries, such as Malaysia,
Pakistan, Bangladesh and the GCC countries, have attained a significant level of development in their Islamic finance
industry.
28
The mapping also considers cross-sectoral systemic risks in the IFSI through the identification of linkages between
banks and institutions in other sectors such as takāful and Islamic capital markets, as well as other non-financial,
household, government and rest of the world (ROW) sectors as per the System of National Accounts (SNA)
classification.
29
Battiston and Martinez-Jaramillo (2018) p.7
30
Four IFSB jurisdictions have been selected for this study. The selection was based on factors such as regional
representation, IFSI development, attainment of systemic importance as per the IFSB’s IFSI Stability Report 2018 and
the availability of comparable data. The initial basis of G20 membership could not be solely used, as relevant data
needed are missing for a particular IFSB jurisdiction among the three IFSB members in the G20. The PSIFIs data for
some jurisdictions that have attained systemic importance are only available in percentage format, which is not
compatible with the type of analysis intended.
31
The data provided by the IFSB PSIFIs database consists of macro-level data of the balance sheet claims and
obligations of Islamic banks in various jurisdictions in an aggregate form, making it suitable for financial network
analyses.

4
there are missing elements in the bilateral exposure matrix developed for analysis.
Nonetheless, it is possible to partially estimate the aggregate sectoral assets and
liabilities based on the PSIFIs Islamic banking data. For this, estimates of the missing
data are generated based on maximum entropy in the first stage and RAS algorithm
in the second.32 This approach is used to distribute assets and liabilities as evenly as
possible among the counterparty sectors based on available data in the IFSB PSIFIs
database.33

In the bilateral exposure matrix created, the unknown elements are estimated
assuming that the marginal distribution of the aggregate of the known assets and
liabilities are independent. Thus, if sector A provides, for instance, 40% of aggregate
funding, and Sector B receives 30% of total financing, the element AB in the matrix
would be (0.4 x 0.3 = 0.12). In addition, for computational purposes, the adjacency
matrix of the bilateral exposure among the sectors comprises country-level data rather
than specific Islamic financial institutions data. As such, funding and financing data are
taken in aggregates, rather than based on instruments or maturities.34

There are seven sectors considered in this study following the SNA classification and
similar studies carried out at the International Monetary Fund (IMF) and a number of
central banks.35 Given that there are four countries sampled, the adjacency matrix
consists of 28 cells for each of the 17 quarters from 4Q13 to 4Q17. As such, a closed
system which posits a complete IFSI network and which results in a dense network is
assumed in this paper, meaning that all sectors are expected to have relations with at
least one other sector. The Financial Network Analytics (FNA) software is used for data
generation, as well as both the financial network measures and visualisation.36

Due to the data limitations, the results obtained are, at best, indicative rather than
inferential.37 Even though the Islamic banking sector of some of the four countries
selected is systemically important on the respective national level, these countries are
by no means a representation of the global IFSI. Moreover, the fact that all jurisdictions
with the exception of two operate a dual financial system would make an analysis of
interconnection between the Islamic financial institutions and their conventional
counterparts relevant. However, due to the lack of data, this fact, as well as possible
cross-border transactions, had to be discounted in this working paper.

1.5 Structure of the Paper

The remainder of this working paper is broadly divided into four sections as follows. In
Section 2, a brief review of the concept of systemic risk, and of various dimensions of

32
“The RAS method is used in a situation when only row and column sums of desired input–output table are known.
The table is then estimated from an older fully-known input–output table in a way that the resulting table is consistent
with given row and column sums.” Holy and Safr (2017) p.2
33
This approach follows numerous financial network analysis research studies conducted at various RSAs and the
IMF in which such missing bilateral exposure data are estimated via a maximum entropy algorithm in the first stage
and a RAS algorithm in the second (see Upper and Worms,2004, 2011; Martinez-Jaramilo et al., 2012; Markose, 2012;
and De Almeida Antoun, 2015), Anand et al (2018)
34
The lack of such required granular data impeded analysis on maturities and instrument bases as was done in similar
financial network analysis in the conventional system. See for instance, Castrén and Rancan (2014).,De Almeida
Antoun (2015), Anand et al (2017)
35
See Upper and Worms (2004, 2011); Martinez-Jaramilo et al. (2012); Markose (2012); and De Almeida Antoun
(2015).
36 www.fna.fi
37
Markose (2012) and Kreis and Leisen (2017) provide evidence for the limitations of using simulated financial network
models.

5
systemic risk, is provided. Section 3 focuses on intersectoral linkages in the IFSI.
Section 4 offers a data-driven analysis of the interrelationship among the various
subsectors of the IFSI and the real sector based on financial network analysis and
visualisation. The dataset used, as well as the entire methodological approach
adopted, are further described in detail. Section 5 concludes with highlights of
macroprudential issues arising from the literature reviewed and the modest network
analysis conducted in Section 4.

SECTION 2: SYSTEMIC RISKS, INTERCONNECTEDNESS AND HIDDEN

VULNERABILITIES IN THE IFSI

Systemic risk reflects the vulnerability of a financial institution to external contagion

and spillover due to financial interconnectedness. Defining systemic risk requires that
the source, transmission channel, sectors affected and the implications for structural
vulnerability are taken into consideration. Systemic risk is thus defined as “the risk of
widespread disruption to the provision of financial services that is caused by an
impairment of all or parts of the financial system, which can cause serious negative
consequences for the real economy”.38

Systemic risk’s contagion effect on the real economy underscores the policy attention
it receives from stakeholders including governments, regulatory and supervisory
agencies (RSAs), operators and financial intermediaries, as well as both corporate and
retail investors. Such risk primarily evolves due to, among other factors, financial
market development, regulatory environment, and behaviour of other market
participants often prompted by regulatory arbitrage, etc.39

For the IFSI, there seems to be a need for a clear-cut view of both what a system is
and what systemic risk implies. From a jurisdictional regulatory perspective, an explicit
statement on whether “system” refers to a sector, country, region or the world is
required. For instance, there is a tendency to view the Islamic financial system as if it
comprises only the Islamic banking system40, due to the relative dominance of that
sector, whereas Islamic capital markets and takāful may also be systemically important
in their own right. Furthermore, the IFSB’s mandate covers all three main sectors –
that is, Islamic banking, the Islamic capital market and takāful.41 As such, systemic risk
from the IFSI perspective may not be restricted to a view from the angle of one sector
only. Rather, it should be from a view that reflects sectoral interconnectivity to reveal
hidden vulnerabilities among the various sectors and their links to the real economy.

Systemic risk may also be viewed from the perspectives of its origin and transmission
channels.42 This is because it can occur within or be transmitted across various
financial markets, instruments, institutions or even infrastructure. There is also a

38
See Cortes et al. (2018).
39
See Smaga (2014).
40
As at 2Q18, Islamic banking accounts for 72% of the global IFSI assets (IFSB IFSI Stability Report 2019)
41
While supranational organisations like the Basel Committee on Banking Supervision (BCBS), IOSCO, IAIS and the
IMF focus more on their particular supranational systems, it is not so in the case of the IFSI.
42
Cortes et al. (2018).

6
possibility that the same sector may be the originator, transmitter and bearer of the
consequence of a financial crisis. In this instance, systemic risk is strictly endogenous,
emanating from either the collective behaviour of small financial institutions due to
asset commonality and strategic complementarities,43 fire sales or activities of a
systemically important financial institution (SIFI) within a given sector or jurisdiction.

Sectors in a financial system are interconnected via a network of many financial

obligations and claims to and from each other whose value is contingent on the
individual and collective financial soundness of interconnected sectors in the
network.44 Numerous studies45 on these sectoral linkages often accentuate the need
for an understanding of the effect of interconnectedness in a financial system,
especially towards developing macroprudential policies to mitigate the reoccurrence of
the systemic financial crisis. Two possible views – vis. the “domino” effects and the
“tsunami” effects – are noted as possible approaches that stimulate contagion. These
effects are depicted in Figure 2.1.

Figure 2.1 Domino vs. tsunami views of interconnectedness.

Source: Nematrian. Adapted from IMF (2016).

The “domino” effects view takes cognisance of the interconnectedness that exists in a
financial system such that, if a firm fails, it can trigger chain defaults among other firms
within the interconnected system. However, a focus on such direct interconnectivity
may cloud the possible “tsunami” effect of hidden vulnerabilities within the system that
may be triggered by sectors not focused on.

For instance, given that compared to the other sectors in the IFSI the Islamic banking
sector is apparently more interconnected, the macroprudential focus may be on that
sector. Nonetheless, other subsectors of the Islamic capital market sector, such as
pension funds, are very popular in some economies and are considered to be very
much related to the real economy given that such funds are both long-term and tied to
a key socioeconomic fundamental – employment. Furthermore, given the fact that it
takes a while to build up vulnerabilities in some instances and in some sectors such
as takāful, there exists a possible susceptibility to misinterpretation which makes
determining whether or not an event is systemic very much open to debate.

43
See Basso and Costain (2016); and De Nicolo, Favara and Ratnovski (2012).
44
Castrén and Rancan (2014).
45
BIS (2018), Annual Economic Report, Castrén and Rancan (2014), De Almeida Antoun (2015)

7
Although systemic risk is often viewed as a “tail event” with low probability of
occurrence, the severity of its consequences requires that a broad view of its
manifestation is taken into consideration for macroprudential policy formulation. This
is especially so given that lack of interconnectedness may not necessarily translate to
avoidance of systemic risks. Recent views of systemic risks transcend those arising
primarily from shock within the financial system. There are possibilities that national
financial systemic risks may be triggered by hyperinflation due to political risks46, or
even climate change47 On the flip side, recent considerations of systemic risks also
factor in the effect of societal change as currently manifested in the disruptive
technological innovation offered by cryptocurrencies, FinTech, TechFins, InsurTechs,
etc.48 In fact, systemic risks can spring up everywhere. However, not all possibilities
are of equal probability.

