Sunteți pe pagina 1din 150

World Economic and Financial Sur veys

Global Financial Stability Report

A Bumpy Road Ahead

APR
18

I N T E R N A T I O N A L M O N E T A R Y F U N D
Wor l d Economi c a nd F i na n ci a l S ur v ey s

Global Financial Stability Report


April 2018

A Bumpy Road Ahead

I N T E R N A T I O N A L M O N E T A R Y F U N D
©2018 International Monetary Fund

Cover and Design: Luisa Menjivar and Jorge Salazar


Composition: AGS, An RR Donnelley Company

Cataloging-in-Publication Data

Joint Bank-Fund Library

Names: International Monetary Fund.


Title: Global financial stability report.
Other titles: GFSR | World economic and financial surveys, 0258-7440
Description: Washington, DC : International Monetary Fund, 2002- | Semiannual | Some issues also have thematic
titles. | Began with issue for March 2002.
Subjects: LCSH: Capital market—Statistics—Periodicals. | International finance—Forecasting—Periodicals. |
Economic stabilization—Periodicals.
Classification: LCC HG4523.G557

ISBN 978-1-48433-829-2 (Paper)


978-1-48434-876-5 (ePub)
978-1-48434-878-9 (Mobipocket)
978-1-48434-877-2 (PDF)

Disclaimer: The Global Financial Stability Report (GFSR) is a survey by the IMF staff published twice
a year, in the spring and fall. The report draws out the financial ramifications of economic issues high-
lighted in the IMF’s World Economic Outlook (WEO). The report was prepared by IMF staff and has
benefited from comments and suggestions from Executive Directors following their discussion of the
report on April 2, 2018. The views expressed in this publication are those of the IMF staff and do not
necessarily represent the views of the IMF’s Executive Directors or their national authorities.

Recommended citation: International Monetary Fund. 2018. Global Financial Stability Report: A
Bumpy Road Ahead. Washington, DC, April.

Please send orders to:


International Monetary Fund, Publications Services
P.O. Box 92780, Washington, DC 20090, U.S.A.
Tel.: (202) 623-7430 Fax: (202) 623-7201
E-mail: publications@imf.org
www.bookstore.imf.org
www.elibrary.imf.org
CONTENTS

Assumptions and Conventions vii


Further Information viii
Preface ix
Foreword x
Executive Summary xi
IMF Executive Board Discussion Summary xiii

Chapter 1 A Bumpy Road Ahead 1


Outlook for Financial Stability 1
Monetary Policy Normalization in Advanced Economies 5
Reach for Yield or Overreach in Risky Assets? 10
Crypto Assets: New Coin on the Block, Reach for Yield, or Asset Price Bubble? 21
Vulnerabilities in Emerging Markets, Low-Income Countries, and China 26
Funding Challenges of Internationally Active Banks 38
Box 1.1. The VIX Tantrum 47
Box 1.2. An Econometric Lens on What Drives Term Premiums 49
Box 1.3. The Changing Investor Base in the US Leveraged Loan Market 51
Box 1.4. Central Bank Digital Currencies 52
Box 1.5. Regulatory Reform—Tying Up the Loose Ends 53
References 54
Chapter 2 The Riskiness of Credit Allocation: A Source of Financial Vulnerability? 57
Summary 57
Introduction 58
The Riskiness of Credit Allocation: Conceptual Framework 60
The Riskiness of Credit Allocation and Its Evolution across Countries 62
The Riskiness of Credit Allocation and Macro-Financial Stability 67
The Role of Policy and Structural Factors 70
Conclusions and Policy Implications 72
Box 2.1. Measuring the Riskiness of Credit Allocation 74
Box 2.2. Credit Allocation in China: Is Credit Flowing to the Most Profitable Firms? 76
Box 2.3. The Joint Dynamics of the Riskiness of Credit Allocation, Financial Conditions,
Credit Expansions, and GDP Growth 78
Box 2.4. The High-Yield Share during a Credit Boom and Output Growth 80
Annex 2.1. Description and Definition of Variables 81
Annex 2.2. The Determinants of the Riskiness of Credit Allocation 82
Annex 2.3. The Riskiness of Credit Allocation and Macro-Financial Outcomes 87
References 91
Chapter 3 House Price Synchronization: What Role for Financial Factors? 93
Summary 93
Introduction 94
House Price Synchronicity: A Conceptual Framework 97

International Monetary Fund | April 2018 iii


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

House Price Synchronization in Countries and Cities 100


Analyzing Contributors to House Price Synchronization 104
House Price Synchronization and Risks to Growth 108
Policy Discussion 109
Box 3.1. Global Investors, House Price Dispersion, and Synchronicity 111
Box 3.2. Housing as a Financial Asset 113
Box 3.3. The Globalization of Farmland 115
Box 3.4. House Price Gap Synchronicity and Macroprudential Policies 117
Annex 3.1. Data Sources and Country Coverage 119
Annex 3.2. Measuring Synchronization and Country-Pair Analysis 121
Annex 3.3. Technical Annex 125
References 130

Tables
Table 1.1. Correlation of Bitcoin with Key Asset Classes and within Crypto Assets 24
Annex Table 2.1.1. Riskiness of Credit Allocation: Economies Included in the Analysis 82
Annex Table 2.1.2. Country-Level Data Sources and Transformations 83
Annex Table 2.2.1. Cyclicality of the Riskiness of Credit Allocation 84
Annex Table 2.2.2. Impact of Financial Conditions and Lending Standards on the Riskiness of
Credit Allocation 85
Annex Table 2.2.3. Impact of Policy and Institutional Settings on the Riskiness of Credit Allocation 86
Annex Table 2.3.1. Panel Logit Analysis: Probability of the Occurrence of a Systemic Banking Crisis 88
Annex Table 2.3.2. Panel Logit Analysis: Banking Sector Equity Stress Risk 89
Annex Table 2.3.3. Impact of the Riskiness of Credit Allocation on Downside Risks to Growth 90
Annex Table 3.1.1. Data Sources 119
Annex Table 3.1.2. Economies and Cities Included in the Analyses 121
Annex Table 3.2.1. House Price Gap Synchronization at Country Level and Bilateral Linkages 123
Annex Table 3.2.2. House Price Gap Synchronization at Country Level and Global Factors 124
Annex Table 3.3.1. Capital Account Openness and Synchronicity 127
Annex Table 3.3.2. Global Investors, House Price Dispersion, and Synchronicity:
Regression Results 129

Figures
Figure 1.1. Global Financial Conditions 2
Figure 1.2. Growth-at-Risk 4
Figure 1.3. Nonfinancial Private Sector Debt 5
Figure 1.4. Market Interest Rates, Central Bank Balance Sheets, and US Financial Indicators 6
Figure 1.5. US Inflation Expectations and Term Premium 8
Figure 1.6. Term Premium Correlations, Spillovers, and Exchange Rate Relationships 9
Figure 1.7. Valuations of Global Equities 11
Figure 1.8. Valuations of Corporate Bonds 13
Figure 1.9. Leveraged Loan Issuance, Quality, and Developments after Regulatory Guidance 14
Figure 1.10. Correlations and Interconnectedness 16
Figure 1.11. Measures of Leverage and Investment Funds with Derivatives-Embedded Leverage 18
Figure 1.12. Strong Inflows into Exchange-Traded Funds Pose Challenges for the Less Liquid
Fixed-Income Markets 20
Figure 1.13. Crypto Assets: Size, Price Appreciation, Realized Volatility, and Sharpe Ratio 23
Figure 1.14. Share of Trading Volumes across Exchanges, Crypto Assets, and Fiat Currencies 25
Figure 1.15. Improving Fundamentals, Increased Foreign Currency Issuance 27
Figure 1.16. Creditor Base and External Financing Vulnerabilities 29
Figure 1.17. Rising Vulnerabilities and More Complex Creditor Composition 31
Figure 1.18. Stylized Map of Linkages within China’s Financial System 33

iv International Monetary Fund | April 2018


CONTENTS

Figure 1.19. Chinese Banking System and Financial Market Developments and Liabilities 34
Figure 1.20. Risks and Adjustment Challenges in Chinese Investment Products 36
Figure 1.21. Chinese Insurers 37
Figure 1.22. Advanced Economy Bank Health 39
Figure 1.23. US Dollar Credit Aggregates and Bank Intragroup Funding Structures 41
Figure 1.24. Non-US Banks’ International Dollar Balance Sheets 42
Figure 1.25. Non-US Banks’ International US Dollar Liquidity Ratios 44
Figure 1.26. Foreign Exchange Swap and Short-Term Bank Funding Markets 45
Figure 1.1.1. US Asset Prices 47
Figure 1.2.1. Estimated Term Premiums 49
Figure 1.3.1. Nonbanks Have Increased Their Credit Exposure in the US Leveraged Loan Market 51
Figure 2.1. Financial Conditions Have Been Loose in Recent Years 58
Figure 2.2. Low-Rated Nonfinancial Corporate Bond Issuance Has Been High in Some Advanced
Economies 59
Figure 2.3. Key Drivers of the Riskiness of Credit Allocation 61
Figure 2.4. The Riskiness of Credit Allocation Is Cyclical at the Global Level 63
Figure 2.5. Selected Economies: Riskiness of Credit Allocation, 1995–2016 64
Figure 2.6. The Riskiness of Credit Allocation Rises When a Credit Expansion Is Stronger 66
Figure 2.7. The Association between the Size of a Credit Expansion and the Riskiness of Credit
Allocation Is Greater When Lending Standards and Financial Conditions Are Looser 66
Figure 2.8. The Riskiness of Credit Allocation Rises to a High Level before a Financial Crisis,
and Falls to a Low Level Thereafter 67
Figure 2.9. Higher Riskiness of Credit Allocation Signals Greater Risk of a Systemic Banking Crisis 68
Figure 2.10. Higher Riskiness of Credit Allocation Signals Greater Risk of Banking Sector Stress 68
Figure 2.11. Higher Riskiness of Credit Allocation Signals Higher Downside Risks to GDP Growth 69
Figure 2.12. The Association of the Riskiness of Credit Allocation with Downside Risks to GDP
Growth Depends on the Size of Credit Expansion 70
Figure 2.13. The Association of a Credit Expansion with the Riskiness of Credit Allocation Depends
on Policy and Institutional Settings 71
Figure 2.1.1. Measuring the Riskiness of Credit Allocation 74
Figure 2.1.2. Histograms of Measures of the Riskiness of Credit Allocation 75
Figure 2.2.1. China: Profitability of Credit Allocation, 1997–2016 76
Figure 2.2.2. China: Profitability of Credit Allocation, by Ownership and Sector 77
Figure 2.3.1. The Riskiness of Credit Allocation and Financial Conditions 78
Figure 2.4.1. Impulse Response of Cumulative Real GDP Growth to a High-Yield Share Shock
Given a Credit Boom 80
Figure 3.1. House Price Gains in Selected Cities and Countries Have Been Widespread 94
Figure 3.2. Widespread House Price Gains Have Accompanied Accommodative Financial Conditions 95
Figure 3.3. Institutional Investor Participation Has Been on the Rise 96
Figure 3.4. Global Financial Conditions, Portfolio Channels, and Expectations Contribute to House
Price Synchronization, as Do Supply Constraints and Local Policy 98
Figure 3.5. Synchronization Has Steadily Increased across Countries and Cities 100
Figure 3.6. The Relative Contribution of the Global Factor Has Grown 101
Figure 3.7. Instantaneous Quasi Correlation of House Price Gaps Shows Financial Cycle Properties 102
Figure 3.8. Relative Contribution of the Global Factor Varies across Regions 103
Figure 3.9. Economies Differ in Their House Price Interconnectedness 104
Figure 3.10. Interconnectedness among Cities’ House Prices Varies 105
Figure 3.11. Average Country-Level Housing Market Spillovers Have Increased 106
Figure 3.12. Bilateral Links between Countries Are Associated with House Price Synchronization 106
Figure 3.13. Greater Financial Openness Is Associated with Higher House Price Synchronization 107
Figure 3.14. On Average, the Global Factor for House Prices Has Increased along with That
for Equities 107

International Monetary Fund | April 2018 v


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 3.15. Global Financial Conditions, as Proxied by Global Liquidity, Have Different
Associations with House Price Synchronization across Countries and Cities 108
Figure 3.16. House Price Synchronization Predicts a Downside Risk to Economic Growth at
Short Horizons 108
Figure 3.1.1. Real House Prices in 40 Largest US Cities by Population 111
Figure 3.2.1. Housing Return Predictability 113
Figure 3.2.2. Predictability of Returns on Housing and Capital Account Openness 113
Figure 3.3.1. Large-Scale Land Acquisitions over Time by Target Region 115
Figure 3.4.1. Macroprudential Tools Indirectly Reduce House Price Synchronicity 118

Online Annexes
Annex 1.1. Option-Implied Volatility: The Quantity and Price of Risk for Stocks and Bonds
Annex 1.2. Bank International Dollar Funding Methodology

&EJUPShT/PUF "QSJM 

5IJTPOMJOFWFSTJPOPGUIF('43IBTCFFOVQEBUFEUPSFGMFDUUIFGPMMPXJOHDIBOHFTUPUIF
QSJOUWFSTJPO
0OQBHF 'JHVSF
UIFEBUBJOQBOFMIBWFCFFODPSSFDUFE
0OQBHF 'JHVSF
UIFYBYJTMBCFMTJOQBOFMIBWFCFFODPSSFDUFE

vi International Monetary Fund | April 2018


1
CHAPTER

ASSUMPTIONS AND CONVENTIONS

The following conventions are used throughout the Global Financial Stability Report (GFSR):
. . . to indicate that data are not available or not applicable;
— to indicate that the figure is zero or less than half the final digit shown or that the item does not exist;
– between years or months (for example, 2016–17 or January–June) to indicate the years or months covered,
including the beginning and ending years or months;
/ between years or months (for example, 2016/17) to indicate a fiscal or financial year.
“Billion” means a thousand million.
“Trillion” means a thousand billion.
“Basis points” refers to hundredths of 1 percentage point (for example, 25 basis points are equivalent to ¼ of
1 percentage point).
If no source is listed on tables and figures, data are based on IMF staff estimates or calculations.
Minor discrepancies between sums of constituent figures and totals shown reflect rounding.
As used in this report, the terms “country” and “economy” do not in all cases refer to a territorial entity that is a state
as understood by international law and practice. As used here, the term also covers some territorial entities that are
not states but for which statistical data are maintained on a separate and independent basis.
The boundaries, colors, denominations, and any other information shown on the maps do not imply, on the part
of the International Monetary Fund, any judgment on the legal status of any territory or any endorsement or
acceptance of such boundaries.

International Monetary Fund | April 2018 vii


FURTHER INFORMATION

Corrections and Revisions


The data and analysis appearing in the Global Financial Stability Report are compiled by the IMF staff at the
time of publication. Every effort is made to ensure their timeliness, accuracy, and completeness. When errors are
discovered, corrections and revisions are incorporated into the digital editions available from the IMF website and
on the IMF eLibrary (see below). All substantive changes are listed in the online tables of contents.

Print and Digital Editions


Print copies of this Global Financial Stability Report can be ordered at https://www.bookstore.imf.org/books/title/
global-financial-stability-report-april-2018.

The Global Financial Stability Report is featured on the IMF website at http://www.imf.org/publications/gfsr.
This site includes a PDF of the report and data sets for each of the charts therein.

The IMF eLibrary hosts multiple digital editions of the Global Financial Stability Report, including ePub,
enhanced PDF, Mobi, and HTML: http://elibrary.imf.org/Apr18GFSR.

Copyright and Reuse


Information on the terms and conditions for reusing the contents of this publication are at http://www.imf.org/
external/terms.htm.

viii International Monetary Fund | April 2018


PREFACE

The Global Financial Stability Report (GFSR) assesses key risks facing the global financial system. In normal times,
the report seeks to play a role in preventing crises by highlighting policies that may mitigate systemic risks, thereby
contributing to global financial stability and the sustained economic growth of the IMF’s member countries.
The analysis in this report has been coordinated by the Monetary and Capital Markets (MCM) Department
under the general direction of Tobias Adrian, Director. The project has been directed by Fabio Natalucci and
Dong He, both Deputy Directors, as well as by Claudio Raddatz and Anna Ilyina, both Division Chiefs. It has
benefited from comments and suggestions from the senior staff in the MCM Department.
Individual contributors to the report are Ali Al-Eyd, Adrian Alter, Sergei Antoshin, Anil Ari, Magally Bernal,
Christian Bogmans, Luis Brandão-Marques, Peter Breuer, Jeroen Brinkhoff, John Caparusso, Qianying Chen,
Sally Chen, Yingyuan Chen, Kevin Chow, Fabio Cortes, Jane Dokko, Dimitris Drakopoulos, J. Benson Durham,
Martin Edmonds, Alan Xiaochen Feng, Rohit Goel, Tryggvi Gudmundsson, Hideo Hashimoto, Sanjay Hazarika,
Frank Hespeler, Henry Hoyle, Mohamed Jaber, David Jones, Mitsuru Katagiri, Will Kerry, Oksana Khadarina,
Ashraf Khan, Divya Kirti, Robin Koepke, Romain Lafarguette, Jiaqi Li, Yang Li, Sheheryar Malik,
Rebecca McCaughrin, Aditya Narain, Thomas Piontek, Jochen Schmittmann, Dulani Seneviratne, Juan Solé,
Ilan Solot, Nour Tawk, Jérôme Vandenbussche, Jeffrey Williams, Peichu Xie, and Akihiko Yokoyama.
Magally Bernal, Claudia Cohen, Breanne Rajkumar, and Han Zaw were responsible for word processing.
Gemma Diaz from the Communications Department led the editorial team and managed the report’s
production with support from Linda Kean and editorial assistance from Sherrie Brown, Lucy Scott Morales,
Nancy Morrison, Katy Whipple, AGS, and Vector Talent Resources.
This particular issue of the GFSR draws in part on a series of discussions with banks, securities firms, asset
management companies, hedge funds, standard setters, financial consultants, pension funds, central banks, national
treasuries, and academic researchers.
This GFSR reflects information available as of March 30, 2018. The report benefited from comments and
suggestions from staff in other IMF departments, as well as from Executive Directors following their discussion of
the GFSR on April 2, 2018. However, the analysis and policy considerations are those of the IMF staff and should
not be attributed to Executive Directors or their national authorities.

International Monetary Fund | April 2018 ix


FOREWORD

S
tarting with this report, Chapter 1 of the interest rates after years of very easy financial conditions
Global Financial Stability Report (GFSR) will and take active steps to reduce these risks. Asset price
regularly provide a quantitative assessment of spillovers have important implications for the housing
the degree to which future GDP growth faces market. As explained in Chapter 3, house price correla-
downside risks from financial vulnerabilities, using a tions across countries and across major cities have been
Growth-at-Risk (GaR) framework. The GaR approach trending up during the past 30 years, suggesting that
links financial conditions to the distribution of future spillovers via the housing sector may play a prominent
GDP growth outcomes and provides a framework role in a future crisis.
for assessing the trade-off between supporting growth A variety of indicators point to vulnerabilities from
in the short term and putting financial stability and financial leverage, a deterioration in underwriting
future growth at risk over the medium term. Our standards, and ever more pronounced reaching for yield
current assessment through the prism of GaR is that, behavior by investors in corporate and sovereign debt
over the past six months, short-term downside risks markets around the world. Chapter 2 presents an inno-
to global financial stability have increased somewhat, vative gauge of the riskiness of credit allocation. The new
reflecting somewhat tighter financial conditions amid metric computes the difference in vulnerability between
investors’ concerns about newly announced trade mea- the firms with the largest and smallest expansions in
sures. Even so, still-accommodative financial condi- debt. This indicator exhibits strong forecasting power
tions continue to be supportive of economic growth. for downside risks to GDP growth, and is currently at
Taking a longer view, downside risks, as measured by medium to elevated levels in several countries. A host
GaR, remain large: easy financial conditions continue of more conventional metrics of corporate debt vulner-
to fuel financial vulnerabilities, leaving the global ability around the world, including a deterioration in
economy exposed to the risk of a sharp tightening nonprice terms and underwriting standards in debt deals,
in financial conditions. Policymakers thus face the suggest that market risks are rising, as easy financial con-
twin challenges of continuing to support growth in ditions support high issuance and strong global capital
the short term by keeping monetary policy accom- flows. In low-income countries, the share of private and
modative as well as reining in rising financial stability non–Paris Club creditors is increasing, and greater use of
risks in the medium term by deploying micro- and collateralized debt exposes borrowing countries to poten-
macroprudential policy tools. tially costly debt restructurings in the future.
Managing the gradual process of monetary policy Over the past year, crypto assets trading has emerged
normalization will be tricky against this backdrop of as a new potential vulnerability. Price volatility of
elevated medium-term risks, and will require careful crypto assets has been much higher than that of com-
communication from central banks and policymakers modities, currencies, or stocks. Financial stability risks
to reduce the risks from a sharp tightening of financial could arise from leveraged positions taken by inves-
conditions. The spike in volatility in global equity mar- tors in this new asset class, infrastructure weaknesses
kets in early February has brought into focus the risk of cryptocurrency exchanges, and fraud, in addition
of abrupt, adverse feedback loops in a period of asset to elevated volatility. Regulators around the world are
price adjustments. The recently increased trade tensions responding to the growing use of crypto assets through
have led to investors’ jitters, and a wider escalation of various measures, including enforcement actions, indi-
protectionist measures could ultimately take a toll on rect interventions via the banking system, and outright
the global economy and on global financial stability. bans. While crypto assets may generate new vulner-
Many markets still have stretched valuations, and may abilities, they also create opportunities and, indeed, a
experience bouts of volatility in the period ahead, in the number of central banks around the world are consid-
context of continued monetary policy normalization in ering the issuance of central bank digital currency.
some advanced countries. Investors and policymakers Tobias Adrian
should be cognizant of the risks associated with rising Financial Counsellor

x International Monetary Fund | April 2018


EXECUTIVE SUMMARY

T
he global economic outlook has continued Valuations of risky assets are still stretched, with
to improve, as discussed in the April 2018 some late-stage credit cycle dynamics emerging,
World Economic Outlook, with the pace reminiscent of the precrisis period. This makes markets
of economic growth picking up and the exposed to a sharp tightening in financial conditions,
recovery becoming more synchronized around the which could lead to a sudden unwinding of risk premi-
world. While still supportive of economic growth, ums and a repricing of risky assets. Moreover, liquidity
global financial conditions have tightened somewhat mismatches and the use of financial leverage to boost
since the October 2017 Global Financial Stability returns could amplify the impact of asset price moves
Report (GFSR). Such a tightening reflects primarily on the financial system. Although no major disrup-
the bout of equity volatility in early February and a tions were reported during the episode of volatility in
decline in risky asset prices at the end of March follow- early February, market participants should not take
ing concerns about a wider escalation of protectionist too much comfort. Investors and policymakers must
measures. remain attuned to the risks associated with higher
Short-term risks to financial stability have increased interest rates and greater volatility. Policymakers should
somewhat relative to the previous GFSR, and medium- address financial vulnerabilities by using more actively
term risks continue to be elevated. Financial vulner- the micro- and macroprudential tools at their disposal
abilities, which have accumulated during years of or by enhancing their toolkits as needed—for example,
extremely low rates and volatility, could make the road to address risks in the nonbank financial sector.
ahead bumpy and could put growth at risk. Indeed, The banking sector has become more resilient since
Growth-at-Risk analysis (described in Chapter 3 of the the global financial crisis. However, it is important
October 2017 GFSR) shows that risks to medium- to ensure that the postcrisis regulatory reform agenda
term economic growth, stemming from easy financial is completed. In advanced economies some weaker
conditions, remain well above historical norms. banks still need to strengthen their balance sheets, and
In advanced economies, stronger growth momen- some institutions operating internationally run dollar
tum and the firming of inflation have eased to some liquidity mismatches. A sudden spell of turbulence in
extent a key challenge facing central banks: maintain- financial markets could expose these mismatches and
ing the monetary accommodation required to support crystallize dollar funding strains.
the economic recovery while addressing medium-term A number of emerging market economies have
financial vulnerabilities. But the firming of inflation taken advantage of an extended period of benign exter-
also brings risks. For example, inflation may pick up nal financial conditions to improve their fundamentals.
faster than currently anticipated, possibly propelled However, they could be vulnerable to a sudden tight-
by significant fiscal expansion enacted in the United ening of global financial conditions or spillovers from
States. Central banks may respond to higher inflation monetary policy normalization in advanced economies,
more aggressively than currently expected, which could resulting in an increase in risk aversion and capital
lead to a sharp tightening of financial conditions. This flow reversals. The severity of such potential shocks
tightening could spill over to risky asset prices, bank will differ across countries, depending on economic
dollar funding markets, and both emerging market fundamentals and the policy responses to those shocks.
economies and low-income countries, as discussed Although regulators in China have taken steps to
below. To minimize these risks, central banks should address risks stemming from the interconnectedness of
continue to normalize monetary policy gradually and the banking and shadow banking sectors, vulnerabili-
communicate their decisions clearly to support the ties remain high. Further regulatory actions are crucial
economic recovery. to continue reducing risks in the financial sector.

International Monetary Fund | April 2018 xi


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

The technology behind crypto assets has the poten- ability of banking stress, in addition to the previously
tial to make the financial market infrastructure more documented signals provided by credit growth. Coun-
efficient. However, crypto assets have been afflicted by try authorities can use the measures introduced in this
fraud, security breaches, and operational failures, and chapter to monitor the buildup of vulnerabilities via
have been associated with illicit activities. At present, risk taking in credit allocation. The chapter discusses
crypto assets do not appear to pose financial stability policies that can mitigate the increase in credit riski-
risks, but they could do so should their use become ness during credit expansions.
more widespread without appropriate safeguards. Chapter 3 documents a striking increase in house
Chapter 2 takes a comprehensive look at the evolu- price synchronization among 40 countries and 44
tion of the riskiness of corporate credit allocation, major cities in advanced and emerging market econo-
given concerns that the continued search for higher mies over the past several decades. The exposure of
yield may have led banks and investors to extend too countries and cities to global financial conditions may
much credit to risky borrowers. The chapter docu- help explain that increase. Rising housing valuations
ments a pattern in which the firms obtaining more since the global financial crisis raise the specter of a
credit are relatively riskier during periods of strong simultaneous decline in house prices should financial
credit expansion, especially when lending standards conditions reverse. The chapter suggests that height-
are loose or financial conditions are easy. An increase ened synchronicity of house prices can signal a higher
in the riskiness of credit allocation signals heightened probability of adverse scenarios for the real economy,
downside risks to GDP growth and a higher prob- especially when credit is high or rapidly expanding.

xii International Monetary Fund | April 2018


IMF EXECUTIVE BOARD DISCUSSION SUMMARY

The following remarks were made by the Chair at the conclusion of the Executive Board’s discussion of the
Fiscal Monitor, Global Financial Stability Report, and World Economic Outlook on April 2, 2018.

E
xecutive Directors broadly shared the key protectionism and inward-looking policies. Record-
messages of the flagship reports and found high levels of global debt, geopolitical tensions, and
the analytical chapters topical, relevant, and climate events also threaten global growth prospects.
insightful. They welcomed the broadbased Against this backdrop, Directors underscored that
recovery of the global economy, supported by a pickup the cyclical upswing provides a golden opportunity to
in investment and trade. Directors observed that global advance policies and reforms to strengthen medium-
growth is expected to rise further in the near term. term prospects and reduce vulnerabilities. Priorities are
Meanwhile, inflation remains muted in many countries. to raise potential output, ensure the gains are widely
Subdued labor productivity growth and population shared, enhance economic and financial resilience, and
aging continue to hold back growth in advanced econo- safeguard debt sustainability. Directors stressed that
mies. While the recent commodity price increase has a multilateral framework that is open, resilient, and
supported a recovery in commodity-dependent emerg- adhered to by all can support growth and benefit the
ing market and developing economies, the ongoing global economy. Enhanced commitment to multilateral
adjustment processes continue to weigh on growth. cooperation is particularly needed to reduce trade bar-
Directors agreed that risks around the short-term riers and distortionary trade practices, and to promote
outlook are broadly balanced, but beyond the next a rule-based multilateral trading system that works for
several quarters, risks are tilted to the downside. On all. Directors also called for multilateral cooperation to
the upside, the cyclical pickup in advanced economy further reduce incentives for cross-border profit shifting
growth may prove stronger than expected as slack in and tax evasion, avoid tax competition, implement
labor markets may be larger than currently assessed. the postcrisis financial regulatory reform agenda, and
On the downside, a sharp tightening of global finan- address other shared challenges such as refugees, secu-
cial conditions could have negative repercussions for rity threats, cyber risks, and climate change. Reducing
growth, while financial vulnerabilities accumulated excess external imbalances requires policy efforts to lift
over years of low interest rates could amplify the the contribution of domestic sources of growth above
impact of asset price movements on the financial sys- overall GDP growth in surplus countries and to boost
tem, putting growth at risk in the medium term. Most potential output and saving in deficit countries.
Directors noted that the tax reform in the United Directors concurred that monetary accommodation
States is procyclical and may trigger inflation pressure should continue in advanced economies with infla-
and a faster-than-anticipated withdrawal of monetary tion below target. Where output is close to poten-
accommodation, as well as widen global imbalances, tial and inflation is rising toward target, a gradual,
although the view was also expressed that the reform data-dependent, and well-communicated withdrawal
would boost investment and efficiency, and thus move of monetary support is warranted. Directors sup-
the US economy to a higher, sustainable growth path. ported the call for fiscal policy to start rebuilding
An abrupt tightening of global financial conditions, buffers now, where appropriate, to create room for an
especially if accompanied by capital flow reversals, eventual downturn and prevent fiscal vulnerabilities
could be challenging for several emerging markets and from becoming a source of stress. Fiscal adjustment is
low-income developing countries, notwithstanding warranted in most countries, calibrated to avoid pro-
improved resilience of their financial systems. Down- cyclicality and anchored on fiscal reforms that increase
side risks are particularly evident from escalating trade productivity and promote human and physical capital.

International Monetary Fund | April 2018 xiii


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

In countries that have ample fiscal space and are ing market and developing economies enhance their
operating at or close to capacity, fiscal policy should be resilience to external shocks. Directors welcomed
used to facilitate growth-enhancing structural reforms. China’s progress in reducing financial vulnerabilities
Directors also saw a role for fiscal policy in promot- and encouraged further efforts to strengthen its regula-
ing equality, and for labor and immigration policies in tory and supervisory frameworks, particularly in the
boosting labor supply. shadow banking sector.
Directors agreed that digitalization presents both Directors noted that low-income developing coun-
opportunities and risks. Digitalization can reduce tax tries face multiple challenges in their effort to progress
compliance costs, improve spending efficiency, and toward the 2030 Sustainable Development Goals.
enhance social protection. At the same time, it cre- They expressed concern over the broad-based increase
ates challenges for fiscal policy and the international in public debt burdens, the increasing number of
tax system. Directors noted that mitigating risks countries at high risk of debt distress, and data gaps.
from digitalization would require a comprehensive These underscore the urgent need for fiscal prudence,
reform agenda, adequate resources, and a coordinated improved debt management capacity, and greater debt
approach toward a long-term vision of the interna- transparency on the part of both debtors and credi-
tional tax architecture. tors, as well as concerted efforts from the international
Directors welcomed the increased resilience of community. Several countries need to make room in
the banking system and stressed the importance of their budgets to accommodate higher spending on
completing and implementing the postcrisis regula- social services, such as health care and education, and
tory reform agenda. They encouraged policymakers public investment, by mobilizing domestic revenues
to develop and deploy micro- and macroprudential and improving spending efficiency. Commodity
tools to address financial vulnerabilities, and to closely exporters and those vulnerable to climate-related events
monitor risks related to credit allocation and increas- face additional complex challenges of diversifying their
ingly synchronized house prices across countries. The economies. While country circumstances differ, com-
global implications of Brexit-related challenges also call mon priorities for promoting economic diversification
for close cross-border cooperation. Directors concurred and employment include increasing access to credit,
that, while crypto assets do not pose an immediate expanding vocational skills training, and improving the
threat to financial stability, if widely used, they may quality of infrastructure.
raise issues about investor and consumer protection, Directors expressed concern over the stalled progress
money laundering, and tax evasion. in the catching-up process of emerging market and
Directors agreed that enhancing the quality of developing economies. They noted that, to facilitate
credit intermediation, avoiding credit booms that income convergence, policies should aim to strengthen
lead to excessive risk taking, and, where feasible, governance, improve educational and health outcomes,
permitting exchange rate flexibility can help emerg- and lower entry barriers for new firms.

xiv International Monetary Fund | April 2018


1
CHAPTER

A BUMPY ROAD AHEAD

Outlook for Financial Stability Global financial conditions have tightened some-
Despite ongoing monetary policy normalization in what, on balance, since the October 2017 Global
some advanced economies and some signs of firming Financial Stability Report (GFSR), reflecting the spike
inflation, global financial conditions are still very in equity market volatility in early February and inves-
accommodative relative to historical norms. Although tors’ jitters in late March about a wider escalation of
supportive of near-term growth, easy financial condi- trade tensions (Figure 1.1, panel 1). Nonetheless, even
tions also continue to facilitate a buildup of financial as the US Federal Reserve has continued to normalize
fragilities, increasing risks to global financial stabil- monetary policy, global financial conditions remain
ity and economic growth over the medium term. broadly accommodative relative to historical norms
across both advanced and emerging market econo-
Still-Easy Financial Conditions Continue to Support mies. Figure 1.1 (panels 1 and 2) shows global and
Economic Growth regional financial conditions indices (FCIs), as well as
their key components.
With global economic recovery now stronger and
Although still-easy financial conditions support
more synchronized (as discussed in the April 2018
economic growth in the near term, they may also
World Economic Outlook [WEO]), monetary pol-
contribute to a buildup of financial imbalances,
icy authorities in advanced economies have started
excessive risk taking, and mispricing of risks. The
to, or are gearing up to, normalize their monetary
growth-at-risk (GaR) approach—which links finan-
policy stance (see “Monetary Policy Normalization in
cial conditions to the distribution of future GDP
Advanced Economies”). Over the years since the global
growth outcomes—provides a framework for assess-
financial crisis, accommodative monetary policy has
ing the intertemporal trade-off between supporting
been crucial to ensuring a sustainable global economic
growth in the near term and putting financial stabil-
recovery. But with inflation well below target and
ity and future growth at risk over the medium term.1
buoyant market sentiment, central banks in advanced
The key steps in this approach are as follows: First, a
economies have faced a difficult balancing act of
model of output growth is estimated as a function of
keeping interest rates low to support the economy
current economic and financial conditions. Second,
and addressing financial vulnerabilities that could put
this model is used to forecast conditional distribu-
growth at risk in the medium term. The recent firming
tions of growth for different horizons. Finally, to
of inflation has provided policymakers with more
gauge the impact of financial conditions on growth
leeway to address financial vulnerabilities, including
prospects, changes in the forecasted severely adverse
by deploying and developing micro- and macropru-
growth outcomes (those that occur with a 5 percent
dential tools.
probability, also called the “tail” of the distribution)
Prepared by staff from the Monetary and Capital Markets Depart- for different horizons are compared with previous
ment (in consultation with other departments): Fabio Natalucci forecasts. Changes in financial conditions that result
(Deputy Director), Anna Ilyina (Division Chief), J. Benson Durham in a deterioration in severely adverse growth forecasts
(Advisor), Hideo Hashimoto (Advisor), Ali Al-Eyd (Deputy Division
Chief), Peter Breuer (Deputy Division Chief), Will Kerry (Deputy
(that is, a leftward shift in the tail) can be interpreted
Division Chief), Sergei Antoshin, Magally Bernal, Jeroen Brinkhoff, as financial vulnerabilities potentially increasing
John Caparusso, Sally Chen, Yingyuan Chen, Kevin Chow, Fabio toward macrocritical levels. This means that these
Cortes, Dimitris Drakopoulos, Martin Edmonds, Rohit Goel, Tryg-
vulnerabilities could magnify the severity of an eco-
gvi Gudmundsson, Sanjay Hazarika, Frank Hespeler, Henry Hoyle,
Mohamed Jaber, David Jones, Ashraf Khan, Robin Koepke, Yang
Li, Sheheryar Malik, Rebecca McCaughrin, Aditya Narain, Thomas
Piontek, Jochen Schmittmann, Juan Solé, Ilan Solot, Nour Tawk, 1See Chapter 3 of the October 2017 GFSR for a description of

Jeffrey Williams, Akihiko Yokoyama, and Han Zaw. the growth-at-risk methodology.

International Monetary Fund | April 2018 1


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 1.1. Global Financial Conditions

Global financial conditions have tightened somewhat, but remain supportive of growth.
1. Global Financial Conditions Index
(Standard deviations)
4
2 Tightening
0
–2 2017:Q3
–4
–6
± one standard deviation from 2017:Q3
–8
–10
2012 13 14 15 16 17 18:Q1

The price of risk is low, markets are buoyant, and leverage is high across both advanced and emerging market economies.
2. Global and Regional FCIs and Their Components
Quintiles
Tighter Easier
Global FCI Components Oct. 2017 GFSR
1. Short-term rates
2. Long-term real rates
3. Term spread
4. Corporate spread
5. Interbank spread
6. Equity price growth
7. Equity return volatility
8. Credit to GDP
9. Credit growth
10. House price growth
Global FCI (1–10)
Regional FCIs
United States
Euro area
Japan
China
EM excluding China
Regional Price of Risk FCIs
United States
Euro area
Japan
China
EM ex-China
1991 92 93 94 95 96 97 98 99 2000 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18

Sources: Bloomberg Finance L.P.; and IMF staff estimates.


Note: Panel 1 shows the Global (Financial Conditions Index) FCI. This was originally presented in GFSR October 2017 (Chapter 3). Higher values of FCIs indicate
tighter conditions. The shaded area denotes ± one standard deviation changes relative to the level of Global FCI at 2017:Q3. Panel 2 shows quintiles of global and
regional FCI series and components relative to their own history. Results are compared with a “Price of Risk” FCI, encompassing price-based information only
(components 1–7). Easing of conditions is shown in blue and tightening in yellow. For FCI components, the shading is based on their contribution to the FCI index,
e.g., a narrowing of credit spreads relative to historical norms would be contributing to the FCI easing, and hence, shown in blue. EM = emerging market;
FCI = financial conditions index; GFSR = Global Financial Stability Report.

2 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

nomic downturn in the future, even without neces- tions, which could provide a separate and substantial
sarily leading to a systemic financial crisis. headwind to growth.