SECTION 3: INTERSECTORAL LINKAGES IN THE IFSI

The underlying philosophy of Islamic financial service rendering that precludes
engaging in speculative, highly leveraged and risky activities may explain why the IIFS
were not affected by the first-wave effect of the GFC. However, the fact that the IFSI
was not spared the second wave of the GFC due to, for instance, interbank activity
and real estate investment49 is a manifestation of the industry’s susceptibility to
systemic risk and systemic linkage among its various subsectors and between it and
the real sector. This may result from external global financial instability and its many
triggers. Internally, sectoral convergence in some regards and significant differences
that result from factors such as regulatory capital frameworks, core business activities,
supervisory approaches, etc. are also pertinent. The need for the IFSB to have a
composite view – for instance, in its standards-setting responsibility in the IFSI – is
also well noted.50

The dynamic nature of the IFSI may also require peculiar effective policy calibration
given that in most jurisdictions where these IIFS operate, the relative dominance of the
Islamic banking sector suggests that not much may be known about cost of policies
and how they affect, for instance, Islamic capital market activities. Similarly, the relative
smaller size of the takāful sector in most jurisdictions contributes to its being
considered as non-systemically important51. Notwithstanding this, the sector provides
a very essential service and also provides funds for both banking and capital market
activities. In addition, systemic risks build-up is slower in the takāful sector (similarly to
conventional insurance) than in the banking sector where liquidity issues cause a
financial crisis to build more quickly.52

46
Bitar, Hassan and Walker (2017).
47
Battiston and Martinez-Jaramillo (2018) p. 2
48
Aurer and Claessens (2018); BIS (2018), Annual Economic Report.
49
IFSB (2014), IFSI Stability Report.
50
IFSB-16: Revised Guidance on Key Elements in the Supervisory Review Process of Institutions offering Islamic
Financial Services [Excluding Islamic Insurance (Takāful) Institutions and Islamic Collective Investment Schemes]. This
technical standard explicitly specifies the entire gamut of Islamic financial service activities that the IFSB oversees.
51
Despite growing at a relative faster rate compared to conventional insurance, takāful accounts for less than 2% of
the global Islamic finance asset worth (IFSB IFSI Stability Report 2019).
52
The experience of Lloyds in the United Kingdom in the early 1990s suggests that systemic risks build up rather
slowly in the insurance sector – on average, for between five and 10 years (Koster, 2012). In fact, Borio and Drehman
(2009) warned about the false sense of security developed by regulators during a period of low volatility given that a
financial crisis is not an instant event but rather one that builds up over time.

8
The propagation of shocks across various financial sectors, notwithstanding the
diverse business models, derives either directly from the interlinkage in contractual
obligations or indirectly through exposure to common triggers and transmission
channels of systemic crisis53. For instance, in the Islamic capital market sector the
diversity of businesses conducted by intermediaries in the market and the
consequential risks they pose necessitate that regulatory attention is focused more on
their business conduct.54 These intermediaries may propagate and transmit shocks to
the entire financial system through asset liquidation, direct exposure55 and common
risk factors channels.56

Open-ended funds such as Islamic mutual funds are faced with redemption risks, as
investors may decide to redeem their shares in the short run, whereas the funds have
been invested by the mutual fund in a longer tenure investment. Other open-ended
funds like the Islamic exchange-traded funds, while not prone to redemption risk, often
do hold assets with higher illiquidity, making them more exposed to liquidity risks. The
effect of these risks on market performance can trigger a run by investors57 given that
flows and performance are co-integrated.58 In the conventional sphere, studies have
shown that such was the situation that played out during the peak of the GFC, resulting
in extensive fire sales of assets by open-ended funds to meet shortfalls in liquidity and
lost sales in individual stocks.59

Money market open-ended funds have also been shown to have the potential to
trigger, or even propagate, systemic risk. This may be through the direct exposure
channel as a result of the significant reduction in their exposure to risky issuers during
market stress. Impairment of the large funding provided by money market funds
infringes on their intermediation capability and their capacity to provide funding to other
financial sectors, especially the banks,60 as occurred in Europe during the European
sovereign debt crisis that began in 2009.61

In recent years, there has been a proliferation of Islamic hedging instruments in both
the Islamic banking and Islamic capital market sectors in many jurisdictions. 62 The
instruments, tools and strategies align not only with the operationalisation of a number
of Islamic contracts for the purpose of minimising risks, but also with one of the
essential principles of Shari’ah , which is to protect wealth. A variety of such Islamic
hedging instruments are being used in several jurisdictions, essentially as Shari’ah-
compliant hedging alternative to conventional derivative instruments. These include

53
For instance, Battiston and Martinez-Jaramillo (2018) p.3 for a brief on default contagion, distress contagion,
common assets contagion and funding liquidity contagion.
54
This is in terms of their relations with and claims made on customers, products offered and disclosure practices.
55
Cortes et al. (2018) noted that banks, for instance, have direct exposure to hedge funds through the former’s prime
brokerage service provided to the latter.
56
Even where systemic risks do not originate from hedged funds, their institutional structure that allows for a swift
switch across investments and markets may amplify cross-sectoral financial shock in their bid to arbitrage for profit
maximisation and funds protection.
57
According to Ferroli, Schoenholtz and Song Shin (2014), risk-averse investors would opt to take the first mover
advantage by redeeming early in anticipation that the net asset value (NAV) of the fund can only fall further.
58
Cortes et al. (2018); and Coval and Stafford (2007).
59
Anand et al. (2015); Cortes et al. (2018).
60
The banking sector may also be affected by the money market funds, especially where the banks provide sponsor
support to their money market funds subsidiary by purchasing a significant portion of the illiquid asset or guaranteeing
the par value of the money market funds.
61
Correa, Sapriza and Zlate (2012).
62
Jurisdictions such as Malaysia, Indonesia, Iran and the UK, to name a few, have accommodated such tools.

9
alternatives to profit rate swaps, foreign currency swaps, foreign currency forwards
and options, etc.63

A fundamental difference between conventional derivatives and their Islamic hedging

alternatives is that the latter cannot be used, without a genuine underlying real
transaction, for the sole aim of generating profits. In this regard, market volatility is
minimised and systemic stability is ensured. However, most of such Islamic hedging
transactions are traded over-the-counter (OTC) rather than through an organised
exchange, which results in opaqueness of the market and unavailability of data, thus
masking hidden vulnerabilities to systemic shock origination and propagation.

During the GFC, the hedge funds’ significant exposure to, and losses from some asset
classes, pointed to the fact that such funds are highly vulnerable to a combination of a
myriad of liquidity, credit and equity risk factors.64 Consequently, numerous investors
exercised their right of redemption of shares with a level of alacrity that had to be
moderated by the hedge funds’ gates in order to meet outflow demand and to protect
the remaining investors.

Due to the nature of the hedge fund transactions, it is important for regulatory and
supervisory authorities to fully know the nature and size of such transactions being
used by IIFS in their market, in their counterparties and in underlying contracts in order
to fully understand the potential risk that such transactions pose to systemic stability.

In the takāful sector, the inverted production business model65 adopted makes it
relatively less susceptible to market volatility compared to the Islamic banking and
Islamic capital market sectors. The takāful sector is thus relatively unlikely to be the
source of a systemic risk in the IFSI, because it is less interconnected and less
dependent on economic cycle. Nonetheless, the takāful sector, like the conventional
insurance sector, may also be faced with various types of risks, including equity market
risks, concentration risks, liquidity risks, technical risks, etc.66

The vulnerability of a takāful firm can be amplified by events such as major

catastrophes or unusually high frequency and magnitude of policy surrender, which
may necessitate fire sales to meet liquidity needs. Furthermore, significant indirect
financial shock propagation can result if a takāful firm draws heavily on its banking line
of credits, thus amplifying liabilities in the Islamic banking sector.67

The Islamic banking sector’s dominance in a financial system is arguably linked to its
pivotal role around which other sectors rotate. This is so, given that as part of its
intermediation role it mobilises funds for financing purposes from the other sectors on
its liabilities side to other sectors on its assets side. Furthermore, the role an Islamic

63
Ibid
64
Le Sourd (2009).
65
Insurance is based on prepaid funding via premiums and penalties for policy discontinuation (Cortes et al., 2018). A
stable cash flow is generated and funds are invested in long-term assets funded by short-term liabilities, which makes
insurance companies not to involve in maturity transformations (IAIS, 2011).
66
For instance, an insurance company may face exposure in the equity market or the property market in its bid to
hedge against inflation. Similarly, concentration risks may result on the asset side from investment in a concentrated
asset class, and from a narrow exposure to a range of underwriting channels on the liability side. Insurance firms are
also susceptible to both technical risk and liquidity risk in the event of an unusually high claim due to a random
fluctuation in business or a catastrophic event.
67
Acharya and Richardson (2010).

10
bank plays in the IFSI is different due to the fact that any financing and funding that it
originates or mobilises is basically a direct linkage and not traded in an exchange,
unlike, for instance, equity securities traded in the Islamic capital market.

The interconnectedness and centrality of banks in a financial network is also

differentiated given that they can participate in investment in the other sectors, such
as the insurance or securities market, on their own account or act as an intermediary.
So, the banking sector, by the nature of the transaction it engages in, is exposed to a
variety of risks. For instance, financing to other sectors such as households,68 non-
financial corporations, etc., often represents the largest asset in a bank’s balance sheet
and may expose it to credit risk, while deposits mobilisation, as the largest liability, may
expose the bank to funding and liquidity risks69.

In the IFSI, the Islamic banks seem to be very much interconnected to other sectors.
The balance sheets of Islamic banks indicate that, to varying degrees across countries,
they are very active in their intermediation role. On the assets side of the balance
sheet, in most instances in most countries, Shari’ah-compliant financing to
households, non-financial corporations, government and the rest of the world accounts
for more than 70% of total assets. Though this may be an indication of an over-
exposure of the Islamic banks to the non-financial sector on the asset side, it
nonetheless also indicates that the Islamic banks’ financing to the real sector is very
high in line with the fundamental philosophy of Islamic finance – that it be linked to the
real economy.

The implication of this may need further exploration, especially to find out whether the
intersectoral link between Islamic banks and the real economy has really strengthened
somewhat due to a wider sectoral diversification of financing and investments. There
may also be further “hidden” intersectoral links if, for example, a large part of consumer
finance is for home furnishing and equipment. An analysis of Islamic banks’ balance
sheets might provide indications of whether such risk clustering (and, hence, a stronger
link of Islamic banks with one segment of the real economy) does in fact exist.

Although the balance sheets of most countries’ Islamic banks indicate that interbank
funding and financing are, on average, about 4% and 6%, respectively,70 there is also
a need to take a closer look at the involvement of Islamic banks in interbank activities
in the money market. This may be a significant asset channel through which
intrasectoral systemic risk may be propagated.

The investment opportunities of Islamic banks are more limited compared to their
conventional counterparts. For this reason, the former’s typical balance sheet reveals
a higher concentration in the real estate and construction industry.71 In some other

68 As much as 46% of the total financing by Islamic banks was to households, while about 12% was to the
construction and real estate sector (IFSB, IFSI Stability Report 2018)
69
Battiston and Martinez-Jaramillo (2018) offers insights into how default contagion can originate and propagate
though unsecured interbank market as well as the possibility of distress contagion even if the obligor does not default.
The authors also highlighted the possibility of common assets contagion arising from the indirect connection among
banks due to investment in common assets, as well as funding liquidity contagion on the liability side if banks decide
to hoard credit rather than make it available to other market players.
70
IFSB, PSIFIs database as at 4Q17.
71
See IFSB (2019) IFSI Stability Report (Forthcoming), and IMF (2017) Country Report No. 17/145.

11
jurisdictions, there is a prominent presence of the IIFS in the small to medium
enterprise (SME) sector, which further exposes them to financial instability due to the
fragility of that sector.

In terms of interconnectivity, a concern from a system stability perspective could be,

for example, the links between Islamic banks and takāful operators as issuers and
holders of ṣukūk. Of particular interest in this regard is corporate ṣukūk and, even more
specifically, ṣukūk issued by banks to strengthen their capital base. The assets side of
the balance sheets of Islamic banks in most countries indicates a steady increase in
ṣukūk financing and other Shari’ah-compliant securities. However, this accounts for
less than 10% of total financing in most jurisdictions. A similar trend is also observed
in terms of ṣukūk funding across countries.

Most ṣukūk are issued by sovereigns. The volume of corporate ṣukūk is approximately
25% of total issuances (and decreasing over time).72 A large proportion, about 30% of
corporate ṣukūk, are issued by Islamic financial institutions, which are also major
investors in ṣukūk. There could be a problematic mutual dependency if bank A holds
ṣukūk which have been issued by bank B to strengthen its capital base, and vice versa.
Investment in ṣukūk can be a tool to diversify risk, but it could also lead to a sectoral
risk concentration if the ṣukūk are issued by industries that are prime recipients of
Islamic bank financing (e.g. construction and real estate).

SECTION 4: SECTORAL FINANCIAL NETWORK ANALYSIS IN IFSI

4.1 A Network Analysis of Sectoral Linkages in IFSI

This section adopts financial network analysis73 and visualisation to investigate
interconnectedness among the various sectors in both the real and Islamic financial
systems of the selected countries.74 Following previous studies,75 the rationale for
adopting a network approach,76 notwithstanding the difficulty of obtaining the granular
bilateral exposure data, is hinged on the possibility of using estimated data to generate
sectoral linkage networks.77

The main goal of network analysis is to build a conceptually appealing network of

intersectoral asset claims and financial obligations that permits a subsequent

72
IFSB (2018), IFSI Stability Report.
73 Hansen (2014) provides a detailed critique of the pros and cons of various systemic risk analysis techniques. Bisias
et al. (2012) also provide a rich literature survey on systemic risk analytics
74
Based on the SNA classification adopted in the PSIFIs, these sectors include the banking, Islamic capital market,
insurance and other financial institutions, household, government and the ROW. The balance sheet is essentially that
of the Islamic banks in the jurisdictions covered in this study. Full details of the data are given in the methodology
section of the paper.
75 Castren and Rancan (2014).
76
Network analysis is an established and useful approach to modelling systemic interconnectedness. It cuts across
numerous fields, although its use is rather limited and is emerging only recently in systemic links analysis in finance
due to data limitations.
77
Studies like De Almeida Antoun (2015), Castren and Rancan (2014), Markose (2012), Martinez-Jaramillo et al.
(2012), Upper (2011), etc., have applied it. However, Markose (2012) and Kreis and Leisen (2017) provide evidence
for the limitations of using simulated financial network models.

12
determination of aggregate network vulnerabilities.78 Unlike the alternate approach of
using stock market prices, evidence has shown that a network approach is more
suitable given that default leading to a systemic crisis originates from changes in the
balance sheets of financial institutions.79 In fact, it is argued that the data availability
argument of using the market-based model is not a strong justification for using it. The
market-based model does not provide early warning signals of a crisis, as “by the time
they spiked, [the] market would have tanked already”.80

The network method is very useful for analysing both direct and indirect, as well as
weighted, interlinkages in order to decipher the structural heterogeneity present in the
linkages across financial subsectors of interest. Moreover, the PSIFIs database used
in this study consists of macro-level data of the balance sheet claims and obligations
of Islamic banks in various jurisdictions in an aggregate form. In addition to its content
relevance, the form of data entry also suits the network analysis as the various sectors
can be represented as nodes or vertex,81 and the interlinkage among them as the link
or the arc.

It is envisaged that at the end of the analysis, sectors from which potential shocks can
originate can be identified in the countries selected. Moreover, since the data cover
about nine semi-annual82 or 17 quarterly observations, the network structure can be
observed on a time-variant basis to capture structural heterogeneity in the trajectory of
the interlinkages across sectors.83 Finally, the various centrality and connectedness
measures analysed can offer insights into the relative strength and importance of
various sectoral linkages identified for systemic risk management and macroprudential
policy designs.84

4.2 Network Construction

This section offers a brief description of how the network used in the analysis section
is developed. There are two distinct routes towards the development of better
information systems on structural data for the financial sector and its interrelations with
the real economy. The first route takes the system of national accounts (SNA) as a
macroeconomic point of departure. The second route takes the disclosure
requirements of regulated financial institutions as a microeconomic point of departure.
This working paper adopts the first approach given that it is less demanding in terms
of data requirements and suits the format of the PSIFIs database towards constructing
a bilateral exposure matrix.

78 Markose (2012), p. 13, comments that intervention and stabilisation policies crucially depend on knowing who is
linked to whom and how.
79
Kreis and Leisen (2017).
80
Markose (2012), p. 1.
81
Nodes or vertex, indicated by a circle in a network diagram, represent a financial sector – for example, Islamic
banking or Islamic capital market. The arc or the links are the arrows that show interconnectedness among the sectors
in a network.
82
The data for Country 4 are available on a semi-annual basis. Other countries are available on a quarterly basis.
83
Adding more subsectors and more detailed lists of financial instruments would come at a high price. Each additional
subsector and instrument requires more input data. A complete SNA requires four different types of accounts (tables)
for each category of instrument. First, a balance sheet for the stocks of assets and liabilities at the beginning of the
period, and a financial account for all transactions between institutional units during the period. Two other requirements
are an account for the impact of “exceptional” events (such as natural catastrophes) that cause changes in the volume
of assets and liabilities (e.g. by destruction), and a revaluation account for changes in the values of assets and liabilities
due to price changes.
84
Martinez-Jaramillo et al. (2012).

13
The seven sectors85 in each of the four selected countries represent the nodes indexed
by “i”. The weighted financial linkage between any two nodes is indexed by W ij. Each
country network W D is derived based on the total amount of assets and liabilities for
each sector in each country. In a closed system like that assumed in this paper, every
asset in one sector has a corresponding liability in one or several other sectors within
the network.

While it is usually possible to know the value of the assets a sector has, based on
aggregate data, it may be impossible to find detailed statistics on how much of these
assets correspond to specific liabilities in the other sectors. If data are not available in
the necessary granularity, several techniques can be employed to “generate” it – that
is, the matrix of (the results of) all bilateral interactions in the network, from the
available aggregate data. If only the sums of the rows and columns on the margins of
a matrix are known and the non-marginal cells are unknown, it is not possible to
calculate the missing cell entries in a deterministic manner because the number of
equations is far less than the number of unknowns. This means that more than one set
of cell entries is compatible with the aggregate marginal sums, and there is probabilistic
uncertainty with respect to the generated network structure and the interlinkages
matrix, respectively. Hence, it has to be decided which of the possible solutions of the
under-identified system is to be chosen. The maximum entropy method offers a
solution to this problem86.