At the same time, easy financial conditions risk fuel-


Short-Term Risks Have Increased Somewhat, while
ing financial vulnerabilities that may put medium-term
Medium-Term Vulnerabilities Remain Elevated
growth at risk. The estimated three-year-ahead growth
Against the backdrop of slightly tighter financial
distribution has a much fatter left tail compared with
conditions, short-term financial stability risks have
the one-year-ahead growth distribution (Figure 1.2,
increased somewhat since the previous GFSR. Even so,
panel 2). Given current conditions, the GaR model
the current broadly accommodative financial condi-
forecasts that, under the severely adverse scenario,
tions appear to have dampened the near-term risks
global growth will be negative three years from
to growth relative to a few years ago. The GaR model
now. The downward slope of the curve (the dashed
forecasts that, under current financial conditions, the
red line in Figure 1.2, panel 3) illustrates the inter-
severely adverse outcome is for global growth to fall to
temporal trade-off between the near-term and the
about 3 percent or less over the following year (the red
medium-term growth prospects amid easy financial
dot in Figure 1.2, panel 1). In comparison, in 2015
conditions. Continued easing of financial conditions
the predicted range of severely adverse growth out-
over the past two years has tilted the curve, improving
comes was notably less favorable.2
economic prospects in the near term while worsen-
This assessment, however, does not mean that the
ing the medium-term growth outlook. In contrast,
global financial system and the real economy are
the severely adverse medium-term growth forecast at
immune to macroeconomic, geopolitical, or policy
the end of 2016 (the dashed blue line in Figure 1.2,
shocks in the near-term:
panel 3), for example, was relatively less negative than
• For example, inflation in the United States may
rise faster than expected, possibly owing to the the current forecast. Finally, a comparison of GaR
recent fiscal expansion. Central banks in response severely adverse medium-term growth forecasts since
may tighten monetary policy more forcefully than the 1990s suggests that risks to medium-term growth
currently anticipated. In such a scenario financial stemming from the current easy financial conditions
conditions could tighten sharply, generating adverse are well above historical norms (Figure 1.2, panel 4).
spillovers to other advanced (see “Monetary Policy As central banks continue to normalize monetary
Normalization in Advanced Economies” section) policy, financial vulnerabilities foreshadow a bumpy
and emerging market economies (see “Vulnerabili- road ahead. High leverage and other balance sheet
ties in Emerging Markets, Low-Income Countries, mismatches tend to amplify the impact of shocks on
and China” section), as well as adversely affecting the financial system and the broader economy. Leverage
the internationally active banks that rely on dollar in the nonfinancial sector has been rising in many major
funding (see “Funding Challenges of Internationally economies, as discussed in the October 2017 GFSR, and
Active Banks” section). remains high (Figure 1.1, panel 2, and Figure 1.3, panel
• Trade tensions and greater protectionism could 1), implying that aggregate debt-service ratios could
affect financial stability via increased uncertainty deteriorate quickly once financial conditions tighten
and lower global growth. As discussed in the April (see “Reach for Yield or Overreach in Risky Assets?”
2018 WEO, a wider escalation of protectionist mea- and “Vulnerabilities in Emerging Markets, Low-Income
sures would take a toll on global output and welfare, Countries, and China” sections). In addition, some
both directly and indirectly by raising geopolitical economies with already-high nonfinancial sector debt are
tensions. This would shift the distribution of global seeing faster growth in house prices (see gray dots in the
growth outcomes to the left, with attendant negative upper right corner in Figure 1.3, panel 2). In contrast,
implications for global financial stability. But even banks have raised their capital and liquidity buffers since
before any impact on trade, there may be a decline the global financial crisis, pointing to increased resil-
in confidence and a tightening in financial condi- ience, though they may still be vulnerable to funding
shocks (see “Funding Challenges of Internationally
2As can be seen in Figure 1.1, global financial conditions have Active Banks” section). At the same time, use of financial
eased significantly since 2015–16. leverage outside the banking sector is on the rise as the

International Monetary Fund | April 2018 3


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 1.2. Growth-at-Risk

Supportive financial conditions tend to dampen the near-term risks, with growth-at-risk forecasting the severely adverse outcome (for example,
with 5 percent probability) for global growth at about 3 percent or less one year ahead.
1. Percentiles of One-Year-Ahead Growth Forecast Densities
(Percent)
6 6
4 4
2 2
0 0
–2 –2
–4 –4
–6 –6
–8 5th percentile –8
–10 Median –10
95th percentile
–12 –12
–14 –14
1991 93 95 97 99 2001 03 05 07 09 11 13 15 17 18
Q1

But easy financial conditions also raise the odds of adverse growth Medium-term risks to growth have increased in recent years ...
outcomes in the medium term—the three-year-ahead growth
distribution has a much fatter left tail than the one-year-ahead growth
distribution.
2. Growth Forecast Densities at 2018:Q1 3. Severely Adverse Growth Outcomes (5th Percentile)
(Probability) at Different Future Horizons
(Percent)
1.2 3.35
One year ahead Forecast path at 2018:Q1
1.0 Three years ahead At 2017:Q3
At 2016:Q4 2.35
0.8

0.6 1.35
5th percentile
0.4
0.35
0.2

0.0 –0.65
–1 0 1 2 3 4 5 6 One year Two years Three years
Percent ahead ahead ahead

... and are well above historical norms, given the current financial conditions.
4. GaR Forecasts of Severely Adverse Growth Outcomes: Percentile Ranks
Quintiles
Worst Best

One year ahead

Three years ahead

1991 92 93 94 95 96 97 98 99 2000 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18

Sources: Bloomberg Finance L.P.; IMF, World Economic Outlook database; and IMF staff estimates.
Note: Growth-at-risk (GaR) refers to the set of outcomes that fall into the 5th percentile of (conditional) forecast densities of global growth. Panel 2 presents forecast
densities for growth, one and three years ahead. In panel 4, the color shading depicts the percentile rank for the 5th percentile threshold of densities for one-year-
and three-year-ahead growth. Red denotes lower growth outcomes.

4 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

prolonged period of low interest rates has fueled search Figure 1.3. Nonfinancial Private Sector Debt
for yield and compressed market risk measures (see
1. Credit-to-GDP Ratio
“Reach for Yield or Overreach in Risky Assets?” section). (Percent)
Although the recent bout of volatility in global 220 Advanced economies without
200
equity markets (Box 1.1) did not lead to any major 180
Advanced economies a crisis in 2007–08
hit by a crisis in
dislocations, the episode underscores the need for 160 2007–08
investors and policymakers to remain attuned to the 140
120
risks associated with rising interest rates after years of 100
low rates and low volatility. 80 China
60
40
Monetary Policy Normalization in 20
Emerging market economies
(excluding China)
Advanced Economies 0
1970 75 80 85 90 95 2000 05 10 15
The buildup of financial vulnerabilities over the past
few years has left financial markets exposed to the risk 2. House Price Growth and Credit, 2017:Q3
of a sharp tightening of financial conditions. In this HUN
10
context, central banks must strike a delicate balance of CAN IRL
CZE ISR AUS SWE

(percent, year over year)


Real house price growth
gradually withdrawing monetary policy accommoda- 5
IND TUR AUT USA GBR CHN NLD
POL DEU PRT NOR
tion while avoiding disruptive volatility in financial MEX COL THA
MYS ESP KOR DNK
JPN BEL CHE
markets. This balancing act highlights the importance 0
IDN CHL FIN
ZAF ITA FRA
of continued clarity in central bank communications. GRC
5
BRA
The Global Economy Faces a Critical Transition as 10
RUS
Monetary Policy Gradually Normalizes
0 50 100 150 200 250 300
Financial markets have thus far adjusted relatively Credit-to-GDP ratio (percent)
smoothly to the gradual pace of monetary policy
normalization, benefiting from clear central bank Sources: Bank for International Settlements (BIS); Haver Analytics; IMF, World
Economic Outlook database; and IMF staff calculations.
communications and historically large central bank Note: In panel 1, advanced economies hit by a crisis are those in the Laeven and
asset holdings (Figure 1.4, panel 1). Indeed, although Valencia (2012) database with a crisis starting in 2007 or 2008 that are also in the
the expected path of policy interest rates in the United Bank for International Settlements credit database. Panel 2 shows data for
2017:Q3 or latest available figures, if earlier. House price growth is calculated over
States points to a faster pace of tightening relative the past three years (annualized). Data for credit-to-GDP ratios are from the BIS.
to other advanced economies, reflecting differences For some countries, these data include inter- and intra-company loans, which can
be significant. Countries with gray dots have a credit-to-GDP ratio above the
in the interest-rate-hiking cycle, it remains consis- median historical ratio for advanced or emerging market economies. The median is
tent with gradual removal of accommodation (Fig- calculated using annual data from 1970 (or first available date, if later) for a pooled
sample of 27 advanced economies and 16 emerging market economies. Green
ure 1.4, panel 2). dots are countries with a credit-to-GDP ratio below the median. Data labels in the
But policymakers may face increasing challenges panel use International Organization for Standardization (ISO) country codes.
to ensuring a smooth normalization path. Substantial
medium-term financial vulnerabilities have built up
during the period of prolonged monetary accommo- potential risk underscores the importance of a smooth
dation. As central banks withdraw accommodation by process to avoid sudden, sharp volatility and disrup-
raising short-term interest rates and shrinking their tions in financial markets.
balance sheets, a decompression of term premiums (the
compensation investors demand for holding bonds in
excess of risk-free short-term interest rates) may cause Why Have Term Premiums Remained Low in the United
an abrupt tightening of financial conditions.3 This States Even as the Federal Reserve Has Started to
Reduce Its Portfolio?
3Significant uncertainty surrounds the magnitude and size of the
In the United States, the Federal Reserve has increased
adjustments, as discussed in the October 2017 GFSR. Indeed, by let- the federal funds rate six times since December 2015.
ting asset holdings mature, central bank balance sheet normalization
will increase net supply to the public, a development expected to put Yet the term premium remains near historical lows, and
upward pressure on term premiums and broader risk premiums. financial conditions have continued to ease, in contrast

International Monetary Fund | April 2018 5


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 1.4. Market Interest Rates, Central Bank Balance Sheets, and US Financial Indicators
Easy global financial conditions are underpinned by advanced Policy rate expectations still point to gradual rate hikes.
economy central banks’ large asset holdings.
1. Change in Central Banks’ Balance Sheet Assets 2. Estimated Average Short-Term Interest Rates through Two Years
(Trillions of US dollars) (Percent)
16 40 3.0
United States Aggregate stock as a share of GDP United States
United Kingdom (percent, right scale) United Kingdom 2.5
12 Euro area Germany 2.0
Japan 30 Japan
Aggregate stock 1.5
8 1.0
0.5
20
4 0.0
–0.5
0 10 –1.0
2005 06 07 08 09 10 11 12 13 14 15 16 17 2013 14 15 16 17 18

Term premiums have remained compressed in major economies and S financial conditions have continued to ease despite policy rate
are near historic lows. hikes ...
3. Estimated 10-Year Term Premiums 4. Financial Conditions and US Federal Funds Rate
(Percent)
3.0 1993–94 0.0
United States
2.5 Germany Tightening of 1999–2000

Financial conditions index


Japan financial 2004–06 –0.2
2.0
conditions 2015–present
1.5 –0.4
1.0
0.5 –0.6
0.0
–0.8
–0.5 Rate hikes

–1.0 –1.0
2007 08 09 10 11 12 13 14 15 16 17 18 0 1 2 3 4 5 6 7
Federal funds rate (percent)

... while the US dollar has weakened ... ... and measures of in ation compensation have remained relatively
muted.
5. US Real Effective Exchange during Federal Reserve Hiking Cycles . Five ear Five ear ea n ation S a
(Index) (Percent)
Mar. 1983 Feb. 1988 Feb. 1994 United States (left scale)
115 Jun. 1999 Jun. 2004 Dec. 2015 3.0 Euro area (left scale) 1.6
Japan (right scale)
Index (month of first hike = 100)

110
1.2
2.5
105
0.8
100 2.0
0.4
95
1.5
90 0.0

85 1.0 –0.4
–6 –3 0 3 6 9 12 15 18 21 24 27 30 2014 15 16 17 18
Months relative to first hike

Sources: Bloomberg Finance L.P.; and IMF staff calculations.

6 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

to previous hiking cycles (Figure 1.4, panels 3 and 4). • The signaling channel of balance sheet reduction may
The tightening cycle so far has not offset the broader be muted, especially compared with the significant
weakness of the dollar since the beginning of the normal- signaling effects associated with implementation of
ization process in the United States (Figure 1.4, panel 5). asset purchases. This is because the Federal Reserve
Moreover, although measures of inflation compensation has mapped out the unwinding of its balance sheet
have moved a bit higher recently with the firming in into the future, with a “high hurdle” for revision.
inflation, they continue to be relatively low in the United At least in the United States so far, the unwinding
States and other countries (Figure 1.4, panel 6).4 of balance sheet measures is less data dependent
Several factors may help explain why the effects of the than the purchase program. Guidance around the
Federal Reserve’s policy actions on term premiums (and initial balance sheet reduction contains much less
thus financial conditions) have been somewhat muted to information about the future path of the traditional
date, compared with the sizable decline following initial tool compared with possible signaling effects of asset
implementation of unconventional balance sheet policies:5 purchases. Indeed, at the nominal lower bound,
• Liquidity considerations point to likely asymmetric unconventional policy tools supplement traditional
responses of term premiums to asset purchases, on levers. But when removing accommodation, policy
the one hand, and shrinkage of balance sheets, on rates have no upper bound.
the other. For example, many studies find that the
Federal Reserve’s first asset purchase program had
a larger effect than subsequent programs. One pos- Are Term Premiums Too Low Given Economic Variables
sible explanation is that the first rounds may have and Other Fundamentals?
ameliorated illiquidity and extinguished potential Term premiums remain very low by historical
fire sales of assets. By contrast, the initial withdrawal standards. But are they “too low” relative to economic
of unconventional accommodation seems unlikely fundamentals? The answer to this question is central to
to have had the concomitant and opposite effect of determining the implications of the prolonged period
boosting liquidity premiums and therefore yields. of monetary policy accommodation for global financial
• Central bank purchase programs may have “structurally” markets. Even though a variety of shocks could push
lowered term premiums, especially in the current envi- term premiums higher, the impact of these shocks on
ronment of lower equilibrium policy rates. Investors financial markets could be sudden and more pro-
likely expect asset purchases to remain in the policy nounced if term premiums are too low given the stage
toolkit in the future, whether or not central banks of the economic cycle.
reduce their asset holdings to near precrisis levels.6 To Model estimates suggest that term premiums are not
commit credibly to abandoning these tools may prove too low. Analysis finds that term premiums are broadly
difficult. Moreover, policymakers are presumably more in line with investors’ expectations for growth, inflation,
likely than before to pull these levers, given new limits and the current stance of monetary policy. As shown in
to conventional measures, because equilibrium or Box 1.2, the estimated term premium has remained near
terminal policy interest rates (the rate that is consistent the lower bound of fitted model values over the past
with full employment and capacity utilization and sta- few years, in line with the large-scale monetary accom-
ble prices) may be lower today as a result of underly- modation needed to support the economic recovery.8
ing structural factors in the economy that keep interest In addition, the gap between the estimated and the
rates nearer their nominal lower bound.7 model-based weighted-average estimated term premi-
ums has been closing recently, suggesting that term
4Inflation compensation, typically referred to as breakeven
premiums are largely in line with investors’ expectations
inflation rates, is defined as the inflation rates that, if realized, would
leave an investor indifferent between holding an inflation-protected for economic fundamentals. However, the model also
Treasury security and a nominal Treasury security. suggests that term premiums are significantly vulnerable
5As such, careful studies of the effects of unconventional policy mea-
to any revisions in those expectations, particularly with
sures (including Gagnon and others 2010) may be less relevant, if not
somewhat misleading, to understanding the reversal of these measures.
6In addition, a possible sustained dearth of global risk-free assets 8Figure 1.2.1 in Box 1.2 shows the average and range of 900
could also durably lower the level of the term premium. model-fitted values for the Adrian, Crump, and Moench (2013)
7These factors include demographic effects and changes in produc- term premium estimate, conditional on various economic and
tivity, among others. See Chapter 2 of the April 2017 GFSR. financial factors, for the United States and Germany.

International Monetary Fund | April 2018 7


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 1. . S n ation ectation an er Pre iu possibly as a result of the recent fiscal expansion at a late
stage of the credit cycle. In response to the revision in
Term premiums have moved lower, and uncertainty about future in ation
has declined. the inflation outlook, the Federal Reserve may withdraw
1. US Term Premium and Dispersion of 10-Year-Ahead monetary policy accommodation at a faster pace than
n ation ectation currently anticipated. In this scenario, term premiums
Percent
4.5 1.0 could suddenly decompress, risk premiums could rise,
Ten-year term premium (left scale)
4.0 Dispersion of 10-year-ahead in ation forecasts 0.9 and global financial conditions could tighten sharply, with
3.5 (three-quarter moving average, right scale) 0.8 possible adverse consequences for the global economy.
3.0
2.5 0.7 Emerging markets are vulnerable to spillovers
2.0 0.6 from an abrupt tightening in global financial condi-
1.5 0.5 tions. Gradual and well-telegraphed normalization of
1.0
0.5
0.4 monetary policy in advanced economies has provided
0.0
0.3 a period of favorable external conditions, and inves-
0.2
–0.5 tor sentiment toward emerging markets has remained
–1.0 0.1
1991 93 95 97 99 2001 03 05 07 09 11 13 15 17
constructive. The favorable conditions have allowed
weaker issuers to access markets, and the creditor base
Markets are not pricing in sharply higher in ation. now includes investors more inclined to turn over their
. ar et lie Probabilit o ig n ation portfolios. If the tightening cycle is accompanied by a
Percent c ance o P n ation 3 ercent over five ear
50
rise in investor risk aversion, portfolio flows to emerg-
ing markets could fall by at least one-quarter under
United States
40 Euro area realistic assumptions (see “Vulnerabilities in Emerg-
ing Markets, Low-Income Countries, and China”
30
section). This drop would increase rollover risks and
20
the cost of funding in these countries. Low-income,
small, non-investment-grade borrowers are particularly
10 exposed to such risks because they have seen a sharp
rise in debt vulnerabilities over the past few years.
0
2010 11 12 13 14 15 16 17 18

Sources: Bloomberg Finance L.P.; Federal Reserve Bank of Minneapolis; and IMF Correlations among Global Term Premiums
staff estimates. (and Expected Rates) Underscore Risks of
Note: In panel 2, euro area options are illiquid. CPI = consumer price index.
International Spillovers
Rapid decompression of term premiums could
respect to uncertainty about the future path of inflation quickly spill over to global financial markets. Key
and growth, or the path for monetary policy. questions are the extent to which movements in term
premiums are correlated across countries today, and
thus primed for contagion, and the direction and
Financial Markets Remain Vulnerable to an intensity of such spillovers. Some evidence indicates
Inflation Surprise that sovereign term premiums among major economies
Although the expected path of policy rates has (Canada, Germany, Japan, United Kingdom, United
recently increased somewhat in some countries, markets States) move very closely together, even as investors’
continue to price in a gradual pace of monetary tight- expectations for policy rate paths in these countries
ening. Uncertainty about future inflation outcomes has have diverged. This trend seems to have preceded the
diminished in tandem with declining term premiums Federal Reserve’s lift-off from the nominal lower bound
(Figure 1.5, panel 1). In addition, market participants in December 2015, and is in line with the view that
are not currently pricing in a risk of sharply higher infla- asset purchases may be a stronger driver of spillovers
tion over the next few years (Figure 1.5, panel 2). than standard monetary policy via short interest rates
An upside surprise to inflation could pose a challenge (Figure 1.6, panel 1).
to investors and policymakers. For example, inflation Moreover, model estimates indicate the impact of
in the United States may increase faster than expected, spillovers between G4 (Germany, Japan, United King-

8 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.6. Term Premium Correlations, Spillovers, and Exchange Rate Relationships

Term premiums in major advanced economies move very closely Spillovers between G4 term premiums are elevated, with the
together even as market expectations of policy rate paths diverge. United States dominating the direction.
1. Term Premiums and Expected Rate Correlations 2. Net Pairwise Spillovers between the US Term Premium and
Percent e laine b fir t rinci al co onent German, Japanese, and UK Premiums
(Percent)
90 Two-year expected rate 25
5x5-year forward premium 20
80
15
70
10
60 5
0
50
–5
40
–10
30 –15
1997 99 2001 03 05 07 09 11 13 15 17 1994 98 2002 06 10 14 18

Short rates dominate movements in exchange rates between the ... and the sterling-US dollar cross rates.
euro-US dollar ...
3. Spot US Dollar Exchange Rate Betas: Euro 4. Spot US Dollar Exchange Rate Betas: British Pound
6 6
Two-year expected rate Two-year expected rate
5 5x5-year forward premium 5x5-year forward premium 5
4 4
3
3
2
2
1
0 1

–1 0
–2 –1
1997 99 2001 03 05 07 09 11 13 15 17 1997 99 2001 03 05 07 09 11 13 15 17

Sources: Bloomberg Finance L.P.; and IMF staff estimates.


Note: The dotted horizontal lines denote unconditional sample estimates. In panel 2, a positive (negative) value indicates that the US term premium is a shock
transmitter (receiver) to German, Japanese, and UK term premiums. G4 = Group of Four (Germany, Japan, United Kingdom, United States).

dom, United States) term premiums is elevated, with In this environment, spillovers from a faster
spillovers from the United States to the other countries withdrawal of US Federal Reserve monetary policy
mainly dominating (Figure 1.6, panel 2).9 Although accommodation in the wake of an inflation surprise
the net impact of such spillovers appears notably lower and associated repricing of inflation risk and term pre-
than at the height of the crisis, alongside higher cor- miums could rapidly tighten US and global financial
relations of term premiums this suggests considerable conditions. This could challenge major central banks,
scope for a rapid rise in interest rates to be transmitted such as the European Central Bank, that are not as far
to global markets. along in the normalization process, perhaps forcing
them to respond through additional accommodation.
Although term premiums may be more correlated
9The methodology, by Diebold and Yilmaz (2012), obtains a
at present, perhaps because of global factors, central
time-varying spillover index using rolling generalized forecast error
variance decompositions in a generalized vector autoregression banks’ strategies regarding conventional policy tools
model. The framework measures directional spillovers by using the remain critical for communicating the stance of mon-
normalized elements of the variance decomposition matrix. The etary policy. For example, term premium differentials
net pairwise spillovers are then calculated by taking the difference
between the total spillovers transmitted from market i to all markets
do not appear to have dominated the transmission
j and the spillovers transmitted from all markets j to market i. of monetary policy through exchange rates, at least

International Monetary Fund | April 2018 9


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

among the G4.10 In fact, the sensitivity of currencies investors to take additional risk, some central banks
to expected short rate differentials has remained ele- around the globe have either been raising policy rates
vated in recent years (Figure 1.6, panels 3 and 4). This or preparing investors for an eventually less accommo-
finding holds both on average over the past 20 years dative stance. And although the share of assets with
and for estimates for the latest sensitivity. negative yields remains sizable globally, this fraction
has ticked down in recent months. So, rather than a
reach for yield prompted by central bank accommo-
Continued Clear Monetary Policy Communication Is dation, there may be outright speculative overreach in
Essential to Avoid Market Disruptions some risky assets.11
Gradual removal of monetary accommodation The key questions are the extent to which finan-
and clear communications will help anchor market cial vulnerabilities have increased since the previous
expectations and prevent undue volatility. To support GFSR, how the constellation of current accommoda-
the recovery and ensure inflation objectives are met, tive financial conditions and vulnerabilities compares
monetary authorities should maintain accommodation, with past episodes of financial stress, and whether
as needed. When normalizing policy, central banks asset valuations appear stretched, given current cyclical
should do so in a gradual and well-communicated conditions. This final determination matters. If asset
manner. They should also provide guidance on pro- valuations are not judged to be significantly out of
spective changes to policy frameworks if such changes line with fundamentals, policymakers can continue to
are warranted. Gradualism and clear communications normalize monetary policy gradually and to implement
are crucial given the confluence of still relatively low macroprudential and other regulatory measures aimed
inflation, easy global financial conditions, and rising at lessening financial stability risks. In contrast, if asset
financial vulnerabilities. To address the buildup in misalignments are significant and may put growth
financial vulnerabilities and avoid putting growth at at risk in the future, a more forceful policy response
risk, policymakers should also deploy and develop may be needed.
appropriate micro- and macroprudential tools. To shed some light on rising financial vulnerabil-
ities, this section focuses on asset price valuations in
equity, corporate bond, and leveraged loan markets; on
Reach for Yield or Overreach in Risky Assets? financial leverage, including that embedded in deriv-
Against a backdrop of mounting vulnerabilities, risky asset ative products; and on liquidity mismatches related to
valuations appear overstretched, albeit to varying degrees the proliferation of certain types of investment funds
across markets, ranging from global equities and credit and strategies (for example, exchange-traded funds).
markets, including leveraged loans, to rapidly expanding
crypto assets (discussed in the next section). Moreover, the
increasing use of financial leverage to boost returns and Equity Valuations Remain Expensive
the growing influence of some passive investment vehicles, The ongoing global economic recovery, strong
particularly exchange-traded funds (ETFs) in less liquid corporate performance, and still-low interest rates have
underlying markets, could amplify the impact of asset supported equity prices, on balance, since the previ-
price moves on the financial system. ous GFSR (Figure 1.7, panel 1). In the United States,
and through the spate of volatility beginning in early
Financial Vulnerabilities Continue to Build amid Easy February, equity market capitalization has risen from
Financial Conditions 95 percent of GDP in 2011 to 155 percent of GDP in
The unconventional monetary policies implemented March 2018. Rising global equity prices have sup-
since the global financial crisis, including both asset
purchases and forward guidance, clearly and by design 11“Reach for yield” may be a dated description of current investor
encouraged investors to reach for yield. But today’s behavior and financial asset price developments. Following Hanson
policy environment differs. Rather than encourage and Stein (2015), the effects are transitory, and the lengths of these
episodes depend on the capacity of so-called return-oriented arbitra-
geurs to take offsetting positions. Insofar as financial conditions are
10Based
on estimated dynamic correlations following Cappiello, very accommodative (for example, the ability and willingness to take
Engle, and Shepphard (2006). on leverage), any reach for yield should not have persisted.

10 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.7. Valuations of Global Equities


Global equities have extended their rally in recent months, on balance, ... supporting a rebound in equity issuance, especially in emerging
since October ... markets.
1. MSCI Equity Indices 2. Gross Equity Issuance
(End-2011 = 100) (Billions of US dollars, quarterly)
300 400
United States EM Asia Japan United States Other advanced EMs excluding China
250 Euro area EM Latam EM EMEA Europe China
United Kingdom 300
200

150 200

100
100
50

0 0
2009 10 11 12 13 14 15 16 17 18 2005 07 09 11 13 15 17
In the United States, valuations are high relative to their historical However, US equity valuations appear low relative to US Treasuries.
averages and pre-GFC peak, and to other countries.
3. Valuation Metrics 4. Valuation Metrics
(Dotted arrows: 2008–12; solid arrows: 2012–18; (Z-scores)
squares: long-term average)
26 3 –3
Emerging markets Japan PE (left scale)
24 United States China CAPE (left scale)
2 –2
Price-to-earnings ratio

22 Europe ERP (right scale, inverted)


20 1 –1
18
0 0
16
14 –1 1
12
–2 2
10
8 –3 3
0.5 1.0 1.5 2.0 2.5 3.0 3.5 1990 92 94 96 98 2000 02 04 06 08 10 12 14 16 18
Price-to-book ratio
Equity valuations have been supported by historically low interest rates The relative equity valuations are closer to their historical averages if
and robust earnings expectations. more sustainable earnings are assumed.
5. Decomposition of Equity Market Performance 6. Equity Risk Premium under Different Assumptions
(Percent contributions to cumulative returns) (Z-scores; inverted scale)
Equity risk premiums IBES EPS growth forecast
80 Earnings (current and projected) Long-term average of EPS growth –2
70 Risk-free rate WEO nominal GDP growth forecast
60 Price return –1
50
0
40
30
1
20
10 IBES with term premium + 50 bps 2
0 IBES with term premium + 150 bps
–10 3
2013 14 15 16 17 18 1990 92 94 96 98 2000 02 04 06 08 10 12 14 16 18

Sources: Bloomberg Finance L.P.; Citigroup Index LLC; ICE Bank of America Merrill Lynch; JPMorgan Chase & Co; IMF, World Economic Outlook (WEO) database;
Morgan Stanley Capital International (MSCI); Standard & Poor’s; Thomson Reuters IBES; and IMF staff calculations.
Note: bps = basis points; CAPE = cyclically adjusted price-to-earnings ratio; EM = emerging market; EMEA = Europe, Middle East, Africa; EPS = earnings per share;
ERP = equity risk premium; GFC = global financial crisis; Latam = Latin merica; PE = price-to-earnings ratio.

International Monetary Fund | April 2018 11


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

ported a moderate rebound in new issuance, especially still high—have edged lower, especially in China and
in emerging markets (Figure 1.7, panel 2). other emerging markets (Figure 1.8, panel 4). Effective
These developments raise questions about val- interest rates paid by the corporate sector moved higher,
uations and potential investor excesses. Standard particularly outside the United States. As a result, inter-
price-to-earnings and price-to-book valuation metrics est coverage ratios have dipped everywhere except China
remain elevated in most regions (Figure 1.7, panel 3). and the United States (Figure 1.8, panel 5).
For the United States, these measures remain rela- Recent US tax reform will have important implica-
tively high compared to both historical levels and tions for the corporate sector. As discussed in the April
current valuations in other countries. Indicators that 2017 GFSR, most US companies will gain from the
incorporate longer-term averages of realized earnings reform. However, historical experience in the United
to capture expectations, such as cyclically adjusted States in the 1980s and with the repatriation tax hol-
price-to-earnings, continue to support this assessment, iday in 2004 suggests that financial risk taking often
even after the volatility spike in February and the slide follows tax policy changes, as evidenced by heightened
in equity prices in March on concerns about trade ten- purchases of financial assets, mergers and acquisitions,
sions (Figure 1.7, panel 4). dividends, and share buybacks. The cap on the tax
Some measures of the US equity risk premium, in deductibility of interest expense will reduce incentives
which equity valuations are conditional on the level of for debt financing, which tends to affect highly lever-
interest rates, suggest that shares have been closer to aged companies disproportionately (Figure 1.8, panel
fair value. Indeed, strong near-term earnings expecta- 6). These firms may face funding pressures because of
tions, as well as historically low interest rates, sustain higher interest expenses, more volatile earnings, and a
comparatively wide equity risk premiums (Figure 1.7, more compressed schedule for adapting their funding
panel 5). However, this approach is highly sensitive structure to the new tax code.
to profit forecasts as well as to different assumptions
about the discount factor. Equity valuations deteriorate
under alternative, less sanguine proxies for earnings, Signs of Overheating Are Evident in the
such as longer-term averages or nominal GDP growth. Leveraged Loan Market
Also, higher projected paths for interest rates similarly The leveraged loan market, consisting of commercial
narrow the equity premium and imply richer valua- loans extended to borrowers who are non–investment
tions (Figure 1.7, panel 6). grade or already have significant amounts of debt, is seen
by market participants as a barometer for broader risk
taking. Global credit markets have grown massively in
Corporate Bond Valuations Are Stretched and Credit recent years. Global leveraged loan issuance hit a record
Quality Is Deteriorating in Risky Segments high in 2017 of $788 billion, surpassing the precrisis high
With central banks in advanced economies continu- of $762 billion in 2007. Most issuance occurred in the
ing to lift policy rates from the nominal lower bound or United States, amounting to $564 billion (Figure 1.9,
signaling a not-too-distant commencement of the nor- panel 1). Since 2007, US institutional leveraged loans
malization process, the share of negative-yielding global outstanding have doubled to almost $1 trillion, compared
bonds has dipped lower since the October 2017 GFSR. with $1.3 trillion in US high-yield bonds outstanding.12
This ratio, however, remains significant (Figure 1.8, While refinancing volumes have been significant given the
panel 1). Against a backdrop of low default rates, corpo- low-interest-rate environment, borrowing to fund mergers
rate spreads remain at very low levels, even in the riskiest and acquisitions, leveraged buyouts, dividends, and share
segments (Figure 1.8, panel 2). Favorable financial con- buybacks still accounts for half of total issuance amid
ditions have boosted corporate bond issuances. Issuance improving global growth (Figure 1.9, panel 2).
of riskier bonds has surged, and the share of lower-grade
bonds (BBB-rated) in the investment-grade universe has
been rising (Figure 1.8, panel 3). 12Institutional leveraged loans include term loans structured

Strong economic growth and corporate restructuring specifically for institutional investors, such as loan funds, collat-
eralized loan obligations, real money investors, and hedge funds,
efforts, particularly in the energy sector, have sup- though there are some banks that buy institutional term loans. These
ported corporate profitability; and debt ratios—while tranches include first- and second-lien loans.

12 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.8. Valuations of Corporate Bonds


A still-high share of negative-yield assets ... ... has supported demand for risky assets and compressed credit spreads.
1. Share of Negative Yielding Global Bonds 2. Credit Spreads per Rating Bucket, 1999–2018
25 (Percent) (Basis points) 4,000
Range
Mar. 2018 3,500
20 $10.6 trillion Median 3,000

15 2,500

$7.6 trillion 2,000


10 1,500
1,000
5
500
0 0

US BBs

US Bs

US CCCs

EU BBs

EU Bs

EM BBs

EM Bs
2013 14 15 16 17

This has spurred the new issuance of risky bonds in


lower-credit-quality buckets. Profitability has helped reduce debt ratios.
3. Share of the Lowest Credit Bucket 4. Debt Ratios
(Percent) (Net debt to EBITDA, percent)

40 Lower-grade (CCC-rated) issuance Lower-grade (BBB-rated) bonds 50 7


as a percent of high-yield bond and in the investment-grade index United States Europe
leveraged loan issuance (left scale) (right scale) China EMs excluding China 6
30 45 5
4
20 40
3

10 35 2
1
0 30 0
1998 2000 02 04 06 08 10 12 14 16 1999 2001 03 05 07 09 11 13 15 17:Q3

As a result, interest coverage ratios have dipped, except for the United Highly levered firms are more likely to be impacted by the S corporate
States and China. tax reform.
5. Interest Coverage Ratios 6. Comparison among High- and Low-Leveraged US Firms
(EBIT to interest expense)
14 80 76 12
United States Europe Low leverage 11
12 China EMs excluding High leverage 10
10 China 60
8
8 6
40 6
6 29
4
4 20
2 2
5
1
0 0 0
1999 2001 03 05 07 09 11 13 15 17:Q3 Number of Gross debt Average maturity
firms proportion where (years)
interest expense
30 percent EBITD
(percent)

Sources: Bloomberg Finance L.P.; ICE Bank of America Merrill Lynch; JPMorgan Chase & Co; Standard & Poor’s; and IMF staff calculations.
Note: In panel 2, the full sample is from 1999 to 2018. In panels 4 and 5, the full sample from 1999 to 2005 from the source is limited. In panel 6, high leveraged is
defined as firms with net debt divided by EBITD four times. Real estate and utilities sectors are not included. EBIT = earnings before interest and taxes;
EBITD = earnings before interest, taxes, depreciation, and amorti ation; EM = emerging market.

International Monetary Fund | April 2018 13


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 1.9. Leveraged Loan Issuance, Quality, and Developments after Regulatory Guidance

Loan issuance reached record highs in 2017 ... ... with the share of proceeds used to fund acquisitions and
shareholder enhancements still large.
1. New Issue Global Leveraged Loan Volume 2. Global Leveraged Loan Issuance by Use of Proceeds
(Billions of US dollars) (Percent)
900 100
800 US issuers Other 90
Non-US issuers
700 80
600 Refinancing 70
500 60
50
400 Dividends and 40
300 buybacks Mergers and
acquisitions 30
200 20
100 10
Leveraged buyouts
0 0
1998 2000 02 04 06 08 10 12 14 16 2005 07 09 11 13 15 17

Covenant protections have weakened over time ... ... leading to potentially lower recovery rates in the next default cycle.
3. New Issue Covenant-Lite US Leveraged Loans and Covenant 4. Average Annual First Lien US Loan Implied Recovery Rates
Quality Index (Percent)
80 4.5 100
Covenant-lite percent of new 82% average
70 issuance (left scale) Global 69% average 90
4.0 financial
60 Moody’s loan covenant quality 80
index score (right scale) crisis
50 3.5 70
40 60
Higher scores 3.0
30 equal weaker 50
20 covenants 40
2.5
10 30
0 2.0 20
2007 08 09 10 11 12 13 14 15 16 17 2003 04 05 06 07 08 09 10 11 12 13 14 15 16

Highly leveraged loan deals have increasingly been arranged by ... while adjustments to earnings expectations have led to less
nonbank lenders ... conservative leverage calculations.
5. Fraction of US Deals with a Nonbank Entity as Lead Agent 6. Percent of US Deals with EBITDA Adjustments
(Percent)
30 50
Greater than six times total leverage Mergers and acquisitions
25 All loan issuance Leveraged buyouts
All deals 40
20
30
15
20
10

5 10

0 0
2005 06 07 08 09 10 11 12 13 14 15 16 17 2001 03 05 07 09 11 13 15 17

Sources: Barclay’s; Moody’s Default and Recovery database; Standard & Poor’s Leveraged Commentary and Data; and IMF staff calculations.
Note: In panel 3, the Moody’s Loan Covenant Quality Index score is a yearly average; data are unavailable from 2008 to 2010 due to lack of rated leveraged loan
issuance. A higher Covenant Quality Index score represents weaker covenant protections. In panel 4, implied recovery rates are based on loan prices one month after
default. EBITDA = earnings before interest, taxes, depreciation, and amortization.

14 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

The global leveraged loan market now offers an (usually in the form of corporate bonds) to protect
interesting example of the extent to which reach for loans in the event of a default. But the average debt
yield has supported issuance and adversely affected cushion of first-lien covenant-lite loans is now only
price and nonprice terms, as well as credit quality, 15 percent, down from about 33 percent before the
despite efforts by regulators to rein in risk taking.13 financial crisis. Although the number of defaults so
Strong issuance and lofty valuations, including a weak- far in this cycle has been limited, weakening investor
ening of nonprice terms such as investor protections, protections and eroding debt cushions have coincided
could exacerbate the next default cycle. A sharp rise in with lower average recovery rates for defaulted loans
defaults following a tightening of financial conditions, (69 percent), compared with the precrisis average of
or a shutdown of the market at the extreme, could 82 percent (Figure 1.9, panel 4).
have large negative implications for the real economy Regulators in the United States and Europe have
given the growing size of the loan market to date and taken actions aimed at curbing market excesses.15 One
the role it plays in channeling funding to corporations. unintended consequence of these actions appears to
Signs of late credit cycle dynamics are already be a migration of activity away from banks toward
emerging in the leveraged loan market and, in some institutional investors, such as collateralized loan obli-
cases, are reminiscent of past episodes of investor gations, bank loan mutual funds, private equity firms,
excesses. Lower-quality companies continue to enjoy and other private funds (Kim, Plosser, and Santos
ample access to credit. Yet at the same time, ratings 2017). (See Box 1.3 for a discussion of the changing
have deteriorated. In the United States, the percentage investor base in the US leveraged loan market.) As
of new loan issuance rated single-B or lower increased noted, institutional leveraged loans outstanding have
from about 25 percent in 2007 to 65 percent in 2017, grown rapidly in recent years, with institutional inves-
although this trend could partly reflect some changes tors increasingly playing an important role in highly
in rating agencies since the crisis.14 Meanwhile, new leveraged loan deals (Figure 1.9, panel 5). In addition,
deals include fewer investor protections, such as looser nonprice terms, which are more difficult to monitor,
covenants and thinner subordination in the capi- have been loosening. Weaker covenants have reportedly
tal structure. For example, covenant-lite loans have allowed borrowers to inflate projections of earnings
evolved from a specialty structured debt instrument before interest expenses, taxes, depreciation, and amor-
before the financial crisis to the largest market segment tization (EBITDA) and to borrow more after the clos-
today. Covenant-lite loans made up 75 percent of new ing of the deal. New loans with EBITDA add-backs or
institutional loan issuance in 2017. In addition, the adjustments that conceal deteriorated leverage metrics
quality of loan covenants has continued to deteriorate have reached new highs (Figure 1.9, panel 6).16
(Figure 1.9, panel 3).
To be fair, weaker covenants may reflect the loan
market’s changing investor base as loans mature into
a widely accepted asset class in investors’ portfolios.
But looser provisions inherently provide fewer warning 15For example, in March 2013 US federal banking agencies issued
signals about a potential default and may thereby result guidance to reduce risk in the leveraged loan market, both for loans
in lower recovery rates. For example, in the recent retained on banks’ balance sheets and for those repackaged for sale to
other parties. More recently, in May 2017 the ECB issued supervi-
past banks typically demanded a first-lien claim on sory guidance concerning expectations around leveraged transactions
collateral as well as sufficient loss-absorption capacity in Europe and the ongoing monitoring of the fundamental credit
quality of leveraged exposures. In particular, US and European
13There is no universal definition for leveraged loans, but the supervisors recommended that banks follow heightened risk man-
term usually refers to a speculative-grade loan for which the obli- agement when a borrower’s debt exceeds six times its earnings before
gor’s postfinancing leverage (as measured by debt-to-assets ratio, interest expense, taxes, depreciation, and amortization.
debt-to-equity ratio, cash-flow-to-total-debt ratio, or other such 16EBITDA add-backs generally consist of pro forma fees and

standards unique to particular industries) significantly exceeds indus- expenses (expected to be eliminated following an acquisition deal) that
try norms. Leveraged borrowers typically have diminished ability increase the pro forma EBITDA during the loan syndication process.
to adjust to unexpected events and changes in business conditions For additional debt after deal closing, most structures include a debt
because of their higher ratio of total liabilities to capital. incurrence clause that allows the borrower to add debt subject to the
14Cohen and Manuszak (2013) find that increased competition satisfaction of certain financial ratios along with fixed-dollar debt bas-
among credit rating agencies from 2002 to 2007 led to lower subor- kets that generally permit the borrower to incur debt without reference
dination levels and less stringent ratings. to a maintenance covenant or other financial ratio.

International Monetary Fund | April 2018 15


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 1.10. Correlations and Interconnectedness

In the United States, correlations between individual stocks and across Within the global equity market, cross-country correlations have
sectors have picked up. rebounded somewhat.
1. Average Correlations 2. Average Correlations
(90-day rolling versus S&P 500; cap weighted) er u t e nite State five ear rolling in o
1.0 Taper US 1.0
Presidential February tantrum election
election turbulence 0.9
0.9 Brexit 0.8
Tax reform 0.7
0.8 bill introduced
0.6
0.5
0.7 0.4
Advanced economies 0.3
0.6 EMs excluding China 0.2
Across sectors
China
Within sectors 0.1
0.5 0.0
Jan. 2015 Jan. 16 Jan. 17 Jan. 18 2001 03 05 07 09 11 13 15 17

The average correlations across major asset classes have remained ... while market turnover has been relatively low.
relatively high ...
3. Average Correlations with S&P 500 4. Corporate Bond Market Turnover
(Ratio of trading volumes to amounts outstanding, percent)
1.0 Commodities 250
0.8 US Treasuries Investment-grade bonds
0.6 US credit High-yield bonds 200
0.4
150
0.2
0.0
100
–0.2
–0.4 50
–0.6
–0.8 0
1990 92 94 96 98 2000 02 04 06 08 10 12 14 16 18 2005 07 09 11 13 15 17 18

Sources: Bloomberg Finance L.P.; Haver Analytics; Market Axess; and IMF staff estimates.
Note: In panel 4, data are as of end-February 2018. EM = emerging market. S&P = Standard & Poor’s.