A comprehensive empirical study87 have surveyed and assessed seven methods for
uncovering financial network structures from partial data. Several methods applied the
maximum entropy principle, with the iterative standard maximum-entropy method
being one of the two best performers for the reproduction of the structure of exposures.

The basic idea is to choose “that solution which injects the minimum amount of
additional and, by definition, non-verifiable information into the data”88 (such as, for
example, assumptions about the behaviour of nodes). Entropy measures the
informational content of a message, and it “is greatest when the message recipient is
the most uncertain as to the outcome of a given event”.89 Maximising entropy thus
means to choose that solution which is associated with the maximum uncertainty or
randomness of the distribution of cell entries.90 If only a minimum of additional
information (such as non-negativity conditions) has to be injected (in order to decide
which solution will be chosen for further analysis), the uncertainty remains at its
maximum. In formal terms, the entropy has to be maximised subject to a number of
constraints, and the resulting optimisation problem is usually solved by an iterative
process. Should additional data become available, they could be fed into this process.

From a practical perspective, an “initial guess” of the network (the matrix with initial cell
entries) is generated by computing the exposure of sector i to sector j as the product

85
The sectors are: Islamic Banking (IB), Islamic Capital Market (ICM), Other Financial Institutions (OFI), Non-
Financial Corporation (NFC), Household (H_H), Government (GOV), and Rest of the World (RoW).
86
Hazan (2018) p. 1
87
Anand et al (2018)
88
Sheldon and Maurer (1998), p. 693.
89
Ibid.
90
Zhou, Cai and Tong (2013).

14
of the row sum of sector i and the column sum of sector j. “This network is subsequently
re-scaled by the aggregate positions, first along the rows and then the columns, until
the aggregate position constraints are satisfied.”91

4.3 Network Measures, Visualisation and Analysis

Although the listed issues regarding the data quality and comparability cannot all be
solved, this working paper adopts an innovative tool via an in-built command in the
FNA software used in similar studies for the sectoral network data analysis and
visualisation. While other methods have been proposed, such as those that minimise
density,92 maximum entropy is preferred in the financial network literature, especially
in contexts similar to that in this study, where there are few interconnected nodes with
the possibility of yielding a dense and almost complete network.93

4.3.1 Network Measures

The FNA software is applied to generate a number of network measures that can be
used to interpret the potential shock transmission and vulnerabilities arising from
bilateral exposures among the various sectors in a network. These measures are
broadly divided into centrality and connectedness algorithms.94 The centrality
algorithms generally help to identify the relative importance of a node or vertex in a
network based on a number of network topological criteria, some of which are
specifically meant for financial applications.95 The connectedness algorithms help to
identify how nodes are connected in the network. In this study, two connectedness
algorithms used are the clustering co-efficient and degree.

Computation of the network centrality and connectedness algorithms is based on the

sectoral data for aggregate assets and liabilities between two points in time, 4Q13 and
4Q17.96 In order to cater for effect size and to mitigate the implications of different
volumes of sectoral activities,97 especially where all the data are extracted from a
predominantly Islamic banking database, the sectoral links are weighted relative to
their values in the network. This is important given that networks are built to depict
financial stability, as drawing links based on absolute value may not be representative
of a sector’s relative bilateral exposure.98

4.3.2 Network Visualisation

A network is basically a graphical representation of an adjacency matrix that consists
of the bilateral exposures of the various nodes in the network. While the position of the
node in the graph is arbitrary, its size indicates the relative net bilateral exposure

91
Anand et al. (2018); see also Upper and Worms (2004).
92
Anand et al. (2014).
93
De Almeida Antoun (2015).
94
www.fnalab.com.
95
The various measures used in this study include: Betweeness, Closeness, Degree, Eccentricity, Farness, PageRank
and Random Walk Betweeness. The specific algorithms for financial network analysis include CheiRank, SinkRank
and SourceRank
96
4Q13 is used as the base period, while 4Q17 is used as the reference period. The choice of the periods, though
mainly arbitrary, is meant to reflect any changes between the two periods as the earliest and latest data points in the
PSIFIs database. Moreover, it suits the inclusion of Country 4, whose data are compiled on a semi-annual basis.
97
The data used are predominantly from the banking sector, which may be more exposed to certain sectors than
others.
98
De Almeida Antoun (2015). FNA Analysts Guide (2018)

15
position of a sector in the network.99 Similarly, the thickness of the arrow that links
nodes indicates the proportional exposure of the sector from which the arrow emanates
to the sector that the arrow connects. Each node in the diagram represents a sector.
Green indicates a sector that has a net claim from the other sectors, while red indicates
a net obligation from a sector to others. The arrows show the direction of flow of funds
between two sectors.

The sectoral links for each of the four selected countries for 4Q17 are presented below.
Figures 4.1 to 4.4 depict the network interconnectedness across the seven sectors
based on Islamic banking data obtained from the PSIFIs database and bilateral
exposures generated via a maximum entropy algorithm.100 The figures should be
interpreted together with the tables in Appendix 1, which show the various network
centrality measures for 4Q13 as the base period and 4Q17 as the reference period.

Based on Figure 4.1, it can be seen that the ‘Other financial institutions’ (OFI) sector
in Country 1 has the top bilateral linkage to the ‘Non-financial corporation’ (NFC),
‘Household’ (H-H) and ‘Islamic capital market’ (ICM) sectors.101 The red colour of the
OFI sector node indicates that it has lower claims to receive from, compared to
obligations it needs to fulfil to the other sectors. This contrasts sharply with Country 2,
where the top bilateral exposures are to the household sector. In terms of size, the H-
H sector’s node is bigger in this instance, denoting larger exposure. The largest
exposure of the Country 2 H-H is to the NFC.102 This denotes that, even though
Country 2 households have claims to the OFI, they nonetheless also have obligations
to the other sectors. However, the fact that the OFI’s node is also red indicates some
consistency with Country 1’s sectoral linkage in that regard.

In the case of Country 3, similar to Country 2, the H-H sector seems to be the sector
most exposed to a bilateral relationship with the other sectors. It is also the only sector
with a red node, indicating that the H-H sector in Country 3 has more obligations to the
other sectors than claims due to it. The greatest obligation is to the NFC sector,
indicated by the thickness of the link between the two sectors. All the financial sectors
and the ROW sector also have claims against the H-H sector. Compared to Country
4, a sharp contrast can be observed given that the most exposed sector is the OFI.
This is similar to the trend in Country 1 especially and, to a lesser extent, in both
Country 2 and Country 3. In Country 4, also, the nodes for both the H-H sector and the
Islamic banks are red, indicating that both sectors also have larger obligations to other
sectors than claims receivable by them.

The general trend observed across the various jurisdictions sampled in this paper is
similar in many regards to those of the Financial Stability Board Report.103 The report
also measured bilateral linkages among the various financial sectors based on balance

99
In the network visualisation, red indicates that the ratio of the liability to the asset in a sector is greater than 1.
100
See De Almeida Antoun (2015) for a description and example of data generation via maximum entropy.
101
The ICM is essentially represented by funding generated by and financing through ṣukūk and other Sharīʻah-
compliant securities. The ṣukūk is further divided into corporate and sovereign ṣukūk. The latter is used together with
related measures to represent the ‘Government’ (GOV) sector.
102 This may be in terms of the H-H investments in shareholding in the various sectors captured under the NFC sector,
as listed in the table in Appendix 2.
103
FSB (2018), Report on Shadow Banking.

16
sheet data. The report noted a low relationship between the OFIs and the banking
sector, as well as an increasing trend in non-bank funding, especially to households.

Figure 4.1 Sectoral Linkages Network, Country 1, 4Q17

IB: Islamic Banks; NFC: Non-Financial Corporation; ICM: Islamic Capital Market; OFI:
Other Financial Institutions; GOV: Government; H-H: Household; ROW: Rest of the
World.

Figure 4.2 Sectoral Linkages Network, Country 2, 4Q17

IB: Islamic Banks; NFC: Non-Financial Corporation; ICM: Islamic Capital Market; OFI:
Other Financial Institutions; GOV: Government; H-H: Household; ROW: Rest of the
World.

17
Figure 4.3 Sectoral Linkages Network, Country 3, 4Q17

IB: Islamic Banks; NFC: Non-Financial Corporation; ICM: Islamic Capital Market; OFI:
Other Financial Institutions; GOV: Government; H-H: Household; ROW: Rest of the
World.

Figure 4.4 Sectoral Linkages Network, Country 4, 4Q17

IB: Islamic Banks; NFC: Non-Financial Corporation; ICM: Islamic Capital Market; OFI:
Other Financial Institutions; GOV: Government; H-H: Household; ROW: Rest of the
World.

18
4.3.3 Network Centrality Algorithms
Centrality algorithms are very crucial in network analysis and have been commonly
used in the various fields of related analysis. A very essential attribute of this algorithm
is that, over the years, some specific variants peculiar to financial networks have been
developed to capture the power, influence, control and independence of the various
nodes in a network.104 To assess the relative importance of a sector in a financial
network, it is suggested in the literature that a number of attributes should be
explored.105

First, it is expected that nodes do not exist in isolation. There should be as many
interlinkages among the various sectors in a network (degree) as possible. Second,
the total monetary value of assets and liabilities upon which the bilateral linkages would
be established should be sufficiently large (strength). Other features mentioned include
closeness, which indicates the likelihood of contagious transmission of risks among
the sectors in a few steps in the network. While it is important that there are many
paths through which a sector can link to any other sector in the network (betweenness),
of very high importance is the cognisance given to the counterpart sectors in a network
relative to a particular sector (eigenvector and PageRank).106 Specifically, the FNA
software also provides algorithms for calculating SinkRank as a measure of systemic
importance in the network.107 Other financial centrality measures are SourceRank and
CheiRank. It is important to note that for all the centrality measures, the basic
interpretation is that the higher or larger the centrality measure, the greater the
importance of a sector in the network.

4.3.3.1 Betweenness Centrality and Closeness Centrality

Of the various general network centrality algorithms identified, the betweenness and
closeness centrality algorithms are interpreted in this paper.108 The betweenness
centrality algorithm signifies the number of the shortest paths between any two sectors
in the network passing through a given sector for which the betweenness centrality is
interpreted. In other words, it indicates the mediating or bridging importance of a sector
in the interlinkages between any other two sectors. Unlike other network centrality
measures, it is captured in its absoluteness. As such, the higher the centrality the
higher the influence a sector can exert as a propagation channel. It is an important
indicator of the systemic risk propagation potentiality of a sector, especially through an
indirect means.

Closeness is a measure of influence denoting how quickly a node can reach other
nodes in a network – in other words, how quickly a sector can reach other sectors in
the network (e.g. in terms of risk propagation). It is also described as an interpretation
of a sector’s relative independence in a network, as a high closeness centrality means

104
Only the relevant financial network centrality measures are analysed and interpreted in this study.
105
Henggeler and Muller (2006), cited in Martinez-Jaramillo et al. (2012).
106
See Saltoglu and Yenilmez (2010) for the use of these measures in financial network analysis.
107
This importance depicts strong ties and interconnectedness to other sectors. Systemic Importance as it is used for
prudential regulation by the RSAs follows a more detailed methodology. See the threshold used for Islamic banking
sector in the IFSB’s IFSI Stability Report 2019.
108
According to Peltonen, Rancan and Sarlin (2015), betweenness and closeness consider both the direct and indirect
linkages, thus denoting the central position that a node occupies in a network.

19
a sector relies less on other sectors for it to be exposed to systemic risk.109 In most
instances, a network that is not strongly connected would not generate a closeness
centrality score.110

As shown in Appendix 1, an assessment of the betweenness centrality and closeness

centrality across the four countries in this study reveals that between 4Q13 and 4Q17,
all sectors in Country 2 except for the IBs and all sectors in Country 4 except for OFIs
recorded increased centrality over time. In both Country 1 and Country 3, the relative
importance of every sector in isolation as a bridge connecting any other two sectors in
the IFSI recorded an increase over time. This implies that while the IBs’ capacity to act
as a channel of shock propagation declined considerably in Country 2, a similar
outcome was recorded in the OFI in Country 4.

In terms of the closeness centrality measures, similar trends are observed in both
Country 1 and Country 3. No measure was generated for Country 4, apparently due to
the weak connectedness in the network.111 In the case of Country 2, the closeness
measure for the NFC declined considerably, suggesting that the NFC over time
recorded declining capacity to quickly propagate systemic risk to any other sector in
the network.

4.3.3.2 Strength Centrality and Degree Centrality112

The degree centrality denotes the importance of a sector in a network depending on

how many other sectors it is connected to. The more connected a sector is, the greater
will be the severity of its failure and the higher the risk of propagation to the other
sectors. One of the main criticisms of this measure is the fact that it does not take into
consideration the importance of other sectors except those to which it is closest, and
the fact that it is based on absolute measures.113 As such, it is suggested that where it
is weighted based on the value of the parameters used in a financial network analysis,
it can be a measure of the strength of a sector in a network.114

Based on the table in Appendix 1, notwithstanding the declining clustering coefficient

score indicating weak connectedness, the weighted degree centrality measure
denotes an increase over the period 4Q13 to 4Q17. Country 1 sectoral connectedness
is confirmed by both its higher weighted degree centrality and clustering coefficients.115
Country 2 and Country 3 had a consistent clustering coefficient over the period of the
study. However, based on the weighted degree centrality, both the OFI and ROW
sectors in Country 2 had a comparatively lower measure in 4Q17 relative to 4Q13. The
case of Country 3 presents a sharp contrast given that its degree centrality scores for
the GOV, H-H, IBs and ROW sectors are all lower in the reference period compared
to the base period.

109
Martinez-Jaramillo et al. (2012).
110
FNA Analyst Guide (2018).
111
This was also confirmed based on the vertex clustering coefficient score for Country 4, which indicated declining
network connectedness (see Appendix 1).
112
Network Centrality Measures: Betweenness, Clustering Co-efficient and Degree (show importance of
interconnectedness and capacity for systemic shock transmission among various sectors).
113
Martinez-Jaramillo et al. (2012).
114
fnalab.com/command
115
Clustering coefficient as a measure of connectedness ranges from 0 to 1. The higher the figure, the more connected
a sector is.

20
4.3.3.3 Financial Network Algorithms116

Four important specific financial network algorithms are considered to capture the
centrality of each sector considered in this study. These include the CheiRank,
PageRank, SourceRank and the SinkRank algorithms.

The CheiRank centrality is essentially calculated on the transposed network to

measure the proportion of time a sector contacts another within a network in an infinite
random walk over the transposed network.117 As a result of the matrix transposition,
the resulting inverted link direction is used by the CheiRank to rank the sectors in a
financial network on the basis of the average link it has to other sectors to which it has
net obligations. The PageRank is similar in many regards to the CheiRank except that
it is an eigenvector-based measure of network centrality.118

The SourceRank and the SinkRank are related centrality metrics used to assess the
systemic importance of a sector in a financial network. In fact, the former is an
equivalent to calculating the latter on the transposed network.119 SinkRank centrality’s
relative importance is hinged on its capacity to measure how central a failing sector is
in a network via its bilateral interlinkages with others. This metric assumes a strongly
connected network and takes a value of between 0 and 1. The higher the SinkRank,
the higher the likelihood that a sector will have a devastating effect on the network in
the event of its failure.120

Based on the table in Appendix 1, the financial network centrality algorithms show a
consistent pattern in each of the various countries. In Country 4, the GOV, H-H, ICM,
and OFI sectors all recorded lower centrality scores in 4Q17 compared 4Q13. In
Country 1, the GOV, IB, NFC and ROW sectors all recorded lower financial centrality,
indicating the reduced likelihood of these sectors creating a devastating effect on the
financial network in the event that a systemic risk originates from any of them. In the
case of Country 2, the two sectors with relatively lower financial centrality measures in
4Q17 compared to 4Q13 are the IBs and the OFI sectors. Finally, in Country 3, while
the IB sector recorded a lower SinkRank, both the NFC and ROW had lower measures
in the other financial network centrality algorithms.

SECTION 5: CONCLUSION AND RECOMMENDATIONS

Drawing policy inferences from the modest financial network analysis conducted in this
study requires some care as the results obtained are only indicative of the likely actual
outcome. This is hinged on the fact that the data upon which the network is constructed
are strictly Islamic banking data. While, based on the balance sheet, we can have an
idea of what the Islamic banks mobilised in funding and what they disbursed as
financing, the bilateral linkages among the other sectors had to be estimated. This may
frustrate testing the various measures developed to assess the impact of
macroprudential policies for timely identification, follow-up and policy efficacy. The lack

116
Financial Network Measures: CheiRank, PageRank, SourceRank and SinkRank (show systemic importance of a
sector in a financial network).
117
fnlab.com/manual
118
Ibid.
119
Ibid.
120
Soramaiki and Cooks (2013).

21
of data may also frustrate clear-cut cost/benefit analyses of the policy implementation
in order to calibrate policy design which is very crucial to the standard setters and
supervisory bodies – for instance, the IFSB as well as the RSAs.

Nonetheless, the various network centrality and connectedness measures computed

in this study have highlighted that sectoral bilateral linkages vary across different
jurisdictions. The exposure to systemic risk seems more likely to originate from the OFI
sector in most of the countries sampled. This, however, is not prejudicial to the relative
importance of other sectors as potential systemic risks originators and propagators.
There may, therefore, be a need to have a look at the existing macroprudential
policy put in place in most jurisdictions with a view to identifying what likely
issues are emerging and what implications they may have for the peculiar nature
of the IFSI pending when the requisite granular data are available to conduct a
sectoral linkage study that strongly guides policy.

Data could be extracted from databases of multilateral organisations like the World
Bank and the IMF, as well as from balance sheets and income statements of financial
institutions. However, the availability of data does not ensure that they are always
comparable. IFSIs in various jurisdictions have different disclosure practices,
apparently due to differences in cross-country supervisory and regulatory approaches
to Islamic banking supervision. For instance, caveats regarding the comparability of
data between Islamic banks (in different GCC countries) due to differences in the
structuring of Sharīʻah-compliant products is noted. The caveat also extends to the
balance sheet and income statement items between Islamic and conventional banks
– in the same country – because of different business philosophies and product
orientations.

Ample arguments have also been provided as to why company accounts may not be
suitable for use in national accounts.121 Both accounts have different regulatory
requirements, level of details required, presentation style and timeliness. At most, such
data from company accounts can be used as a proxy for national accounts after due
adjustments to the variables extracted.