The Price of Volatility Remains Low But realized volatility and underlying forecasts
Another critical issue is how much risk investors of future volatility also increased. Both near- and
perceive around asset valuations. Indeed, valuations for longer-term equity options now appear close to, if not
options, on the one hand, and underlying securities, below, levels consistent with volatility forecasts. There-
on the other, are distinct, strictly speaking. During fore, the premium investors require to compensate
the turmoil in global equity markets in early February, for volatility risks was little changed, on net, which is
implied volatilities derived from equity options, which ultimately consistent with persistently accommodative
reflect information not only about investors’ expecta- financial conditions, as well as a renewed willingness
tions for volatility but also the premium they require on the part of investors to sell volatility. A broadly sim-
to bear volatility risk, spiked sharply from subdued ilar picture arises across other asset classes, including
levels. The VIX term structure, based on short- to US dollar swaptions.
longer-dated option expiration dates, not only shifted Correlations within and across asset classes are also
higher but also inverted briefly (see Online Annex 1.1 important indicators of financial conditions. Indeed,
on implied volatility pricing).17 correlations among even typically unrelated assets tend
to increase sharply during financial crises, making
17See Online Annex 1.1 at www.imf.org/en/Publications/GFSR diversification difficult as investors’ overall portfolio risk
for more details. increases. Correlation measures have rebounded of late,

16 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

albeit from subdued levels relative to historical norms. However, other forms of financial leverage appear to
Within the US stock market, correlations between be on the rise:
individual stocks and across sectors picked up some- • Synthetic collateralized debt obligations (CDOs):
what after the completion of major tax legislation and Analysts estimate synthetic CDO issuance to have
increased further after the spike in volatility in February surged to between $80 billion and $100 billion in
2018 (Figure 1.10, panel 1). Global equity market cor- 2017—well below precrisis levels but up from about
relations also rebounded in recent months, even before $20 billion a year in 2014–15 (Figure 1.11, panel 1).
the drop in global share prices (Figure 1.10, panel 2). Some market participants also speculate that insti-
Finally, broader correlations across asset classes have tutional investors are actively increasing leverage to
increased, which suggests that global diversification has boost yields using total return swaps and asset swaps,
become somewhat more difficult (Figure 1.10, panel 3). although little evidence is available at this point.
It should be noted, however, that trends in realized • Margin debt: The margin debt from stock borrowing
statistical correlations may understate the prospects of stands at a record $580 billion in the United States,
contagion risk. Indeed, both correlations and volatility about 2 percent of overall market capitalization as of
tend to increase at precisely the most inopportune and the end of 2017 (Figure 1.11, panel 2).18 Although
unforeseeable times; namely, when prices of risky assets this share is below the peak in 2008, it is still quite
swoon. In addition, market turnover has been relatively elevated. Also worrisome, the current net exposure19
low, especially for high-yield bonds, which may com- of investors involved in stock margin borrowing is
pound price discovery distortions and illiquidity in the at record negative highs relative to overall market
future (Figure 1.10, panel 4). capitalization compared with the past 25 years (Fig-
Beyond asset price correlations and volatility, the ure 1.11, panel 3).
ongoing structural changes in the investment man- • Use of financial leverage by investment funds: Mean-
agement industry affect interconnectedness and the while, assets under management of large regulated
potential for spillovers across markets. For example, bond investment funds that actively use derivatives
broker dealers’ intermediation role has declined in have increased to more than $1.5 trillion, about
recent years, leading to a greater role for the non- 17 percent of the world’s bond fund sector (Fig-
bank sector. Institutional investors include both firms ure 1.11, panel 4). The use of embedded leverage
dedicated to high-frequency trading across markets, through derivatives is increasing as fund managers
which have become more prominent, and also other seek to enhance low yields. The lack of sufficient data
market participants, such as insurance companies and collection and oversight by regulators compounds
pension funds, which may be using less procyclical the risks.20 Gross notional exposure of bond funds
investment strategies. In any case, these new market to derivatives is worrisome. The average derivatives
structures have not been tested during a significant leverage (defined as gross notional exposure) of an
market downturn. asset-weighted sample of more than 200 US- and
European-domiciled bond funds has risen from
Increasing Use of Financial Leverage May Amplify Risks 215 percent to 268 percent of assets over the past
four years (Figure 1.11, panel 5). The level of deriv-
As the financial crisis illustrates, leverage can amplify
negative shocks through pernicious feedback loops.
Sharp price declines can lead to investor runs and fire 18Stock margin borrowing data can also include fixed-income

sales of liquid and safe assets to cover redemptions and securities, but most transactions are related to stocks. The data
include both retail and institutional investor transactions.
margin requirements. 19Defined as the difference between debit balances and free credit

There have been some noteworthy developments balances in customers’ security margin and cash accounts.
20No disclosure requirements for detailed leverage information
since the crisis. For example, lower volumes of repur-
for regulated investment funds are in place in the United States, and
chase agreements (repos), at least relative to market cap- requirements are in place only on a selected basis in some European
italization, may be reflective of less financial leverage. countries. Implementing comprehensive and globally consistent
In the years before the global financial crisis, investors reporting standards across the asset management industry would
give regulators better data with which to locate leverage risks. For
widely used repos and leverage to boost returns. But
example, reporting standards should include enough information on
stricter regulations, as well as changes in bank business derivatives to show funds’ sensitivity to large moves in underlying
models, have significantly reduced repo activity. rate and credit markets.

International Monetary Fund | April 2018 17


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 1.11. Measures of Leverage and Investment Funds with Derivatives-Embedded Leverage
Although well below precrisis levels, synthetic CDO issuance is coming ... with elevated stock borrowing by margin indicating greater leverage
back ... risk in US equity markets ...
1. Synthetic CDO Issuance 2. Margin Debt from Stock Borrowing
(Billions of US dollars)
700 700 Margin debt (billions of US dollars, left scale) 2.9
600 600 Margin debt (percent of market
capitalization, right scale) 2.5
500 500
2.1
400 400
1.7
300 300
1.3
200 200
100 100 0.9

0 0 0.5
2005 06 07 08 09 10 11 12 13 14 15 16 17 1992 94 96 98 2000 02 04 06 08 10 12 14 16

... and net exposures in margin accounting for stock borrowing There is strong growth in the AUM of selected large regulated bond
reaching record negative levels. funds that use derivatives ...
3. Net Exposure of Investors with Margin Debt 4. Assets under Management of Selected Bond Funds
(Percentage of market capitalization) (Trillions of US dollars)
2.0 2.0
Investor positive net exposure Funds active in derivatives with no reported leverage
1.5 (margin debt < free cash + credit balance in margin accounts) Funds with reported leverage
1.0 1.5

0.5
1.0
0.0
–0.5 0.5
–1.0 Investor negative net exposure
(margin debt > free cash + credit balance in margin accounts)
–1.5 0.0
1992 94 96 98 2000 02 04 06 08 10 12 14 16 2004 05 06 07 08 09 10 11 12 13 14 15 16 17

... with rising gross notional exposures ... ... that can exceed multiples of fund net asset values.
5. Average Derivatives Leverage of Selected Bond Funds 6. Derivatives Leverage and AUM across the Different Funds
(Gross notional exposure as a percentage of net asset value) (Gross notional exposure percentage; billions of US dollars)
400 Asset-weighted leverage 25th percentile 75th percentile 3,000

Leverage
(percent)
350 2,000
1,000
300
0
250
0
of US dollars)
AUM (billions

200 20
40
150
60
80 180 80
100
2013–14 2014–15 2015–16 2016–17

Sources: Annual reports of selected regulated investment funds; Bloomberg Finance L.P.; Federal Reserve; Financial Industry Regulatory Authority; ICE Bank of
America Merrill Lynch; and IMF staff estimates.
Note: Selected EU-domiciled investment funds report a gross notional exposure of their derivative positions in their annual report. Funds with reported leverage in
derivatives positions in the sample account for over $1 trillion of these assets, including the assets of the US-domiciled version of the same EU-domiciled funds that
report leverage. Although these funds are separate investment vehicles, they share the same mandate and portfolio manager and therefore have closely matched
portfolios, exhibiting a high correlation of returns. The remaining $500 billion of assets correspond to a group of selected funds that do not report leverage in
derivatives positions but are known to be active in derivatives (the funds’ latest annual reports list at least 15 derivatives positions). In panels 2 and 3, margin debt
data may also include nonequity securities such as bonds. In panel 6, the data are as of the latest annual reports. AUM = assets under management;
CDO = collateralized debt obligation.

18 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

atives notional exposure in the sample ranges quite is no conclusive evidence about the broader impact
significantly: the bottom 25th percentile of funds has of large outflows from less liquid bond ETFs on
embedded derivatives leverage in the 100 to 150 per- underlying markets,25 the fast growth of these ETFs is
cent range. But the top 25th percentile shows funds worth monitoring, given their potential for increasing
with average leverage between 300 and 2,800 percent contagion risks:
(Figure 1.11, panel 6).21 Some investors may enter • Frequent trading: Investors in ETFs appear to trade
derivatives contracts to hedge unwanted risk. How- more actively than market participants in the under-
ever, others may do so to boost returns, which, in lying asset class, which may increase contagion risk.
turn, can amplify shocks during periods of stress.22 To start, unlike flows into retail mutual funds, ETF
flows are very volatile (Figure 1.12, panel 4). Less
liquid bond markets, such as high-yield bonds, lack
Growth in Less Liquid Bond ETFs May Raise Financial the depth and breadth to accommodate large and
Stability Concerns frequent transactions.26 Even during the financial
The assets under management of ETFs invested crisis, outflows from high-yield bond investment
in less liquid assets—bank loans and high-yield and funds were limited, with a maximum monthly
emerging market bonds—have risen rapidly to more outflow of 2.5 percent (of assets) in October 2008.
than $140 billion (Figure 1.12, panel 1). Although the Monthly ETF outflows now often exceed 3 percent
share of high-yield bond and emerging market bond (of assets), which may become more of a concern as
ETF assets is still small (less than 5 percent of the total the market share of ETFs rises.
market value of underlying bond markets), it more than • Sensitivity to changes in risky asset prices: As evidenced
tripled from 2010 to 2017 (Figure 1.12, panel 2).23 during the February episode of volatility in equity
ETFs offer several benefits to investors: they enhance markets, the sensitivity of high-yield and emerging
price discovery, provide an alternative source of market bond ETFs to S&P 500 returns is higher
liquidity through exchange trading, facilitate hedging than the sensitivity of their underlying indices to
and diversification, and charge lower fees than other S&P 500 returns. This suggests that the rise in
investment funds.24 Indeed, ETFs can provide addi- ETFs, particularly those investing in relatively illiq-
tional liquidity to less liquid bond markets: only about uid assets, may increase contagion risk and possibly
one-fifth of transactions in high-yield and emerging amplify price moves across asset markets during
market bond ETFs prompt a corresponding transac- periods of stress. Greater investment in passive
tion in the underlying market as a result of outflows investment strategies, such as ETFs, may be related
from ETFs; that is, a destruction of ETF shares (Fig- to the rise in cross-asset correlations during periods
ure 1.12, panel 3). of stress, one of the main attributes of contagion.
However, the extension of ETFs to less liquid bond Benchmark-focused investors are more likely to be
markets may pose risks related to liquidity mismatches driven by common shocks than by the idiosyncratic
between ETFs and underlying assets. Although there fundamentals of assets they invest in.27

21While US and European regulated investment funds are subject to 25There is some evidence that the largest holdings of high-yield

explicit leverage limits, derivatives exposures may mask implicit lever- bond ETFs are increasingly and more systematically underperforming
age since there is less explicit regulation on leverage, particularly in the the broader market during days of large outflows. During these days
United States. See Chapter 1 of the April 2015 GFSR for more detail. (top 5th percentile of daily shares destroyed), the largest 10 bond
22There are 14 funds in the sample (with about $25 billion in holdings of US high-yield bond ETFs showed significantly greater
assets under management) that report both the gross notional underperformance to the market in the 2015–17 period as compared
exposure of their derivatives and a leverage exposure, adjusted for with the 2010–11 period, when their ownership of the underlying
discretionary hedging and netting. About two-thirds of their gross market was less than a quarter of what it is today. There is no evi-
derivatives notional exposure is not dedicated to hedging and net- dence, however, of large redemptions from these ETFs having a signif-
ting, but to boosting returns and additional risk. icant impact on the pricing of the broader underlying market. This is
23Data on the share of the loan market are not included owing to not at all surprising given that their share of the underlying high-yield
lack of data availability. and emerging market bond markets is still less than 5 percent.
24ETFs are generally index-tracking funds that are traded on 26This limitation is reflected in the lower trading volumes, smaller

exchanges and allow investors to gain exposure to several asset classes trading size, smaller share of large trades, and less frequent trading
on a real-time basis at a relatively low cost compared with higher-fee of the less liquid fixed-income markets. Some high-yield bonds do
regulated investment funds that do not offer intraday liquidity. ETFs not even transact on a daily basis. See Chapter 1 of the Octo-
thereby enhance price discovery and offer liquid and transparent ber 2014 GFSR.
investment and hedging alternatives. 27See Chapter 1 of the April 2015 GFSR.

International Monetary Fund | April 2018 19


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 1.1 . Strong n o into c ange ra e Fun Po e allenge or t e e i ui Fi e nco e ar et


ETFs invested in less liquid bond markets are receiving strong ... and ETFs are owning a growing share of the underlying markets.
in ows ...
1. et un er anage ent o F nve te in lobal ig iel . on F ol ing a a S are o otal ar et alue
an oan an erging ar et on Percent
illion o S ollar
150 5
Global bank loans EM ETFs
EM bonds US high-yield ETFs 4
100 Global high yield
3

2
50
1

0 0
2007 08 09 10 11 12 13 14 15 16 17 2008 09 10 11 12 13 14 15 16 17

Although ETFs can provide additional liquidity to the less liquid bond ... their investor base is significantly more ight prone ...
markets ...
3. atio o verage ra ing olu e to S are De tro e or reate or . Flo a a Percentage o Net et alue or ig iel on
S ig iel an on F F an egulate nve t ent Fun
Si ont oving average Percent
8 ETFs 20
US high-yield bonds EM bonds
Regulated investment funds
7 15

10
6
5
5
0
4 –5
3 percent (of NAV) out ows
3 –10
2012 13 14 15 16 17 2008 09 10 11 12 13 14 15 16 17

... and their greater sensitivity to major liquid markets increases contagion risks.

. verage D na ic on itional eta it S P

0.6
EM and US high-yield ETFs
0.5 Underlying indices

0.4

0.3

0.2

0.1

0.0
2008 09 10 11 12 13 14 15 16 17 18

Sources: Bloomberg Finance L.P.; EPFR Global; Haver Analytics; ICE Bank of America Merrill Lynch; and IMF staff estimates.
Note: The market value of underlying bonds in panel 2 is calculated using ICE Bank of America Merrill Lynch indices. EM = emerging market; ETF = exchange-traded
fund; NAV = net asset value. S&P = Standard & Poor’s.

20 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Risks Arising from the Buildup of Financial Crypto Assets: New Coin on the Block, Reach for
Vulnerabilities Should Be Managed Yield, or Asset Price Bubble?
Regulators and financial market participants should Amid stretched valuations in many risky asset classes,
avoid complacency and be mindful of the risk of crypto assets have erupted onto the financial landscape
sudden bouts of extreme volatility. Although financial and their prices have skyrocketed. Some of the technolog-
markets functioned well during the turbulence in early ical advances behind them have the potential to increase
February, the episode was largely confined to global the efficiency of payment systems and the financial
equity markets. Asset valuations remain stretched, and infrastructure. There has been a notable proliferation of
rising interest rates may be accompanied by a repricing crypto assets in recent years and major US exchanges have
of risky assets and further spikes in volatility. Regula- launched futures contracts. However, crypto assets have
tors should, therefore, ensure that financial institutions also been afflicted by notorious cases of fraud, security
maintain robust risk management standards, including breaches, and operational failures and have been asso-
through the close monitoring and assessment of expo- ciated with illicit activities. At present, crypto assets do
sures to asset classes deemed to be overvalued. Finan- not appear to pose macrocritical financial stability risks.
cial market participants should remain attuned to the Policymakers, however, will need to be nimble, innova-
risks associated with rising interest rates and monetary tive, and cooperative to tackle potential financial stability
policy normalization. challenges should crypto assets be used more widely.
Given signs of late-stage credit cycle dynamics,
policymakers should use the macroprudential tools at Crypto Assets: A New Asset Class and Means of Payment?
their disposal more actively. In addition to deploying Crypto assets have the potential to combine the
standard capital- and borrower-based macroprudential benefits of traditional currencies and commodities.28
instruments, regulators should improve credit risk Like fiat money, they can potentially be exchanged for
monitoring, also focusing on deterioration of nonprice other currencies, be used for payments, and store val-
terms and investor protection. Regulators should also ue.29 As investment products, they may offer portfolio
be mindful of the unintended consequences of regula- diversification, although their ability to do so is still
tory measures, including migration of activity toward limited by their short track record, regulatory uncer-
more opaque segments of the financial system. tainty, and primitive market infrastructure.
Finally, the macroprudential toolkit needs to be The technology underlying crypto assets—distributed
expanded to address risks in the nonbank financial sec- ledger technology (DLT)—could also lead to more
tor. For example, regulators should do the following: efficient market infrastructure (IMF 2016a and CPMI
• Endorse a clear and common definition of financial lever- 2017). This technology differs from traditional payment
age in investment funds: This definition would improve systems, which require a clearing entity, such as a central
transparency, particularly for derivatives positions. bank, that settles transactions and distributes funds
Lack of progress on regulation covering the use of between participants. DLT, in contrast, uses multiple
derivatives is also a concern that should be addressed. copies of the central ledger, which are kept by individual
• Continue to strengthen supervisory frameworks for entities. Blocks of transactions are subsequently validated
liquidity risk management in investment funds: and recorded, forming a historical chain—hence the
Although the International Organization of Securities name blockchain. New units of the major crypto assets
Commissions’ latest report on liquidity risk manage- are supplied by “miners” who solve a cryptographic
ment for collective investment funds (IOSCO 2018) puzzle as part of the validation process and receive a
provides welcome guidance on this front, there is new coin in return. This procedure, however, is costly
scope for the country authorities to monitor further
the effectiveness of existing liquidity risk management 28The term “crypto asset” is used here to refer to digital currencies

that rely on encryption techniques to regulate the generation of units


tools used by fund managers. More broadly, it is
and verification of transfers. Digital currencies are often referred to as
important that the authorities across different juris- “cryptocurrencies” in the popular press. Although tokens and initial
dictions agree on a harmonized and coherent mac- coin offerings (ICOs) are discussed at times in the section, the main
roprudential approach to the financial stability risks focus is on crypto assets. ICOs are issuances of digital currencies sold
via auction or investor subscription in return for crypto assets.
stemming from investment fund activities, including 29Some jurisdictions, however, have forbidden the use of crypto

the possibility of conducting stress test exercises. assets as a medium of exchange for payments.

International Monetary Fund | April 2018 21


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

in terms of both energy and time. The supply process been correlated with other assets, and therefore could
differs somewhat across crypto assets and allows for provide diversification benefits to investors, on balance.
some flexibility. For example, there is an upper limit on The unconditional correlation between Bitcoin and
the eventual outstanding amount of Bitcoins. But crypto other asset classes was close to zero between September
assets can be designed without such an upper limit, thus 2015 and March 2018 (Bank of America Merrill Lynch
mimicking more closely the money supply dynamics in 2017; Burniske and White 2017) (Table 1.1, panel
traditional fiat money systems. 1).31 Even during the most recent bout of volatility,
Crypto assets have been touted as a new form of the correlation of Bitcoin with most mainstream assets
money. However, they are still far from fulfilling the did not appear to change significantly. Pairwise correla-
three basic functions of money. While they may serve tions between different crypto assets are comparatively
as a store of value, their use as a medium of exchange subdued, again despite tremendous variance in returns
has been limited, and their elevated volatility has (Table 1.1, panel 2). Although these correlations are pos-
prevented them from becoming a reliable unit of itive, they are somewhat lower than correlations with G4
account. These shortcomings could change with wider sovereign yields and equities.32 However, it is important
adoption and technological improvements, and some to note that these correlations may change over time. So
crypto assets may be able to perform the functions of while some investors are beginning to investigate whether
money better, thus putting competitive pressure on crypto assets could be an asset class in their own right, it
fiat currencies (Box 1.4). is too early to draw clear conclusions.
Even after accounting for recent price corrections, Dedicated crypto-asset exchanges (CEs) provide
crypto assets have experienced spectacular apprecia- liquidity, leverage, and custodial services. More than
tion over the past year, spurred by the global reach for 180 CEs are transacting in thousands of different coins
yield. Nonetheless, they represent only a small share of across jurisdictions, adding up to an average daily vol-
the global financial system. Their total market value is ume of $30 billion. Still, liquidity tends to be highly
less than 3 percent of the combined G4 central bank concentrated in a select few coins and exchanges.
balance sheets (Figure 1.13, panel 1). Bitcoin alone The top 14 CEs account for more than 80 percent
accounts for 47 percent of crypto assets’ market value, of reported volume (Figure 1.14, panel 1), and the
while the next two largest crypto assets, Ethereum top 10 crypto assets account for 82 percent of the
and Ripple, account for 15 percent and 8 percent, total reported volume (Figure 1.14, panel 2). Among
respectively. As such, crypto assets currently pose currency pairs with fiat currency on one side, the US
limited challenges to fiat currencies or to the conduct dollar dominates with 71 percent of volume, followed
of monetary policy. The dramatic growth in the sector, by the yen and the euro with about 14 percent and
however, may pose risks to financial stability in the 11 percent, respectively (Figure 1.14, panel 3).
future and thus warrants vigilance by regulators. In December 2017, the Chicago Mercantile
Much attention has been devoted to the skyrocket- Exchange (CME) and Chicago Board Options
ing prices of crypto assets in 2017, which has invited Exchange (CBOE) introduced Bitcoin futures con-
comparisons with past speculative bubbles (Fig- tracts. For now, however, futures volumes represent a
ure 1.13, panel 2). However, after accounting for price small fraction of overall trading activity on the CME
volatility, risk-adjusted returns have not dramatically and CBOE and only 2.3 percent of reported trading in
exceeded those of mainstream assets over the medium the Bitcoin cash market on CEs (Figure 1.14, panel 4).
term, though they have in the most recent year (Fig- However, CEs are a major source of risk for inves-
ure 1.13, panel 3).30 For example, the Sharpe ratio tors, given their opaque and often unregulated nature.
of crypto assets was relatively close to the risk-reward Security breaches and exchange failures have led to
ratio of the S&P 500 over the past three years, and it periods—albeit short-lived—of high volatility and
was below what investing in so-called FANG stocks
(Facebook, Amazon, Netflix, Google) would yield 31September 2015 is used as the starting point of the sample
(Figure 1.13, panel 4). However, crypto assets have not because of data availability limitations before then.
32To assess conditional correlations, another multivariate GARCH

(asymmetric, generalized autoregressive conditional heteroscedastic-


30Admittedly, some of the volatility in crypto assets followed the ity) model was estimated, which found no clear trend during the
consideration of regulatory measures in various countries. recent period of sharp crypto-asset appreciation.

22 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.13. Crypto Assets: Size, Price Appreciation, Realized Volatility, and Sharpe Ratio
Crypto assets account for a small fraction of G4 central bank balance Comparison with historical bubbles.
sheets.
1. Market Capitalization of Crypto Assets 2. Price Changes
(Billions of US dollars) (Number of times the starting price)
700 6 60

Number of times the starting price


Market capitalization—Bitcoin (billions of South Sea
600 US dollars, left scale) 5 Mississippi 50
500 Market capitalization—Other Altcoins Bitcoin
(billions of US dollars, Ieft scale) 4 40
Nasdaq Composite
400 Share of G4 central bank balance sheets
3 Japan TOPIX 30
(percent, right scale)
300
2 20
200
100 1 10

0 0 0
–0.8 –0.6 –0.3 0 0.2 0.5 0.8
Dec. 2016
Jan. 17
Feb. 17
Mar. 17
Apr. 17
May 17
Jun. 17
Jul. 17
Aug. 17
Sep. 17
Oct. 17
Nov. 17
Dec. 17
Jan. 18
Feb. 18
Mar. 18
Number of years around the peak

Bitcoin s reali ed volatility is much higher than that of other asset Risk-adjusted returns of crypto assets have not dramatically exceeded
classes. those of other mainstream assets.
3. Volatility Bitcoin 4. Annualized Sharpe Ratio of the Selected Asset Classes
(Percent) EM FX (95th percentile)
350 EM FX 3.0
Oil
300 Past one year 2.5
Gold
Ripple Past three years 2.0
250
Ethereum
1.5
200
1.0
150
0.5
100 0.0
50 –0.5
0 –1.0
FANGs

Crypto assets

US real estate

S&P 500

US bonds

Gold

Oil

EM FX
2015 16 17 18

Sources: Bloomberg Finance L.P.; CoinDance; CoinMetrics; European Central Bank; Haver Analytics; national central banks; Yale International Center for Finance; and
IMF staff estimates.
Note: Panel 3 is based on 90-day realized volatility. In panel 4, crypto assets is an average across Bitcoin, Ethereum, Litecoin, and Ripple. The Sharpe ratio is the
average return earned in excess of the risk-free rate per unit of total risk. EM = emerging market; FANGs = equal-weighted index of highly traded stocks of
technology and tech-enabled companies such as Facebook, ma on, Net ix, and lphabet s Google; F = foreign exchange; G4 = Group of Four (euro area, Japan,
United Kingdom, United States); TOPIX = Tokyo Stock Price Index.

International Monetary Fund | April 2018 23


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Table 1.1. Correlation of Bitcoin with Key Asset Classes and within Crypto Assets
The unconditional correlation between Bitcoin and other asset classes has been close to zero.
1. Unconditional Covariance Matrix of Daily Returns within Selected Asset Classes
Long US Chinese
Bitcoin S&P 500 Treasury ETF Euro Renminbi Gold
Bitcoin 1.00 0.02 0.02 –0.04 0.04 0.03
Standard & Poor’s 500 0.02 1.00 –0.32 –0.05 –0.09 –0.14
Long US Treasury ETF 0.02 –0.32 1.00 0.11 –0.07 0.39
Euro –0.04 –0.05 0.11 1.00 –0.37 0.42
Chinese renminbi 0.04 –0.09 –0.07 –0.37 1.00 –0.28
Gold 0.03 –0.14 0.39 0.42 –0.28 1.00
Pairwise correlations among the various crypto-asset pairs remain low.
2. Unconditional Covariance Matrix of Daily Returns within Selected Crypto Assets
Bitcoin Monero Ethereum Ripple Litecoin
Bitcoin 1.00 0.36 0.35 0.28 0.49
Monero 0.36 1.00 0.40 0.23 0.29
Ethereum 0.35 0.40 1.00 0.22 0.30
Ripple 0.28 0.23 0.22 1.00 0.33
Litecoin 0.49 0.29 0.30 0.33 1.00
Sources: Bloomberg L.P.; and IMF staff estimates
Note: Correlations are calculated over September 2015–March 2018. ETF = exchange-traded fund.

severe losses. Data on trading volumes can be unreli- • Leveraged trading: CEs have set generous limits on
able, especially since CEs operate under heterogeneous leveraged positions, in some cases reportedly 15
rules with different fee structures, investor bases, and times, 25 times, and even 100 times (Deutsche
levels of regulatory oversight. Bank 2017).33 As in any exchange, sudden deprecia-
tions prompt margin calls and amplify price moves.
Separately, concerns have also been raised about
Financial Stability Risk Assessment futures contracts traded on the CME and CBOE,
It is impossible to know the extent to which crypto given that clearing members in these exchanges bear
assets may transform the financial infrastructure and the risk associated with these contracts through their
whether most new crypto assets are likely to disap- obligation to the guarantee fund, even if they do not
pear as in past episodes of technological innovation participate directly in the market.34 Still, the com-
(as many tech companies did during the boom of the bination of low asset return correlations discussed
late 1990s, for example). Before they can transform previously and crypto assets’ small footprint within
financial activity in a meaningful and lasting manner, the financial system suggests that the risk of spill-
crypto assets will first need to earn the confidence overs from idiosyncratic price moves in crypto assets
and support of consumers and financial authorities. to the wider market may be limited at this point.
The initial step in this process will involve coming to • Integration into mainstream financial products: The
a consensus within the global regulatory community proliferation of crypto-asset-related investment funds,
about what crypto assets are—for example, a secu- ETFs, and futures contracts increases the opportu-
rity or a currency—and the role they can play in the
33Leverage limits have been reported at 15 times an investor’s cash
financial system. Although Bitcoin was indeed created
deposits in Japan’s bitFlyer exchange (“Bitcoin feeding frenzy fuelled
to circumvent a lack of trust among trading parties by 15 times leverage, says exchange,” https://www.ft.com/content/
(Nakamoto 2008), a series of notorious fraud cases has 7f02cdba-dbd6-11e7-a039-c64b1c09b482). Other exchanges offer
undermined this goal, suggesting increased prudential even more extreme leverage opportunities of up to 100 times (see
www.bitmex.com). In practice, however, industry contacts indicate
regulation may be needed. At present, crypto assets do
that actual average leverage tends to be between 3 and 8 times.
not appear to pose risks to financial stability. However, 34“Open letter to CFTC chairman Giancarlo regarding the listing

regulators should be vigilant to the potential for finan- of crypto-asset derivatives,” from the US Futures Industry Associa-
cial stability challenges that could arise should crypto tion to the Commodity Futures Trading Commission regarding the
introduction of futures contracts on crypto assets (https://fia.org/
assets be used more widely. A few aspects that deserve articles/open-letter-cftc-chairman-giancarlo-regarding-listing-crypto
monitoring are highlighted below. -asset-derivatives).

24 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.14. Share of Trading Volumes across Exchanges, Crypto Assets, and Fiat Currencies

Trading volume is highly concentrated, with 80 percent of volume Volume share across crypto assets is led by Bitcoin, Ethereum, Ripple,
traded on just 14 exchanges. and Tether.
1. Cumulative Market Share across the Various Cryptoexchanges, 2. Reported Volumes by Cryptocurrency, March 2018
March 2018 (Percent)
(Percent)
100 Others 19%
95 percent volume = 36 exchanges
Cumulative market share

80 (20 percent of total) Storm 1%


Bitcoin 39%
TRON 2%
60
Ethereum Classic 2%
Bitcoin Cash 3%
40 80 percent volume = 14 exchanges EOS 3%
(8 percent of total) Litecoin 3%
20
60 percent volume = 7 exchanges Ripple 4%
(4 percent of total)
0 Ethereum 11% Tether 14%
1 21 41 61 81 101 121 161 181
Number of cryptoexchanges, sorted by size

Composition of reported volumes has shifted away from the Chinese Bitcoin futures volumes remain low.
exchanges.
3. Bitcoin: Reported Volumes by Fiat Currency 4. Bitcoin Futures as a Proportion of Cash Volume
(Percent) (Percent)
100 6.0

80 Ratio of volumes traded in futures 5.0


versus cash
4.0
60
Average 3.0
40 People’s Bank of China
crackdown on 2.0
cryptoexchanges
20 1.0

0 0.0
Dec. 28, 2017

Jan. 1, 18

Jan. 15, 18

Jan. 29, 18

Feb. 12, 18

Feb. 26, 18

Mar. 12, 18

Mar. 26, 18
Jun. 2016 Sep. 16 Dec. 16 Mar. 17 Jun. 17 Sep. 17 Dec. 17 Mar. 18
Chinese renminbi Euro Japanese yen
Korean won Others US dollar

Sources: Bitcoinity; Bloomberg Finance L.P.; CoinMarketCap; and IMF staff estimates.

nities for mainstream investors to incorporate these financial stability risks may become more promi-
assets into their portfolios. However, this broadening nent because the critical prudential and safety-net
of the investor base could result in increased correla- functions of existing banking systems (for example,
tion between crypto assets and traditional assets over consumer protection, resolution regulations, and
time, increasing the potential for transmission of systemic liquidity management by the central bank)
shocks, especially during episodes of risk aversion. would safeguard a smaller segment of the financial
• Partial disintermediation of the banking system: A system, and the ability of central banks to function as
large shift away from fiat money toward crypto assets a lender of last resort may also be curtailed.
could add challenges to banks’ business models. Such • Cross-border considerations: The lack of transparency in
a shift, if on a broad scale, would result in a more the markets and the rapid pace of growth could cause
decentralized financial system in which banks would market disruptions. Those disruptions could be trans-
play a smaller role in traditional lending business and mitted across national boundaries given the borderless
in payment systems. In such a decentralized system, nature of the underlying transaction mechanisms, a

International Monetary Fund | April 2018 25


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

development that could be further facilitated by the have issued guidelines with the intent to regulate
differing national regulatory approaches. ICOs based on economic function and the purpose for
which the token is issued, its tradability, and its trans-
Investor Protection and Anti-Money-Laundering Aspects ferability. In contrast, China, and Korea have cracked
down on some trading activities.
Crypto assets also present concerns for investor and
Future policymaking will need to be nimble, inno-
consumer protection, as highlighted by the Interna-
vative, and cooperative. The IMF can help advance the
tional Organization of Securities Commissions and in
agenda on regulation of crypto assets by offering advice
related forums. In this regard, securities regulators have
and by serving as a forum for discussion and interna-
drawn attention to the risks around ICOs, mostly on
tional collaboration. National authorities and inter-
the back of the increasing targeting of ICOs to retail
national standard setters are encouraged to intensify
investors by parties located outside the investor’s home
cooperation on the monitoring of crypto assets and on
jurisdiction, thus escaping the purview of the relevant
the consistency of the regulatory approach. Immediate
securities regulator. Risks around ICOs include the
action is needed to close data gaps that inhibit effective
heightened potential for fraud, cross-border distribu-
monitoring of potential risks and their links to the core
tion risks relating to heterogeneous regulatory regimes,
financial system; support systemic risk assessment and
information asymmetries, technological flaws, and
timely policy responses; and underpin measures to pro-
liquidity risks partly caused by the lack of reliable mar-
tect consumers, investors, and market integrity. And
ket makers and opaque trading practices.35
given the borderless nature of crypto assets and risks of
By design, crypto-asset transactions entail a high
regulatory arbitrage, drawing out common elements of
degree of anonymity. This results in a potentially major
effective regulatory approaches to facilitate consistent
new vehicle for money laundering and the financing of
international cooperation is essential. Such common
terrorism. Therefore, regulators and supervisors will have
elements could include good practices and regulatory
to be particularly vigilant regarding money laundering
requirements to promote the transparency and integ-
and the financing of terrorism when it comes to design-
rity of ICOs and to strengthen the risk management
ing the appropriate environment for crypto assets (IMF
and robustness of crypto-asset exchanges.
2016a). Preventive measures such as reporting require-
ments, customer due diligence, and transaction moni-
toring could be employed to ensure that crypto assets
provide similar safeguards to traditional money against
Vulnerabilities in Emerging Markets, Low-
money laundering and the financing of terrorism.
Income Countries, and China
A number of emerging market economies have taken
Policy Response advantage of benign external financial conditions to
address imbalances and build buffers; in others, however,
Ultimately, regulators need to decide what role
vulnerabilities have continued to build. Monetary policy
crypto assets could play in the financial system. So
normalization in advanced economies could result in a
far, views have varied widely, often within the same
tightening of global financial conditions and a reduction in
jurisdiction (see FATF 2015). In the United States, the
capital flows, increasing rollover risk and adversely affecting
Commodity Futures Trading Commission sees crypto
productive investment. With weaker issuers increasingly
assets as a commodity, whereas the Internal Revenue
able to access capital markets and with fickle investors
Service considers them property, and the Securities
playing a larger role in recent years, stress amplifiers have
and Exchange Commission (SEC) has acted on a
risen. In addition, a considerable number of low-income
case-by-case basis, including by halting some ICOs.36
countries and other small non-investment-grade issuers have
Discrepancies also appear across countries. After host-
experienced a sharp deterioration in debt sustainability.
ing a large share of recent ICOs, the Swiss authorities
Meanwhile, the creditor composition in these countries has
35Fordetails, see IOSCO (2017). become more complex, posing policy challenges for ongoing
36Forexample, the SEC ruled last year that tokens issued by a vir- and prospective debt restructuring. In China, regulators
tual organization known as “The DAO” were securities, hence sub- have taken a number of steps to reduce risks in the finan-
ject to federal regulation. More recently, the SEC halted some ICOs
and launched a probe into several crypto assets. The SEC’s chairman cial system. Despite these efforts, however, vulnerabilities
has promised increased scrutiny to prevent fraud in this area. remain elevated. The use of leverage and liquidity transfor-

26 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.15. Improving Fundamentals, Increased Foreign Currency Issuance

Portfolio ows have been robust. Reserve coverage has improved, but a tail of weak countries remains.
1. Nonresident Portfolio Flows to EMs 2. Reserves as a Share of External Financing Needs
(Three-month rolling sum in billions of US dollars) (Percent of countries)
Portfolio debt Portfolio equity
140 60
<50% 50–100%
120
100 50
80 40
60
40 30
20
0 20
–20 10
Taper RMB US
–40 tantrum shock election
–60 0
2013 14 15 16 17 18 2005 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20

EMs were less vulnerable to dollar appreciation during 2014–15 than Sovereign issuers had a record year in 2017, and the share of
in the late 1990s. non-investment-grade issuers is on the rise.
3. Change in Emerging Markets’ Net Foreign Assets during 4. Sovereign Issuance of International Bonds by Rating
Periods of Dollar Appreciation (Percent of total)
40 100 220
IDN Issuance (billions of US dollars,
12-month rolling sum, right scale) 200
Change in net foreign assets

30 80 180
ZAF 160
THA
(percent of GDP)

20 THA
60 140
SAU POL MYS
RUS 120
10 IND IDN
SAU BRA 100
CHN MEX TUR 40 80
0
MYS TUR 60
MEX IND 20
–10 CHN POL RUS Weaker credit quality 40
ZAF BRA issuers accessing markets 20
–20 0 0
–20 0 20 40 60 80 2011 12 13 14 15 16 17 18
Exchange rate depreciation Not rated CCC or below
(percentage change versus US dollar) B BB
Linear trend (period 1998–99) Period 1998–99 BBB A or above
Linear trend (period 2014–15) Period 2014–15

Sources: Bloomberg Finance L.P.; Bond Radar; Lane and Milesi-Ferretti (2007); Institute of International Finance; IMF, World Economic Outlook database; and IMF
staff estimates.
Note: Panel 2 sample includes 54 EM economies that are part of an external debt sovereign benchmark index. Data labels in panel 3 use International Organization
for Standardization (ISO) country codes. EM = emerging market; RMB = renminbi.

mation in risky investment products remains widespread, Nonresident portfolio flows to emerging market
with risks residing in opaque corners of the financial system. economies rose to an estimated $240 billion during
2017—twice the pace observed in the previous two
Has the Prolonged Search for Yield Made Emerging years (Figure 1.15, panel 1). Although market interest
Market Economies More Vulnerable? rates in advanced economies have risen notably over
Investor sentiment toward emerging markets has the past six months, emerging market assets have gen-
remained favorable since the previous GFSR, under- erally performed well over the same period, even after
pinned by improving growth prospects and robust accounting for the episode of volatility in global equity
portfolio flows. Real GDP growth in emerging market markets in early February.
economies is projected to reach 4.9 percent in 2018, The gradual and well-telegraphed normalization of
the fastest pace since 2013 (see the April 2018 WEO). monetary policy in advanced economies has pro-

International Monetary Fund | April 2018 27


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

vided a window of opportunity for emerging market countries more susceptible to a reversal in capital
economies. Current account deficits have generally flows. The growing role of fickle investors is evidenced
narrowed since 2013 but remain large in a number of by an upward trend in the “investor base risk index”
emerging markets and are projected to widen, espe- based on Arslanalp and Tsuda (2012) (Figure 1.16,
cially for commodity-importing countries (see the panel 1).37 Foreign investor participation helps
April 2018 WEO). Strong capital inflows have enabled deepen capital markets, but high shares of foreign
some countries to strengthen reserve buffers, leaving ownership can also increase vulnerability to interest
a smaller tail of countries with low reserve adequacy rate and rollover risks; for example, in the event of a
(Figure 1.15, panel 2). Corporate fundamentals have risk aversion episode. Foreign ownership of sovereign
also been improving (see “Reach for Yield or Over- bonds remains high among several emerging market
reach in Risky Assets?” section and Figure 1.8, panels economies (Figure 1.16, panel 2). Among nonbank
4 and 5). A strong recovery in earnings has improved investors, mutual funds and ETFs stand out as poten-
interest coverage, and corporate debt levels have fallen tial sources of volatility because they are associated
somewhat recently but remain elevated in several coun- with increased sensitivity of flows to global financial
tries (see October 2017 GFSR). conditions (for example, Cerutti, Claessens, and Puy
A sharp appreciation of the US dollar could pose 2015; Converse, Levy-Yeyati, and Williams 2018).
challenges to some countries, even as external balance These investment funds now own nearly one-sixth
sheets at an aggregate level have become less vulnerable of fixed-income assets included in emerging market
to exchange rate depreciations. Against the backdrop of benchmark indices, and more than a third in some
an increase in foreign currency sovereign and corporate countries (Figure 1.16, panel 3).
issuance, a stronger US dollar could put pressure on The reduction in portfolio flows to emerging
emerging markets. Borrowers that obtained credit in markets expected to result from monetary policy
foreign currency would see the domestic currency value normalization in the United States in the coming
of their liabilities rise, making it more challenging to years could put countries with weak fundamentals
service and repay debt. A sudden episode of risk aver- at risk. Assuming the Federal Reserve’s balance sheet
sion could be accompanied by capital outflows, reduce normalization proceeds as announced and the federal
productive investment, and put growth at risk in some funds rate is raised to 3.6 percent by early 2020, as
emerging markets. However, many emerging market projected in the April 2018 WEO, portfolio flows
economies have continued to improve their net foreign to emerging markets are estimated to be reduced by
currency positions, thus reducing their exposures an average of $40 billion a year in 2018–19.38 This
to currency depreciations. Indeed, when the dollar estimate assumes a smooth normalization process
appreciated in 2014–15, net foreign asset positions in which there is no increase in investor risk aver-
improved in most emerging markets, a reflection of sion. If, instead, the policy tightening process were
increased foreign currency assets and higher reliance on accompanied by a rise in risk aversion on the order
both equity liabilities and domestic currency borrow- of magnitude observed after the renminbi deval-
ing (Figure 1.15, panel 3; also see IMF 2016b). uation of August 2015, portfolio flows could be
Aggregate measures of net external balances may, reduced by a total of $60 billion a year over the same
however, mask vulnerabilities arising from offset- period, equivalent to one-quarter of annual inflows
ting gross positions and imbalances at a sectoral
level. Indeed, gross issuance of foreign currency
37The investor base risk index aims to capture the likelihood
corporate and sovereign debt securities rose to new
of sudden outflows, given the different types of investors that
highs in 2017, allowing even weaker issuers to hold sovereign debt (Arslanalp and Tsuda 2012). The measure is
access markets (Figure 1.15, panel 4). The share of calculated based on the historical relationship between changes in
non-investment-grade issuance has risen to more than investor holdings of sovereign debt and sovereign bond yields. The
index ranges from 0 to 100. The higher the score, the greater the
40 percent over the past 12 months, boosted by the likelihood of a sudden investor outflow. According to this measure,
return to bond markets of issuers such as Egypt and the most fickle investor type is foreign nonbanks, followed by foreign
smaller issuers in sub-Saharan Africa. banks, foreign central banks, domestic nonbanks, domestic banks,
and the domestic central bank.
Furthermore, exposure to less committed, poten- 38Estimates are based on an econometric model discussed in the
tially “flighty,” investors is growing, which makes October 2017 GFSR.