The IFSB PSIFIs should be able to solve this data availability dilemma in the future;
however, for now, the issue of data limitation remains pertinent in conducting a
granular data-driven analysis of the prevalence, transmission and effect of cross-
sectoral systemic linkage in IFSIs. It is hoped that ongoing efforts by the IFSB to
provide the data for the IFSI to capture both the takāful and Islamic capital
market sectors, in addition to the already existing data on Islamic banking,
would include, sometime in the future, requisite flow of funds data at a granular
level. Such data are needed for constructing a for-whom-to-whom matrix as a
basis for understanding the systemic linkages that exist within the Islamic
financial services industry.

The exposure to systemic risk seems higher for the other (Islamic) financial institutions
comprising takāful, Islamic microfinance institutions, cooperatives and the various non-
bank Islamic financial institutions. However, this is not prejudicial to the relative
importance of other sectors such as households and non-financial corporations as

121
Mahajan (2013).

22
potential systemic risks originators and propagators. Therefore, there may be a need
to have a look at both the existing and likely systemic linkage issues, and the
implications they may have for the peculiar nature of the IFSI in various
jurisdictions pending when the requisite granular data are available to conduct
a sectoral linkage and cross-country study that guides policy.122

The IFSI, it would seem, rarely or marginally features in the prudential policy, as well
as in the complementary fiscal, monetary and structural policy, formulation in most
jurisdictions. As such, the Islamic banking, Islamic capital market and takāful sectors
and their conventional counterparts are often subjected to the same prudential policy
that lacks due cognisance of the specificities of institutions offering Islamic financial
services. Arguably, macroprudential policies are often not clearly articulated to reflect
such specificities. Key drivers of pro-cyclicality and activities of Islamic financial
intermediaries leading to the build-up of systemic risk that are often microprudential in
nature may perhaps not be comprehensively captured.

The BCBS, IOSCO, IAIS and International Association of Deposit Insurers (IADI),
based on their core mandates, focus on the identification, measurement, management
and monitoring of certain aspects of specific systemic risks and linkages peculiar to
their specific supranational domain. The conglomerate structure of the IFSIs
necessitates that a broad spectrum of Islamic financial services is provided
across numerous sectors. This peculiarity underscores the need for a
composite supervisory outlook.

The IFSI is not absolutely immune from prudential and structural impediments that may
have both intrasector and intersector stability implications across jurisdictions given
the interconnectedness of the institutions within the IFSI and with the real sector. There
are now 12 jurisdictions in which Islamic banking is considered as systematically
important. Although less prominent today, a likely extension of their services across
geographical borders creates the potential for contagion and spillover effects in the
years ahead. Obviously, to achieve an optimal outcome from the
macroprudential policies for the IFSI, there may be a need to have a cross-
sectoral approach to IFSI regulation and supervision. This of course would
require an investigation into the institutional supervisory arrangements across
the IFSB jurisdictions, as well as a set of prudential standards and effective
review process that ensures achieving reasonable degree of convergence
across sectors and minimizing moral hazard123 and arbitrage opportunities.

No doubt, the financial regulation and supervision from a microprudential point of view
are important in many regards. What the GFC has shown, however, is the insufficiency
of such policy for ensuring sustainable financial stability. Complementing the
microprudential policy with a macroprudential policy seems a more effective means of
averting financial crisis. Based on the indicative findings in this study, and from extant
related studies on systemic linkages among the various sectors in a financial system,

122
Although a few Islamic finance corporations have established subsidiaries in different jurisdictions, cross-border
transactions seem to be of very limited size, at least for now. The use of a network analysis may be hinged on its
perceived usefulness to build on this study for future cross-border contagion spillover analysis as and when data
become available.
123
A highly interconnected financial system may incentivize collective moral hazard manifest in excessive risk-taking
against the notion of being too-interconnected to fail.

23
a common conclusion is that the positive effect of the financial soundness of individual
institutions may be attenuated by the complexity of the interactions and
interdependency in a financial network. Therefore, averting financial instability
would require a blend of both the micro- and macroprudential framework in a
way that aligns the objectives of both policies.

The data used in this study do not cater for spillover effects that are quite well-
researched in the conventional sphere. This may be due to the fact that most Islamic
banks are as yet non-systemically important on a global scale. There is evidence to
suggest that, in the short term, macroprudential policy may generate unintended
systemic risk spillovers across both sectoral and country divides, especially due to
regulatory arbitrage.124

In addition, there is an increasing impetus to understand the implication of climate risk

on financial stability and resilience due to financial system interconnectedness. This
specifically relates to the effect of changing climate not only on physical assets and
implication for insurance liabilities and real sector default on financial contracts, but
also the consequential various policies and advocacies across the globe promoting
transition to a low carbon economy. Such transition is estimated to result in stranded
assets as much as 82% of global coal reserves, 49% of global gas reserves and 33%
of global oil reserves.125 For the IFSI, most of the jurisdictions that have attained
systemic significance derive their revenue from the oil and gas sector. The
implication of the implementation of the climate policies for financial stability
and resilience of the IFSI may warrant further investigation126 through among
other means a financial network approach due to the interconnectedness among
the sectors in the IFSI.

While there are presently no global systemically important institutions in the IFSI, the
sectoral interlinkages noted in the network analysis raise concerns of systemic risks.
This is more so viewed from the perspective of potential origination and propagation
of risks emanating from a common behaviour among non-systemically important
intuitions. There may, therefore, be a need in future related work of the IFSB to
have an activity-focused approach to supplement the entity-based approach that
is presently prevalent, in order to capture potential systemic risks that may
derive from asset commonality or behavioural commonality archetypal of non-
systemically important institutions.

124
FSB (2018), Report on Shadow Banking, p. 36.
125
Battiston and Martinez-Jaramillo (2018) p.10.
126
This suggestion is not prejudicial to the availability of requisite data for this purpose which is also still a major
challenge in the conventional financial system.

24
References
Acharya, V.V. and Richardson, M. (2010). Is the Insurance Industry Systemically
Risky? in V. Acharya, T. Cooley, M. Richardson and I. Walter (eds),
Regulating Wall Street: The Dodd-Frank Act and the Architecture of
Global Finance (Chapter 9). John Wiley & Sons.

Alqahtani, F., Mayes, D. G. and Brown, K. (2015). Economic Turmoil and Islamic
Banking: Evidence from the Gulf Cooperation Council. Paper presented at the
the 10th International Conference on Islamic Economics and Finance, Doha-
Qatar.

Anand, K., Craig, B. and von Peter, G. (2014). Filling in the Blanks: Network Structure
and Interbank Contagion, BIS Working Papers No. 445. Basel: Bank for
International Settlements.

Anand, K., van Lelyveld, I., Banai, A., Friedrich, S., Garratt, R., Halaj, G., Fique, J.,
Hansen, I., Jaramillo, S. M., Lee, H., & Others (2018). The Missing Links: A
Global Study on Uncovering Financial Network Structures from Partial Data,
Journal of Financial Stability, 35, 107–19.

Aurer, R. and Claessens, S. (2018). Regulating Cryptocurrencies: Assessing Market

Reactions, BIS Quarterly Review, September, 51–65.

Aziz, Z. (2008). Enhancing the Resilience and Stability of the Islamic Financial System.
A Keynote Address by the Governor of the Central Bank of Malaysia, at the
Islamic Financial Services Board and Institute of International Finance
Conference: Enhancing the Resilience and Stability of the Islamic Financial
System, Kuala Lumpur, 20 November.

BIS (2018). Annual Economic Report. Bank for International Settlements,

www.bis.org/publ/arpdf/ar2018e.htm.

Basso, H. S. and Costain, J. S. (2016). Macroprudential Theory: Advances and

Challenges, Banco de Espana Occasional Paper No. 1604. Available at:
https://ssrn.com/abstract=2756361 or http://dx.doi.org/10.2139/ssrn.275636
1.

Battiston, S. and Martinez-Jaramillo, S. (2018). Financial Networks and Stress Testing:

Challenges and New Research Avenues for Systemic Risk Analysis and
Financial Stability Implications. Journal of Financial Stability, 35, 6–16.

Bisias, D., Flood, M., Lo, A. W. and Valavanis, S. (2012). A Survey of Systemic Risks
Analytics. Office of Financial Research Working Paper No. 0001.

Bitar, M., Hassan, M. K. and Walker, T. (2017). Political Systems and the Financial
Soundness of Islamic Banks, Journal of Financial Stability, 31, 18–44.

Borio, C., Drehmann, M., (2009). Towards an operational framework for financial
stability: “fuzzy” measurement and its consequences. BIS Working papers no
284.

25
Bruno, V., Shim, I. and Shina, H. S. (2015). Comparative Assessment of
Macroprudential Policies, BIS Working Papers No. 502. Basel: Bank for
International Settlements.

Castrén, O. and Rancan, M. (2014). Macro-Networks: An Application to the Euro Area

Financial Accounts, Journal of Banking & Finance, 46, 43–58.

Claessens, S. and Kose, A. M. (2018). Frontiers of Macrofinancial Linkages, BIS

Papers No. 95. Basel: Bank for International Settlements.

Correa, R., Sapriza, H. and Zlate, A. (2012). Liquidity Shocks, Dollar Funding Costs,
and the Bank Lending Channel During the European Sovereign Crisis,
International Finance Discussion Papers No. 2012-1056, Federal Reserve
Board.

Cortes, F., Lindner, P., Malik, S. and Segoviano, M. A. (2018). A Comprehensive Multi-
Sector Tool for Analysis of Systemic Risk and Interconnectedness (SyRIN),
IMF Working Paper WP/18/14. Washington, D.C.: International Monetary Fund.

Coval, J. and Stafford, E. (2007). Asset Fire Sales (and Purchases) in Equity Markets,
Journal of Financial Economics, 86, 479–512.

Cussen, M. (2017). Identifing Inter-sectoral Exposures in Ireland Using Network

Analysis, Economic Letter Series, 3, 1–9.

De Almeida Antoun, L. (2015). A Network Analysis of Sectoral Accounts: Identifying

Sectoral Interlinkages in G-4 Economies, Working Paper No. 15/111.
Washington, D.C.: International Monetary Fund.

De Nicolo, G., Favara, G. and Ratnovski, L. (2012). Externalities and Macroprudential

Policy. IMF Staff Discussion Note.

Depository Trust and Clearing Corporation (DTCC) (2018). The Next Crisis Will Be
Different: Opportunities to Continue Enhancing Financial Stability 10 Years
After Lehman’s Insolvency. Available at: www.dtcc.com/systemic-risk/white-
papers.

Ferroli, M., Schoenholtz, K. L. and Song Shin, H. (2014). Market Tantrums and
Monetary Policy, U.S. Monetary Policy Forum Report 8, Initiative on Global
Markets. Chicago: University of Chicago Booth School of Business.

Hansen, L. P. (2014). Challenges in Identifying and Measuring Systemic Risk, in M.

Brunnermeier and A. Krishnamurthy (eds), Risks Topography: Systemic
Risk and Macro Modelling (pp. 15–30). Chicago: University of Chicago Press.

Hashem, S. Q. and Giudici, P. (2016). Systemic Risk of Conventional and Islamic

Banks: Comparison with Graphical Network Models, Applied
Mathematics, 7, 2079–96.

Hazan, A. (2018). A maximum entropy network reconstruction of macroeconomic

models. Physica A: Statistical Mechanics and its applications, Elsevier,
10.1016/j.physa.2018.12.020.hal-01812746v2

26
Holy, V. and Safr. K. (2017). The Use of Multidimensional RAS Method in Input–Output
Matrix Estimation. https://arxiv.org/pdf/1704.07814.pdf

IFSB (2011). Revised Compilation Guide on Prudential and Structural Islamic Financial
Indicators. Kuala Lumpur: Islamic Financial Services Board.

IFSB (2014). Islamic Financial Services Industry Stability Report. Kuala Lumpur:
Islamic Financial Services Board.

IFSB (2018). Islamic Financial Services Industry Stability Report. Kuala Lumpur:
Islamic Financial Services Board.

IFSB (2019). Islamic Financial Services Industry Stability Report. Kuala Lumpur:
Islamic Financial Services Board.

IMF (2016). Global Financial Stability Report. Washington, D.C.: International

Monetary Fund.

IMF (2017). Ensuring Financial Stability in Countries with Islamic Banking – Country
Case Studies, IMF Country Report No. 17/145, Washington, D.C.: International
Monetary Fund.

IMF (2018). Core Principles for Islamic Finance Regulation and Assessment
Methodology, IMF Policy Paper (pp. 1–38). Washington, D.C: International
Monetary Fund.

Kammer, A., Norat, M., Pinon, M., Prasad, A., Towe, C. and Zeidane, Z. (2015).
Islamic Finance: Opportunities, Challenges, and Policy Options,
SDN/15/05. Washington, D.C.: International Monetary Fund.

Koster, P. (2012). Crisis Management in the Islamic Financial Services Industry.

Paper presented at the Fifth Islamic Financial Stability Forum on Eff-
ective Crisis Management and Resolution Framework, Manama, Bahrain,
29 March.

Kreis, Y., Leisen and D. P. J. (2017). Systemic Risk in a Structural Model of Bank
Default Linkages, Journal of Financial Stability.
http://dx.doi.org/10.1016/j.jfs.2017.05.010.

Landau, J. (2009). Procyclicality: What it means and what could be done. Being
Remarks delivered by Deputy Governor of the Bank of France, at the Bank of
Spain's conference on Procyclicality and the Role of Financial Regulation;
Madrid, 4 May.

Le Sourd, V. (2009). Hedge Funds Performance in 2008. EDHEC Risk and Asset
Management Research Centre. February.

Loepfe, L., Cabrales, A. and Sánchez, A. (2013). Towards a Proper Assignment of

Systemic Risk: The Combined Roles of Network Topology and Shock
Characteristics, PLoS ONE, 8(10): e77526.
doi:10.1371/journal.pone.0077526.

Mahajan, S. (2013). Challenges of Company Accounts based data in National

Accounts. A Paper prepared for the Joint ESSnet Workshop on Consistency

27
 covering Work Packages 2 and 3, ‘On the Way to better Consistency in
European Business Statistics’ Rome, Italy, 11-12 June 2013.

Markose, S. M. (2012). Systemic Risk from Global Financial Derivatives: A Network

Analysis of Contagion and its Mitigation with Super-Spreader Tax, IMF
Working Papers No. 12/282. Washington, D.C.: International Monetary
Fund.

Martínez-Jaramillo, S., Alexandrova-Kabadjova, B., Bravo-Benítez, B. and

Solórzano-Margain, J. P. (2012). An Empirical Study of the Mexican
Banking System's Network and its Implications for Systemic Risk, Working
Papers No. 2012-07, Mexico City: Banco de México.

Peltonen, T., Rancan, M. and Sarlin, P. (2015). Interconnectedness of the Banking

Sector as a Vulnerability to Crises, ECB Working Paper.

Saltoglu, B. and Yenilmez, T. (2010). Analyzing Systemic Risk with Financial

Networks: An Application during a Financial Crash, MPRA Paper, University
Library of Munich, Germany.

Sheldon, G. and Maurer, M. (1998). Interbank Lending and Systemic Risk: An

Empirical Analysis for Switzerland, Swiss Journal of Economics and Statistics,
134(4.2), 685–704.

Silva, C.T., da Silva, M.A, and Tabak, B.M. (2017). . Systemic Risks in Financial
Systems: A Feedback Approach. Working Paper 461. Banco do Brasil. P. 1-
49.

Smaga, P. (2014). The Concept of Systemic Risk, Systemic Risk Centre Special
Paper No. 5. London School of Economics and Political Science. August.
http://eprints.lse.ac.uk/61214/1/sp-5.pdf.

Soramaiki, K. and Cook, S. (2013). SinkRank: An Algorithm for Identifying

Systemically Important Banks in Payment Systems, Economics: The
Open-Access, Open-Assessment E-Journal, 7 (2013-28), 1–27.
http://dx.doi.org/10.5018/economics-ejournal.ja.2013-28.
Upper, C. (2011). Simulation Methods to Assess the Danger of Contagion in
Interbank Markets, Journal of Financial Stability, 7(3), 111–25.

Upper, C. and Worms, A. (2004). Estimating Bilateral Exposures in the German

Interbank Market: Is There a Danger of Contagion?, European Economic
Review, 48, 827–49

van Dijk, M. A. (2013). The Social Costs of Financial Crises. SSRN Working Paper
no:2278526.

Zhou, R., Cai, R. and Tong, G. (2013). Applications of Entropy in Finance: A Review,
Entropy, 15, 4909-31.

28
APPENDICES

Appendix 1

Various Network Centrality and Connectedness Measures Based on Maximum

Entropy

Country 1

Chei Clustering Page Sink Source

Base Period Sector Betweenness Rank Co-efficient Degree Rank Rank Rank
2013-12-31 GOV 0 0.0283 0 0 0.0283 0.099 0.0895
2013-12-31 H_H 0 0.087 0 2397 0.087 0.1457 0.0999
2013-12-31 IB 6 0.0548 0.3333 1118 0.0548 0.1207 0.1077
2013-12-31 ICM 3 0.0578 0.3333 1232 0.0578 0.1173 0.1146
2013-12-31 NFC 0 0.406 0.5 14935 0.406 0.1105 0.6032
2013-12-31 OFI 0 0.3377 0 12629 0.3377 0.4103 0.0989
2013-12-31 ROW 0 0.0283 0 0 0.0283 0.099 0.0895

Reference Chei Clustering Page Sink Source

Period Sector Betweenness Rank Co-efficient Degree Rank Rank Rank
2017-12-31 GOV 10.5 0.0214 1 1.0508 0.0214 0.0214 0.0756
2017-12-31 H_H 0 0.125 1 7398 0.125 0.3388 0.1466
2017-12-31 IB 13.5 0.043 1 1618 0.043 0.0834 0.0992
2017-12-31 ICM 0 0.0888 1 4520 0.0888 0.1152 0.1359
2017-12-31 NFC 0 0.2983 1 16653 0.2983 0.0214 0.4195
2017-12-31 OFI 0 0.4015 1 22689 0.4015 0.7521 0.194
2017-12-31 ROW 0 0.0219 1 26.2712 0.0219 0.0214 0.076

Country 2

Chei Clustering Page Sink Source

Base Period Sector Betweenness Rank Co-efficient Degree Rank Rank Rank
2013-12-31 GOV 4.3333 0.0266 1 4133 0.0266 0.0214 0.0824
2013-12-31 H_H 0 0.4505 1 335809 0.4505 0.8216 0.3609
2013-12-31 IB 20.6667 0.0893 1 80848 0.0892 0.1022 0.1516
2013-12-31 ICM 0 0.0933 1 57059 0.0933 0.0214 0.1523
2013-12-31 NFC 0 0.1057 1 66835 0.1057 0.0214 0.166
2013-12-31 OFI 0 0.1907 1 136435 0.1907 0.3862 0.0992
2013-12-31 ROW 0 0.044 1 17876 0.044 0.0214 0.0996