28 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.16. Creditor Base and External Financing Vulnerabilities

The investor base for EM sovereign debt has become more fickle in Vulnerabilities in public debt structure remain high for several
recent years. countries.
1. nve tor a e i n e ggregate ro erging ar et . Public Debt el b Nonre i ent an Deno inate in Foreign urrenc
cono ie (Share of total)
50 25th–75th percentile 80

Public debt held by nonresidents


Median 70
45 DOM
Average IDN
POL 60
40
HUN UKR 50
35 TUR URY 40
MEX ARG
ZAF CHL PER COL 30
30
THA PHL 20
25 RUS
BRA 10
IND
20 0
2004 06 08 10 12 14 16 0 10 20 30 40 50 60 70 80
Public debt in foreign currency

Fund investors account for a growing share of EM portfolio asset Countries with high financing needs could be vulnerable to a
holdings. tightening of global financing conditions.
3. Exposure to Fund Investors 4. Reserve Adequacy and External Financing Needs
S are o bon in benc ar in ice o ne b inve t ent
funds, percent)
18 50 RUS THA 250
Total (left scale) Range of countries (right scale) UKR 45
16 EGP ig e ternal financing re uire ent
IDN 40 PHL 200
MAL

Reserves/ARA metric
HUN ARG
ROU 35 BRA KAZ IND ROU
14 PHL PER 30 150
MEX IDN MEX Reserve adequacy range
12 COL 25 COL POL
RUS 20 CHL 100
10 TUR 15 CHN HUN ZAF MYS
POL TUR
8 10 PAK 50
ZAF THA CHL BRA
5
IND CHN
6 0 0
2012 13 14 15 16 17 0 5 10 15 20 25 30 35 40
verage external financing requirements, 2018–20

Sources: EPFR Global; JPMorgan Chase & Co.; and IMF staff estimates.
Note: Panel 1 uses the data set from rslanalp and Tsuda (2012 updated). Panel 2 uses 2017 data or latest available. In panel 3, exposures to fund investors are
measured as assets under management at investment funds (based on EPFR data) divided by the amount outstanding of bonds in EM local currency and hard
currency benchmark indices (based on JPMorgan data). Panel 4 uses the IMF s ssessing Reserve dequacy ( R ) metric estimates for 2017, not ad usted for
capital controls. Data labels in the figure use International Organi ation for Standardi ation (ISO) country codes. EM = emerging market.

in 2010–17.39 Countries that have not addressed that rising interest rates could weigh on growth and
vulnerabilities (such as low reserve adequacy) during aggravate financial stability risks. Commodity pro-
the favorable period could be particularly at risk of ducers could be further affected if monetary tighten-
a reversal in capital flows from rapid tightening of ing is accompanied by weakening commodity prices
global financial conditions (Figure 1.16, panel 4). (Husain and others 2015).
Moreover, countries with fixed exchange rates at
different stages of the economic cycle face the risk
Countries Should Prepare for Tighter Financial
Conditions by Pursuing Adequate Policies
39Specifically, a 100-basis-point increase in the spread on US
Policymakers in emerging markets should use current
BBB-rated corporate bonds was assumed, which is on par with the
favorable conditions to prepare for a potential retrench-
increase observed from July 2015 to February 2016.

International Monetary Fund | April 2018 29


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

ment in capital flows that may be accelerated by mon- In recent years, however, public debt vulnerabili-
etary policy normalization in advanced economies and ties have increased because of revenue declines for
associated tightening in financial conditions. Vulnerable commodity-exporting countries, exchange rate depre-
countries could be disproportionately affected and should ciations, consolidation of previously unaccounted for
strengthen fundamentals in preparation for less benign state-owned enterprise debt, and rising interest rate
external financial conditions (Sahay and others 2014; costs attributable to higher shares of nonconcessional
Chen, Mancini-Griffoli, and Sahay 2014). To build debt.40 More than 45 percent of low-income coun-
resilience and reduce the likelihood of outflows, coun- tries were at high risk of, or already in, debt distress
tries should maintain sound macroeconomic, structural, as measured by debt sustainability ratings in 2017
financial, and macroprudential policies, taking into (Figure 1.17, panel 1),41 while several countries have
account their cyclical position, balance sheet vulnerabil- debt-to-GDP levels close to what they were when debt
ities, and macroeconomic policy space. Because capital relief was granted (see April 2018 Fiscal Monitor).
outflow pressures can be driven by external rather than In addition, vulnerabilities are on the rise not just
domestic factors, it is also necessary to build appropriate in the current set of low-income countries but also
buffers that can be used during stress, possibly by taking in a wider set of small non-investment-grade issuers,
advantage of low interest rates to borrow long term. which includes countries that have “graduated” from
Buffers include building international reserves to support low-income country status (Figure 1.17, panel 2).
exchange rate regimes in periods of stress or taking steps The increase in private and non–Paris Club creditors
toward making the exchange rate regime more flexible has led to a substantial change in creditor composi-
where appropriate. Monitoring firms’ foreign exchange tion over the past decade. Among countries recently
exposures and ensuring their capacity to absorb exchange surveyed by the IMF, the combined share of external
rate risks would also help emerging market economies financing provided by commercial creditors increased
cope with a reduction in capital inflows. from 7.5 percent to 15 percent (Figure 1.17, panel 3)
If external financial conditions deteriorate sharply, between 2007 and 2016, and financing from non–
a rapid and appropriate macroeconomic and finan- Paris Club creditors has risen from 18.5 to 37 percent.
cial policy response to capital outflow pressures is Among non–Paris Club creditors, China has taken a key
particularly important (IMF 2012, 2015, 2016c). role in providing external financing. Since 2010, China
Exchange rates often serve as a critical shock absorber, has provided commitments of more than $100 billion
but in countries with sufficient reserve buffers, foreign a year, on average, in financing to emerging market
exchange intervention can be useful for preventing economies,42 over $30 billion of which has been to
disorderly market conditions and allowing the econ- low-income countries. This change in debt composition
omy to gradually adjust to a new equilibrium. In the has been more pronounced in several heavily indebted
context of outflow pressures, capital flow manage- poor countries (HIPCs) that have received debt relief
ment measures should only be implemented in crisis and are now in debt difficulty (Figure 1.17, panel 4).43
situations or when a crisis is considered imminent, and The shift to a more diverse composition of creditors
should not substitute for any needed macroeconomic can facilitate faster accumulation of debt and can also
adjustment. When warranted, such measures should make debt resolution more complex. The involvement
be transparent, temporary, and nondiscriminatory and of new non–Paris Club official, as well as private, credi-
should be lifted once crisis conditions abate. tors remains relatively untested. There is less experience

40See IMF (2018) for some stylized facts on debt accumulation in

recent years.
Rising Public Debt Vulnerabilities in Low-Income 41The group of low-income countries refers to countries eligible
Countries and Small Non-Investment-Grade Sovereigns for concessional financing through the Poverty Reduction Growth
Trust. For a definition of low-income developing countries,
Debt burdens have increased and affordability has
see IMF 2018.
deteriorated over the past few years among low-income 42According to estimates using data from AidData at the College

borrowers and other small non-investment-grade of William & Mary (http://aiddata.org/data/chinese-global-official


issuers. Public and external debt burdens for many -finance-dataset).
43To date, 36 countries have received the full amount of debt
borrowers decreased from 2007 to 2014, especially relief for which they were eligible through the HIPC initiative and
in countries that benefited from debt relief efforts. the Multilateral Debt Relief Initiative.

30 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.17. Rising Vulnerabilities and More Complex Creditor Composition

More than 45 percent of LICs are at high risk or in debt distress. Debt-servicing costs have risen across small non-investment-grade
EM issuers.
1. Debt Sustainability Ratings 2. Debt Service as a Share of Exports
(Percent of PRGT-eligible countries with DSAs) (Percent of small high-yield EMs)
50 90
45 In debt distress High risk >25 percent 20–25 percent 15–20 percent 80
40 70
35 60
30
50
25
40
20
15 30
10 20
5 10
0 0
2007 08 09 10 11 12 13 14 15 16 17 2007 08 09 10 11 12 13 14 15 16 17 18

Composition of creditors of public debt has changed substantially ... ... particularly in several post-HIPC countries.

3. External Debt Creditor Composition . ternal re itor o o ition o Po t P ountrie in Debt Di ficult
(Percent of public and publicly guaranteed external debt) (Percent of public and publicly guaranteed external debt)
WB, IDB, IMF, AfDB, AsDB, and Paris Club Other multilateral Non–Paris Club excluding China China
Bonds Commercial banks Other commercial
100 100

80 80

60 60

40 40

20 20

0 0
2007 08 09 10 11 12 13 14 15 16 2007 16

Sources: Bank for International Settlements; IMF-Organisation for Economic Co-operation and Development-WB Joint External Debt Statistics; 2017 survey of IMF
country desks; World Bank International Debt Statistics; IMF, International Financial Statistics database; and various IMF staff reports.
Note: Panel 1 debt sustainability assessment ratings for LICs are based on the Debt Sustainability Framework for Low-Income Countries. Panel 2 sample includes
35 non-investment-grade relatively small issuers that are part of JP Morgan’s EMBIG index. Each has a weight of less than 2 percent of the index. Panel 3 is based
on 37 LIDCs where continuous data are available from 2007 to 2016. Panel 4 countries included are Cameroon, Chad, Republic of Congo, Ethiopia, Ghana,
Mauritania, Mo ambique, and ambia. Figures are in simple averages and may be overly in uenced by the experience of countries with very high levels of debt.
Other commercial creditors include oil traders. AfDB = African Development Bank; AsDB = Asian Development Bank; DSA = Debt Sustainability Analysis;
EM = emerging market; EMBIG = JPMorgan’s Emerging Market Bond Index Global; HIPC = highly indebted poor country; IDB = Inter-American Development Bank;
LIC = low-income country; LIDC = low-income developing country; PRGT = Poverty Reduction and Growth Trust; WB = World Bank.

with their engagement before and during debt distress to pay debt in lieu of cash. Both commercial and bilat-
than with traditional official lenders. Many of them eral lenders have resorted to collateralized lending, as
have not been part of debt resolution in the past, but highlighted in recent debt distress cases (for example,
they could be called on to provide support in such cases. Chad, Republic of Congo, Venezuela) that are yet to
The use of collateralized debt can further complicate be resolved.44 Apart from such cases, however, details
debt resolution and lower recovery rates for creditors on collateralized deals remain scant.45 Given that sover-
with unsecured claims. Some commodity-producing
44For details on recent debt distress cases in low-income develop-
countries offer their exports as collateral; for example,
ing countries, see IMF (2018).
by issuing senior loans through state-owned enterprises 45Bräutigam, Gallagher, and Hwang (2016) find that one-third

or by pledging commodity shipments that can be used of Chinese loans to Africa are secured by commodity exports,

International Monetary Fund | April 2018 31


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

eigns have significant protections from seizure of assets, Official creditors, when needed, should emphasize
most creditors are reliant on good faith negotiations timely resolution of debt distress cases to avoid poten-
to secure recovery in distress.46 The direct claim on an tial spillovers and to minimize the costs for both the
asset or a revenue stream, however, can grant holders issuer and creditors. Transparent and broad creditor
of collateralized debt favorable treatment. Thus, collat- coordination should be encouraged, especially when
eralized claims could impair the ability of the sovereign the set of lenders is diverse. New official creditors
to offer more generous terms in a renegotiation of its should consider the benefits of adopting sustainable
unsecured debt, and require a more significant haircut lending rules, such as those endorsed by the Group
on remaining debt to ensure debt sustainability. of 20. Finally, borrowers and official creditors should
The higher share of private sector creditors could ensure transparency of the contractual terms for new
make low-income countries and other vulnera- debt, including debt that is issued by entities related to
ble emerging market borrowers more sensitive to a the sovereign.
tightening of global financial conditions. The increase
in the share of Eurobonds and commercial loans
with shorter maturities can expose issuers to higher Shadow Banking Reform and Risk in China
rollover and interest rate risk. These new avenues of The large-scale and opaque interconnections of the
financing are untested, and it is unclear whether they Chinese financial system continue to pose stabil-
will remain available if financial conditions tighten ity risks (Figure 1.18). China’s RMB 250 trillion
significantly, particularly for first-time and low-rated (300 percent of GDP) banking system is tightly
issuers. Part of this new debt is held by investors linked to the shadow banking sector through its
who do not specialize in this sector and may choose exposure to off-balance-sheet investment vehicles.
to allocate their funds elsewhere if higher-yielding These vehicles are largely funded through the issu-
opportunities become more abundant in more tra- ance of investment products (RMB 75 trillion),
ditional hard currency assets (for example, US high with roughly half sold to multiple investors as
yield). In addition, the anticipation of complex debt high-yielding alternatives to bank deposits and half
resolutions and potentially lower recovery rates could held by single investors, including banks.47 They
trigger more rapid market repricing at the first sign of invest in various assets, such as bonds, bank depos-
sovereign stress. its, and nonstandard credit assets, as well as in other
investment products. Insurance companies also have
considerable exposure to these vehicles because they
Policies Should Address Rising Debt Vulnerabilities invest in their products and use them as a source of
To ensure a sustainable debt burden, policymakers funding. These little-regulated vehicles have played
should reduce vulnerabilities related to the structure of a critical role in facilitating China’s historic credit
their debt and attract a stable investor base, including boom and have helped create a complex web of expo-
through local bond market development. Debt man- sure between financial institutions.
agers should minimize risks emanating from rollovers, Banks are exposed to investment vehicles along
potential foreign exchange mismatches, and collateral- many dimensions—as investors, creditors, borrow-
ization. Countries should explore state contingent debt ers, guarantors, and managers. These vehicles rely
instruments that may offer some protection against on banks’ short-term financing to use leverage and
unforeseen shocks such as natural disasters, assuming manage their maturity mismatches. Banks, in turn,
these instruments are priced at reasonable cost for the receive significant flows from these vehicles in the
issuer by investors (IMF 2017b).
47Bank-issued non-principal-guaranteed wealth management

and Bräutigam and Gallagher (2014) find that roughly half of the products account for the majority of products sold to multiple inves-
$132 billion in Chinese financing to Latin America and Africa is tors. As used herein, investment products include asset management
commodity backed. products issued by securities brokers, fund companies and their
46Sovereign states are typically granted immunity for noncom- subsidiaries, trust companies, and insurers. Money market funds
mercial activities in international courts. The two jurisdictions most and other public mutual funds are more strictly regulated and not
commonly used for international debt issuance formalize this immu- included. Other forms of nonbank credit activities also carry risks
nity under the Foreign Sovereign Immunities Act (United States) but are not considered in this section; for instance, money market
and under the State Immunity Act (United Kingdom). funds, other public mutual funds, and exposures between firms.

32 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.18. Stylized Map of Linkages within China’s Financial System

Financial intermediaries and products


Ultimate Ultimate
borrowers Banks (RMB 250 trillion) creditors
Big 5
Small and medium banks
Implicit guarantees

Investment vehicles
Assets Liabilities
Households, Bank deposits and debt Repo and other borrowing
firms, government Direct
Corporate and government bonds
financing vehicles Investment investment Households
Credit and other illiquid assets products and firms
Governments, (RMB 75 trillion)
policy banks To: Investment products

Cross-product leverage and


investment flows (double-counting)

Black arrows indicate


Other financial institutions (RMB 24 trillion) fund ows
Fund management Blue arrows indicate
Securities Trust Insurance investment product
companies and
brokers companies companies issuance
subsidiaries

Sources: Asset Management Association of China; CEIC; ChinaBond; People’s Bank of China; and IMF staff estimates.
Note: Investment products include non-principal-guaranteed bank-issued wealth management products and asset management products issued by other financial
institutions depicted. Numbers shown are total on-balance-sheet assets for banks and financial institutions, and total investment products outstanding as of
end-2017 or latest available reporting period. Numbers for other financial institutions do not include fund management companies and their subsidiaries due to lack
of data. See also Ehlers, ong, and hu (2018). RMB = renminbi.

form of deposits and bond investments. Banks and ment vehicles and other borrowing between financial
other financial institutions are also direct investors institutions. Since the summer of 2016, regulators have
in investment products. Small and medium-sized incorporated bank-sponsored investment vehicles in
banking institutions and insurance companies are the macroprudential framework and have taken other
particularly exposed, with investment products steps to curb financial sector leverage and intercon-
accounting for one-fifth and one-third of their assets, nectedness.48 Proposed asset management rules would
respectively. About one-quarter of investment vehicle also overhaul the investment product market beginning
assets, in turn, are invested in other vehicles, leading in 2018. In addition to limits on investment vehicle
to opaque cross-holding and leverage structures that leverage and complexity, banks would be gradually
are difficult for regulators and investors to monitor. restricted from investing in these vehicles or providing
Banks in particular are seen as implicitly guarantee- them with financial support. This restriction would
ing the RMB 25 trillion in investment products they limit their ability to implicitly guarantee investment
manage, which allows them to package high-risk products’ fixed-yield returns, effectively converting
credit investments as low-risk retail savings products. roughly half of the market from deposit-like products
Investment vehicles managed by nonbank financial into mutual funds. In addition, the insurance regulator
institutions are perceived to be higher risk, but in has clamped down on the sale of short-term invest-
most cases banks still bear some risk as creditor, end ment products by life insurers.
investor, or guarantor.
The authorities have substantially tightened the 48For more details on China’s financial system stability assessment

regulatory framework to reduce risks related to invest- and associated policy recommendations, please refer to IMF (2017a).

International Monetary Fund | April 2018 33


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 1.19. Chinese Banking System and Financial Market Developments and Liabilities
Regulatory tightening has sti ed growth in financial sector leverage ... but capital positions are becoming more stretched and underlying
and in risky investment products ... profitability continues to fall.
1. China: Three-Month Changes in Investment Products Outstanding 2. Chinese Banks: Tier 1 Capital Ratio and Preprovision Operating
and Small and Medium Banks’ Financial Sector Claims Profit to et
(Trillions of renminbi) (Percent)
CET1 ratio Preprovision operating profits ratio
8 12.0 2.4
Investment products
Bank claims on 11.5 Small and 2.3
6
financial sector medium banks
11.0
Bank liabilities to 2.2
4
financial sector 10.5
2.1
2 10.0
2.0
9.5
0 Big 5 banks
9.0 1.9
–2 1.8
8.5
–4 8.0 1.7
Jun. 2015 Dec. 15 Jun. 16 Dec. 16 Jun. 17 Dec. 17 2011 12 13 14 15 16 17:Q3 2011 12 13 14 15 16 17:Q3

Bank vulnerabilities remain elevated as funding costs rise ... ... and liquidity remains stretched.
3. Chinese Bond Prices: Corporate Credit Spreads and Short-Term 4. Chinese Small and Medium Banks: Short-Term Nondeposit Funding
Unsecured Bank Funding Costs an ig ualit i ui et
HQLA (percent of bank assets, left scale)
Short-term nondeposit funding (percent of bank assets, left scale)
6.0 Yield spread: Corporate (AAA) 1.9 25 Ratio of short-term nondeposit funding to HQLA (right scale) 2.6x
and government three-year bonds
5.5 (percentage points, right scale) 1.7
20 2.4x
5.0 Three-month negotiable 1.5
4.5 certificate of deposit yield 1.3
(percent, left scale) 15 2.2x
4.0 1.1
3.5 0.9 10 2.0x

3.0 0.7
5 1.8x
2.5 0.5
2.0 0.3 0 1.6x
Dec. 2015

Feb. 16

Apr. 16

Jun. 16

Aug. 16

Oct. 16

Dec. 16

Feb. 17

Apr. 17

Jun. 17

Aug. 17

Oct. 17

Dec. 17

Feb. 18

2015 16 17:H1

Sources: Asset Management Association of China; Bloomberg Finance L.P.; CEIC; People’s Bank of China; S&P Global Market Intelligence; and IMF staff calculations.
Note: Panel 4 is based on a sample of 10 banks that disclose HQLA and contractual maturities. Short-term is less than three-month contractual maturity or on
demand. China s HQL definition disallows required reserves held at the People s Bank of China. CET1 = common equity Tier 1; HQL = high-quality liquid assets.

Chinese Banks Have Made Progress in Deleveraging, but annual basis in 2016 to less than 20 percent at the end
Risks Remain Elevated of 2017, and banks’ holdings of investment products
Tighter regulations have lowered growth in banks’ issued by other banks has also declined sharply.
use of risky short-term funding and in investment Financial stability risks nonetheless remain high, and
products, slowing the buildup of bank vulnerabilities. smaller banks are particularly vulnerable. Bank buffers
Lending by small and medium-sized banks through continue to thin at many of the country’s commercial
investment vehicles has slowed, as has their use of banks. In addition to still-elevated investment vehi-
wholesale short-term financing and the overall volume cle exposures, core Tier 1 capital ratios are declining
of investment products outstanding (Figure 1.19, and remain near minimum levels for many small and
panel 1). Notably, growth of banks’ exposure to other medium-sized banks, while preprovision profitabil-
financial institutions fell from about 80 percent on an ity also continues to weaken (Figure 1.19, panel 2).

34 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Following tighter regulatory constraints, money market (Figure 1.20, panel 3). Without bank-guaranteed fixed
rates have risen sharply, leading to wider corporate bond yields on investment products, the generally risk-averse
spreads, particularly for weaker borrowers (Figure 1.19, retail investor base is likely to shift toward less risky
panel 3). Highlighting liquidity risks faced by small and instruments, a development that would reduce net
medium-sized banks, reliance on short-term nondeposit demand for already illiquid corporate bonds.50 Banks
funding remains high, and short-term wholesale liabil- will also need to gradually recognize some portion of
ities are still more than double the available liquidity the corporate credit exposure held through investment
buffers at smaller banks (Figure 1.19, panel 4). vehicles as loans or bonds, requiring capital and pro-
visioning costs that will cut into loan growth capacity.
For small and medium-sized banks, even absorbing
Reforming China’s Investment Product Market— half of these exposures over two years would reduce net
An Important Conduit for Shadow Credit—Poses new loan growth from 17 percent to 6 percent, unless
Challenges to Financial Stability banks raise new capital (Figure 1.20, panel 4) (see also
A key challenge for the reform agenda will be phas- the October 2017 GFSR).
ing out implicit guarantees for investment vehicles.
Because they primarily hold illiquid and long-term
assets, such as corporate bonds and nonstandard credit China’s Insurance Sector Has Grown Rapidly, Increased
assets, these vehicles rely on guarantees to borrow and Its Risk Profile, and Become Closely Linked with Other
to meet maturing short-term liabilities to product Parts of the Financial System
holders. As a result, investment vehicles are now the Chinese life insurers have grown rapidly, and their
largest net borrower in China’s repurchase market, share prices have been volatile. Insurers’ assets have
driving overall market activity, often with relatively more than tripled in size over the past seven years,
illiquid collateral (Figure 1.20, panel 1). Furthermore, growing in line with the rest of the Chinese financial
direct lending by large banks to their sponsored vehi- system (Figure 1.21, panel 1). Growth has been fueled
cles amounts to about 10 percent of their investment by “universal life insurance,” flexible savings products
product liabilities, on average.49 (in 2015 and 2016), and more traditional life policies
Without such financial support, investment vehi- (in 2017) (Figure 1.21, panel 2). At the same time,
cles would need to hold safer, more liquid asset insurers’ share prices have risen sharply, accompanied
portfolios to avoid rollover and refinancing risks. by an increase in volatility reflecting perceived elevated
Yet allocations to such assets have recently decreased risks (Figure 1.21, panel 3). Recently, the regulator
among bank-sponsored investment vehicles, falling to took control of a large insurance group that had
one-third in 2017, from about half in 2015 (Fig- financed a rapid expansion into other business areas
ure 1.20, panel 2). Rising use of illiquid assets and with short-term high-guarantee investment products.
borrowing suggests dependence on implicit guarantees The shift into riskier investments entails vulnera-
is still trending up, underscoring the difficulty of prog- bilities for insurers and the system at large. To attain
ress in this critical area. the high guaranteed returns of their long-term policies
Reducing risks in the investment product market (4 percent, in many cases) amid the relatively small
will require further slowing credit growth in the near and illiquid corporate bond market, insurers have
term, which is necessary to ensure financial stability shifted their investments from bonds and deposits to
and sustainable growth in the medium term. Invest- equity, funds, and “other assets” (Figure 1.21, panel
ment vehicles have bought nearly all the net increase in 4). These other assets include asset and wealth man-
corporate and financial bond issuance in the past three agement products, debt and equity products, and
years and hold 70 percent of such bonds outstanding participations in joint ventures (Figure 1.21, panel 5).
These large investments in infrastructure, real estate,
49Eight banks (including four of the Big Five lenders) disclose
and loan portfolios concentrate credit risks, including
active direct lending to their investment vehicles, which account for for insurers with limited expertise in credit assessment.
nearly half of the bank-managed investment product market (more
than RMB 10 trillion in non-principal-guaranteed wealth manage- 50More than 80 percent of outstanding wealth management

ment products). This lending was equivalent to 15 percent of these products are billed as low risk, rated as 1 or 2 on an industry group–
banks’ core Tier 1 capital as of mid-2017. defined scale to 5 (with 5 being riskiest).

International Monetary Fund | April 2018 35


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 1.20. Risks and Adjustment Challenges in Chinese Investment Products


Investment vehicles are borrowing more ... ... and holding more illiquid assets.
1. Estimated Investment Vehicle Net Repo Borrowing and Interbank 2. Chinese Investment Vehicles: Bank-Issued Product Portfolio Allocation
Gross Repo Position, by Institution Type (Percent)
(Trillions of renminbi)
3.5 Investment vehicles: Small and medium banks 14 100
90
3.0 Estimated average net repo Large commercial banks 12
borrowing outstanding Other NBFIs Relatively 80
2.5 Investment vehicles 10 liquid 70
and funds 60
2.0 8
50
1.5 6 40
1.0 4 30
Relatively
illiquid 20
0.5 2
10
0.0 0 0
Jun. 2015
Sep. 15
Dec. 15
Mar. 16
Jun. 16
Sep. 16
Dec. 16
Mar. 17
Jun. 17
Sep. 17
Dec. 17

Jun. 2013

Jun. 14

Jun. 15

Jun. 16

Jun. 17 Dec. 2014 Jun. 15 Dec. 15 Jun. 16 Dec. 16 Jun. 17 Dec. 17


Deposits, interbank, and cash Sovereign and policy bank bonds
Corporate and financial bonds NSCA and other
Sources: CEIC; China Central Clearing Depository Corporation; National Sources: Bank Wealth Management Registration and Trusteeship Center; and
Interbank Funding Center; People’s Bank of China; Shanghai Clearing House; IMF staff calculations.
WIND; and IMF staff calculations. Note: Due to the lack of available data, data for June 2015 are the interpolation
Note: Gross repo position includes the sum of outstanding month-end cash of December 2014 and December 2015 data. “NSCA and other” includes mostly
borrowing and lending positions. “Investment vehicles and funds” includes repo illiquid credit assets but also has derivatives and investment fund shares.
positions by mutual funds (which are net lenders) and other NBFIs not captured NSCA = nonstandard credit asset.
in the “Other NBFIs” category. Estimated average repo borrowing outstanding is
the People’s Bank of China—reported quarterly net repo borrowing volume for
all funds, divided by the ratio of nonbank repo volume to month-end position,
minus the reported net repo position of public mutual funds and other NBFIs.
NBFI = nonbank financial institution.

Reforming investment products will further slow credit growth by ... and limiting small banks’ ability to increase lending without fresh
weakening demand for corporate and financial bond issuance ... capital.
3. China Bond Market: Corporate and Non-Policy-Bank Financial Bonds 4. Chinese Small and Medium Banks: Two-Year Estimated Loan Growth
Outstanding, by Holder Capacity Given Shadow Credit Recognition Assumption1
(Trillions of renminbi) (Percent)
35 20
Banks and other NBFIs 17.5 percent
30 Public mutual funds 15
Investment vehicles 11 Annual loan growth
25 (mid-2017)
10
20 6
Estimated two-year loan growth2
5
15 0
0
10
5 –5
–5
0 –10
Dec. 2014 Jun. 15 Dec. 15 Jun. 16 Dec. 16 Jun. 17 Dec. 17 25 50 75 100
Percentage of shadow credit recogni ed as loans
Sources: China Clearing and Depository Corporation; National Interbank Funding Sources: Bank financial reports; S P Global Market Intelligence; and IMF staff
Center; People’s Bank of China; Shanghai Clearing House; WIND; and IMF staff calculations.
1
calculations. Shadow credit is defined as 100 percent of banks investments in third-party
Note: Public mutual fund holdings shown are interpolated semi-annual data. unconsolidated structured products and 20 percent of their sponsored non-principal-
NBFI = nonbank financial institution. guaranteed wealth management products. Based on a sample of 25 listed banks
with available disclosures. Assumes banks receive no external capital and maintain
static capital and profitability ratios. Shadow credit recognition entails raising risk
weightings for selected assets to 100 percent from initial weightings of 25 and 0
percent for structured products and wealth management products, respectively.
2
Growth rates shown are annuali ed. Negative number indicates loan book would
need to shrink to initially accommodate existing shadow credit. Loan growth shown
is net of loans converted from existing shadow credit.

36 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.21. Chinese Insurers

Chinese insurers have grown rapidly ... ... fueled by life insurance sales.
1. Insurance Sector Total Assets 2. Annual Insurance Premiums
(Trillions of renminbi)
20 25 4.0
Total (trillions of renminbi, left scale) Property insurance
Percent of GDP (right scale) Traditional life 3.5
Percent of financial system (right scale) 20 Universal life
15 3.0

15 2.5
10 2.0
10 1.5
5 1.0
5
0.5
0 0 0.0
2010 11 12 13 14 15 16 17 2013 14 15 16 17

Insurers shares have risen sharply, accompanied by high volatility. Increased revenues have been invested in higher-risk assets but
capital has not been raised.
3. Equity Performance and Volatility 4. Non-Fixed-Income Assets and Capital
China insurance China banks Heng Seng index (Percent of total assets)

Equity performance olatility


(2014 = 100) (percent)
180 60 60

160 50 50
Equity, funds, and
140 40 40
“other assets”
120 30 30

100 20 20
Capital
80 10 10

60 0 0
2014 15 16 17 18 2014 15 16 17 18 2013 14 15 16 17

Other assets are mainly portfolios of infrastructure pro ects, real ariation of alternative investments and capital buffers within the
estate, and loans provided by asset managers. sector is large.
5. Decomposition of “Other Assets,” 2016 6. Capital and Alternative Investments, 2016
(Percent of other assets) (Percent of assets)
55
Largest Rest of
15 life 45

Alternative investments
sector
insurers
35
Equity funds and schemes
25%
Asset management 25
products and wealth
management Associates and Real 15
products Debt schemes oint ventures estate
30% 27% 12% 6%
5
4 6 8 10 12 14 16
Capital
Sources: nnual reports; Bloomberg Finance L.P.; China Insurance Regulatory Commission; Morgan Stanley Capital International; and IMF staff calculations.
Note: In panel 3, volatility is calculated as the annuali ed standard deviation of the relative price change for the 60 most recent trading days closing price. In panel 5,
an associate is an entity in which the company group has a long-term interest of generally not less than 20 percent of the equity voting rights and over which it is in
a position to exercise significant in uence. Panels 5 and 6 are based on annual reports of the 15 largest life insurers. These companies cover two-thirds of the total
assets of the Chinese insurance sector. In panel 6, the si e of the bubbles denotes total assets.

International Monetary Fund | April 2018 37


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Furthermore, the uncertain and volatile returns on As recommended by the IMF’s recent Financial Sector
these assets may not match the minimum yields prom- Stability Assessment (IMF 2017a), authorities should
ised to policyholders. Increased illiquid assets covered prioritize strengthening policy frameworks and financial
by deposit-like insurance products raise exposure institutions’ liquidity and capital buffers to prevent the
to redemptions at short notice.51 When faced with dismantling of implicit guarantees from inadvertently
net cash outflows, insurers may need to sell off their bringing forward stability risks. Equally important,
illiquid assets, potentially adding to market volatility. authorities must address the wide range of nonregula-
In addition, insurers are in some instances part of tory factors that have driven the proliferation of risky
financial conglomerates encompassing several sectors.52 investment products and excessive demand for credit
While these links give insurers the ability to sell their more broadly; for instance, GDP growth targets.54
products within their own networks, they bring risks The insurance supervisory regime should continue
of spillovers across sectors. to evolve toward a transparent, market- and risk-based
Whether all insurers have sufficient resilience against regime that includes close cooperation with other
these vulnerabilities is uncertain. Current regulations authorities. The authorities have strengthened over-
require relatively low capital charges for infrastructure sight of insurers by curtailing the sale of “universal
investments, joint ventures, and real estate compared life” policies and addressing duration mismatches. The
with, for instance, corporate bonds. Moreover, capital introduction of a stronger prudential standard in the
requirements for investments in funds are fixed and not China Risk-Oriented Solvency System in 2016 was
based on the risks of the underlying assets.53 Despite the another important step. Nevertheless, the increase in
elevated risks, capital levels have remained unchanged insurers’ “other assets” suggests further work is needed.
(Figure 1.21, panel 4). Medium-sized and smaller insurers Additional transparency on the nature, credit quality,
have invested more heavily in alternative assets and have and valuation of these investments, as well as a thorough
weaker capability to manage related risks (Figure 1.21, review of prudential treatment to adequately reflect the
panel 6). In addition, risk assessments are clouded by risks of the underlying assets, are needed. The profile of
complex and opaque company structures and uncertainty liabilities—including duration and surrenders—should
about the exact nature and credit quality of the underly- be closely monitored, and further action to curb unusual
ing investments, including implicit guarantees. liquidity risks should be considered. Finally, the size,
complexity, and interconnectedness of the largest life
insurers require enhanced group supervision, strong
Authorities Should Continue to Reform the Investment
cross-sector coordination, and a framework for recovery
Product Market and Enhance the Insurance
and resolution should one of them fail. The recently
Supervisory Regime
announced merger of the China Insurance Regulatory
Addressing remaining financial risks is key to pro- Commission and the China Banking Regulatory Com-
moting financial stability in China. The proposed asset mission should facilitate closer cooperation with respect
management reforms are a promising blueprint for to insurance and banking supervision.
gradually taming risks within the investment product
sector. Regulators should, however, further limit leverage
for lower-risk products and eventually require that Funding Challenges of Internationally
implicitly guaranteed off-balance-sheet business carry the Active Banks
same capital and liquidity buffers as on-balance-sheet Although banks have strengthened their consolidated
business. Careful sequencing of reforms is also critical. balance sheets over the past decade, dollar balance sheet
liquidity remains a source of vulnerability. International
51About one-fifth of life insurers’ liabilities are deposits and dollar lending continues to increase, dominated by
policyholders’ investments, which are presumed to be more easily non-US banks operating through international branch
withdrawn by policyholders than traditional life insurance products.
52One-third of the consolidated balance sheets of the five
networks. Most rely heavily on short-term wholesale
largest insurance groups consists of banking, asset management, or dollar funding and, at the margin, on volatile foreign
other activities.
53The risk factor applied to infrastructure equity plans is 12 per- 54For example, budget constraints at state-owned enterprises and

cent, to real estate it is 8 to12 percent, and for 10-year AA-rated local governments should be tightened, and the system’s vulnerabil-
corporate bonds it is 15 percent. The risk factor applied to bond ity to slower credit growth should be reduced via improvements to
funds is 6 percent. insolvency and debtor workout regimes.

38 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.22. Advanced Economy Bank Health


Equity market signals are mixed. Bank balance sheet metrics have improved ...
1. an Price to oo atio an Profitabilit 2. Bank Balance Sheet Health
(Percent of sample bank assets)
2.5
Euro area
Other Europe Outer ring: 8%
2.0 North America 2017
13%
Price-to-book ratio

Asia-Pacific
(March 2018)

1.5 38% Buffer and loan-


Inner ring: to-deposit ratios
52% 2007 above target?
1.0 40%
Neither
49%
0.5 One
Both
0.0
0.0 0.5 1.0 1.5 2.0 Targets (percent): Buffer ratio = 4.0; loan-to-deposit ratio = 100
nalysts forecast 2019 return on assets (percent)

... including capital buffers and funding profiles. But work to fortify balance sheets should continue.
3. Buffer and Loan-to-Deposit Ratios 4. Euro Area Nonperforming Loans
(Percent of sample bank assets) (Billions of euros)
Total sia-Pacific Euro area North America Other Europe
Bank buffer ratio (percent): <4 >4 Others Cyprus, Greece, and Portugal Ireland, Italy, and Spain
100 1,200
80
60 1,000
40
20 800
0
Loan-to-deposit ratio (percent): <100 >100 600
100
80 400
60
40 200
20
0 0
2007 17 2007 17 2007 17 2007 17 2007 17 2009 10 11 12 13 14 15 16 17

Sources: Bloomberg Finance L.P.; IMF, Financial Soundness Indicators database; S&P Global Market Intelligence; and IMF staff calculations.
Note: In panel 1, US bank assets have been adjusted for derivatives netting. Panels 2 and 3 are based on a sample of 691 banks headquartered in advanced
economies. In panels 2 to 4, where 2017 data are unavailable, the latest published figures are used. sia-Pacific = ustralia, Japan, orea, New ealand, and
Singapore; North merica = Canada and the nited States; Other Europe = Denmark, Iceland, Norway, Sweden, Swit erland, and the nited ingdom; Bank
buffers = Tier 1 common capital plus loan loss provisions less 60 percent of nonperforming loans as a percentage of tangible assets (ad usted for derivatives netting
at S banks). The 60 percent figure is an assumption used for this analysis and not a regulatory requirement.

exchange swap markets. A sharp tightening of financial some banks’ business models, as discussed in previous
conditions could expose structurally vulnerable liquidity GFSRs (Figure 1.22, panel 1).
positions and trigger forced asset sales or even defaults, But balance sheet metrics suggest that banks’
amplifying and transmitting market turbulence. consolidated financial positions have been fortified
over the past decade.55 In 2007 almost 40 percent
of the sample, by assets, had weak buffers and high
Banks Have Bolstered Their Balance Sheets, but These loan-to-deposit ratios, but this proportion is now less
Efforts Need to Continue, Especially at Weaker Institutions than 10 percent (Figure 1.22, panel 2). This improve-
Markets are providing mixed signals about
55The October 2017 GFSR looked at global banks, but this
the health of the banking sector. Equity market
analysis is based on a sample of almost 700 advanced economy
price-to-book ratios vary across banks, likely reflect- banks. Also see BIS (2018) for a discussion on the enhanced resil-
ing investor concerns about the sustainability of ience of banks.

International Monetary Fund | April 2018 39


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

ment has been achieved by increasing capital and The International Dollar Banking System Faces a
liquidity, raising provisions, and improving funding Structural Liquidity Mismatch
profiles in response to enhanced prudential standards, Demand for US dollar–denominated assets from
stricter supervision, better risk management practices outside the United States continues to grow rapidly.
at banks, and pressure from investors. Demand remains robust, since the US dollar is often
Although bank buffers have increased in aggregate the default currency for commodities, energy, trade
(Figure 1.22, panel 3), there is a tail of weaker banks, credit, and corporate borrowers (especially in emerg-
representing about 20 percent of sample assets, with ing market economies). Banks and other institutional
lower levels of capital and provisions against non- investors in low-interest-rate advanced economies also
performing loans (NPLs).56 These banks are mainly seek dollar assets to enhance yields. Although dollar
concentrated in Europe (inside and outside the euro bonds outstanding have increased rapidly, loans remain
area) and would be more susceptible to shocks such the largest form of credit (Figure 1.23, panel 1). Banks
as a sudden bout of market turmoil or an unexpected are central to this system through both lending and
economic downturn. The combination of a pickup derivatives market activities.
in economic growth, actions taken to reduce these Non-US banks occupy a dominant position in the
NPLs, and policy measures by the European author- provision of US dollar credit (Figure 1.23, panel 2).
ities have contributed to a decline in the stock of Banks intermediate dollars internationally through
NPLs in recent quarters (Figure 1.22, panel 4), but branches in the United States and elsewhere; these
NPL levels remain high at some banks.57 So while branches are relatively free to transfer funds across
the economic recovery will certainly help reduce borders. Non-US banks’ branches in the United States
NPLs, a comprehensive strategy—involving strict have been dollar borrowers from overseas, on net, since
supervision, ambitious NPL reduction targets, mod- 2011, but the gross flows in each direction remain
ernizing insolvency and foreclosure frameworks, and considerable (Figure 1.23, panel 3). By contrast,
further developing distressed debt markets—needs to US subsidiaries of foreign banks gather retail dollar
be fully implemented to address the NPL problem deposits but are limited in their flexibility to transfer
at its root. funds intragroup across borders or legal entities, so
Banks have also improved their funding profiles; they play little role in the international dollar system
nonetheless, more could be done to bolster resil- (Figure 1.23, panel 4) (McCauley, McGuire, and von
ience against liquidity risks in some institutions.58 Peter 2010; McCauley and von Peter 2012).
About one-third of sample banks, by assets, still have This section, therefore, assesses funding and liquidity
loan-to-deposit ratios in excess of 100 percent (Fig- across non-US banks’ international US dollar balance
ure 1.22, panel 3). This does not necessarily mean sheets, defined to include non-US banks’ dollar posi-
that these banks will fail to meet regulatory mea- tions outside the United States plus their US branches,
sures, such as the liquidity coverage ratio.59 But these but excluding their US subsidiaries. The discussion
results do suggest that attention should continue to focuses on country banking systems, and is based on
be paid to liquidity risks, particularly with respect top-down country aggregate balance sheet information
to the dollar-funding profiles of banks operating combined with a bottom-up aggregate of non-US
internationally. banks’ branches in the United States (see Online
Annex 1.2).60
Overall, non-US banks’ international US dollar
balance sheets rely more on short-term or wholesale
56The buffer ratio is Tier 1 common capital and provisions minus dollar funding than do their consolidated balance
60 percent of NPLs as a percentage of tangible assets (adjusted for sheets (Figure 1.24, panel 1). These short-term whole-
derivatives netting at US banks). The 60 percent figure is an
sale instruments—interbank deposits, commercial
assumption used for this analysis rather than a regulatory requirement.
57In March, the European Commission and European Central paper, and certificates of deposit—along with relatively
Bank proposed new measures targeting NPLs. unstable (corporate, nontransactional, and uninsured)
58See also Chapter 3 of the October 2013 GFSR for a discussion

of changes in bank funding structures over time.


59Data for the liquidity coverage ratio are not available over time 60See Online Annex 1.2 at www.imf.org/en/Publications/GFSR

for the full sample of banks. for more details.

40 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.23. US Dollar Credit Aggregates and Bank Intragroup Funding Structures

While dollar bonds outstanding have increased rapidly, loans remain ... dominated by non-US banks operating through international branch
the largest form of credit ... networks.
1. Dollar Credit Extension by Non-US Entities, by Instrument 2. Banks’ US Dollar-Denominated Claims Outside the United States
(Trillions of US dollars)
14 Bonds issued by nonbank sector (trillions of US dollars, left scale) 70 18
Bank loans to nonbanks (trillions of US dollars, left scale) Non-US banks
12 US banks 16
Bank loan share (percent, right scale)
65 14
10
12
8 10
60
6 8
6
4 55 4
2 2
0 50 0
2000 02 04 06 08 10 12 14 16 2000 02 04 06 08 10 12 14 16

Non-US banks’ international branches are key dollar intermediation ... while subsidiaries play a very limited role.
channels ...
3. US Branches of Non-US Banks: Intragroup Borrowing and Lending 4. US Subsidiaries of Non-US Banks: Intragroup Borrowing and Lending
(Billions of US dollars) (Billions of US dollars)
800 Due to overseas—intragroup Due to overseas—intragroup 70
Borrowing Borrowing
600 Due from overseas—intragroup Due from overseas—intragroup 60
Net—intragroup Net—intragroup
400 50
200 40
0 30
–200 20
–400 10
–600 Lending 0
Lending
–800 –10
2005 07 09 11 13 15 17 2005 07 09 11 13 15 17

Sources: Bank for International Settlements; Federal Financial Institutions Examination Council; Federal Reserve call reports; and IMF staff calculations.

deposits are prone to outflows and can generate refi- funding ratio.62 The aggregate stable funding ratio is
nancing risk under stressed conditions. lower for US dollar international balance sheets than
This use of short-term funding makes international for consolidated (aggregate position in all curren-
US dollar balance sheets structurally vulnerable to cies) balance sheets, and the international US dollar
liquidity risks. This vulnerability can be assessed using liquidity ratio is lower than the reported LCRs for
two indicators—a liquidity ratio61 that approximates banks’ consolidated positions (Figure 1.24, panel 2).63
the Basel Liquidity Coverage Ratio (LCR) and a stable US dollar liquidity ratios vary widely between banking

61The liquidity ratio is estimated high-quality liquid assets divided 62The stable funding ratio is stable funding (total deposits plus

by estimated funding outflows over a short stress period (see Online long-term securities and swap funding) divided by loans (see Online
Annex 1.2 for more details). This mimics the Basel framework’s Annex 1.2 for more details). This is intended to be broadly analo-
liquidity coverage ratio but relies on more limited disclosure of gous to the Basel framework’s net stable funding ratio but probably
assets (to measure high-quality liquid assets) and liabilities (to generates higher estimates since it does not apply available stable
measure one-month stress outflow). Analysis of the sensitivity of the funding haircuts to wholesale deposits. For Japan, 70 percent of
liquidity ratio to changes in the underlying assumptions (in Online swap funding is greater than one year in duration and is therefore
Annex 1.2) suggests that the estimates shown here may be somewhat treated as stable, based on Bank of Japan data; for other countries,
overstated; that is, dollar liquidity ratios as measured by the Bank for 50 percent of swap funding is included in stable funding.
International Settlements Liquidity Coverage Ratio would probably 63Global systemically important banks now meet the consoli-

be somewhat lower than shown here. dated, Basel LCR.

International Monetary Fund | April 2018 41


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 1.24. Non-US Banks’ International Dollar Balance Sheets

Non-US banks tend to rely on short-term or wholesale US dollar funding. Their US dollar liquidity is usually weaker than their overall positions.
1. Funding Mix, 2016 2. Liquidity Indicators: Consolidated and US Dollar Balance Sheet
(Percent) (Percent)
Swaps Interbank Other ST market Bonds Deposits Dollar stable funding ratio Consolidated stable funding ratio
100 150
90 Stable
80 deposit 125
70 100
60
Less
50 stable 75
40 deposit
30 Dollar liquidity ratio 50
20 Consolidated liquidity
coverage ratio (reported) 25
10
0 0
Consolidated International dollar 2006 08 10 12 14 16
(all currencies) Total = Non-US + US
operations branches

And S dollar funding ratios vary significantly between banking systems.


3. Liquidity Ratio 4. Stable Funding Ratio
120 30
Change in liquidity ratio 2006–17

100 Germany

Change in stable funding ratio


2006–17 (percentage points)
United United Kingdom 20
80 Germany Kingdom
(percentage points)

60 Canada 10
Canada
40 France
0
20 Japan
Switzerland
0 –10
France
–20
–20
–40 Japan
Switzerland
–80 –30
50 60 70 80 90 100 110 120 130 140 50 75 100 125 150 175
Liquidity ratio 2017 (percent) Stable funding ratio 2017 (percent)

Sources: Bank for International Settlements; bank financial statements; Bank of Japan; Federal Financial Institutions Examination Council; S P Global Market
Intelligence; and IMF staff estimates and analysis.
Note: Measurement of the liquidity ratio and stable funding ratio is explained in the text and in more detail in Online nnex 1.2. International dollar = dollar
claims liabilities in non- S offices and in S branches of non- S banks. ST = short-term.

systems—the French and German aggregate liquidity Only the Japanese banking system’s liquidity ratio
ratios are somewhat lower than their peers’, though declined over the same period, although it currently
they have been rising over the past few years, and stands at about 100 percent (Figure 1.25, panel
the German banking system’s stable funding ratio is 3). This decline reflects a rise in interbank liabili-
below the levels in some other countries (Figure 1.24, ties used to fund an increase in loans and securities
panels 3 and 4). (Figure 1.25, panel 5).
Overall, US dollar liquidity ratios have improved Aggregate US dollar stable funding ratios, however,
since the global financial crisis. This improvement are largely unchanged over 2006–17 (Figure 1.25,
has largely been driven by large increases in High panel 2). Individual banking systems have shown
Quality Liquid Assets (HQLA, reserves at central little progress in strengthening stable funding ratios,
banks and holdings of official sector bonds), proba- and in some the ratio has actually fallen (Figure 1.25,
bly in response to intensifying regulatory scrutiny of panel 4). These declines reflect rapid growth in dollar
short-term liquidity positions (Figure 1.25, panel 1). loans—particularly in the Canadian, French, and Japa-

42 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

nese banking systems—that has exceeded banks’ ability tite to supply dollars may be more procyclical than
or willingness to source deposits (Figure 1.25, panel banks’. Because these new players place the yen they
6). This situation is perhaps due to a reach for yield receive in swap transactions in Japanese government
in banks looking to boost profitability by expanding bills, their ability to provide dollar funding in the
lending across borders through an increased matu- yen-dollar market may also be constrained by the
rity mismatch. Systems whose stable funding ratios scarcity of high-quality yen assets in the market;
have improved (UK and German banking systems) about 85 percent of short-term Japanese government
accomplished this only by shrinking dollar loans (Fig- bills are now held by non-Japanese investors and the
ure 1.25, panel 6). Bank of Japan.

Banks Use Foreign Exchange Swaps to Meet Short-Term Several Forces Are Tightening Dollar Funding Conditions
Currency Funding Mismatches, but This Market May Not
Be a Reliable Backstop in Periods of Stress US dollar funding markets have begun to tighten.
Market participants have pointed to a number of
Non-US banks use foreign exchange swap markets factors behind this, including an expected rise in
to meet short-term currency needs. While some banks Treasury bill issuance, US companies changing their
have lengthened the tenor of their swap positions, investment patterns ahead of repatriating offshore
banks still plan to tap swap markets when liquidity is assets, and continued central bank normalization.
tight. Non-US banks’ dependence on cross-currency This tightening can be illustrated by the widening
swaps varies, but two facts stand out: their use has of the dollar LIBOR-OIS spread (the difference
increased overall over the past decade, and Japanese between the London interbank offered rate and the
banks rely relatively heavily on these instruments overnight indexed swap rate) in recent months (Fig-
(Figure 1.26, panel 1). These developments are con- ure 1.26, panel 6).
cerning, because cross-currency basis swap spreads have Moreover, country-specific liquidity regulations,
moved sharply in the past (Figure 1.26, panel 2) and while helping to strengthen national financial systems,
because swap markets have been more volatile than may inadvertently introduce frictions in international
other short-term funding sources such as repo and funding markets. Some regulators have increased
interbank markets (Figure 1.26, panel 3). This suggests restrictions on or surveillance of cross-border intra-
that swap markets may not be a reliable backstop in group liquidity flows in recent years and are extend-
periods of stress. ing the perimeter of their liquidity requirements to
Furthermore, the yen-dollar market—a crucial foreign banks operating in their country (Buch and
source of bank funding—may have become more Goldberg 2015; Gambacorta, van Rixtel, and Schi-
procyclical because of changes in market structure. As affi 2017; Goldberg and Gupta 2013; Reinhardt and
sovereign yields have fallen below policy guaranteed Riddiough 2014).
return targets, Asian life insurers have sought yield The combination of balance sheet vulnerabilities
in dollar-denominated securities. The need to hedge and market tightening could trigger funding prob-
currency risk has driven a surge in demand for swaps lems in the event of market strains. Market turbu-
(Figure 1.26, panel 4). US banks’ dollar swap supply lence may make it more difficult for banks to manage
has not kept up with this growing demand.64 Non- currency gaps in volatile swap markets, possibly
traditional lenders, such as hedge funds and sovereign rendering some banks unable to roll over short-term
wealth funds, have stepped in to meet this demand dollar funding. Banks could then act as an amplifier
and now account for about 70 percent of the supply of market strains if funding pressures were to compel
of foreign currency derivatives to Japanese financial banks to sell assets in a turbulent market to pay their
institutions (Figure 1.26, panel 5). But their appe- liabilities that are due. Funding pressure could also
induce banks to shrink dollar lending to non-US bor-
64The size of US banks’ short-tenor dollar swap supply is esti-
rowers, thus reducing credit availability. Ultimately,
mated by their holdings of claims on the Japanese official sector, as
non-Japanese investors receiving yen in swap transactions typically there is a risk that banks could default on their dollar
invest the yen in short-term Japanese government bills. obligations.

International Monetary Fund | April 2018 43


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 1.25. Non-US Banks’ International US Dollar Liquidity Ratios


The aggregate liquidity ratio has improved ... ... but the stable funding ratio is little changed.
1. Decomposition of International Dollar Liquidity Ratio, 2006–17 2. Decomposition of International Dollar Stable Funding Ratio, 2006–17
(Percent) (Percent)
100 Level Changes Level Level Changes Level 180
90 160
80 140
70 120
60 100
50
80
40
60
30
20 40
10 20
0 0
LR 2006 HQLA Interbank Interbank Other Deposits LR 2017 SFR 2006 Deposits LT securities Swaps Loans SFR 2017
assets liabilities STL

The drivers of changes in these ratios vary across banking systems.


3. Drivers of Change in Dollar Liquidity Ratios 4. Drivers of Change in Stable Funding Ratio
(Percentage points, 2006–17) (Percentage points, 2006–17)
200 Deposits 400
150 LT securities 300
Swaps
100 200
Loans
50 Change in SFR 100
0
0
–50 HQLA
Net interbank assets –100
–100
–150 Out ows Other STL –200
Deposits
–200 Change in LR –300
–250 –400
GBR CAN DE FRA CHE JPN Total GBR DE CAN JPN FRA CHE Total

Rapid growth of dollar claims is a key challenge ... ... as is rapid loan growth.
5. Dollar Claims and HQLA Growth 6. Dollar Loans and Deposits Growth
(Percent, annualized growth between 2006 and 2017) (Percent, annualized growth between 2006 and 2017)
30 Bubble si e: 15
Total S 10
dollar claims Bubble si e: JPN

Growth in deposits
20 DE Loans as percent
FRA CAN 5
of total claims
Growth in HQL

CAN GBR FRA


0
10 GBR ine
ee l CHE
JPN
45-degr –5

0 –10
e li ne Challenged
45-degre DE –15
CHE Challenged
–10 –20
–10 –5 0 5 10 15 –10 –5 0 5 10 15
Growth in total S dollar claims Growth in loans

Sources: Bank financial statements; Bank for International Settlements; Bank of Japan; Federal Financial Institutions Examination Council; S P Global Market
Intelligence; and IMF staff estimates and analysis.
Note: Dollar claims are loans and securities denominated in dollars. Data labels in the figure use International Organi ation for Standardi ation (ISO) country codes.
HQL = high-quality liquid assets; LR = liquidity ratio; LT = long-term; SFR = stable funding ratio; STL = short-term liabilities.

44 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Figure 1.26. Foreign Exchange Swap and Short-Term Bank Funding Markets

Some non-US banks are reliant on cross-currency funding via swaps. Cross-currency basis swap spreads have widened sharply in the past ...
1. US Dollar Cross-Currency Funding Ratio 2. Three-Month Cross-Currency Basis Swaps
(Net cross-currency derivatives as percent of total assets) (Basis points)
50 Non-US banks CAN CHE DEU FRA GBR JPN 0
40 –20
–40
30
–60
20 Sterling
–80
10 Yen
–100
0 –120
–10 Euro
–140
–20 –160
2000 02 04 06 08 10 12 14 16 2011 12 13 14 15 16 17 18

... and foreign exchange swaps are more volatile than other short-term Demand to hedge foreign currencies by sian financial institutions is
funding sources. increasing ...
3. Sixty-Day Realized Volatility of Three-Month US Dollar Funding Rates 4. Demand for Foreign Exchange Derivatives by Asian Financial Institutions
(Basis points a day) (Billions of US dollars)
10 2,000
FX swap South Korea
9
GC repo Taiwan Province of China
8
OIS Japanese institutional investors 1,500
7 LIBOR Japanese banks
6
5 1,000
4
3
500
2
1
0 0
2011 13 15 17 2010 11 12 13 14 15 16 17
... while the supply is shifting from banks to nontraditional financial US dollar LIBOR-OIS spreads have widened recently.
institutions.
5. Supply of Foreign Exchange Derivatives to Japanese Financial 6. Three-Month LIBOR-OIS Spreads
Institutions (Basis points)
(Billions of US dollars)
1,400 100
Non-Japanese banks US dollar
90
1,200 Non-Japanese, nontraditional lenders Euro
80
Yen
1,000 70
British pound
60
800 50
68% 71%
600 40
61% 75% 70% 65% 30
51% 59%
400 20
32% 29% 10
200 35%
49% 41% 39% 25% 30% 0
0 –10
2010 11 12 13 14 15 16 17 2011 12 13 14 15 16 17 18

Sources: Annual reports; Bank for International Settlements; Bank of Japan; Bloomberg Finance L.P.; Financial Supervisory Commission (Taiwan Province of China);
the Korean Life Insurance Association; and IMF staff estimates.
Note: In panels 4 and 5, the latest data are as of September 2017. In panel 4, data for Korea life insurers and the National Pension Service are estimated assuming
100 percent hedging of their foreign investments. For Taiwan Province of China life insurers, the assumption is a 50 percent hedging of their foreign investments.
Data labels in panel 1 use International Organization for Standardization (ISO) country codes. FX swap = foreign exchange swap (average of euro-dollar and
yen-dollar); GC repo = general collateral repurchase agreement; LIBOR = London interbank offered rate; OIS = overnight indexed swaps.

International Monetary Fund | April 2018 45


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Funding Market Risks Call for Disclosure as Well help investors and analysts better assess international
as Gradual and Coordinated Implementation liquidity and maturity mismatches.
of Regulations • Regulators should develop or maintain
The Basel liquidity framework, centered on the currency-specific liquidity risk frameworks, includ-
LCR, has significantly improved banks’ consolidated ing stress tests, emergency funding strategies, and
balance sheet resilience against short-term funding resolution planning. Coordination and sharing
shocks, and both capital and liquidity regulations have of information among regulators are crucial to
driven considerable improvement in banks’ longer-term reduce any unintended cross-border spillovers from
funding stability. But there is still a need to address jurisdiction-specific liquidity requirements.
risks from foreign currency liquidity mismatches. • Central bank swap lines should be retained to
• Banks should ensure that currency-specific mis- provide foreign exchange liquidity in periods of
matches within individual entities in their banking systemic stress. This should help prevent foreign
groups continue to be managed effectively to reduce currency funding difficulties from spilling over to
the risk of funding strains. other parts of the financial system.
• Consideration should be given to enhancing disclo-
sure of foreign currency funding risks.65 This would Finally, while implementation of the Basel III
package of reforms has helped strengthen the bank-
ing sector, there is still some ground to be covered,
65The Basel Committee’s 2008 Principles for Sound Liquidity and completing the postcrisis reform agenda is vital
Risk Management and Supervision contained guidance on managing (Box 1.5). Ensuring the independence of supervision
liquidity risk, including in different currencies. This guidance
included a principle on the public disclosure of information on will be crucial in this effort, as will be addressing the
liquidity risk. new challenges posed by technology.

46 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Box 1.1. The VIX Tantrum


Global equity markets experienced a bout of about whether other investment strategies, based on
renewed volatility on February 5–6, 2018 (Fig- momentum, risk parity, volatility targeting, or artificial
ure 1.1.1). Equity losses were heavy, with a 7 percent intelligence, may have also exacerbated the initial
cumulative drop in the S&P 500 over the first seven volatility spike. But by the end of the episode, the
trading days of February. The Chicago Board Options VIX, which should reflect investors’ expectations and
Exchange Volatility Index (VIX) of implied equity attitudes toward equity risk, was about in line with
volatility surged, jumping from below 15 at the open forecasts of underlying stock market volatility (see
on February 5 to an intraday peak of 50 on February Online Annex 1.1).1
6, the highest level since August 2015, when China Although technical factors may have exacerbated
devalued its currency. volatility at times, they do not seem to have trig-
Market participants indicated that technical factors gered the initial shock. Mounting fears about higher
in options products and short-volatility strategies inflation in preceding days reportedly soured investor
amplified market moves. For example, the implied sentiment. However, observed moves in market-based
volatility spike forced VIX-related exchange-traded measures of inflation compensation, term premiums,
products to buy large volumes of VIX futures to and implied volatility derived from interest rate
cover short VIX positions, creating a feedback loop swaptions do not appear to be consistent with any
that exacerbated the rise in the VIX. Some of these concurrent, meaningful revision in inflation expec-
exchange-traded products closed with very heavy tations or related risks precisely during the equity
losses. In addition, the evidence to date is inconclu- market swoon.
sive, but debate persists among market participants
1See Online Annex 1.1 at www.imf.org/en/Publications/GFSR

This box was prepared by J. Benson Durham and Will Kerry. for more details.

Figure 1.1.1. US Asset Prices


Equity implied Yield (percent)
volatility Equity index (Feb. 1 = 3)
50 VIX index 3.1
(left scale)
45 S&P 500 index
(right scale)
3.0
40

35
2.9

30
Ten-year yield
(right scale) 2.8
25

20
2.7
15

10 2.6
Feb. 1 Feb. 2 Feb. 5 Feb. 6 Feb. 7 Feb. 8 Feb. 9
2018

Source: Bloomberg Finance L.P.


Note: VIX = Chicago Board Options Exchange Volatility Index.

International Monetary Fund | April 2018 47


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Box 1.1 (continued)


The fall in US equities spilled over to other equity liquidity mismatches as well as rising leverage that may
markets, which fell by about 5–9 percent during amplify market turbulence down the road. The extent
February 1–9. Despite the large price moves, equity of institutional investors’ exposure to short volatility
markets functioned well, with very high trading positions remains unclear. Yet estimates of the price
volumes; liquidity conditions were reportedly reason- of risk, based on volatility projections, are now very
able other than in futures markets; and there was no close to the levels observed before the episode, which
apparent disorderly portfolio unwinding. Declines broadly implies that investors’ willingness to sell vola-
in other risky assets were more modest than the fall tility remains robust today despite the tremors in early
in equities. February. Moreover, valuations remained stretched,
In the aftermath of the VIX tantrum, and after amid a sustained increase in correlations across asset
years of prolonged low interest rates, investors and classes since the episode (as discussed in “Reach for
central bankers are faced with increasing maturity and Yield or Overreach in Risky Assets?” section).

48 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Box 1.2. An Econometric Lens on What Drives Term Premiums


The term premium on a zero-coupon government Figure 1.2.1. Estimated Term Premiums
bond is the extra compensation investors demand
1. United States
for holding government bonds in excess of risk-free
5 Ten-year term premium
short-term interest rates. Specifically, it is the dif-
eighted-average fitted
ference between its yield and the average expected 4 term premium
risk-free short rate over the maturity of the bond. Like
3
equity risk premiums, term premiums are unobserv-
able and must be estimated. Policymakers and inves- 2
tors routinely decompose bond yields into expected
1
rates and term premiums to better understand the
information embedded in the yield curve. 0
To determine what affects term premiums, research-
–1
ers commonly estimate the econometric relationship 1996 98 2000 02 04 06 08 10 12 14 16 18
between these estimates and observable macroeco-
nomic and financial “factors” (Wright 2011; Li and 2. Germany
Wei 2013). The return on a government bond should 4 Ten-year term premium
conceivably correlate with any variable that captures eighted-average fitted term premium
3
some component of either the quantity or the price of
risk around the path of risk-free rates. Relevant factors 2
include forecasts of economic growth and inflation, as
well as measures of uncertainty around those pro- 1
jections; budget deficit forecasts and supply factors
related to “special demand” for safe assets; estimates of 0
the volatility of bond returns; estimated covariance of
bond and stock returns, to assess hedging value; and –1
1996 98 2000 02 04 06 08 10 12 14 16 18
broad measures of financial market stress, including
the VIX (Chicago Board Options Exchange Volatility Sources: Bloomberg Finance L.P.; and IMF staff estimates.
Index) or equity market volatility, to capture so-called Note: Ten-year term premium estimates follow the Adrian, Crump,
flight-to-quality episodes.1 and Moench (2013) model. The weighted-average fair value
estimate is the average of all estimated conditional term premium
Rather than report the result from a single model models. The shaded area denotes the range of fitted values from
and risk false precision, the estimates that follow these models.
average over hundreds of monthly regression models,
based on alternative proxies for the underlying factors,
to enhance robustness. In addition, the approach
emphasizes weighted averages (based on the overall fit through March 2018. For example, for the United
of the models) and ranges rather than a single point States, the models largely capture the so-called conun-
estimate of the fair value of term premiums; that is, drum period during the mid-2000s. Finally, consider-
the required returns statistically commensurate with ing the current environment, as referenced in the main
underlying macroeconomic and financial variables. text, the weighted-average estimate of the fair value
This method not only conveys warranted uncertainty of the 10-year term premium from these hundreds of
around the estimates but also provides a sharper sense monthly regression models was about −10 basis points,
of which factors affect required returns, all else equal. near its sample low, compared with the actual term
Importantly, the models generally track estimated premium estimate of −30 basis points. After closing a
10-year term premiums for Canada, France, Germany, meaningful gap over the past year or so, the reported
Japan, the United Kingdom, and the United States estimated term premium is largely within the range
reasonably well over the sample from February 1996 of all 900 models, and the latest reading is small by
historical comparison (Figure 1.2.1, panel 1).
This box was prepared by J. Benson Durham.
Outside the United States, estimated term premi-
1For a broader discussion of default risk premiums, see the ums on 10-year German bunds are close to historical
April 2018 Fiscal Monitor. lows. The latest fitted value, about −15 basis points,

International Monetary Fund | April 2018 49


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Box 1.2 (continued)


is less than the observed estimate, about 15 basis the estimates imply significant increases in term pre-
points, which strictly speaking suggests that required miums should, say, investors become more uncertain
returns more than compensate for the current about the outlooks for inflation, growth, and the
constellation of risks (Figure 1.2.1, panel 2). Finally, path for monetary policy. Also, naturally this formal
estimated term premiums are similarly close to their time-series approach has shortcomings. Other key vari-
fitted values across Canada, France, Japan, and the ables are hard to capture with formal statistics, includ-
United Kingdom. ing some of the phenomena discussed in the main text
Considering the coefficients of the models, as well and other regulatory restrictions that affect investors’
as the current levels of the underlying factors, the most demand for government paper or debt-management
recent low fitted values of term premiums are owing considerations.
to low survey-based uncertainty about near-term Nonetheless, these statistical results are consistent
GDP growth and inflation, subdued volatility of US with the view that the overall level of longer-dated
Treasury returns, and a persistently lower correlation yields is appropriate given the stance of monetary pol-
between Treasury and risky asset returns. Notably, icy, which, in turn, should remain largely accommoda-
however, the models say nothing about the future tive to support growth and to bring inflation closer to
direction of any of these underlying factors. Indeed, central banks’ targets.

50 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Box 1.3. The Changing Investor Base in the US Leveraged Loan Market
With the US leveraged loan market experiencing Figure 1.3.1. Nonbanks Have Increased Their
impressive growth over the past several years, the buyer Credit Exposure in the US Leveraged Loan Market
base has shifted further toward institutional investors
The US leveraged loan investor base has shifted further
(Figure 1.3.1, panel 1). Similar to the precrisis period,
toward CLOs and asset managers.
structured financial products, such as collateralized
1. US Leveraged Loan Investor Base
loan obligations (CLOs), are an important source of (Percent of new issuance)
demand for low-quality credit. Since 2014, CLOs CLO Loan mutual funds
have purchased more than half of total issuance of Hedge, distressed, and Insurance companies
leveraged loans. US CLOs accounted for 57 percent of high-yield funds Banks and securities firms
leveraged loans outstanding in 2017, with $495 billion Finance companies
in assets under management. CLO issuance (sale of 2%
CLO tranches to outside investors to fund purchases
of loans) reached $118 billion in 2017, above precrisis 2006 48% 10% 14% 6% 20%
levels. Loan mutual funds (including exchange-traded
funds) are another important institutional investor 6% 1%
class. They have grown from roughly $20 billion
in 2007 to $170 billion in assets in 2017, and now 2017 57% 21% 5% 10%
account for more than 20 percent of the institutional
loan market (Figure 1.3.1, panel 2).
Increased holdings of leveraged loans by institu- 0 10 20 30 40 50 60 70 80 90 100
tional investors such as loan mutual funds and CLOs
at the expense of banks may affect market dynamics CLO formation and loan fund growth remain robust.
during times of stress. The migration of loan assets 2. US CLO Issuance and Bank Loan Fund Assets under
to open-end loan mutual funds offering daily liquid- Management
(Billions of US dollars)
ity may exacerbate price moves in the event of large
140 CLO issuance (left scale) 180
investor redemptions under distress (Braithwaite and
120 Bank loan mutual fund and 160
others 2014). Furthermore, market participants cite an ETF AUM (right scale) 140
increase in demand for CLO tranches by asset man- 100
120
agers, insurance companies, and pension funds, which 80 100
now account for 45 percent of AAA CLO market
60 80
share. In the years leading up to the financial crisis,
60
AAA CLO tranches were routinely funded in the 40
40
repurchase agreement (repo) market and through other 20 20
means, essentially using financial leverage to boost
0 0
meager AAA spreads. The unwinding of such lever- 2001 03 05 07 09 11 13 15 17
aged positions reportedly amplified loan price moves
when investors became uncertain about the safety and Sources: EPFR Global; S&P Leveraged Commentary and Data; and
liquidity of higher-rated structured products. At this IMF staff calculations.
Note: AUM = assets under management; CLO = collateralized loan
point, the use of financial leverage to fund CLO posi- obligation; ETF = exchange-traded fund.
tions appears to be limited. Similarly, investors do not
seem to be widely using total return swaps as a vehicle
for gaining leveraged exposure to the loan market
(another common instrument employed in 2006–07).

This box was prepared by Tom Piontek.

International Monetary Fund | April 2018 51


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Box 1.4. Central Bank Digital Currencies


Crypto assets provide challenges and opportunities payment networks or to address the inability to ensure
to central banks. As argued earlier, they are still far the full stability and safety of private cryptocoins.
from fulfilling the three basic functions of money, and From a retail point of view, gradually replacing
their underlying technology still has to develop further notes and coins with a CBDC could yield savings to
before it unequivocally offers the benefits it prom- the state for the costs of maintaining and replacing
ises. Nonetheless, central banks can learn from the notes and coins. It may also reduce transaction costs
properties of cryptocoins and underlying technologies for individuals and small enterprises that have little
to make the use of fiat currencies more attractive. As a or costly access to banking services in some countries
medium of exchange, cryptocoins have certain proper- or regions, and it may facilitate financial inclusion.
ties that central bank money in its current forms (cash Central banks would also be able to tailor the level of
and commercial bank reserves) does not have. Unlike anonymity of a CBDC, ensuring cash-like anonymity
reserve transfers, cryptocoin transactions can be cleared for small-value payments, yet allowing for more tai-
and settled instantaneously without an intermediary, lored regulatory compliance for larger-value payments.
and transacting parties can enjoy anonymity; unlike From a monetary policy perspective, CBDCs could
with cash, transacting parties do not need to be in the help maintain the demand for central bank money in
same place, and the technology offers more flexibility the digital age. Central bank seigniorage would continue
in designing the denomination structure of the crypto- with CBDCs. This, in turn, would allow central banks
coin. These properties make cryptocoins attractive for to continue to finance their operations and distribute
cross-border payments and micro payments in the new profits to government. CBDCs, along with the aboli-
sharing, service-based digital economy. tion of cash, might also allow central banks to overcome
Building on these developments, central banks such the zero lower bound, facilitating truly negative interest
as the Bank of Canada, the People’s Bank of China, rates when necessary, though the benefit of enhanced
the Monetary Authority of Singapore, and the Swedish monetary policy effectiveness may need to be traded
Riksbank have started to explore a new form of central off against a potential cost to financial stability. Making
bank money: central bank digital currency (CBDC). the CBDC a potential competitor to commercial bank
Although approaches vary by institution, and a single deposits could, for instance, lead to volatility in fund
definition is lacking, a CBDC could be defined as flows between commercial banks and the central bank,
a digital form of central bank money that can be potentially resulting in bank runs toward CBDCs and
exchanged, peer to peer, in a decentralized manner. thereby hampering financial stability.
A CBDC would be a token representation of, or an In summary, some central banks have expressed
addition to, cash in physical form (banknotes and interest in exploring the idea of a CBDC.1 Given the
coins) and/or electronic deposits. It could be issued uncertainties described above, a gradual and cautious
by the central bank directly to commercial banks and approach that builds on experience and takes into
other payment services providers or to individuals, account evolving and maturing financial technologies
and would be exchanged at par with the central bank’s seems warranted. Risks to financial stability could
other monetary liabilities. potentially be reduced if the design of the CBDC
Payment system efficiency and stability seem to be is such that it respects the current two-tier banking
important objectives in considering CBDCs. CBDCs system (that is, the separation of commercial banking
could be used to counter the monopoly power that from central banking) and merely creates a digital
strong network externalities might confer on private form of cash.

This box was prepared by Dong He and Ashraf Khan. 1See CPMI (2018).

52 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

Box 1.5. Regulatory Reform—Tying Up the Loose Ends


The postcrisis regulatory reform agenda has been international efforts can now move to full, timely,
successful in enhancing the resilience of the major and consistent implementation, which has already
banks. This resilience has been achieved primarily been delayed and is lagging in important areas such as
through implementation of the Basel III package. cross-border resolution frameworks for banks.
However, the excessive variation in the output of A major challenge for effective implementation
internal models used by banks to compute regulatory is shortcomings in the operational independence of
capital led to concerns that these models were being supervisors from political and market influence. IMF
gamed to reduce regulatory requirements without a Financial Sector Assessment Programs have found that
corresponding reduction in risk exposures. only a handful of the nearly 40 countries that have
To address these concerns, the Basel Commit- been assessed since the global financial crisis are in full
tee on Banking Supervision proposed a package of compliance with the Basel Core Principles on indepen-
enhancements to Basel III in 2014, which was finally dence and accountability. Policymakers must ensure
agreed to in December 2017, bringing closure to a that supervisors have the resources and power to take
critical piece of the regulatory reform agenda. These timely, preemptive, and corrective actions to address
measures limit risk-weighted assets, based on the emerging threats.
internal-ratings-based approach, to a minimum of What else remains on the agenda? The Financial
72.5 percent of the amount calculated using the sim- Stability Board recommendations to transform shadow
pler standardized approach. banking into resilient market-based finance are now
These measures also aim to achieve a better balance being translated into operational guidance to facili-
between simplicity, risk sensitivity, and comparability. tate consistent national implementation. Resolution
In this vein, the agreed-on implementation of the efforts for nonbanks, including central counterparties,
Fundamental Review of the Trading Book has been remain a work in progress, while the reform agenda
postponed to 2022, in response to practical challenges for insurers has not kept pace with planned timelines.
reported by countries, and the standardized approach The issue of tackling incentives for excessive risk taking
to credit risk has been revised to make it more risk has moved away from regulating remuneration to
sensitive (for example, varying risk weights for real reforming governance, addressing misconduct, seeking
estate exposures using loan-to-value ratios). to reinforce individual accountability, and creating a
Agreement on the Basel III enhancements has supportive institutional culture. The difficult decision
come at the cost of: a less conservative risk-weighted on better incorporating sovereign risks into the regula-
assets floor, from the 80 percent proposed initially; tory framework has been shelved for the time being.
further extending the implementation timeline All in all, even though much has been achieved
for these reforms to 2022–27, 20 years since the through the regulatory reforms, there is still some
start of the crisis; an annual cap on any increase in ground to be covered. Given the backdrop of calls for
risk-weighted assets resulting from the measures; and rolling back the reforms, it is vital that the postcrisis
lowering some minimum risk weights in the stan- agenda be completed and implemented to allow super-
dardized approach. visors to focus on emerging challenges, including those
Despite these adjustments, the outcome has brought from rapid developments in financial technology and
certainty to market participants. The focus of the the threats posed by cyberattacks.

International Monetary Fund | April 2018 53


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

References Converse, Nathan, Eduardo Levy-Yeyati, and Tomas Williams.