Reference Chei Clustering Page Sink Source

Period Sector Betweenness Rank Co-efficient Degree Rank Rank Rank
2017-12-31 GOV 10.0526 0.0297 1 10140 0.0297 0.0214 0.0642
2017-12-31 H_H 0 0.471 1 529120 0.471 0.8714 0.5071
2017-12-31 IB 1.7 0.0827 1 92369 0.0827 0.0942 0.1192
2017-12-31 ICM 58.8795 0.1114 1 114511 0.1114 0.0367 0.1421
2017-12-31 NFC 0 0.173 1 186867 0.173 0.0214 0.1957
2017-12-31 OFI 11.3882 0.0814 1 89076 0.0814 0.3564 0.0789
2017-12-31 ROW 1.5882 0.0508 1 36155 0.0508 0.0214 0.0814

29
Country 3

Clustering
Chei Co- Page Sink Source
Base Period Sector Betweenness Rank efficient Degree Rank Rank Rank
2013-12-31 GOV 13.6526 0.0237 1 6017 0.0237 0.0214 0.0796
2013-12-31 H_H 0 0.471 1 1116200 0.471 0.8714 0.3857
2013-12-31 IB 39.3895 0.0619 1 127444 0.0619 0.2275 0.1174
2013-12-31 ICM 10.7143 0.0696 1 148758 0.0696 0.2334 0.1216
2013-12-31 NFC 0 0.293 1 732037 0.2931 0.0214 0.2689
2013-12-31 OFI 0 0.0374 1 43112 0.0374 0.0214 0.0877
2013-12-31 ROW 0 0.0433 1 58832 0.04326 0.0214 0.0912

Clustering
Reference Chei Co- Page Sink Source
Period Sector Betweenness Rank efficient Degree Rank Rank Rank
2017-12-31 GOV 15.1526 0.023 1 4771 0.023 0.0214 0.0782
2017-12-31 H_H 0 0.471 1 1309504 0.471 0.8714 0.3857
2017-12-31 IB 41.8895 0.0476 1 103387 0.0476 0.1822 0.1087
2017-12-31 ICM 10.7143 0.0938 1 260260 0.0937 0.2795 0.1341
2017-12-31 NFC 0 0.2922 1 848270 0.2922 0.0214 0.2694
2017-12-31 OFI 0 0.0382 1 52403 0.0382 0.0214 0.0873
2017-12-31 ROW 0 0.0343 1 40414 0.0343 0.0214 0.085

Country 4

Chei Clustering Page Sink Source

Base Period Sector Betweenness Rank Co-efficient Degree Rank Rank Rank
2013-12-31 GOV 0 0.0283 0 0 0.0283 0.025 0.0879
2013-12-31 H_H 0 0.0692 1 128560 0.0692 0.3298 0.98
2013-12-31 IB 0 0.0607 1 105020 0.0607 0.1309 0.1159
2013-12-31 ICM 5 0.0659 1 119967 0.0659 0.0462 0.1271
2013-12-31 NFC 0 0.3706 1 1081536 0.3705 0.0299 0.5801
2013-12-31 OFI 3 0.3770 1 1094502 0.377 0.5841 0.0976
2013-12-31 ROW 0 0.0283 0 0 0.0283 0.025 0.0879

Reference Chei Clustering Page Sink Source

Period Sector Betweenness Rank Co-efficient Degree Rank Rank Rank
2017-12-31 GOV 0 0.0243 0 0 0.0244 0.0791 0.0894
2017-12-31 H_H 0 0.0655 0.67 229803 0.0655 0.1262 0.0992
2017-12-31 IB 0 0.0632 0.67 222218 0.0632 0.1151 0.1085
2017-12-31 ICM 6 0.0492 0.67 142614 0.0492 0.0927 0.1142
2017-12-31 NFC 0 0.4065 0.75 2144567 0.4065 0.0871 0.6024
2017-12-31 OFI 0 0.3668 0 1911236 0.3668 0.6589 0.0988
2017-12-31 ROW 8 0.0244 0.75 18 0.0243 0.0871 0.0894

30
Appendix 2
PSIFIs as a Bridge to Flow of Funds

In summary, the following is a “translation” of the International Standard Industrial

Classification of All Economic Activities- (ISIC-) inspired classification of activities into an
SNA-reminiscent sector and subsector scheme:

 non-financial corporations sector = i + ii + iv + v + vi + vii +viii;

o manufacturing (ii)
o real estate (vii + viii)
 commercial (vii)
 residential (viii)
o other non-financial corporations (i + iv + v + vi);
 financial corporations sector = iii;
 general government sector = x;
 households sector (including NPISH) = ix;
 rest of the world = xi.

The PSIFIs database provides for a number of jurisdictions figures that can add
quantities to the categories. Two structural Islamic financial indicators and one
additional prudential Islamic financial indicator are of relevance here:

Indicator
PSIFI code

ST01 Total assets

ST01a Total Sharīʿah compliant financing (excluding interbank financing)

ST01b Ṣukūk holdings

ST01c Other Sharīʿah-compliant securities

ST01d Interbank financing

All other assets

ST02 Total funding/liabilities and equities

ST02a Profit-sharing investment accounts (PSIA)

ST02b Other remunerative funding (murābaḥah, commodity murābaḥah, etc.)

ST02c Non-remunerative funding (current account, wadīʿah)

ST02d Ṣukūk issued

ST02e Other Sharīʿah-compliant securities issued

ST02f Interbank funding/liabilities

ST02g All other liabilities

31
ST03 Capital and reserves

ST03a Total revenues

ST03b Financing based

ST03c Investment based (ṣukūk, other Sharīʿah-compliant securities, etc.)

ST03d Fee based

AD06 Value of Sharīʿah-compliant financing by economic activity

AD06a Agriculture, forestry, hunting and fishing

AD06b Mining and quarrying

AD06c Manufacturing

AD06d Electricity, gas, steam and air-conditioning supply

AD06e Water supply; sewerage and waste management

AD06f Construction

AD06g Wholesale and retail trade; repair of motor vehicles and motorcycles

AD06h Transportation and storage

AD06i Accommodation and food service activities

AD06j Information and communication

AD06k Financial and insurance activities

AD06l Real estate activities

AD06m Professional, scientific and technical activities

AD06n Administrative and support service activities

AD06o Public administration and defence; compulsory social security

AD06p Education

AD06q Human health and social work activities

AD06r Arts, entertainment and recreation

AD06s Other service activities (export)

AD06t Activities of households as employers

AD06t* Other financing of households

AD06u Activities of extraterritorial organisations and bodies

AD06u* Financing to non-residents

32
 The PSIFI indicators are assigned to the SNA-inspired sectors and subsectors as
follows:

Allocation of PSIFI Codes to Sectors and Subsectors

Sector/subsector PSIFI code

NFC Non-financial corporations AD06a+b+c+d+e+f+g+h+i+j+l+m+n+p+q+r+s

NF01 Manufacturing AD06c

NF02 Real Estate AD06l

NF09 Other non-financial corporations AD06a+b+d+e+f+g+h+i+j+m+n+p+q+r+s

FC Financial corporations AD06k

GG General government AD06o

HH Households and non-profit institutions serving h. AD06t + t*

RW Rest of the world AD06u + u*

The PSIFI database collects stability indicators for 24 jurisdictions. Not all jurisdictions
provide the indicators that are of relevance here. It is possible to compile sectoral
tables for the Islamic banks in 20 jurisdictions and for Islamic banking windows in
seven jurisdictions.127

127 One jurisdiction (AFG) provides only data for Islamic windows, and four jurisdictions (KWT, NGA, PAK, TUR) do
not provide a breakdown of Sharīʻah-compliant financing by economic activity. One country (EGY) provides neither
assets (financing) nor liabilities (funding) in absolute amounts, which is needed for the correspondence check; see
below.

33
Appendix 3

Data Classification from PSIFIs Database

Source Assets Liabilities

Additional (a) agriculture, forestry, hunting and
Prudential fishing – NFC
Islamic (b) mining and quarrying – NFC
Financial (c) manufacturing – NFC
Indicators
(PIFIs) – (d) electricity, gas, steam and air-
AD06 conditioning supply – NFC
(e) water supply; sewerage and
waste managemen – NFC
(f) construction – NFC
(g) wholesale and retail trade; repair
of motor vehicles and motorcycles –
NFC
(h) transportation and storage –
NFC
(i) accommodation and food service
activities – NFC
(j) information and communication –
NFC
(k) financial and insurance activities
– OFI
(l) real estate activities – NFC
(m) professional, scientific and
technical activities – NFC
(n) administrative and support
service activities – NFC
(o) public administration and
defence; compulsory social security
– GOV
(p) education – NFC
(q) human health and social work
activities – NFC
(r) arts, entertainment and
recreation – NFC
(s) other service activities (export) –
NFC
(t) activities of households as
employers – H-H
(t*) other financing of households –
H-H
(u) activities of extraterritorial
organisations and bodies – ROW
(u*) financing to non-residents –
ROW
Structural Ṣukūk holdings – ICM
Islamic Other Sharīʻah-compliant securities –
Financial ICM
Indicators Interbank financing – IB
(SIFIs) –
ST03
Structural Profit-sharing investment accounts
Islamic (PSIA) – OFI?

34
Financial Other remunerative funding
Indicators (murābaḥah, commodity murābaḥah,
(SIFIs) – etc.) – H-H?
ST04 Non-remunerative funding (current
account, wadīʻah) – H-H?
Ṣukūk issued – ICM
Other Sharīʻah-compliant securities
issued – ICM
Interbank funding/liabilities – IB

35