2018. “How ETFs Amplify the Global Financial Cycle in
Adrian, Tobias, Richard Crump, and Emanuel Moench. 2013.
Emerging Markets.” Working Paper 2018–1, Institute for
“Pricing the Term Structure with Linear Regressions.” Federal
International Economic Policy, Washington, DC.
Reserve Bank of New York Staff Report 740.
Deutsche Bank. 2017. “Who Is Propping Up the Bitcoin
Arslanalp, Serkan, and Takahiro Tsuda. 2012. “Tracking Global
Market?” https://www.db.com/newsroom_news/2017/who-is
Demand for Emerging Market Sovereign Debt.” IMF Working
Paper 12/284, International Monetary Fund, Washington, DC. -propping-up-the-bitcoin-market-en-11752.htm.
Bank of America Merrill Lynch. 2017. “Introducing Cryp- Diebold, Francis, and Kamil Yilmaz. 2012. “Better to Give Than
tocurrencies–What Are They Good For?” Exchanging to Receive: Predictive Directional Measurement of Volatility
Views, October 16. Spillovers.” International Journal of Forecasting 28 (1): 57–66.
Bank for International Settlements (BIS). 2018. “Structural Ehlers, Torsten, Steven Kong, and Feng Zhu. 2018. “Mapping
Changes in Banking after the Crisis.” CGFS Paper 60, Com- Shadow Banking in China: Structure and Dynamics.” BIS
mittee on the Global Financial System, Basel. Working Paper 701, Bank for International Settlements,
Braithwaite, Tom, Tracy Alloway, Michael Mackenzie, and Basel. https://www.bis.org/publ/work701.pdf.
Gina Chon. 2014. “Fed Looks at Exit Fees on Bond Funds.” The Financial Action Task Force (FATF). 2015. “Guidance for a
CNBC website. https://www.cnbc.com/2014/06/16/fed-looks Risk-Based Approach to Virtual Currencies.” http://www.fatf-
-at-exit-fees-on-bond-funds.html. gafi.org/publications/fatfgeneral/documents/guidance-rba-vir-
Bräutigam, Deborah, and Kevin P. Gallagher. 2014. “Bartering tual-currencies.html.
Globalization: China’s Commodity Backed Finance in Africa Gagnon, Joseph, Matthew Raskin, Julie Remache, and Brian
and Latin America.” Global Policy 5 (3): 346–52. Sack. 2010. “Large-Scale Asset Purchases by the Federal
———, and Jyhjong Hwang. 2016. “Eastern Promises: Reserve: Did They Work?” Staff Report 411, Federal Reserve
New Data on Chinese Loans in Africa, 2000 to 2014.” Bank of New York.
Working Paper 4, Chinese Africa Research Initiative, Gambacorta, Leonardo, Adrian van Rixtel, and Stefano Schiaffi.
Washington, DC. 2017. “Changing Business Models in International Bank
Buch, Claudia, and Linda Goldberg. 2015. “International Bank- Funding.” BIS Working Paper 614, Bank for International
ing and Liquidity Risk Transmission: Lessons from across Settlements, Basel.
Countries.” IMF Economic Review 63 (3): 377–410. Goldberg, Linda, and Arun Gupta. 2013. “Ring-Fencing and
Burniske, Chris, and Adam White. 2017. “Bitcoin: Ringing ‘Financial Protectionism’ in International Banking.” Liberty
the Bell for a New Asset Class.” ARK/Coinbase Research Street Economics, Federal Reserve Bank of New York.
White Paper. Hanson, Samuel G., and Jeremy C. Stein. 2015. “Monetary
Cappiello, Lorenzo, Robert Engle, and Kevin Shepphard. 2006. Policy and Long-Term Real Rates.” Journal of Financial Eco-
“Asymmetric Dynamics in the Correlations of Global Equity nomics 115 (3): 429–48.
and Bond Returns.” Journal of Financial Econometrics 4 Husain, Aasim M., Rabah Arezki, Peter Breuer, Vikram Haksar,
(4): 537–72. Thomas Helbling, Paulo Medas, Martin Sommer, and an
Cerutti, Eugenio, Stijn Claessens, and Damien Puy. 2015. IMF Staff Team. 2015. “Global Implications of Lower Oil
“Push Factors and Capital Flows to Emerging Markets: Why Prices.” IMF Staff Discussion Note 15/15, International
Knowing Your Lender Matters More Than Fundamentals.” Monetary Fund, Washington, DC.
IMF Working Paper 15/127, International Monetary Fund, International Monetary Fund (IMF). 2012. “The Liberalization
Washington, DC. and Management of Capital Flows: An Institutional View.”
Chen, Jiaqian, Tommaso Mancini-Griffoli, and Ratna Sahay. IMF Policy Paper, Washington, DC.
2014. “Spillovers from United States Monetary Pol- ———. 2015. “Managing Capital Outflows: Further Opera-
icy on Emerging Markets: Different This Time?” IMF tional Considerations.” IMF Policy Paper, Washington, DC.
Working Paper 14/240, International Monetary Fund, ———. 2016a. “Virtual Currencies and Beyond: Initial
Washington, DC. Considerations.” IMF Staff Discussion Note 16/03,
Cohen, Andrew, and Mark D. Manuszak. 2013. “Ratings Com- Washington, DC.
petition in the CMBS Market.” Journal of Money, Credit and ———. 2016b. “2016 External Sector Report.” IMF Policy
Banking 45 (4): 93–119. Paper, Washington, DC.
Committee on Payments and Market Infrastructures (CPMI). ———. 2016c. “Capital Flows—Review of Experience with the
2017. “Distributed Ledger Technology in Payment, Clearing Institutional View.” IMF Policy Paper, Washington, DC.
and Settlement.” Bank for International Settlements, Basel. ———. 2017a. “People’s Republic of China, Financial Sector Assess-
https://www.bis.org/cpmi/publ/d157.pdf. ment Program.” IMF Country Report 17/358, Washington, DC.
———. 2018. “Central Bank Digital Currencies.” Bank for ———. 2017b. “State-Contingent Debt Instruments for Sover-
International Settlements, Basel. eigns.” IMF Policy Paper 032317, Washington, DC.

54 International Monetary Fund | April 2018


CHAPTER 1 A BUMPY ROAD AHEAD

———. 2018. “Macroeconomic Developments and Prospects Laeven, Luc, and Fabian Valencia. 2012. “Systemic Banking
in Low-Income Developing Countries.” IMF Policy Paper, Crises Database: An Update.” IMF Working Paper 12/163,
Washington, DC. International Monetary Fund, Washington, DC.
International Organization of Securities Commissions (IOSCO). Lane, Philip R., and Gian Maria Milesi-Ferretti. 2007. “The
2013. Principles of Liquidity Risk Management for Collective External Wealth of Nations Mark II: Revised and Extended
Investment Schemes. Madrid. Estimates of Foreign Assets and Liabilities, 1970–2004.”
———. 2017. “IOSCO Board Communication on Concerns Journal of International Economics 73 (2): 223–50.
Related to Initial Coin Offerings (ICOs).” Madrid. https:// Li, Canlin, and Min Wei. 2013. “Term Structure Modeling with
www.iosco.org/news/pdf/IOSCONEWS485.pdf. Supply Factors and the Federal Reserve’s Large-Scale Asset
———. 2018. Recommendations for Liquidity Risk Manage- Purchase Programs.” International Journal of Central Banking
ment for Collective Investment Schemes. Madrid. 9 (1): 3–39.
Kim, Sooji, Matthew C. Plosser, and Joao A. C. Santos. 2017. Nakamoto, Satoshi. 2008. “Bitcoin: A Peer-to-Peer Electronic
“Macroprudential Policy and the Revolving Door of Risk: Cash System.” https://bitcoin.org/en/bitcoin-paper.
Lessons from Leveraged Lending Guidance.” Staff Report Reinhardt, Dennis, and Steven Riddiough. 2014. “The Two Faces
815, Federal Reserve Bank of New York. of Cross-Border Banking Flows: An Investigation into the Links
King, Michael, and Dagfinn Rime. 2010. “The $4 Trillion Ques- between Global Risk, Arms-Length Funding and Internal Capi-
tion: What Explains FX Growth since the 2007 Survey?” BIS tal Markets.” Bank of England Working Paper 498, London.
Quarterly Review (December). Sahay, Ratna, Vivek Arora, Thanos Arvanitis, Hamid Faruqee,
McCauley, Robert, Patrick McGuire, and Goetz von Peter. 2010. Papa N’Diaye, Tommaso Mancini-Griffoli, and an IMF
“The Architecture of Global Banking: From International to Team. 2014. “Emerging Market Volatility: Lessons from the
Multinational.” BIS Quarterly Review (March). Taper Tantrum.” IMF Staff Discussion Note 14/09, Interna-
McCauley, Robert, and Goetz von Peter. 2012. “After the tional Monetary Fund, Washington, DC.
Global Financial Crisis: From International to Multi- Wright, Jonathan H. 2011. “Term Premia and Inflation Uncer-
national Banking?” Journal of Economics and Business tainty: Empirical Evidence from an International Panel
64 (1): 7–23. Dataset.” The American Economic Review 101 (4): 1514–34.

International Monetary Fund | April 2018 55


2
CHAPTER
THE RISKINESS OF CREDIT ALLOCATION: A SOURCE
OF FINANCIAL VULNERABILITY?

Summary

T
he prolonged period of loose financial conditions in recent years has raised concerns that financial
intermediaries and investors in search of yield may have extended too much credit to risky borrowers,
potentially jeopardizing financial stability down the road. These concerns are related to recent evidence
for selected countries that periods of low interest rates and easy financial conditions may lead to a decline
in lending standards and increased risk taking.
Against this backdrop, this chapter takes a comprehensive look at the evolution of the riskiness of corporate
credit allocation—that is, the extent to which riskier firms receive credit relative to less risky ones, its relationship
to the strength of credit expansions, and its relevance to financial stability analysis for a large number of advanced
and emerging market economies since 1991. The chapter focuses on the allocation of credit across firms rather
than the aggregate volume of credit or credit growth.
The chapter finds that the riskiness of credit allocation rises during periods of fast credit expansion, especially
when loose lending standards or easy financial conditions occur concurrently. Globally, the riskiness of credit
allocation increased in the years preceding the global financial crisis and peaked shortly before its onset. It declined
sharply after the crisis and rebounded to its historical average in 2016, the latest available year for globally compa-
rable data. As financial conditions loosened in 2017, the riskiness of credit allocation might have risen further.
An increase in the riskiness of credit allocation signals heightened downside risks to GDP growth and a higher
probability of banking crises and banking sector stress, over and above the previously documented signals provided
by credit growth. Thus, a riskier allocation of corporate credit is an independent source of financial vulnerability.
The results highlight the importance of monitoring the riskiness of credit allocation as an integral part of
macro-financial surveillance. The new measures constructed in this chapter are simple to compute, rely mostly
on firm-level financial statement data that are available in many countries, and can be readily replicated for use
in macro-financial surveillance. For this purpose, policymakers would benefit from collecting these data in a
timely manner.
The chapter shows that various policy and institutional settings may help policymakers mitigate the increase in
the riskiness of credit allocation that takes place during relatively fast credit expansions. A tightening of the macro-
prudential policy stance, greater independence of the supervisory authority from banks, a smaller government
footprint in the corporate sector, and greater minority shareholder protection are all related to a smaller increase in
the riskiness of corporate credit allocation during these episodes.

International Monetary Fund | April 2018 57


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Introduction Figure 2.1. Financial Conditions Have Been Loose in Recent


Years
After years of accommodative monetary policy, (Financial conditions index; various percentiles of the cross-country
financial conditions remain loose in most advanced distribution)
and emerging market economies. Although withdrawal
of monetary policy stimulus has begun in several 3.5
10th percentile Tighter financial conditions
advanced economies and is expected to keep proceed- 3.0
25th percentile
Median
ing at a gradual pace in the United States, and despite
75th percentile
a recent rebound in financial market volatility, financial 2.5
90th percentile
conditions have remained loose, and spreads (includ- 2.0
ing corporate spreads) have remained compressed by
historical standards in both advanced and emerging 1.5
market economies (see Figure 2.1 and Chapter 1).
1.0
Meanwhile, corporate credit-to-GDP ratios remain at
or near their historical highs in both advanced econo- 0.5
mies and emerging markets.1
0.0
This environment has raised concerns among policy-
makers and market analysts that nonfinancial corporate –0.5
credit might have been excessively allocated to risky
–1.0
firms, especially in advanced economies, jeopardiz-

17
1991
92
93
94
95
96
97
98
99
2000
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
ing financial stability down the road. As described in
Chapter 1, persistently easy financial conditions may
Source: IMF staff estimates.
lead to a continued search for yield with too much Note: higher level of the financial conditions index (FCI) means financial
money chasing too few yielding assets, pushing inves- conditions are tighter. The sample comprises 41 advanced and emerging market
economies. For methodology and variables included in the FCI, refer to nnex 3.2
tors beyond their traditional risk tolerance into riskier of the October 2017 Global Financial Stability Report.
investments. Indeed, the share of bond issuance by
nonfinancial corporations with low ratings (high-yield
and BBB-rated bonds) has rebounded from its crisis credit relative to less risky firms. Empirical studies dating
trough in the United States and is at or near an all-time to the mid-1990s for the United States provide evidence
high in the euro area and the United Kingdom (Fig- that the riskiness of corporate credit allocation increases
ure 2.2). At the same time, the October 2017 Global during economic expansions and declines during
Financial Stability Report (GFSR) highlighted that some recessions (for example, Lang and Nakamura 1995;
indicators of nonfinancial corporate vulnerability had Bernanke, Gertler, and Gilchrist 1996).2 More recently,
picked up in several major economies. Although greater Greenwood and Hanson (2013) offer further evidence
risk taking by financial intermediaries could be part of a of such behavior in the United States during the past
healthy economic recovery, it may breed vulnerabilities few decades: the riskiness of corporate credit allocation
that could harm future growth if excessive. rises when credit growth is stronger, the short-term
Country-level studies have documented that the com- Treasury bill yield is lower, the term spread is lower,
position of corporate credit flows changes with financial or high-yield bond returns are higher. Corroborating
conditions and that the riskiness of corporate credit evidence comes from Spain, where riskier firms had
allocation is procyclical. The riskiness of corporate credit nearly the same access to the bank loan market as less
allocation is the extent to which riskier firms receive risky firms in the years preceding the global financial
crisis, but significantly less access during the crisis and
early recovery period (Banco de España 2017). In the
Prepared by a staff team led by Jérôme Vandenbussche and com-
posed of Luis Brandão-Marques, Qianying Chen, Oksana Khadarina, euro area, riskier firms increased their borrowing more
and Peichu Xie under the general guidance of Claudio Raddatz than less risky firms following the rally in euro area sov-
and Dong He. The chapter benefited from contributions by Divya ereign bonds triggered by the European Central Bank’s
Kirti and Jiaqi Li. Claudia Cohen and Breanne Rajkumar provided
editorial assistance.
1See IMF (2016) and the October 2015 GFSR for recent analyses 2A decline in the riskiness of credit allocation during recessions

of the evolution of corporate debt across countries. has sometimes been referred to as a “flight to quality.”

58 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

announcement in 2012 that it stood ready to conduct Figure 2.2. Low-Rated Nonfinancial Corporate Bond Issuance
Outright Monetary Transactions (Acharya and others Has Been High in Some Advanced Economies
(Percent of total nonfinancial corporate bond issuance)
2016). Analyses of granular data from Spain and the
United States also reveal a positive association between Euro area Japan United Kingdom United States
low short-term interest rates and the probability of 80
extending loans to risky borrowers (Jiménez and others
70
2014; Dell’Ariccia, Laeven, and Suarez 2017).
Against this backdrop, this chapter takes a com- 60
prehensive look at the evolution of the riskiness of
corporate credit allocation, its relationship to the size 50
of credit expansions, and its relevance to financial
stability analysis. 40

• No cross-country measures are readily available


30
that capture the riskiness of total credit flows across
firms. To fill this gap, this chapter constructs several 20
measures that map the flow of credit across firms to
the distribution of various firm-level vulnerability 10
indicators for 55 economies since 1991.3 Existing
0
methodologies for assessing firm-level vulnerability

17
1997
98
99
2000
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
or default risk may be more or less suitable to differ-
ent market and data environments. For this reason,
Sources: Dealogic; and IMF staff estimates.
the chapter discusses four options for measuring the Note: “Low-rated” refers to high-yield and BBB-rated bonds; the simple three-year
riskiness of corporate credit allocation—henceforth, moving average is shown. Shaded areas indicate periods during which global real
the “riskiness of credit allocation.” In constructing GDP growth was less than 2.5 percent.

these measures, this chapter provides the most com-


prehensive cross-country analysis of the riskiness of flowing to riskier firms can provide additional
credit allocation to date. information on future macro-financial outcomes,
• Financial stress and growth-at-risk models in the this proposition has remained, at best, a matter of
empirical literature have focused on changes in conjecture in the financial stability literature.5 Fur-
aggregate credit volumes as the key vulnerability thermore, standard indicators of aggregate corporate
measure.4 Although it may seem intuitive that a vulnerability, which are discussed in most financial
measure capturing the extent to which credit is stability reports around the world, do not take
firm-level credit flows into consideration.6
3Some studies have relied on indirect measures such as bond

issuance data by level of credit rating (for example, Kirti 2018).


Others have focused on the share of credit flowing to distressed Following a conceptual discussion of the relationship
(“zombie”) firms. The former measures ignore a significant source of between the riskiness of credit allocation and credit
credit (loans) and are not well suited to most emerging markets and growth, this chapter addresses the following questions:
advanced economies of relatively small size, where domestic bond
market development is low. The latter are partial because they focus • How has the riskiness of credit allocation evolved
only on two categories of firms (distressed and nondistressed). in recent years across a broad spectrum of advanced
4See Schularick and Taylor (2012), Gourinchas and Obstfeld
economies and emerging markets?
(2012), Dell’Ariccia and others (2016), Baron and Xiong (2017),
and Chapters 2 and 3 of the October 2017 GFSR. Gourinchas and
Obstfeld (2012) also emphasize the importance of external imbal- evidence that low credit spreads by themselves forecast poor future
ances, especially in emerging markets. Jordà, Schularick, and Taylor economic performance in the United States.
(2016b) find that in advanced economies financial crises are not 5In the conclusion to their paper, Jiménez and others (2014)

more likely when public debt is high. However, they show that high conjecture that the compositional change in the supply of credit
levels of public debt tend to exacerbate the effects of private sector with respect to risk is more important for financial stability than the
deleveraging after financial crises, as does IMF (2016). Recent papers volume of credit. Kirti (2018) shows that an increase in the share of
also suggest that credit spreads—the extra yield paid by bonds issued high-yield bond issuance during a credit boom predicts lower future
by firms with low credit ratings relative to firms with the best credit growth (see also Box 2.4).
ratings—are particularly low before a financial crisis (Krishnamurthy 6For a conceptual framework of financial stability monitoring, see

and Muir 2017). López-Salido, Stein, and Zakrajšek (2017) provide Adrian, Covitz, and Liang (2015).

International Monetary Fund | April 2018 59


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

• How does the riskiness of credit allocation relate to The remainder of the chapter is organized as follows:
measures of financial conditions over time? Does it The chapter first lays out a stylized conceptual frame-
generally rise during periods of high credit growth? work for macro-financial shocks and the riskiness of
Is it more likely to increase when high credit growth credit allocation. It then describes the construction of
is associated with strong risk appetite? the new measures, their evolution at the global level
• To what extent does the riskiness of credit allocation and in selected economies, their cyclical properties,
help predict financial sector stress and downside and their relationship to various indicators of financial
risks to GDP growth? How far in advance can it conditions. Next, the chapter turns to the empirical
predict these occurrences? Do the predictive prop- analysis of the relationship between the new indicators
erties of the riskiness of credit allocation reinforce and future financial instability as well as downside risks
those of credit growth documented in the exist- to GDP growth. The last core section further explores
ing literature? determinants of the riskiness of credit allocation and
• How is the dynamic of the riskiness of credit alloca- its cyclicality, including macroprudential policies and
tion affected by the regulatory, supervisory, and legal aspects of the supervisory, legal, and institutional
environments? What is the link between the cyclical- frameworks. The last section concludes and presents
ity of the riskiness of credit allocation and common policy implications.
indicators of banking sector soundness?

The Riskiness of Credit Allocation:


The main findings of the chapter follow: Conceptual Framework
• Taking the riskiness of credit allocation into account
helps better predict full-blown banking crises, The theoretical literature has identified various
financial sector stress, and downside risks to growth mechanisms through which the riskiness of credit allo-
at horizons up to three years. Thus, the riskiness cation is related to financial conditions. Variations over
of credit allocation is an indicator of financial time in the riskiness of credit allocation may happen
vulnerability. for separate yet complementary reasons (see Figure 2.3
• A period of high credit growth is more likely to be for a schematic representation of the main channels).
followed by a severe downturn over the medium In the canonical view of the business cycle with
term if it is accompanied by an increase in the financial frictions, the availability of credit to riskier,
riskiness of credit allocation. By contrast, when more vulnerable firms is procyclical, leading to a rise
credit is stagnant or falling, the riskiness of credit in the riskiness of credit allocation during economic
allocation has a negligible effect on downside risks expansions. A driver of fluctuations in the quantity
to GDP growth. and riskiness of credit is the time-varying effect of
• The riskiness of credit allocation at the global level financing frictions attributable to changes in borrowers’
has followed a cyclical pattern over the past 25 years, net worth. Following a positive macroeconomic shock,
has rebounded since its post-global-financial-crisis or when interest rates fall, a firm’s short-term prospects
trough, and was slightly below its historical average and its net worth—the difference between the eco-
at the end of 2016 (the latest data point). nomic value of its assets and its liabilities—increase,
• At the country level, the riskiness of credit allocation reducing the scope of problems related to asymme-
is more strongly associated with credit growth when tries of information between lenders and borrowers,
lending standards are easier, when domestic financial and allowing firms with high leverage easier access
conditions are looser, when credit spreads are lower, to credit markets. Conversely, following a negative
and when global risk appetite is higher. shock, or when interest rates rise, firms with relatively
• A period of credit expansion is less likely to be weak balance sheets find it relatively harder to obtain
associated with a riskier credit allocation when mac- credit (Bernanke and Gertler 1989; Kiyotaki and
roprudential policy has been tightened, when the Moore 1997).7
banking supervisor is more independent, when the
7Various versions of this mechanism are described in the so-called
government has a smaller footprint in the nonfinan-
financial accelerator literature. In this literature, the relaxation of the
cial corporate sector, and when minority shareholder borrowing constraints applies to all firms, not only to riskier ones.
protection is greater. However, borrowing constraints are binding only for the riskiest

60 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

Fluctuations in credit quantity and the riskiness of Figure 2.3. Key Drivers of the Riskiness of Credit Allocation
credit allocation can also be driven by variations over
time in investor beliefs, risk appetite, or perceptions Corporate net worth Risk appetite
of economic uncertainty, which directly affect credit
spreads and expected volatility. In good times, those
most optimistic about asset values can borrow exten- Price of risk
sively to acquire these assets, thereby pushing up asset
prices. Following bad news, uncertainty and volatility
Lending standards
rise, leading lenders to require higher margins, trig-
gering deleveraging and fire sales (Geanakoplos 2010).
To the extent that optimism is positively correlated Credit constraints
with risk, this mechanism can also generate procyclical
variations in the riskiness of credit allocation.8 It is also (In a boom)
possible that in good times investors form unduly opti- Credit demand Credit supply

mistic beliefs about future economic prospects, leading


them to extend credit to more vulnerable firms and Credit volume Riskiness of credit allocation
allowing borrowers to increase their leverage excessively
(Minsky 1977; Kindleberger 1978; Bordalo, Genna-
ioli, and Shleifer 2018). Finally, the risk appetite of
Source: IMF staff.
financial intermediaries with long-term liabilities and Note: The diagram abstracts from the role of bank capital and leverage, feedback
short-term assets is likely to make them search for yield loops, and possible heterogeneity in credit demand.
when monetary conditions are loose, resulting in risk-
ier firms getting easier access to credit (Rajan 2006).
Banks’ capacity and incentives to screen borrow- this role also depends on their own capital levels. Thus,
ers are likely to deteriorate in periods of significant through this channel, an increase in bank capital may
credit expansions, reinforcing the procyclical nature lead to an expansion of credit to firms with poorer
of lending standards and of lending to relatively more fundamentals (Holmstrom and Tirole 1997).9 Yet the
vulnerable firms. The longer a credit expansion lasts, relationship between short-term interest rates, bank
the lower the screening ability of the pool of loan offi- leverage, and bank risk taking is ambiguous in theory,
cers becomes because of a loss of institutional memory because it is the result of the combination of several
about bad credit risks (Berger and Udell 2004). In effects that work in opposite directions (see Dell’Aric-
addition, faced with the need to intermediate larger cia, Laeven, and Marquez 2014; Dell’Ariccia, Laeven,
volumes of credit than usual during a credit boom, and Suarez 2017).10
financial intermediaries do not find it profitable to The balance of these mechanisms will also determine
properly screen borrowers or maintain lending stan- how the riskiness of credit allocation relates to future
dards (Dell’Ariccia and Marquez 2006).
Bank capital can also play an important role in 9Such an increase can, at least in the short term, be the result of

determining the riskiness of credit allocation and its a positive macroeconomic or financial shock, which strengthens the
asset side of banks’ balance sheets. Adrian and Shin (2014) show that
cyclicality through several channels. Banks gather the Holmstrom and Tirole (1997) model translates into a model of
and generate information about the creditworthiness procyclicality.
of potential borrowers and thus can provide credit 10Traditional portfolio allocation models predict that a higher

interest rate on safe assets leads to a reallocation from riskier secu-


to firms that are too risky to tap financial markets
rities toward safe assets (Fishburn and Porter 1976). In contrast,
directly. But banks’ ability to raise funds to perform risk-shifting models of monetary policy predict that an increase in
the interest rate that banks must pay on deposits exacerbates the
firms. Thus, relatively riskier firms benefit disproportionately from agency problem associated with limited liability and increases bank
the cyclical relaxation of these constraints in good times. risk taking, especially for poorly capitalized banks (Matutes and
8Caballero and Simsek (2017) argue that the degree of opti- Vives 2000). Finally, banks may be induced to switch to riskier assets
mism is a critical state variable in the economy, not only because with higher expected yields when monetary easing compresses their
optimism has a direct impact on asset valuations, but also because margins by lowering the yield on their short-term assets relative to
it weakens the dynamic feedback between asset prices, aggregate that on their long-term liabilities, especially if they are poorly capi-
demand, and growth. talized (Dell’Ariccia, Laeven, and Suarez 2017).

International Monetary Fund | April 2018 61


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

macro-financial stability. Directing an increased share of borrower vulnerability: the leverage ratio, the inter-
lending to riskier firms may be fully rational and prof- est coverage ratio (ICR), and the debt-to-profit ratio
itable and reflect the normal functioning of a healthy (or debt overhang). All three ratios have a strong
financial system in some phases of the business and monotonic relationship with credit ratings (Moody’s
credit cycles, or it may reflect improvements in inter- 2006). The ICR is also sometimes used as a proxy
mediaries’ risk-management technologies. Alternatively, for a credit rating (for example, Damodaran 2014).
it may reflect poorer screening of borrowers, excessive A market-based indicator of credit risk, the expected
risk taking (or neglect of risk), and misallocation of default frequency (EDF), is also used.14
financial resources and may therefore have widespread • Starting from information on a firm-level vulnera-
detrimental consequences on the soundness of financial bility indicator, a raw measure is computed as the
intermediaries and the economic performance of the average of this indicator among firms whose debt
economy down the road.11 Furthermore, in the latter (the sum of loans and bonds) increases the most
case, higher riskiness is much more a reflection of com- minus the average computed among firms whose
positional shifts in lending toward riskier firms than a debt increases the least—or declines the most. This
reflection of an aggregate buildup of leverage. raw measure is then transformed into the final
measure by subtracting its country-specific mean to
remove any influence of the country-specific sectoral
The Riskiness of Credit Allocation and Its composition and to ensure both cross-country and
Evolution across Countries cross-measure comparability. An increase in the mea-
A first step in the chapter’s analysis is the construc- sure signals that the vulnerability of firms getting
tion of new measures capturing the riskiness of corpo- relatively more credit has risen relative to the vulner-
rate credit allocation. ability of firms getting relatively less credit. A pos-
• The riskiness of credit allocation cannot be assessed itive (negative) value of the measure indicates that
from aggregate macroeconomic or financial data the riskiness of credit allocation is above (below) its
because they do not reflect the heterogeneity of country sample average. Box 2.1 provides a detailed
firms. The chapter builds on work by Greenwood explanation of how the measure is constructed and
and Hanson (2013) to construct such new measures how to interpret its magnitude.15
based on various indicators of firm vulnerability
for a set of 55 economies (26 advanced econo- The evolution of the riskiness of credit allocation
mies and 29 emerging market economies) over the across countries suggests clear global patterns (Fig-
1991–2016 period using data for listed firms.12 ure 2.4). Its dynamic at the global level is broadly the
• Four firm-level vulnerability indicators are consid- same across the four borrower vulnerability indicators
ered to construct the measures. Methodologies for used. Starting from elevated levels in the late 1990s,
assessing default risk generally rely on accounting it fell in 2000–04 in the aftermath of the Asian
information or on a combination of accounting and Russian crises and of the burst of the dot.com
and market information.13 In the chapter, several equity bubble, reached its historical low in 2004, rose
common accounting-based ratios are used to capture steeply during 2004–08, and hit a peak at the onset

11In addition, excessive borrowing is a source of negative externali-

ties (see Farhi and Werning 2016 and references therein). designed sophisticated rating methodologies that also incorporate
12Data are sourced from the Worldscope database, which provides judgment (for example, Standard and Poor’s 2013).
a rich set of annual financial variables for listed firms. Annex 2.1 14In their study of credit quality in the United States, Greenwood

provides details on the sample and explanations on the data and Hanson (2013) focus the core of their analysis on the EDF
cleaning process. and demonstrate the robustness of their result when using leverage
13Scoring methods are based on a small set of accounting ratios. or the ICR. Acharya and others (2016) measure riskiness using the
These include the Z-score (Altman 1968, 2013) and the O-score (Ohl- ICR. Banco de España (2017) includes leverage and the ICR in its
son 1980). Other methods add market-based variables and use more small set of indicators aimed at capturing financial soundness. See
advanced statistical techniques to compute relative weights (Shumway Annex 2.1 for a precise definition of the firm-level indicators used in
2001; Campbell, Hilscher, and Szilagyi 2011). Other approaches have the chapter.
instead focused on using Merton’s (1974) option pricing formula 15While it is challenging to establish a “neutral” level for the risk-

as the basis for modeling to construct measures of expected default iness of credit allocation, its average over an extended period could
frequency (such as Moody’s KMV model). Credit rating agencies have be a good proxy.

62 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

Figure 2.4. The Riskiness of Credit Allocation Is Cyclical at the Global Level
(Index; global median)

1. Leverage-Based Measure 2. Interest Coverage Ratio–Based Measure


0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
–0.2 –0.2
–0.4 –0.4
–0.6 –0.6
–0.8 –0.8
1995 97 99 2001 03 05 07 09 11 13 15 1995 97 99 2001 03 05 07 09 11 13 15

3. Expected Default Frequency–Based Measure 4. Debt Overhang–Based Measure


0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
–0.2 –0.2
–0.4 –0.4
–0.6 –0.6
–0.8 –0.8
1995 97 99 2001 03 05 07 09 11 13 15 1995 97 99 2001 03 05 07 09 11 13 15

Sources: Worldscope; and IMF staff estimates.


Note: The panels show the simple two-year moving average of the median economy in the unbalanced subsample. Shaded areas indicate periods during which
global real GDP growth was less than 2.5 percent. See Annex 2.1 for the list of economies included in the analysis.

of the global financial crisis. It then declined sharply • The dynamics in the United States (Figure 2.5,
over the next two years and was slightly below its panel 1) and Japan (Figure 2.5, panel 2) are very
precrisis level at the end of 2016, the latest available similar in both cyclicality and magnitude.17 The
data point. most recent period (2014–16), however, suggests
This global dynamic is reflected at the country level, a divergence: the riskiness of credit allocation
with some country-specific nuances. Figure 2.5 shows decreased in the United States to a relatively low
the evolution of the riskiness of credit allocation in level, while in Japan it remained at a level that is
eight major economies using the leverage-based mea- relatively high in historical perspective.18
sure and the EDF-based measure during 1995–2016. • Figure 2.5, panels 3 and 4, show contrasting devel-
The two measures display similar patterns in the first opments in two of the largest euro area countries.
six countries, but sometimes provide contrasting sig- Spain (Figure 2.5, panel 3) had a credit boom
nals in the last two countries, documenting a degree of
complementarity across measures in some countries or 17The pattern in the United States closely resembles that in Green-

wood and Hanson (2013). The decline in Japan in the first half of
periods:16
the 2000s is consistent with the findings of Fukuda and Nakamura
(2011) in their study of zombie lending.
18In the United States, corporate leverage increased across the

board during 2010–16. Since increases are similar across groups of


16While the correlation of the four measures is generally firms, the relative comparisons between groups used in this chapter
high, it is the smallest between the leverage-based and the to track the distribution of credit allocation may not rise over this
EDF-based measures. period (see Box 2.1).

International Monetary Fund | April 2018 63


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 2.5. Selected Economies: Riskiness of Credit Allocation, 1995–2016


(Index)

Leverage-based measure Expected default frequency–based measure


1. United States 2. Japan
1.0 1.0

0.5 0.5

0.0 0.0

–0.5 –0.5

–1.0 –1.0
1995
96
97
98
99
2000
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16

1995
96
97
98
99
2000
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
3. Spain 4. Germany
2.5 1.0

1.5
0.5
0.5
0.0
–0.5
–0.5
–1.5

–2.5 –1.0
1995
96
97
98
99
2000
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16

1995
96
97
98
99
2000
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
5. India 6. China
1.2 1.2
0.8 0.8
0.4 0.4
0.0 0.0
–0.4 –0.4
–0.8 –0.8
–1.2 –1.2
1995
96
97
98
99
2000
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16

1995
96
97
98
99
2000
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16

7. Korea 8. United Kingdom


2.5 1.0

1.5
0.5
0.5
0.0
–0.5
–0.5
–1.5

–2.5 –1.0
1995
96
97
98
99
2000
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16

1995
96
97
98
99
2000
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16

Sources: Worldscope; and IMF staff estimates.


Note: The panels show the simple two-year moving average. Shaded areas indicate periods of growth below the 15th percentile of the growth distribution.
See Box 2.1 for details on the construction of the measures.

64 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

from the late 1990s to the mid-2000s, which was directions in recent years, with the leverage-based
followed by a deep recession during the global measure at a low level at the end of 2016. As in
financial crisis and the euro area sovereign debt Korea, there is a disconnect between the dynam-
crisis. Measures of the riskiness of credit alloca- ics of the two measures for the United Kingdom
tion for this country reflect these developments (Figure 2.5, panel 8) during the 1990s and the
quite well: a steep rise in riskiness took place in 2010s. This disconnect could be due to the effect
the mid- to late 1990s, leading to very high levels of the volatility of firm-level equity prices on the
of riskiness until the crisis of 2008, which trig- EDF-based measure but is a little puzzling given the
gered a sudden and large fall of the indicator. This depth of financial markets in that country. None-
pattern is consistent with the findings of Banco de theless, the two measures point to rising riskiness of
España (2017) mentioned in the introduction to credit allocation before the global financial crisis in
the chapter. By contrast, variations in the riskiness Korea and the United Kingdom.
of credit allocation in Germany (Figure 2.5, panel
4), a country that did not have a credit boom These patterns raise several questions regarding the
during the 20-year period, have remained within cyclicality of the riskiness of credit allocation. Does it
the same narrower range as the United States and systematically rise when GDP growth and credit
Japan, and the measure has moved into positive growth are strong? If so, does this increase depend on
territory in recent years, suggesting a higher level other measures of financial conditions that can signal
of risk taking. expansions in credit supply, such as credit spreads or
• The evolution of the riskiness of credit allocation a broad financial conditions index? To shed light on
in India (Figure 2.5, panel 5) has broadly followed these questions, the econometric analysis that follows
global patterns, and the measure was at a relatively focuses on the relationship between the riskiness of
low level in 2016. The synchronization of China credit allocation, the state of the business cycle, and
(Figure 2.5, panel 6) with global developments is financial conditions using standard cross-country
weaker—peaks and troughs appear to occur with panel regressions (see Annex 2.1 for data sources and
a two- to three-year lag. The finding of a peak in Annex 2.2 for details on methodology).
2009–10 is consistent with recent evidence that the Periods of faster economic and credit expansion are
implementation of a large stimulus plan beginning associated with riskier credit allocations. Regression
at the end of 2008 led to a misallocation of credit analysis indicates that the riskiness of credit alloca-
(Cong and others 2017). Most of the recent liter- tion is procyclical: it increases when GDP growth
ature on credit allocation in China has focused on or changes in the domestic credit-to-GDP ratio are
the link between credit and firm-level productivity stronger. The first finding is consistent with standard
of capital (or profitability) rather than firm-level financial accelerator mechanisms, and the second
credit risk. Using China as an example, Box 2.2 points to mechanisms in which credit supply shocks
illustrates how a set of new profitability-based affect macro-financial outcomes through a risk-taking
indicators, constructed similarly to the new vulnera- channel. The association of credit expansion with
bility indicators discussed in the core of this chapter, greater riskiness of credit allocation is statistically
can provide additional insights into the quality of significant for all four measures. A one standard devia-
credit allocation. tion increase in the change of the credit-to-GDP ratio
• Developments in Korea (Figure 2.5, panel 7) (equivalent to an increase of 5.5 percentage points) is
highlight that only the accounting-based measure associated with an increase in the riskiness of credit
indicated high riskiness before this country’s crisis allocation of 0.12–0.25 standard deviation, depending
in the late 1990s. The EDF-based measure, con- on the exact measure (Figure 2.6). Results are sim-
structed using equity market information, does not ilar for advanced and emerging market economies,
signal any potential problem related to the riskiness although the dispersion of the estimated relationship
of credit allocation at that time, suggesting that is larger in the latter, most likely because of their
equity market investors were too optimistic and that smaller sample size.
accounting-based measures better reflected funda- The association between larger credit expansions
mentals. Also, the two measures point in different and riskier allocations is stronger when financial con-

International Monetary Fund | April 2018 65


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 2.6. The Riskiness of Credit Allocation Rises When a Figure 2.7. The Association between the Size of a Credit
Credit Expansion Is Stronger Expansion and the Riskiness of Credit Allocation Is Greater
(Standard deviations of the riskiness of credit allocation) When Lending Standards and Financial Conditions Are Looser
(Standard deviations of the riskiness of credit allocation)
0.30
0.45
Loose conditions Tight conditions
0.25 0.40

0.35
0.20
0.30

0.25
0.15
0.20

0.10 0.15

0.10
0.05
0.05

0.00
0.00
All economies Advanced economies Emerging market –0.05
economies Lending Financial Corporate VIX
standards conditions index spreads
Sources: Worldscope; and IMF staff estimates.
Note: The figure shows the range of impact of a contemporaneous increase in the Sources: Worldscope; and IMF staff estimates.
change in the credit-to-GDP ratio by one standard deviation on the four (leverage-, Note: The figure shows the range of impact of a contemporaneous increase in the
interest coverage ratio–, debt overhang–, and expected default frequency–based) change in the credit-to-GDP ratio by one standard deviation on the four (leverage-,
measures of the riskiness of credit allocation. Dark-colored (light-colored) bars interest coverage ratio–, debt overhang–, and expected default frequency–based)
indicate that the effects are statistically significant at the 10 percent level or higher measures of the riskiness of credit allocation when lending standards or financial
for four (three) measures out of four. See Annex 2.2 for details on methodology. conditions (financial conditions index, corporate spreads, and I ) are “loose” or
“tight.” The level of a variable is defined as loose (tight) when it is equal to the
25th percentile (75th percentile) of its distribution. Dark-colored (light-colored)
bars indicate that the effects are statistically significant at the 10 percent level or
higher for four (one) measures out of four. See Annex 2.2 for details on
methodology. VIX = Chicago Board Options Exchange Volatility Index.
ditions are loose. A credit expansion accompanied by
loose financial conditions or loose lending standards is
more likely to be driven by shifts in credit supply and
higher risk appetite of financial intermediaries. Regres- of financial conditions on the riskiness of credit allo-
sion analysis provides evidence of such a channel: cation (Box 2.3).19
both variables amplify the cyclicality of the riskiness These trends and properties of the riskiness of
of credit allocation. Specific components of financial credit allocation are generally confirmed when using
conditions appear to matter more than others. In a different sample that covers both listed and unlisted
particular, low corporate credit spreads (or high global firms. The robustness of the results discussed above is
risk appetite, proxied by the Chicago Board Options checked by constructing similar measures using data
Exchange Volatility Index [VIX]) during credit that cover a wider universe of firms (both listed and
expansions result in allocations that are riskier than unlisted), but for a smaller set of countries and over
those observed when the expansions are accompanied fewer years.20 The similarity is very reassuring consid-
by high credit spreads (or low global risk appetite) ering the significant differences in the cross-sectional
(Figure 2.7). Furthermore, a higher stock market coverage of the two databases.
price-to-book ratio is associated with a higher level of
the riskiness of credit allocation. Additional analysis
studying the joint dynamics of the riskiness of credit 19Measurement of these effects assumes that the financial condi-

allocation, financial conditions, credit expansions, and tions index responds contemporaneously to all other variables, while
the riskiness of credit allocation responds with a lag.
economic growth using a panel vector autoregression 20This robustness analysis is based on the Orbis database and

confirms these findings and shows a significant effect covers only 50 economies from 2000. See Annex 2.1 for details.

66 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

Figure 2.8. The Riskiness of Credit Allocation Rises to a High Level before a Financial Crisis, and Falls to a
Low Level Thereafter
(Index; median across all crisis episodes; 11-year window)

1. Leverage-Based Measure 2. Interest Coverage Ratio–Based Measure


1.0 0.04 0.6 3.0
Leverage-based Median leverage 2.5
riskiness of credit (right scale) 0.03 0.4
0.5 Median ICR 2.0
allocation
0.02 0.2 (right scale)
1.5
0.0 0.01 0.0 1.0
0.5
0.00 –0.2 ICR-based
–0.5 riskiness of 0.0
–0.01 –0.4 credit allocation –0.5
–1.0 –0.02 –0.6 –1.0
–5 –4 –3 –2 –1 0 1 2 3 4 5 –5 –4 –3 –2 –1 0 1 2 3 4 5

3. Expected Default Frequency–Based Measure 4. Debt Overhang–Based Measure


1.0 6 1.0 0.20
0.8 5 Median debt overhang
(right scale) 0.15
0.6 4 0.5
EDF-based Median EDF
0.4 riskiness of (right scale) 0.10
credit allocation 3
0.2 0.0
2 0.05
0.0
–0.2 1 –0.5 Debt overhang–based
0.00
–0.4 0 riskiness of credit
allocation
–0.6 –1 –1.0 –0.05
–5 –4 –3 –2 –1 0 1 2 3 4 5 –5 –4 –3 –2 –1 0 1 2 3 4 5

Sources: Laeven and Valencia (forthcoming); Worldscope; and IMF staff estimates.
Note: Systemic banking crises are defined as in Laeven and alencia (forthcoming). The crisis occurs at time 0. Data are demeaned at the country level. The panels
show the median across all crisis countries in a balanced panel. The riskiness measures are constructed as explained in Box 2.1. Median leverage (EDF) refers to the
median of the firm-level leverage (EDF) variable. Median ICR refers to the negative of the median of the firm-level ICR. Median debt overhang refers to the negative
of the median of the EBITD -to-debt ratio. EBITD = earnings before interest payments, taxes, depreciation, and amorti ation; EDF = expected default frequency;
ICR = interest coverage ratio.

The Riskiness of Credit Allocation and Information on the econometric framework is pro-
Macro-Financial Stability vided in Annex 2.3.
Does the riskiness of credit allocation help predict The riskiness of credit allocation has a very clear
episodes of financial instability and downside risks to inverted-U shape around systemic financial crisis epi-
growth? To answer these questions, the econometric sodes. The dynamic of the riskiness of credit allocation
analysis builds on the existing empirical literature on in the period at the start of a crisis is unambiguous: it
the determinants of risks to the financial sector and rises gradually during the five years preceding the crisis,
real activity, and augments the literature’s specifica- reaches a relatively high level, and then falls following
tions with the riskiness of credit allocation. Specif- the onset of the crisis. This is true regardless of the
ically, using cross-country regressions, this section firm-level indicator chosen to construct the riskiness
analyzes whether this new measure constitutes an measure (Figure 2.8). Interestingly, the riskiness of credit
early warning indicator of a systemic financial crisis allocation signals a forthcoming crisis much better than
and of banking sector stress, and whether it is a
predictor of low realizations of future GDP growth.21 age, and to the inclusion of a measure of the high-yield share of bond
issuance. The results, however, are weaker if the post-2008 period is
21The results described in this section are robust to the inclusion of excluded from the sample. The analysis of predictive performance is
standard corporate vulnerability indicators, such as median firm lever- in-sample (all available observations are used to estimate the models).

International Monetary Fund | April 2018 67


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 2.9. Higher Riskiness of Credit Allocation Signals Figure 2.10. Higher Riskiness of Credit Allocation Signals
Greater Risk of a Systemic Banking Crisis Greater Risk of Banking Sector Stress
(Proportional increase in the odds of a banking crisis) (Proportional increase in the odds of banking sector stress)

5.0 2.5

2.0

4.5
1.5

1.0
4.0

0.5

3.5 0.0
Leverage Interest Debt overhang Expected default t t+1 t+2 t+3
coverage ratio frequency
Source: IMF staff estimates.
Source: IMF staff estimates. Note: The figure shows the multiplicative effect of a one standard deviation
Note: The figure shows the multiplicative effect of a one standard deviation increase of the riskiness of credit allocation on the odds of bank equity stress in a
increase in the riskiness of credit allocation on the odds of a systemic banking time window from t to t + h, in which h = 0, 1, 2, 3. Bank equity stress is defined
crisis, as defined in Laeven and alencia (forthcoming). See nnex 2.3 for as annual bank equity excess return over the short-term government bond yield
methodology. that is lower than the country-specific mean by at least one standard deviation.
Each bar shows the minimum and maximum effects across the four (leverage-,
interest coverage ratio–, debt overhang–, and expected default frequency–based)
measures. Dark-colored (light-colored) bars indicate that the effects are
the underlying conventional corporate vulnerability indi- statistically significant at the 10 percent level or higher for four (two) measures out
of four. See nnex 2.3 for methodology.
cators when considered individually (see the blue lines
in Figure 2.8): these more traditional indicators pick up
significantly only when the crisis has already struck.
The riskiness of credit allocation also helps fore-
Regression analysis confirms that a greater riski-
cast banking sector equity stress up to three years in
ness of credit allocation increases the odds of a future
advance. Because the identification and timing of the
systemic banking crisis (Figure 2.9). The effect in the
occurrence of a systemic financial crisis are somewhat
crisis model is measured in addition to the effect of the
subjective and crises are rare events, it is useful to seek
change in credit volumes, which has been emphasized
confirmation of the results obtained in a crisis model
in the literature, and the effect of financial conditions.
by using a banking sector equity stress model for
Thus, for a given size of credit expansion, a greater
which the number of events is larger and the timing is
riskiness of credit allocation implies a higher proba-
completely objective.23 Regression analysis shows that
bility of a financial crisis. A one standard deviation
the riskiness of credit allocation adds predictive power
increase in the riskiness measure increases the odds of a
to such a model for any horizon from zero to three
crisis by a factor of about four.22 The gain in explana-
years (Figure 2.10). A one standard deviation increase
tory power when adding the riskiness variable, between
in the riskiness of credit allocation increases the odds
11 and 25 percentage points, is also reasonably large.
by a factor of 1.3 to 2, making banking sector stress up
22The odds of a crisis refer to the ratio of the probability of

observing a crisis to the probability of not observing it. For instance, 23A banking sector equity stress episode occurs when the

in the sample used in the estimation, the probability of observing annual excess equity return of the banking sector is lower than
a crisis is about 5 percent. Thus, the probability of not observing a the country-specific mean by at least one standard deviation. Such
crisis is about 95 percent, and the odds of a crisis are 5.3 percent episodes are relevant for macro-financial stability because they are typ-
(100*5/95). A fourfold increase from this level would raise the odds ically followed by significant negative credit supply shocks, which, in
to 21 percent. turn, can translate into declines in economic activity and employment.

68 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

Figure 2.11. Higher Riskiness of Credit Allocation Signals Higher Downside Risks to GDP Growth
(Percentage points of GDP growth)

20th percentile 50th percentile


1. Leverage-Based Measure 2. Interest Coverage Ratio–Based Measure
0.2 0.2

0.0 0.0

–0.2 –0.2

–0.4 –0.4

–0.6 –0.6

–0.8 –0.8

–1.0 –1.0
t+1 t+2 t+3 t+1 t+2 t+3

3. Debt Overhang–Based Measure 4. Expected Default Frequency–Based Measure


0.2 0.2

0.0 0.0

–0.2 –0.2

–0.4 –0.4

–0.6 –0.6

–0.8 –0.8

–1.0 –1.0
t+1 t+2 t+3 t+1 t+2 t+3

Source: IMF staff estimates.


Note: The panels show the impact of a one unit increase in the riskiness of credit allocation on the 20th and 50th percentiles of the distribution of future cumulative
GDP growth from year t to year t + h, with h = 1, 2, 3. Solid colored bars indicate that the effects are statistically significant at the 10 percent level or higher. n
empty bar indicates absence of statistical significance. See nnex 2.3 for methodology.

to two times more likely, depending on the measure financial conditions. The effect on the downside
and the horizon. risks to growth is significant when measures of
A riskier credit allocation signals downside risks the riskiness of credit allocation are constructed
to growth in the short to medium term. The analysis based on a sample that covers unlisted as well as
examines the predictive power of the riskiness of listed firms.
credit allocation on two percentiles (20th and 50th) The effects of a riskier credit allocation complement
of cumulative real GDP growth one to three years those of credit expansions on growth-at-risk over the
into the future.24 The riskiness of credit allocation medium term. One might expect that credit booms
is strongly related to the median and left tail of the that are accompanied by a rise in the riskiness of credit
growth distribution over all horizons. In line with allocation pose stronger downside risks to growth than
the findings described previously on banking sector those that are not. The analysis indicates that they do.
stress risk, the new measure provides information on This simultaneous rise in credit volumes and riskiness
downside risks to growth over the short to medium signals elevated risks to growth two and three years
term (Figure 2.11). These effects are in addition to ahead. This finding is consistent with recent evidence
those of changes in the credit-to-GDP ratio and showing that an increase in the high-yield share of
bond issuance in advanced economies during credit
24The approach builds on Adrian, Boyarchenko, and Giannone booms is associated with lower future mean GDP
(2016) and Chapter 3 of the October 2017 GFSR. growth (see Box 2.4 and Kirti 2018).

International Monetary Fund | April 2018 69


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 2.12. The Association of the Riskiness of Credit Allocation with Downside Risks to GDP Growth Depends on the Size
of Credit Expansion
(Percentage points of GDP growth)

1. 20th Percentile 2. 50th Percentile


0.5 0.5

0.0 0.0

–0.5 –0.5

–1.0 –1.0

–1.5 –1.5

–2.0 –2.0

–2.5 –2.5
Low credit growth Mean credit growth High credit growth Low credit growth Mean credit growth High credit growth

Source: IMF staff estimates.


Note: The panels show the range of impact of a one unit increase in the riskiness of credit allocation on the 20th and 50th percentiles of the distribution of future
cumulative GDP growth from year t to year t + 3 across the four (leverage-, interest coverage ratio–, debt overhang–, and expected default frequency–based)
measures. The impact is conditional on high, mean, and low credit growth. High (low) credit growth is defined as one standard deviation above (below) mean credit
growth. Dark-colored (light-colored) bars indicate that the effects are statistically significant at the 10 percent level or higher for four (two) out of four measures. n
empty bar indicates no statistically significant impact of any of the four measures. Further details on the methodology are in nnex 2.3.

Conversely, during credit contractions or relatively selected aspects of the supervisory, legal, and institu-
soft credit expansions, a higher riskiness of credit allo- tional frameworks—come into play. The determinants
cation does not increase downside risks to future GDP of the level and credit cyclicality of the riskiness of
growth. When the change in the credit-to-GDP ratio credit allocation vary somewhat depending on which
is well below its historical average—for example, in the underlying firm-level vulnerability indicator is used.
aftermath of a recession or a creditless recovery—the The analysis that follows focuses on determinants
association between higher riskiness of credit allocation whose robustness is apparent across all four measures.25
and downside risks to GDP growth is weaker, and its The quantitative effects of these structural determinants
sign can reverse if the credit expansion is sufficiently on the cyclicality of the riskiness of credit allocation
weak. Figure 2.12 shows that at a three-year horizon, are summarized in Figure 2.13.
when the change in the credit-to-GDP ratio is low by Bank capital appears to have little significant effect
historical standards, an increase in risk taking has no on the cyclicality of the riskiness of credit alloca-
significant impact on downside risks to growth. This tion. Recent empirical studies on how bank capital
finding indicates that a rise in the riskiness of credit
allocation is harmless in some phases of the cycle. 25The effect on the credit cyclicality of the riskiness of credit allo-

cation is estimated through an interactive term between the policy


or institutional variable and the change in the credit-to-GDP ratio.
The Role of Policy and Structural Factors Variables capturing financial sector depth and financial openness are
not found to have any robust effect across measures and are therefore
Having established that the riskiness of credit alloca- omitted from the discussion. A finding is defined as robust when
tion is a vulnerability indicator, the chapter now turns the regression coefficient is significant for at least two of the four
measures and when the sign is identical across all four measures.
to an analysis of more structural determinants of its
Consistency of the signs of the effects in level and in interaction
level and cyclicality. Three sets of variables—banking is also required. See Annex 2.1 for definitions of the variables and
sector soundness, macroprudential policies, and Annex 2.2 for methodology.

70 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

affects the relationship between financial conditions Figure 2.13. The Association of a Credit Expansion with the
and credit flows to risky firms provide contrasting Riskiness of Credit Allocation Depends on Policy and
results.26 This literature indicates that the link between Institutional Settings
(Standard deviations of the riskiness of credit allocation)
credit conditions, firm riskiness, and bank risk taking
is likely to depend on country circumstances. There- 0.4
Lower setting Higher setting
fore, it may not be surprising that only suggestive
evidence is found that conventional measures of 0.3
banking system capitalization or leverage matter for 0.2
the cyclicality of the riskiness of credit allocation:
greater buffers are generally associated with greater 0.1
cyclicality of the riskiness of credit allocation, but not
0.0
in a robust manner.
However, macroprudential policy tightening reduces –0.1
the cyclicality of the new vulnerability measure. An
–0.2
increase in regulatory capital requirements curtails
domestic banks’ risk-bearing capacity by reducing the –0.3
availability of free capital that banks can use to provide
–0.4
loans. Regression analysis confirms that tightening
of the macroprudential policy stance dampens the –0.5
increase in the riskiness of credit allocation associated Leverage Ceilings and Supervisor Rareness of Minority
ratio penalties on independence state-owned shareholder
with faster credit growth. The result holds for changes constraint credit growth enterprises protection
in minimum leverage ratio and changes in ceilings
and penalties related to credit growth.27 Increases in Sources: Worldscope; and IMF staff estimates.
Note: The figure shows the range of impact of a contemporaneous increase in the
capital conservation buffers also reduce the level of the change in the credit-to-GDP ratio by one standard deviation on the four (leverage-,
riskiness of credit allocation. The capital conservation interest coverage ratio–, debt overhang–, and expected default frequency–based)
measures of the riskiness of credit allocation when policy and institutional settings
buffer and the minimum leverage ratio are policy (leverage ratio constraint, ceiling and penalties on bank credit growth, indepen-
instruments that were introduced as part of the regu- dence of supervisory authority from banks, rareness of state-owned enterprises,
and minority shareholder protection) are at a “lower” setting or a “higher” setting.
latory changes following the global financial crisis. The A lower (higher) setting for macroprudential policy means no policy change (one
findings of the chapter thus suggest that postcrisis reg- tightening action during the year). A lower (higher) setting for the other variables
ulatory tightening has had an impact on the evolution means a level equal to the 25th percentile (75th percentile) of their distribution.
Dark-colored (light-colored) bars indicate that the effects are statistically
of the riskiness of credit allocation and has played a significant at the 10 percent level or higher for four (three) measures out of four.
Empty bars indicate that the effects are statistically insignificant at the 10 percent
level for the four measures. See Annex 2.2 for details on the methodology.

26On the one hand, Jiménez and others (2014) show that in Spain

during 2002–08 a lower overnight interest rate induced relatively


less capitalized banks to grant more loan applications and to commit role in limiting the size of the rebound in the measure
larger loan volumes to risky firms. Acharya and others (2016) find
that, in contrast with relatively highly capitalized banks, relatively
documented in Figure 2.4.28
less capitalized banks in the euro area increased their lending to Greater supervisory independence is associated with
very risky firms following the European Central Bank’s announce- reduced cyclicality of the riskiness of credit allocation.
ment in 2012 that it stood ready to conduct Outright Monetary
A more independent supervisor is likely to be more
Transactions. On the other hand, Dell’Ariccia, Laeven, and Suarez
(2017) find evidence consistent with traditional risk shifting by less empowered to exert its oversight throughout the finan-
capitalized banks, while Schivardi, Sette, and Tabellini (2017) find cial cycle. Accordingly, when the supervisory author-
that undercapitalized banks were less likely to cut credit to zombie ity enjoys greater legal protection from the banking
firms during the recent crisis years in Italy.
27Tightening of minimum capital requirements is found to be industry, the quality of credit allocation is less sensitive
associated with a nonrobust increase in the riskiness of credit allo- to domestic credit growth.
cation, suggesting reverse causality. Loan provisioning requirements
are not found to have any significant effects, either in level or when
interacted with the change in the credit-to-GDP ratio. Jiménez and 28Only one change in minimum leverage requirements was imple-

others (2017) and Uluc and Wieladek (2017) provide evidence that mented before the global financial crisis in the sample. Changes to
tightening capital or provisioning requirements can result in greater ceilings and penalties related to credit growth occur in only four
risk taking by banks. countries in the sample.

International Monetary Fund | April 2018 71


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

The sensitivity of the riskiness of credit allocation to policymakers engage in efforts to collect these granular
domestic credit growth also responds to some aspects data as swiftly as possible.30
of the institutional and legal environments. A smaller Various institutional and policy settings may
government footprint in the nonfinancial corporate help policymakers tame the increase in the riskiness
sector reduces the cyclicality of the new measure. of credit allocation that occurs during large credit
Greater protection of minority shareholders has an expansions. A more independent banking supervisor
effect in the same direction. This latter finding high- can better exert control over lending and origina-
lights the importance of sound corporate governance tion standards during good times, when risks appear
frameworks for financial stability, as documented in contained. Sounder corporate governance standards—
Chapter 3 of the October 2016 GFSR. which may reduce the ability of vulnerable firms’
managers to “gamble for resurrection” or engage in pet
projects—should be promoted. And several macro-
Conclusions and Policy Implications prudential policies, such as the tightening of some
A riskier credit allocation is a source of vulnerability regulatory capital requirements, may reduce the ability
that may threaten financial stability. Policymakers and or willingness of banks to lend to vulnerable firms.31
supervisors should pay close attention to its evolution. Furthermore, policymakers could also address the
Both the volume and allocation of credit matter for potential consequences of an increase in the riskiness
financial stability. A period of high credit growth is of credit allocation during a period of strong credit
more likely to be followed by a severe downturn or growth through increased provisioning requirements
financial sector stress over the medium term if it is and thicker countercyclical capital buffers. The cali-
accompanied by an increase in the riskiness of credit bration of capital buffers should arguably consider the
allocation. Thus, while policymakers should be alert to riskiness of credit allocation.32 Finally, policies aimed
periods of rapid credit expansion or increasing riskiness at directing credit to certain firms or sectors of the
of credit allocation, they should pay special attention economy without due consideration of underlying
when they take place together. Supervisors should credit risk should be discouraged in periods of strong
monitor credit origination standards and the riskiness credit growth.
of credit allocation on a continuous basis, intensify The riskiness of credit allocation at the global level
supervisory scrutiny during episodes of large credit has rebounded since its post-global-financial-crisis
expansion and loose financial conditions, and require trough and was back to its historical average at the end
corrective action if needed.29 of 2016. The relatively mild credit expansion in recent
The riskiness of credit allocation can be mea- years, combined with postcrisis regulatory tighten-
sured using firm-level financial statement data that ing, contributed to a softer rebound in the riskiness
are available in many countries and used for finan- of credit allocation than might be expected given
cial surveillance. The measures of the riskiness of the very loose financial conditions. However, global
credit allocation constructed for this chapter exploit patterns hide relevant country-level heterogeneity, and
cross-sectional information on firm-level net debt the rise of the riskiness of credit allocation in certain
issuance and firm-level vulnerability. Several firm-level
30Financial statement data for domestically listed firms is often
indicators of vulnerability (including leverage, interest
available to policymakers quarterly or semiannually. Therefore,
coverage ratio, debt overhang, and expected default policymakers in many countries should be able to easily construct
frequency) can be used to construct a measure. Each the measures introduced in the chapter for their own country with
is suitable to specific country and data environments. shorter lags and at higher frequency than those reported in the
chapter based on internationally comparable data.
The measures are simple to compute and can be readily 31The evidence provided in the chapter is tentative. Further
used for macro-financial surveillance. Of course, the research needs to be performed to better understand the effect of
usefulness of these indicators for surveillance purposes macroprudential policy on the riskiness of credit allocation.
32Exploring issues related to calibration and timing of macropru-
will depend on the speed with which the underlying
dential policy actions as well as associated GDP growth trade-offs
data become available. It is important, therefore, that are, of course, essential and should be concrete next steps in the
analysis. In particular, delving into the role thicker capital buf-
fers could play in improving macro-financial outcomes following
29In
periods when credit is stagnant or falling, a higher riskiness of a rise in the riskiness of credit allocation to a high level would
credit allocation is less of a vulnerability. seem warranted.

72 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

countries has been more pronounced. As financial on the corporate sector, the riskiness of credit alloca-
conditions loosened further in 2017, the riskiness of tion to households may also be relevant and may not
credit allocation might have continued to rise, which necessarily follow the same patterns. Monitoring this
warrants close monitoring and heightened vigilance. dimension of credit allocation is difficult, especially for
Furthermore, relatively low credit allocation riskiness is a broad set of countries, but evidence from selected
not inconsistent with a large increase in conventional household surveys reported in the October 2017
corporate vulnerability indicators, such as average GFSR suggests that the indebtedness of lower-income,
leverage, as has been observed in some major econo- more vulnerable households has increased in recent
mies in recent years. Finally, while this chapter focuses years in various countries.

International Monetary Fund | April 2018 73


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Box 2.1. Measuring the Riskiness of Credit Allocation


The chapter measures the riskiness of credit allocation Figure 2.1.1. Measuring the Riskiness of Credit
using the approach proposed by Greenwood and Han- Allocation
son (2013). The measure is constructed for four differ-
ent firm-level vulnerability indicators—leverage (total Average vulnerability
debt to total assets), debt overhang (total debt to earn- 10
ings before interest, taxes, depreciation, and amortiza- Riskiness of allocation
9 increases
tion [EBITDA]), interest coverage ratio (ICR; EBITDA
to interest expenses), and expected default frequency. 8
For each firm-level vulnerability indicator, the mea-
7 Top issuers’
sure is built as follows: first, for every year each firm average vulnerability
is assigned the value (from 1 to 10) of its decile in the 6
Riskiness of credit
distribution of the indicator in the country where it is 5 allocation
located. A higher decile represents a larger value of the
Bottom issuers’
underlying vulnerability. Second, firms are similarly 4
average vulnerability
sorted by the changes in net debt to lagged total assets 3
into five equal-size bins. Firms in the bin with the
2
largest increases in debt are called “top issuers,” and Riskiness of
firms in the bin with the largest decreases in debt are 1 allocation increases
the “bottom issuers.” Finally, the measure is computed
as the difference between the average vulnerability Source: IMF staff.
decile for the top issuers and the corresponding aver- Note: Top (bottom) issuers are firms in the top (bottom) quintile of
age for the bottom issuers. the distribution of the ratio of change in net debt to lagged total
assets. firm s vulnerability is measured by its decile in the
Changes in the measure over time help answer the distribution of a vulnerability indicator (for example, expected
following question: what is the evolution of the vul- default frequency).
nerability profile of firms that are accumulating debt
the fastest relative to that of firms that are reducing
debt the fastest? The sign of the measure for some
indicators is adjusted so that it rises when the vulnera- indicator matters. The measure is computed for all
bility of firms whose total debt issuance is the largest is country-year pairs that meet minimum sample size
increasing.1 Figure 2.1.1 summarizes this computation requirements (see Annex 2.1). It reflects the broadest
process graphically. possible measure of debt (notably, it includes both
An example might be useful to provide a better loan and bond financing) and is therefore not affected
understanding of the measure. Suppose that firm lever- by secular shifts in the relative size of bond and loan
age increases by 5 percentage points for all firms and markets. It also reflects the continuous nature of firm
that firm-level issuance increases in equal proportion. vulnerability and default risk.
Mean leverage will increase by 5 percentage points, Using deciles rather than the raw values of a
but the measure of allocation riskiness will not change. vulnerability indicator provides several advantages: it
Conversely, if leverage increases by 5 percentage points minimizes the influence of outliers, avoids the possi-
for top issuers, decreases by 5 percentage points for bility of picking up secular trends, makes the com-
bottom issuers and remains unchanged for all other parison across measures based on different indicators
firms, mean leverage will not change, but the measure straightforward, and provides a way to normalize the
of allocation riskiness will rise. measure across countries. A downside of transforming
Because it abstracts from changes in the mean and into deciles is that information about changes in the
shape of the distribution of the vulnerability indicator, cross-sectional dispersion of the indicator is lost.
only the ranking of a firm in the distribution of that Figure 2.1.2 presents information on the distribu-
tion of the four measures, which helps give a sense
of their magnitude in the sample. Because the focus
This box was prepared by Jérôme Vandenbussche.
1For debt overhang, the deciles of EBITDA to debt (instead of the chapter is on the dynamics of the riskiness of
of debt to EBITDA) are used to avoid classifying firms with credit allocation within countries and not on its vari-
negative earnings as low-vulnerability firms. ation across countries, the measures are demeaned at

74 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

Box 2.1 (continued)


Figure 2.1.2. Histograms of Measures of the Riskiness of Credit Allocation
1. Leverage-Based Measure 2. Interest Coverage Ratio–Based Measure
0.25 0.25

0.20 0.20

0.15 0.15
Fraction

Fraction
0.10 0.10

0.05 0.05

0.00 0.00
–5 –4 –3 –2 –1 0 1 2 3 4 5 –4 –2 –1 0 1 2 4

3. Debt Overhang–Based Measure 4. Expected Default Frequency–Based Measure


0.25 0.25

0.20 0.20

0.15 0.15
Fraction

Fraction
0.10 0.10

0.05 0.05

0.00 0.00
–5 –4 –3 –2 –1 0 1 2 3 4 5 –5 –4 –3 –2 –1 0 1 2 3 4 5

Sources: Worldscope; and IMF staff estimates.


Note: The panel covers 55 economies for the period 1991–2016. Data are demeaned at the country level. The value of
the riskiness of credit allocation is shown on the x-axis.

the country level to construct the histograms shown in allocations.2 Their distributions have the shape of a
the figure and in the analysis. Differences in the aver- bell curve and have a standard deviation of about one.
age value of the indicator across countries may reflect
differences in the industrial composition of their 2The long-term average of the measure in each country could
corporate sectors, so these differences cannot be inter- also be interpreted as representing the neutral allocation of credit
preted to mean that some countries have riskier credit in the absence of cyclical fluctuations.

International Monetary Fund | April 2018 75


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Box 2.2. Credit Allocation in China: Is Credit Flowing to the Most Profitable Firms?
Because nonfinancial corporate debt in China has Figure 2.2. . China rofita ilit o Credit
continued to expand at a brisk pace, understanding Allocation 1997–2016
how credit has been allocated may help assess the (Index)
extent to which vulnerabilities are building.1 Concerns
regarding credit allocation and related medium-term 1.2
macro-financial risks in China have recently focused Profitability
on productivity and profitability rather than on credit 0.8 of credit allocation
risk because of the strong presence of the state in the
corporate and financial sectors and the associated
risk transfers to the sovereign (Song and Xiong 2018; 0.4
Cong and others 2017). This box constructs a new
measure of credit allocation quality that compares the 0.0
profitability of firms whose credit is growing the fastest
to the profitability of firms whose credit is growing
the slowest—henceforth, the profitability of credit –0.4 Riskiness
of credit allocation
allocation—in the same way as described in Box 2.1 to
evaluate these concerns.2,3 –0.8
Although the riskiness of credit allocation has
markedly declined in China since 2012, the profit-
–1.2
ability of credit allocation has experienced only a
1997
98
99
2000
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
mild recovery and remained relatively low at the end
of 2016 (Figure 2.2.1). The profitability of credit
allocation rose significantly in the early 2000s fol- Sources: WIND data; and IMF staff estimates.
Note: Data are demeaned and shown as simple three-year moving
lowing the reforms to state-owned enterprises (SOEs)
averages. The riskiness of credit allocation is based on leverage
in the 1990s, but it started declining just before the data. See nnex 2.1 for definition of the variables.
global financial crisis along with an acceleration in
the credit-to-GDP ratio. This indicator continued
declining during and after the global financial crisis
as a large stimulus plan was put in place in 2009–10. attention for their relatively high share of credit flows
The riskiness of credit allocation also started climb- in recent years (IMF 2017a), their role as policy
ing in that period, but declined significantly after tools for achieving growth targets and development
2011–12, while the profitability of credit allocation goals (Maliszewski and others 2016; Song and Xiong
experienced only a mild recovery and remains low by 2018), and their low relative profitability (Dollar
historical standards. and Wei 2007). From 2007 to 2011, the decline in
The decline in the profitability of credit allocation the profitability of credit allocation took place both
over the past decade has been stronger among SOEs within the universe of SOEs and within the universe
and firms in traditional sectors. SOEs have drawn of private firms. However, while the decline has
continued since then within the group of SOEs, the
This box was prepared by Qianying Chen and Peichu Xie, profitability of credit allocation has improved among
with assistance from Juno Xinze Yao. private firms (Figure 2.2.2, panel 1). Furthermore,
1For concerns about the expansion of credit in China, see
within some sectors considered to be the new engines
IMF (2017a, 2017b). The outstanding stock of corporate debt in
China reached about 163 percent of GDP at the end of 2017.
of Chinese growth (IMF 2017b) the profitability of
2The credit risk dimension of the quality of credit allocation in credit allocation has stabilized or improved over the
China may also have more implications for medium-term growth past 10 years. This is in contrast with more traditional
and the fiscal sector than for short-term financial stability (Song sectors in which a sharp fall has taken place. These
and Xiong 2018). The literature is typically focused on the share sectors used to play a key role as China’s drivers of
of credit to firms with public ownership or with relatively poor
fundamentals in total credit (Lam and others 2017).
economic growth and have the most severe overcapac-
3See Annex 2.1 for details on data sources. ity issues and contain a large share of distressed, or

76 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

Box 2.2 (continued)


Figure . . . ina Profitabilit o re it llocation b ner i an Sector

1. S ver u Non S 1 . Sector 1


n e ange in in e oint
1.0 2
2006 New drivers of growth
2007 Traditional drivers of growth
Other sectors
2008 2005 1

Change in profitability of credit allocation


0.5
State-owned enterprises

2009 0
2004

0.0 –1
2010

2011 2013
–2
2012
–0.5 2014
2016 2015 –3

–1.0 –4
–1.0 –0.5 0.0 0.5 1.0 0 2 4 6 8 10 12 14 16
Non-state-owned enterprises Rank

Sources: WIND data; and IMF staff estimates.


Note: Data are demeaned and shown as a simple three-year moving average. “New drivers” refer to sectors identified as
the new drivers of growth (IMF 2017a). These sectors are information and communication technology (ICT), technology
hardware and equipment, health care equipment and services, pharmaceuticals, biotechnology, and life sciences.
Traditional drivers are automobiles and components, energy, and materials. Other sectors are transportation, retail,
capital goods, media, software and services, consumer goods and services, real estate, and utilities. The simple
three-year moving average of the indicator is used to compute the change between 2006 and 2016. To have at least
40 firms for each industry in 2006, the one-year moving average is used for the ICT sector, and the two-year moving
average is used for the energy, media, and software and services sectors. See nnex 2.1 for variable definitions.
SOEs = state-owned enterprises.

“zombie” firms (Lam and others 2017). The quality are consistent with those of Lam and others (2017)
of credit allocation within other industries has also and call for a sectoral approach to the analysis of
declined since 2006, but to a much lesser extent (Fig- financial vulnerabilities and associated medium-term
ure 2.2.2, panel 2). These findings complement and financial stability risks in China.

International Monetary Fund | April 2018 77


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Box 2.3. The Joint Dynamics of the Riskiness of Credit Allocation, Financial Conditions, Credit
Expansions, and GDP Growth
This box analyzes the joint dynamics of the horizon, as well as to credit expansion and higher GDP
leverage-based measure of riskiness of credit allocation, growth (Figure 2.3.1).1 The response of the riskiness of
financial conditions, credit growth, and the business
cycle. The results of a panel vector autoregression 1The panel VAR includes the financial conditions index

(VAR) using annual data for 41 countries from 1991 to (FCI), GDP growth, change in credit to the private sector to
2016 suggest that loosening financial conditions leads GDP, and the leverage-based measure of the riskiness of credit
to riskier credit allocation over a two- to three-year allocation, as well as country fixed effects, and uses one lag. The
VAR is estimated using Abrigo and Love’s (2016) generalized
method of moments package for Stata (pvar), which in this case
This box was prepared by Luis Brandão-Marques. is warranted because of the relatively short time series available

Figure 2.3.1. The Riskiness of Credit Allocation and Financial Conditions


1. Response of Riskiness to Credit Growth Shock 2. Response of Credit Growth to Riskiness Shock
0.20 1.4
1.2
0.15 1.0
0.8
0.10 0.6
0.4
0.05 0.2
0.0
0.00 –0.2
–0.4
–0.05 –0.6
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10

3. Response of FCI to Riskiness Shock 4. Response of Riskiness to FCI Shock


0.05 0.35
0.00 0.30
–0.05 0.25
–0.10 0.20
–0.15 0.15
–0.20 0.10
–0.25 0.05
–0.30 0.00
–0.35 –0.05
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10

Source: IMF staff.


Note: The figure shows the responses of a given variable to an orthogonal shock to another variable. The responses are
estimated using a panel vector autoregression ( R) of the financial conditions index (FCI), GDP growth, credit growth,
and the leverage-based measure of riskiness of credit allocation, using yearly data (1991–2016) for 41 countries. The
R includes country fixed effects and one lag. The responses of the FCI (panel 3) and the riskiness of credit allocation
(panels 1 and 4) are in standard deviations. The responses of credit growth (panel 2) are in percent of GDP. rise in the
FCI means a loosening of financial conditions. The x-axis in all panels is years after the shock. The dark-green lines are
the average response, and the light-green lines are confidence bands at the 90 percent level.

78 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

Box 2.3 (continued)


credit allocation to shocks to credit and GDP growth also show that credit growth increases significantly after
are like those documented in the chapter and corrob- an increase in the riskiness of credit allocation. This
orate the chapter’s findings about the cyclicality of this response is likely caused by an unobserved loosening
measure. Importantly, an increase in the riskiness of of credit standards that also leads to a more immediate
credit allocation is followed by a tightening (decline) deterioration in credit quality.2 Results from a similar
in financial conditions. The results of the panel VAR panel VAR augmented to include lending standards
(not shown), albeit with a much smaller sample size,
(on average, 19 years per country). The responses of each variable
seem to support this hypothesis.3
in the VAR to orthogonal shocks to the variables are measured
using a simple Cholesky decomposition. The ordering is such 2Looser lending standards imply that lenders increase credit

that the riskiness of credit allocation comes first, followed by to previously credit-constrained firms with low creditworthiness.
credit growth and GDP growth, and the FCI is last. Although Therefore, the perceived increase in the riskiness of the allocation
economic theory does not provide clear guidance for which of credit across firms is followed by higher credit growth.
variables should come first, these results assume that the FCI 3The panel VAR augmented with lending standards also

responds contemporaneously to all other variables. In addition, shows that GDP first rises, but then declines after an increase in
the analysis assumes that the riskiness of credit allocation does the riskiness of credit allocation. This could be consistent with
not respond contemporaneously to credit growth, GDP growth, the higher riskiness of credit allocation feeding the trade-off
or financial conditions; it responds only with a lag. Changing between current economic and financial conditions and future
the ordering of the other variables or including more lags in the financial vulnerabilities (Adrian and Liang 2018). However,
specification does not materially affect results. higher-frequency data are probably needed to tease out all effects.

International Monetary Fund | April 2018 79


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Box 2.4. The High-Yield Share during a Credit Boom and Output Growth
This box focuses on an alternative measure of the Figure 2.4.1. Impulse Response of Cumulative
riskiness of credit allocation, the high-yield (HY) share Real GDP Growth to a High-Yield Share Shock
of bond issuance (see Kirti 2018 for details). The HY Given a Credit Boom
share is based solely on information from bond mar- (Percent)
kets. It provides a simple, complementary approach to
the main metrics used in this chapter. The HY share 1.0
95 percent confidence interval
can be constructed for a sample of 38 countries, with 0.5
Impulse response
coverage for some starting in 1980. Greenwood and
Hanson (2013) construct the HY share for the United 0.0
States and show its relevance for predicting excess
–0.5
bond returns; López-Salido, Stein, and Zakrajšek
(2017) also show that it has macroeconomic relevance –1.0
for the United States.
For the analysis in this box, the HY share is based –1.5
on issuance by nonfinancial corporations and govern-
–2.0
ments. It is procyclical: it rises when recent economic
performance has been good and falls when recent –2.5
economic performance has been bad. A procyclical HY
share suggests extrapolative dynamics, consistent with –3.0
the narratives of Minsky (1977, 1986) and Kindle-
–3.5
berger (1978). The HY share also moves in line with
t+1 t+2 t+3 t+4 t+5
survey measures of bank lending standards.
Focusing on a set of 25 advanced economies, this Sources: Bank for International Settlements; Dealogic; IMF, orld
box examines whether credit booms with a rising HY Economic Outlook database; and IMF staff calculations.
Note: The sample comprises 25 advanced economies. Time
share are followed by lower GDP growth in subse-
periods are shown on the x-axis.
quent years. Credit booms are defined here as episodes
in which the change in the credit-to-GDP ratio
over the previous five years is high relative to recent
international experience. To examine the role of the during a credit boom lowers cumulative GDP growth
HY share, local projection specifications that interact over the next three years by 2 percentage points. The
dummies for credit booms with the average change in HY share also helps separate good from bad credit
the HY share over the course of the boom are used. booms: the probability of growth being low following
Credit booms with a rising HY share are followed a credit boom is very low given a “good” HY indicator
by lower growth over the subsequent three to four and substantially higher given a “bad” HY indicator.
years. Figure 2.4.1 shows the impulse response for an These results suggest that issuance quality (using the
increase in the HY share over the course of the boom. HY share as a proxy) during a credit boom contains
A one standard deviation increase in the HY share information about growth out to three or four years
and that credit booms with a rising HY share merit
This box was prepared by Divya Kirti. special attention from policymakers.

80 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

Annex 2.1. Description and Definition The WIND database, which covers listed Chi-
of Variables nese firms, is used for the analysis in Box 2.2. The
The core of this chapter uses firm-level data from advantage of using WIND over Worldscope is that it
the Worldscope database, which covers the universe provides annual information on ownership. Obser-
of listed firms in many economies around the world. vations are dropped if (1) key financial variables (net
The sample is first cleaned by dropping financial debt issuance, total assets, leverage, EBITDA, interest
sector firms (except those in the real estate sector). expenses, and market capitalization) are missing; (2)
Second, observations are dropped if their values are values are incompatible with the economic content
incompatible with the economic content of the data; of the data (such as negative values of total assets,
for example, when market capitalization, total assets, total liabilities, market capitalization, or interest
total debt, total liability, or interest expenses are expenses); (3) values deviate from accounting identi-
strictly negative or when the operating profit margin ties (for example, the sum of total liability and equity
or the ratio of short-term debt to total debt exceeds book value is greater than total assets by 5 percent or
100 percent. Third, observations are kept only if full more); or (4) the inception date is missing or invalid.
information on net debt issuance; leverage; earnings In addition, only one observation a year is kept for
before interest, taxes, depreciation, and amortization firms listed on several stock markets. Only years
(EBITDA); and market capitalization is available. with at least 50 nonfinancial (including real estate)
Then, only economy-year pairs with no fewer than 40 firms are kept. In the end, about 37,000 firm-year
firms and available information on aggregate credit pairs from 1995 to 2016 are used in the analysis.
to the private sector are kept.33 After all cleaning, Ownership information is available for most firms
about 500,000 nonfinancial firm-year observations only from 2004.
from 55 economies during 1991 to 2016 are left The leverage ratio is defined as the ratio of total
in the sample. debt to total assets. The interest coverage ratio (ICR)
The Orbis database is used for the robustness anal- is defined as the ratio of interest expenses to EBITDA.
ysis. It covers both listed and unlisted firms. The data The debt overhang measure is defined as the ratio of
are cleaned following the guidance in Kalemli-Özcan total debt to EBITDA. The expected default frequency
and others (2015). In addition, only observations (EDF) is computed using the Black-Scholes-Merton
with full information on net debt issuance, leverage, model as in Vassalou and Xing (2004). The ingre-
earnings before interest and taxes (EBIT), loans, and dients in the model are the value of equity, the sum
long-term debt are kept. Then, only economy-year of short-term debt and half of long-term debt and
pairs with at least 50 nonfinancial (including real interest payments, expected returns, the risk-free rate,
estate) firms are kept. In the end, the Orbis sam- and the volatility of the price of equity. The return
ple covers 50 economies. Data availability in several on assets (used in Box 2.2) is defined as the ratio
economies is relatively poor for the 1990s, and panels of EBITDA to total assets. Because availability of
are very unbalanced in most economies before 2005. A EBITDA is poor for some countries in the Orbis
balance is struck by choosing 2000 as the start date for database, EBIT is used instead to compute the debt
the Orbis-based analysis. overhang indicator. Availability of data on interest
expenses is also poor for several countries in Orbis,
so the ICR is not used in this robustness exercise.
33For the construction of the interest coverage ratio–based indi- Computing the EDF requires firm-level equity
cator, a minimum of 40 observations for interest expenses is also market information and therefore cannot be done for
required. An exception is made for one borderline case (Ireland),
for which some years have only 38 or 39 observations. For the unlisted firms.
construction of the debt overhang–based indicator, a minimum The list of economies included in the analysis is
of 40 observations for non-zero debt is also required. For the provided in Annex Table 2.1.1. Other data sources,
construction of the expected default frequency–based indicator,
definitions, and transformations used in this chapter’s
a minimum of 40 observations for expected default frequency is
also required. analysis are summarized in Annex Table 2.1.2.

International Monetary Fund | April 2018 81


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Annex Table 2.1.1. Riskiness of Credit Allocation: Economies Included in the Analysis
Start Year Start Year
Worldscope Orbis Worldscope Orbis
Advanced Economies Emerging Market Economies
Australia* 1991 2000 Argentina* 2000 n.a.
Austria* 1991 2002 Brazil* 1992 2000
Belgium* 1991 2000 Bulgaria* 2006 2000
Canada* 1991 2000 Chile* 1995 2000
Czech Republic* 1997 2000 China* 2000 2000
Denmark* 1991 2000 Croatia 2006 2000
Estonia n.a. 2000 Egypt 2006 n.a.
Finland* 1991 2000 Hungary* n.a. 2000
France* 1991 2000 India* 1993 2002
Germany* 1991 2000 Indonesia* 1992 2002
Greece* 1994 2000 Jordan 2006 n.a.
Hong Kong SAR 1991 2002 Kuwait 2006 2009
Iceland n.a. 2000 Malaysia* 1991 2000
Ireland* 1999 2000 Mexico* 1995 2000
Israel* 2000 2002 Morocco 2009 n.a.
Italy* 1991 2000 Oman 2006 n.a.
Japan* 1991 2000 Pakistan 1995 2000
Korea* 1993 2000 Peru* 2001 2002
Netherlands* 1991 2000 Philippines* 1996 2002
New Zealand* 1999 n.a. Poland* 2000 2000
Norway* 1991 n.a. Romania 2006 2000
Portugal* 1996 2000 Russia* 2005 2000
Singapore 1991 2000 Saudi Arabia 2006 n.a.
Slovak Republic n.a. 2000 Serbia 2010 2000
Spain* 1991 2000 South Africa* 1991 2000
Sweden* 1991 2000 Sri Lanka 2006 2006
Switzerland* 1991 2000 Thailand* 1993 2000
United Kingdom* 1991 2000 Turkey* 1997 n.a.
United States* 1991 2000 Ukraine 2008 2000
Vietnam* 2007 2005
Source: IMF staff.
Note: End year for Worldscope is 2016; for Orbis, 2015. n.a. = data not available.
* The financial conditions index is available (see the October 2017 Global Financial Stability Report for the methodology).

Annex 2.2. The Determinants of the Riskiness gap) and credit cycle (real credit growth), and to esti-
of Credit Allocation mating two-way clustered standard errors at country
This annex provides a general overview of the empir- and year levels.
ical methodologies used in this chapter to analyze the
cyclical determinants of the riskiness of credit allo- Cyclicality of the Riskiness of Credit Allocation
cation and its relationship to institutional and policy The empirical specification is as follows:
variables. A finding is defined as robust across measures
when the regression coefficient is significant for at least Riskinessi,tX = α Xi + γ Xt + β X1 ∆ Crediti,t + β X2 ∆GDPi,t
two of the four measures and when the sign is identical
+ β X3 Appreciationi,t + ε Xi,t , (A2.2.1)
across all four measures. Consistency of the signs of the
effects in level and in interaction is also required. in which X ∈ {leverage, interest coverage ratio, debt
The results are robust to using alternative data overhang, expected default frequency} represents a
sources for credit, including credit data compiled borrower vulnerability or credit risk indicator and, cor-
by the Bank for International Settlements (both for respondingly, Riskinessi,tX measures the riskiness of credit
total credit to the nonfinancial private sector and for allocation based on that indicator for country i at time
credit to the nonfinancial corporate sector), to using t. ∆Credit is the change in the ratio of bank credit to
different ways to capture the business cycle (output the nonfinancial private sector to nominal GDP, and

82 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

Annex Table 2.1.2. Country-Level Data Sources and Transformations


Variable Description Source Transformation
Macroeconomic Variables
Real GDP Gross domestic product, constant prices in national currency IMF, World Economic Outlook
database
Current Account Current account balance, in US dollars IMF, World Economic Outlook
database
Exchange Rate National currency per US dollar IMF, International Financial
Statistics and World Economic
Outlook databases
Macro-Financial Variables
Lending Standards Cumulative net percentage balance (or diffusion index) of the weighted Haver Analytics; IMF staff Z-Score at country level
percentage of surveyed financial institutions reporting tightened estimates
credit standards minus the weighted percentage reporting eased
credit standards. An increase in this index implies a net tightening.
Financial Conditions For methodology and variables included in the FCI, refer to Annex 3.2 IMF staff estimates Z-Score at country level
Index (FCI) of the October 2017 Global Financial Stability Report. Positive
values of the FCI indicate tighter-than-average financial conditions.
Corporate Spreads Corporate yield of the country minus sovereign yield of the benchmark Bloomberg Finance L.P.; Z-Score at country level
country; JPMorgan Corporate Emerging Markets Bond Index Broad Thomson Reuters Datastream
is used for emerging market economies where available.
Private Credit-to-GDP The credit provided to the private sector by domestic money banks as IMF, International Financial Demeaned at country level
Ratio a share of GDP Statistics and World Economic
Outlook databases
Stock Price-to-Book Ratio Yearly averages of price-to-book ratios Thomson Reuters Datastream Z-Score at country level
VIX Chicago Board Options Exchange Volatility Index Bloomberg Finance L.P. Logarithm; demeaned
across time
Financial Stress Variables
Systemic Banking Crisis Dummy for systemic banking crisis start Laeven and Valencia
(forthcoming)
Banking Sector Equity Dummy variable for banking sector stress is equal to 1 when the Thomson Reuters Datastream;
Stress annual excess equity return of the banking sector (relative to a Bloomberg Finance L.P.; IMF,
zero-coupon government bond yield with short maturity) is below International Financial Statistics
the country-specific mean by at least one standard deviation in any database
year within the time frame. Equity return is defined as the change
in the logarithm of the equity price index of the banking sector (or
financial sector if a banking sector price index is not available).
Money market rate or interbank lending rate is substituted for
government bond yield if not available.
Banking Sector Characteristics
Buffers from Banking The buffer of a country’s banking system (capitalization and returns) World Bank, Global Financial Demeaned at country level
Default relative to the volatility of returns. It is defined as (ROA+(Equity/ Development Database (2017)
Assets))/sd(ROA), where ROA is return on assets. sd(ROA)
is the standard deviation of ROA. ROA, Equity, and Assets are
country-level aggregate figures.
Policy and Institutional Variables
Independence of The degree to which the supervisory authority is protected by the legal Barth, Caprio, and Levine (2013) Country average across
Supervisory Authority system from the banking industry. Higher values indicate greater years
from Banks independence.
Rareness of State-Owned The negative of the scope of state-owned enterprises (SOEs), which Organisation for Economic Country average across
Enterprises (SOEs) is the pervasiveness of state ownership across 30 business sectors Co-operation and Development, years
measured as the share of sectors in which the state controls at Economy-wide Product Market
least one firm. Regulation Database
Minority Shareholder Minority Shareholder Rights Protection Index Guillén and Capron (2016) Country average across
Protection Index years
Net Tightening Capital Net tightening of macroprudential instrument regarding capital Alam and others (forthcoming) Demeaned at country level
Conservation Buffers conservation buffers
Net Tightening Ceilings Net tightening of macroprudential instrument regarding ceilings and Alam and others (forthcoming) Demeaned at country level
and Penalties on Bank penalties on overall bank credit growth
Credit Growth
Net Tightening Minimum Net tightening of macroprudential instrument regarding leverage ratio Alam and others (forthcoming) Demeaned at country level
Leverage Ratio
Source: IMF staff.

International Monetary Fund | April 2018 83


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Annex Table 2.2.1. Cyclicality of the Riskiness of Credit Allocation


(1) (2) (3) (4) (5)
Robustness
Variables Dependent Variable: Riskiness of Credit Allocation Based on Leverage Sign Significance
Change in Credit-to-GDP Ratio 0.05*** 0.04*** 0.06*** 4 4
(0.01) (0.01) (0.01)
Real GDP Growth 0.08*** 0.12*** 0.05* 4 4
(0.02) (0.02) (0.03)
Appreciation against the US Dollar –0.04*** –0.02* –0.05***
(0.01) (0.01) (0.01)
Country Group All AE EM
Country Cluster Yes Yes Yes
Country Fixed Effect Yes Yes Yes
Year Fixed Effect Yes Yes Yes
Observations 986 563 423
Number of Countries 55 26 29
R2 0.31 0.34 0.37
Source: IMF staff estimates.
Note: Dependent variable = riskiness of credit allocation based on leverage for columns (1)–(3). For robustness, the cyclicality of the other three measures of
the riskiness of credit quality (based on interest coverage ratio, debt overhang, and expected default frequency) is investigated in the full sample. The number
of measures (out of four) that have the same sign and that are significant at the 10 percent level or higher is reported in columns (4) and (5). See Annex
Table 2.1.1 for countries and years in the sample. See Annex Table 2.1.2 for definitions and source of all variables. In all specifications, standard errors are
clustered at the country level. Standard errors are in parentheses. AE = advanced economies; EM = emerging market economies.
***p < 0.01; **p < 0.05; *p < 0.1.

∆GDP is real GDP growth. Domestic currency appre- are clustered at the country level, as before. The term
ciation against the US dollar is included to control for FCi,t represents a financial conditions index (FCI),
a potential mechanical valuation effect on the riskiness financial variables representing specific components of
of credit allocation from debt denominated in foreign the broad index, or a measure of lending standards.
currency. Both country (α Xi ) and year (γ Xt ) fixed effects Both ∆Credit and FC are demeaned at the country
X
are included. The standard errors are clustered at the level. The estimated coefficient δˆ measures the level
country level for all specifications. effect of FCi,t on the riskiness of credit allocation
Results are provided in Annex Table 2.2.1. The when demeaned ∆Credit is 0. The estimated coeffi-
X
results also hold if the fiscal position is controlled for cient θˆ captures the marginal effect on the credit
through the general government structural balance. cyclicality of the riskiness of credit allocation caused
In addition, the results are robust (and coefficients by a change in the FCI, financial variables, or lend-
quantitatively very similar) when instrumenting GDP ing standards.
growth and the change in the credit-to-GDP ratio The results for lending standards, FCI, corporate
by their lagged values to account for their potential spreads, stock market price-to-book ratio, and log
endogeneity. VIX (Chicago Board Options Exchange Volatility
Index) are shown in Annex Table 2.2.2. Columns
Financial Conditions, Lending Standards, and the (1)–(5) show the results obtained when each financial
Riskiness of Credit Allocation variable enters the regression individually. The impact
The following equation is estimated: of other financial variables, such as stock market vola-
tility, a credit boom dummy (as defined in Dell’Aric-
Riskinessi,tX = α Xi + γ Xt + β X Controlsi,t + δ X FCi,t cia and others 2016), length of credit boom, a
dummy to capture different phases of a credit boom,
+ θ X × FCi,t × ∆ Crediti,t + ε X , (A2.2.2)
i,t cross-border bank-flows-to-GDP ratio, and housing
in which Controlsi,t is a vector of control variables price inflation is also investigated. However, none of
including change in the credit-to-GDP ratio, real these variables has a robust significant impact on the
GDP growth, and domestic currency appreciation as riskiness of credit allocation, so they are not included
discussed in the previous section. The standard errors in the table.

84 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

Annex Table 2.2.2. Impact of Financial Conditions and Lending Standards on the Riskiness of Credit Allocation
(1) (2) (3) (4) (5) (6) (7)
Robustness
Variables Dependent Variable: Riskiness of Credit Allocation Based on Leverage Sign Significance
Change in Credit-to-GDP Ratio 0.05*** 0.05*** 0.05*** 0.05*** 0.05***
(0.02) (0.01) (0.01) (0.01) (0.01)
Bank Lending Standards –0.10 4 1
(0.07)
Change in Credit-to-GDP Ratio –0.03* 4 4
Bank Lending Standards (0.02)
Financial Conditions Index (FCI) –0.05 3 0
(0.07)
Change in Credit-to-GDP Ratio –0.01** 4 4
FCI (0.00)
Corporate Credit Spreads –0.07 4 0
(0.06)
Change in Credit-to-GDP Ratio –0.02** 4 2
Corporate Credit Spreads (0.01)
Stock Price-to-Book Ratio 0.20*** 4 4
(0.06)
Change in Credit-to-GDP Ratio 0.01 4 0
Stock Price-to-Book Ratio (0.01)
Change in Credit-to-GDP Ratio –0.04** 4 3
Log (VIX) (0.02)
Controls Yes Yes Yes Yes Yes
Country Cluster Yes Yes Yes Yes Yes
Country Fixed Effect Yes Yes Yes Yes Yes
Year Fixed Effect Yes Yes Yes Yes Yes
Observations 266 824 663 949 986
Number of Countries 21 41 37 51 55
R2 0.39 0.34 0.33 0.33 0.31
Source: IMF staff estimates.
Note: Real GDP growth and domestic currency appreciation against the US dollar are controlled for in all regressions. Increase in bank lending standards
means stricter bank lending standards. Increase in financial conditions index means tighter financial conditions. Dependent variable = riskiness of credit allo-
cation based on leverage for columns (1)–(5). For robustness, the cyclicality of the other three measures of the riskiness of credit allocation (based on interest
coverage ratio, debt overhang, and expected default frequency) is investigated in the full sample. The number of measures (out of four) that have the same
sign and that are significant at the 10 percent level or higher is reported in columns (6) and (7). See Annex Table 2.1.1 for countries and years in the sample.
See Annex Table 2.1.2 for definitions and source of all variables. In all specifications, standard errors are clustered at the country level. Standard errors are in
parentheses. VIX = Chicago Board Options Exchange Volatility Index.
***p < 0.01; **p < 0.05; *p < 0.1.

Policy and Institutional Settings and the Riskiness of clustered at the country level. The term Zi,t represents
Credit Allocation different measures of financial market depth, banking
system soundness, macroprudential policy, the legal
This analysis investigates the role played by financial
and institutional framework, and banking supervi-
market depth, banking system soundness, macropru-
sion quality. All financial development and financial
dential policy, selected aspects of the legal and insti-
soundness variables enter the regression in the form
tutional framework, and banking supervision quality
of a one-year lag to eliminate potential endogene-
on the riskiness of credit allocation. The following
ity concerns. The estimated coefficient ρˆ X measures
equation is estimated:
the level effect of Zi,t on the riskiness of credit allo-
Riskinessi,tX = α Xi + γ Xt + β X Controlsi,t + ρ X Zi,t cation when demeaned ∆Credit is 0. The estimated
coefficient φˆ X captures the marginal effect on the
+ φ X × Zi,t × ∆ Crediti,t + ε X , (A2.2.3)
i,t credit cyclicality of the riskiness of credit allocation
in which Controlsi,t is the same set of control variables with respect to a change in each of the Zi,t variables.
as in the previous section. The standard errors are Because of lack of sufficient time series variation, data

International Monetary Fund | April 2018 85


Annex Table 2.2.3. Impact of Policy and Institutional Settings on the Riskiness of Credit Allocation

86
Dependent Variable: Riskiness of Credit Allocation Based on Leverage
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Legal and Robustness
Financial Macroprudential Supervision Institution
Variables Soundness Policy Quality Aspects Sign Significance
Change in Credit-to-GDP Ratio 0.05*** 0.05*** 0.05*** 0.05*** 0.05*** 0.12*** –0.00 0.15***
(0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.02) (0.03)
Lag Buffers from Banking Default 0.01 3 0
(0.01)
Change in Credit-to-GDP Ratio Lag Buffers from Banking Default 0.005** 3 2
(0.002)
Net Tightening of Capital Conservation Buffers –0.45** 4 3

International Monetary Fund | April 2018


(0.21)
Change in Credit-to-GDP Ratio Net Tightening of Capital Conservation Buffers –0.09*** 4 1
(0.03)
Net Tightening of Minimum Leverage Ratio –0.29 –0.30 4 0
(0.20) (0.20)
Change in Credit-to-GDP Ratio Net Tightening of Minimum Leverage Ratio –0.09* –0.09* 4 2
GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

(0.05) (0.05)
Net Tightening on Ceilings and Penalties on Bank Credit Growth –0.57 –0.57 4 2
(0.54) (0.54)
Change in Credit-to-GDP Ratio Net Tightening on Ceilings and Penalties on Bank Credit Growth –0.07** –0.07** 4 4
(0.03) (0.03)
Change in Credit-to-GDP Ratio Independence of Supervisory Authority from Bank –0.09*** 4 3
(0.02)
Change in Credit-to-GDP Ratio Rareness of State-Owned Enterprises –0.01* 4 4
(0.01)
Change in Credit-to-GDP Ratio Minority Shareholder Protection Index –0.02*** 4 3
(0.01)
Controls Yes Yes Yes Yes Yes Yes Yes Yes
Country Cluster Yes Yes Yes Yes Yes Yes Yes Yes
Country Fixed Effect Yes Yes Yes Yes Yes Yes Yes Yes
Year Fixed Effect Yes Yes Yes Yes Yes Yes Yes Yes
Observations 861 976 976 976 976 929 739 898
Number of Countries 55 54 54 54 54 52 37 46
R2 0.33 0.31 0.31 0.31 0.32 0.32 0.34 0.34
Source: IMF staff estimates.
Note: Dependent variable = riskiness of credit allocation based on leverage. Real GDP growth and domestic currency appreciation vis-à-vis the US dollar are controlled for in all regressions. Column (4) is a horse race between
different macroprudential policies. The number of measures (out of four) that have the same sign and that are significant at the 10 percent level or higher is reported in columns (9) and (10). For the macroprudential policies, the
robustness information is based on the horse race. See Annex Table 2.1.1 for countries and years in the sample. See Annex Table 2.1.2 for definitions and sources of all variables. In all specifications, standard errors are clustered at
the country level. Standard errors are in parentheses.
***p < 0.01; **p < 0.05; *p < 0.1.
CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

for all variables related to the legal and institutional The Impact of the Riskiness of Credit Allocation on
framework and supervisory quality are averaged at Systemic Banking Crisis Risk
the country level and enter the regression only as an The logarithm of the odds ratio of the start of a
interaction term. systemic banking crisis is analyzed using the following
The results are shown in Annex Table 2.2.3. Col- panel logit model:
umns (1)–(3) and (5)–(7) show the results obtained
when each variable found to be robustly significant P[Crisisstartt = 1| Xi,t−1]
log _________________
enters the regression individually.34 Column (4) pres- P[Crisisstartt = 0| Xi,t−1]
ents the results of a horse race between the macropru- = α i + β ∆ Crediti,t−1
mv3 + γRiskiness mv3
i,t−1
dential measures that are significant when entering
individually. + δ Controlsi,t−1
mv3 + u ,
i,t (A2.3.1)

in which Crisisstart is a dummy variable equal to 1


Annex 2.3. The Riskiness of Credit Allocation at the start of a systemic banking crisis, as defined in
and Macro-Financial Outcomes Laeven and Valencia (forthcoming) and equal to 0
This annex discusses the empirical methodologies otherwise. X refers to the vector of explanatory vari-
used to analyze how the riskiness of credit allocation ables. α i is a country fixed effect. ∆ Credit is the change
affects the occurrence of systemic banking crises, in the ratio of bank credit to the nonfinancial private
banking sector stress, and downside risks to GDP sector to nominal GDP.35 Riskiness is the riskiness of
growth. The results are robust to using alternative data credit allocation, based on the leverage indicator, the
sources for credit, including credit data compiled by interest coverage ratio indicator, the debt overhang
the Bank for International Settlements (for both total indicator, or the expected default frequency indica-
credit to the nonfinancial private sector and credit to tor. Controls include controls for the macroeconomic
the nonfinancial corporate sector). The results are also and financial environment; that is, the change in the
robust to the inclusion of corporate spreads, median current-account-balance-to-GDP ratio, real GDP
firm leverage (or median interest coverage ratio), and growth, and a financial conditions index. All explan-
share of high-yield bond issuance as an additional atory variables enter the equation as the lag of their
control variable. simple three-year moving average and are demeaned
at the country level. The selection of macroeconomic
variables follows the specification of Jordà, Schularick,
and Taylor (2016a).36 An extended version of this
exercise includes interaction terms between the change
in the credit-to-GDP ratio and the riskiness of credit
34This analysis also investigates (1) measures of financial depth, allocation. The results, presented in Annex Table 2.3.1,
including the ratios of private credit to GDP, bank assets to GDP, are robust to using alternative estimators for the panel
bank credit to deposits, and external loans and deposits to domestic logit model (including two-way-clustered standard
deposits; and capital account openness; (2) other measures of
banking sector soundness, including bank concentration; probability errors of the coefficients).
of default of the banking sector; and the ratios of bank capital to
total assets, bank regulatory capital to risk-weighted assets, and bank
return on equity; (3) an additional 11 types of macroprudential The Effect of the Riskiness of Credit Allocation on
instruments, including countercyclical capital buffers and minimum
capital requirements; (4) other measures of supervisory quality, such
Banking Sector Equity Stress Risk
as a dummy for high supervisory quality based on Basel Core Prin- The importance of the riskiness of credit allocation
ciples assessments, restructuring power of the supervisory authority,
and the degree of independence of the supervisory authority from
for financial stability is explored in a further dimen-
political influence; and (5) other legal and institutional indicators,
such as anti-self-dealing (Djankov and others 2008), burden of
proof and disclosure index (La Porta, Lopez-de-Silanes, and 35The change in the credit-to-GDP ratio is winsorized at the

Shleifer 2006), corruption index (La Porta, Lopez-de-Silanes, 1 percent level to reduce the influence of outliers.
and Shleifer 2006), and the corporate governance opacity index 36This specification differs from Jordà, Schularick, and Taylor

(Brandão-Marques, Gelos, and Melgar 2013). None of these (2016a) in that it uses real GDP growth instead of real GDP
are found to have a robust significant impact on the riskiness of growth per capita. The results are robust to using real GDP
credit allocation. growth per capita.

International Monetary Fund | April 2018 87


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Annex Table 2.3.1. Panel Logit Analysis: Probability of the Occurrence of a Systemic Banking Crisis
(1) (2) (3) (4) (5) (6)
Variables Dependent Variable: Start of a Systemic Banking Crisis
Change in Credit-to-GDP Ratio 0.202*** 0.141* 0.0565 0.0745 0.0808 –0.0902
(0.0699) (0.0737) (0.0849) (0.100) (0.108) (0.131)
Financial Conditions Index –1.742** –2.536*** –2.686*** –2.907*** –4.441***
(0.682) (0.611) (0.604) (0.724) (0.854)
Riskiness_Leverage 1.924***
(0.674)
Riskiness_Interest Coverage Ratio 2.533***
(0.861)
Riskiness_Debt Overhang 2.087***
(0.461)
Riskiness_Expected Default Frequency 2.113***
(0.734)

Observations 443 443 443 443 431 361


Number of Countries 21 21 21 21 20 17
Country Cluster Yes Yes Yes Yes Yes Yes
Country Fixed Effects Yes Yes Yes Yes Yes Yes
Pseudo R 2 0.243 0.353 0.465 0.487 0.515 0.606
Source: IMF staff estimates.
Note: Standard errors are in parentheses. Explanatory variables enter the regression as the lag of their simple three-year moving average and are demeaned at
the country level; the change in credit-to-GDP ratio is winsorized at 1 percent. Controls include the change in current-account-to-GDP ratio and the real GDP
growth rate.
***p < 0.01; **p < 0.05; *p < 0.1.

sion: the risk of a banking sector equity price stress model (including two-way-clustered standard errors of
event. This risk is examined using the following frame- the coefficients).
work, as in Baron and Xiong (2017):

P[stresst,t+h = 1| Xi,t−1] The Impact of the Riskiness of Credit Allocation on


log ________________ = α i + β ∆ Crediti,t−1
mv3
P[stresst,t+h = 0| Xi,t−1] Downside Risks to GDP Growth
+ γRiskinessi,t−1
mv3
The following equation is estimated:
+ δ Controlsi,t−1
mv3 + u
i,t+h , ∆ yi,t,t+h = β ∆ Crediti,t−1
mv3 + γRiskiness mv3
i,t−1
(A2.3.2) + δ ∆ Crediti,t−1
mv3 × Riskiness mv3
i,t−1

in which stress is a dummy variable equal to 1, if there mv3 + u ,


+ Controlsi,t−1 i,t (A2.3.3)
is a stress event of banking sector equity prices in the
in which ∆ yi,t,t+h is the cumulative real GDP growth
time window from t to t + h, h = 0,...,3. A stress event
rate over the future h years (from t to t + h), in
is defined as an episode in which the annual excess
which h = 1,...,3. Riskiness and the change in the
equity return on the banking sector (relative to a
credit-to-GDP ratio are defined as in the previously
zero-coupon government bond yield of short maturity)
described analyses. Controls include real GDP growth
is below the country-specific mean by more than one
and a financial conditions index. The financial con-
standard deviation. The other variables are defined in
ditions index includes the sovereign spread, which
the same way as in the crisis model described previ-
partially captures the impact of fiscal policies.37 All
ously. Controls include a financial conditions index.
explanatory variables enter the equation as the lag
An extended version of this exercise includes interac-
of their simple three-year moving average and are
tion terms between the change in the credit-to-GDP
demeaned at the country level. The model is esti-
ratio and the riskiness of credit allocation. Annex
Table 2.3.2 presents the results. The results are robust 37Fiscal policies are found to affect economic recoveries in a differ-

to using alternative estimators for the panel logit ent empirical framework by IMF (2016).

88 International Monetary Fund | April 2018


CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

Annex Table 2.3.2. Panel Logit Analysis: Banking Sector Equity Stress Risk
(1) (2) (3) (4) (5) (6) (7) (8)
Variables Dependent Variable: Bank Equity Crash between t and t + h (h = 1,3)
t+1 t+3 t+1 t+3 t+1 t+3 t+1 t+3
Change in Credit-to-GDP Ratio –0.000975 0.0129 0.0294 0.0306 0.0316 0.0317 0.0345 0.0253
(0.0381) (0.0433) (0.0309) (0.0357) (0.0365) (0.0427) (0.0437) (0.0464)
Riskiness_Leverage 0.898*** 0.727***
(0.246) (0.246)
Riskiness_Interest Coverage
Ratio 0.690*** 0.717**
(0.256) (0.320)
Riskiness_Debt Overhang 0.569** 0.440
(0.223) (0.271)
Riskiness_Expected Default 0.451* 0.321
Frequency (0.274) (0.296)

Observations 573 573 573 573 552 552 505 505


Number of Countries 36 36 36 36 34 34 33 33
Country Cluster Yes Yes Yes Yes Yes Yes Yes Yes
Country Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes
Pseudo R 2 0.0882 0.130 0.0495 0.115 0.0517 0.102 0.0388 0.0950
Source: IMF staff estimates.
Note: Standard errors are in parentheses. Explanatory variables enter the regression as the lag of their simple three-year moving average, and are demeaned at
the country level; the change in credit-to-GDP ratio is winsorized at 1 percent. Each estimation controls for financial conditions.
***p < 0.01; **p < 0.05; *p < 0.1.

mated using quantile regressions with nonadditive of the riskiness of credit allocation on growth is also
fixed effects to examine the relationship between examined using a logit regression with a low-growth
the riskiness of credit allocation and the 20th and outturn dummy as the dependent variable. In that
50th percentiles of the future growth distribution. exercise, low-growth outturn is equal to 1 when the
Regressions with and without the interaction between cumulative real GDP growth rate over the future h
the change in the credit-to-GDP ratio and riskiness years (from t to t + h) is below the 20th percentile
are both estimated. The results are shown in Annex of its country-specific distribution and equal to zero
Table 2.3.3. Similar results are obtained for the otherwise. The findings confirm those obtained in the
leverage-based measure using Orbis data. The impact quantile regression framework.

International Monetary Fund | April 2018 89


90
Annex Table 2.3.3. Impact of the Riskiness of Credit Allocation on Downside Risks to Growth (20th and 50th percentiles of growth distribution)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)
Variables 20 pt 50 pt 20 pt 50 pt 20 pt 50 pt 20 pt 50 pt 20 pt 50 pt 20 pt 50 pt 20 pt 50 pt 20 pt 50 pt
Cumulative Real GDP Growth Rate over Future Three Years (t, t + 3)
Change in –0.232*** –0.268*** –0.239*** –0.254*** –0.228*** –0.291*** –0.172*** –0.231*** –0.224*** –0.268*** –0.192*** –0.207*** –0.219*** –0.272*** –0.213*** –0.128***
Credit-to-GDP (0.0335) (0.0358) (0.0364) (0.0471) (0.0278) (0.0260) (0.0274) (0.0380) (0.0349) (0.0367) (0.0233) (0.0472) (0.0294) (0.0315) (0.0369) (0.0328)
Ratio
Riskiness_Leverage –0.468*** –0.480*** –0.494*** –0.444***
(0.144) (0.107) (0.159) (0.127)
Change in –0.0549** –0.0820***
Credit-to-GDP (0.0253) (0.0288)
Ratio
Riskiness_

International Monetary Fund | April 2018


Leverage
Riskiness_Interest –0.927*** –0.421*** –1.306*** –0.391***
Coverage Ratio (0.207) (0.118) (0.237) (0.0948)
(ICR)
Change in –0.237*** –0.217***
Credit-to-GDP (0.0467) (0.0360)
GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Ratio
Riskiness_ICR
Riskiness_Debt –0.406* –0.328** –0.522*** –0.229*
Overhang (0.237) (0.140) (0.161) (0.132)
Change in –0.146*** –0.204***
Credit-to-GDP (0.0297) (0.0277)
Ratio
Riskiness_Debt
Overhang
Riskiness_ –0.879*** –0.383** –0.942*** –0.397**
Expected Default (0.243) (0.161) (0.233) (0.190)
Frequency
Change in –0.0798 –0.171***
Credit-to-GDP (0.0749) (0.0199)
Ratio
Riskiness_
Expected Default
Frequency
Observations 602 602 602 602 602 602 602 602 592 592 592 592 532 532 532 532
Number of 41 41 41 41 41 41 41 41 41 41 41 41 39 39 39 39
Countries
Source: IMF staff estimates.
Note: Standard errors are in parentheses. Explanatory variables enter the regressions as the lag of their simple three-year moving average and are demeaned at the country level; the change in credit-to-GDP ratio is winsorized at 1 percent.
Controls include real GDP growth and a financial conditions index. pt = percentile.
***p < 0.01; **p < 0.05; *p < 0.1.
CHAPTER 2 THE RISKINESS OF CREDIT ALLOCATION: A SOURCE OF FINANCIAL VULNERABILITY?

References Brandão-Marques, Luis, Gaston Gelos, and Natalia Melgar.


2013. “Country Transparency and the Global Transmission of
Abrigo, Michael R. M., and Inessa Love. 2016. “Estimation
Financial Shocks.” IMF Working Paper 13/156, International
of Panel Vector Autoregression in Stata.” Stata Journal 16
Monetary Fund, Washington, DC.
(3): 778–804.
Caballero, Ricardo J., and Alp Simsek. 2017. “A Risk-Centric
Acharya, Viral V., Tim Eisert, Christian Eufinger, and Christian
Model of Demand Recessions and Macroprudential Policy.”
W. Hirsch. 2016. “Whatever It Takes: The Real Effects of
NBER Working Paper 23614, National Bureau of Economic
Unconventional Monetary Policy.” SAFE Working Paper 152,
Research, Cambridge, MA.
Research Center SAFE—Sustainable Architecture for Finance
Campbell, John Y., Jens Dietrich Hilscher, and Jan Szilagyi. 2011.
in Europe, Goethe University, Frankfurt.
“Predicting Financial Distress and the Performance of Dis-
Adrian, Tobias, Nina Boyarchenko, and Domenico Giannone.
2016. “Vulnerable Growth.” Federal Reserve Bank of New tressed Stocks.” Journal of Investment Management 9 (2): 14–34.
York Staff Report 794, New York. Cong, Lin, Haoyu Gao, Jacopo Ponticelli, and Xiaoguang Yang.
Adrian, Tobias, Daniel Covitz, and Nellie Liang. 2015. 2017. “Credit Allocation under Economic Stimulus: Evidence
“Financial Stability Monitoring.” Annual Review of Financial from China.” Unpublished, University of Chicago.
Economics 7 (1): 357–95. Damodaran, Aswath. 2014. Applied Corporate Finance, 4th
Adrian, Tobias, and Nellie Liang. 2018. “Monetary Policy, edition. Hoboken, NJ: Wiley.
Financial Conditions, and Financial Stability.” International Dell’Ariccia, Giovanni, Deniz Igan, Luc Laeven, and Hui Tong.
Journal of Central Banking 14 (1): 73–131. 2016. “Credit Booms and Macrofinancial Stability.” Economic
Adrian, Tobias, and Hyun Song Shin. 2014. “Procyclical Leverage Policy 31 (86): 299–355.
and Value-at-Risk.” Review of Financial Studies 27 (2): 373–403. Dell’Ariccia, Giovanni, Luc Laeven, and Robert Marquez. 2014.
Alam, Zohair, Adrian Alter, Jesse Eiseman, Gaston Gelos, Hee- “Real Interest Rates, Leverage, and Bank Risk-Taking.” Jour-
don Kang, Machiko Narita, Erlend Nier, and Naixi Wang. nal of Economic Theory 149 (1): 65–99.
Forthcoming. “Digging Deeper—Evidence on the Effects of Dell’Ariccia, Giovanni, Luc Laeven, and Gustavo A. Suarez.
Macroprudential Policies from a New Database.” IMF Work- 2017. “Bank Leverage and Monetary Policy’s Risk-Taking
ing Paper, International Monetary Fund, Washington, DC. Channel: Evidence from the United States.” Journal of
Altman, Edward I. 1968. “Financial Ratios, Discriminant Anal- Finance 72 (2): 613–54.
ysis and the Prediction of Corporate Bankruptcy.” Journal of Dell’Ariccia, Giovanni, and Robert Marquez. 2006. “Lend-
Finance 23 (4): 189–209. ing Booms and Lending Standards.” Journal of Finance 61
———. 2013. “Predicting Financial Distress of Companies: (5): 2511–46.
Revisiting the Z-Score and ZETA Models.” In Handbook of Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes,
Research Methods and Applications in Empirical Finance, edited and Andrei Shleifer. 2008. “The Law and Economics of
by Adrian R. Bell, Chris Brooks, and Marcel Prokopczuk, Self-Dealing.” Journal of Financial Economics 88 (3): 430–65.
428–56. Cheltenham, UK: Edward Elgar Publishing. Dollar, David, and Shang-Jin Wei. 2007. “Das (Wasted) Kapital:
Banco de España. 2017. “Financing and Investment Decisions of Firm Ownership and Investment Efficiency in China.”
Spanish Non-Financial Corporations.” 2016 Annual Report. IMF Working Paper 07/9, International Monetary Fund,
Baron, Matthew, and Wei Xiong. 2017. “Credit Expansion and Washington, DC.
Neglected Crash Risk.” Quarterly Journal of Economics 132 Farhi, Emmanuel, and Iván Werning. 2016. “A Theory of Mac-
(2): 713–64. roprudential Policies in the Presence of Nominal Rigidities.”
Barth, James R., Gerard Caprio Jr., and Ross Levine. 2013. “Bank Econometrica 84 (5): 1645–704.
Regulation and Supervision in 180 Countries from 1999 to Fishburn, Peter C., and R. Burr Porter. 1976. “Optimal
2011.” Journal of Financial Economic Policy 5 (2): 111–219. Portfolios with One Safe and One Risky Asset: Effects of
Berger, Allen N., and Gregory F. Udell. 2004. “The Institutional Changes in Rate of Return and Risk.” Management Science 22
Memory Hypothesis and the Procyclicality of Bank Lending (10): 1064–73.
Behavior.” Journal of Financial Intermediation 13 (4): 458–95. Fukuda, Shin‐ichi, and Jun‐ichi Nakamura. 2011. “Why Did
Bernanke, Ben S., and Mark Gertler. 1989. “Agency Costs, Net ‘Zombie’ Firms Recover in Japan?” World Economy 34
Worth and Business Fluctuations.” American Economic Review (7): 1124–37.
79 (1): 14–31. Geanakoplos, John. 2010. “The Leverage Cycle.” In NBER
———, and Simon Gilchrist. 1996. “The Flight to Quality and Macroeconomics Annual 2009, Vol. 24, edited by Darren Ace-
the Financial Accelerator.” Review of Economics and Statistics moglu, Kenneth Rogoff, and Michael Woodford. Cambridge,
78 (1): 1–15. MA: National Bureau of Economic Research.
Bordalo, Pedro, Nicola Gennaioli, and Andrei Shleifer. 2018. Gourinchas, Pierre-Olivier, and Maurice Obstfeld. 2012. “Stories
“Diagnostic Expectations and Credit Cycles.” Journal of of the Twentieth Century for the Twenty-First.” American
Finance, published online. Economic Journal: Macroeconomics 4 (1): 226–65.

International Monetary Fund | April 2018 91


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Greenwood, Robin, and Samuel G. Hanson. 2013. “Issuer Qual- Lam, W. Raphael, Alfred Schipke, Yuyan Tan, and Zhibo Tan.
ity and Corporate Bond Returns.” Review of Financial Studies 2017. “Resolving China’s Zombies: Tackling Debt and Rais-
26 (6): 1483–525. ing Productivity.” IMF Working Paper 17/266, International
Guillén, Mauro F., and Laurence Capron. 2016. “State Capacity, Monetary Fund, Washington, DC.
Minority Shareholder Protections, and Stock Market Devel- Lang, William W., and Leonard I. Nakamura. 1995. “‘Flight
opment.” Administrative Science Quarterly 61 (1): 125–60. to Quality’ in Banking and Economic Activity.” Journal of
Holmstrom, Bengt, and Jean Tirole. 1997. “Financial Inter- Monetary Economics 36 (1): 145–64.
mediation, Loanable Funds, and the Real Sector.” Quarterly López-Salido, David, Jeremy C. Stein, and Egon Zakrajšek.
Journal of Economics 112 (3): 663–91. 2017. “Credit-Market Sentiment and the Business Cycle.”
International Monetary Fund (IMF). 2016. “Debt: Use It Quarterly Journal of Economics 3 (1): 1373–426.
Wisely.” Fiscal Monitor, Washington, DC, October. Maliszewski, Wojciech, Serkan Arslanalp, John Caparusso, José
———. 2017a. “Financial Sector Stability Assessment Report, Garrido, Si Guo, Joong Shik Kang, W. Raphael Lam, and
People’s Republic of China.” Country Report 17/280, others. 2016. “Resolving China’s Corporate Debt Problem.”
Washington, DC. IMF Working Paper 16/203, International Monetary Fund,
———. 2017b. “People’s Republic of China—Staff Report for Washington, DC.
the 2017 Article IV Consultation.” Country Report 17/205, Matutes, Carmen, and Xavier Vives. 2000. “Imperfect Compe-
Washington, DC. tition, Risk Taking, and Regulation in Banking.” European
Jiménez, Gabriel, Steven Ongena, José-Luis Peydró, and Jesús Economic Review 44 (1): 1–34.
Saurina. 2014. “Hazardous Times for Monetary Policy: What Merton, Robert C. 1974. “On the Pricing of Corporate Debt:
Do Twenty-Three Million Bank Loans Say about the Effects The Risk Structure of Interest Rates.” Journal of Finance 29
of Monetary Policy on Credit Risk-Taking?” Econometrica 82 (2): 449–70.
(2): 463–505. Minsky, Hyman P. 1977. “The Financial Instability Hypothesis:
———. 2017. “Macroprudential Policy, Countercyclical Bank An Interpretation of Keynes and an Alternative to ‘Standard’
Capital Buffers and Credit Supply: Evidence from the Span- Theory.” Challenge 20 (1): 20–27.
ish Dynamic Provisioning Experiments.” Journal of Political ———. 1986. Stabilizing an Unstable Economy. New Haven,
Economy 125 (6): 2126–77. CT: Yale University Press.
Jordà, Òscar, Moritz Schularick, and Alan M. Taylor. 2016a. Moody’s. 2006. “The Distribution of Common Financial Ratios
“The Great Mortgaging: Housing Finance, Crises and Busi- by Rating and Industry for North American Non-Financial
ness Cycles.” Economic Policy 31 (85): 107–52. Corporations: July 2006.” Special Comment.
———. 2016b. “Sovereigns versus Banks: Credit, Crises, and Ohlson, James A. 1980. “Financial Ratios and the Probabilistic
Consequences.” Journal of the European Economic Association Prediction of Bankruptcy.” Journal of Accounting Research 18
14 (1): 45–79. (1): 109–31.
Kalemli-Özcan, Şebnem, Bent Sorensen, Carolina Rajan, Raghuram G. 2006. “Has Finance Made the World Risk-
Villegas-Sanchez, Vadym Volosovych, and Sevcan Yesiltas. ier?” European Financial Management 12 (4): 499–533.
2015. “How to Construct Nationally Representative Schivardi, Fabiano, Enrico Sette, and Guido Tabellini. 2017.
Firm-Level Data from the ORBIS Global Database.” NBER “Credit Misallocation during the European Financial Crisis.” BIS
Working Paper 21558, National Bureau of Economic Working Paper 669, Bank for International Settlements, Basel.
Research, Cambridge, MA. Schularick, Moritz, and Alan M. Taylor. 2012. “Credit Booms
Kindleberger, Charles Poor. 1978. Manias, Panics, and Crashes: A Gone Bust: Monetary Policy, Leverage Cycles, and Financial Cri-
History of Financial Crises. New York: Basic Books. ses, 1870–2008.” American Economic Review 102 (2): 1029–61.
Kirti, Divya. 2018. “Lending Standards and Output Growth.” Shumway, Tyler. 2001. “Forecasting Bankruptcy More Accu-
IMF Working Paper 18/23, International Monetary Fund, rately: A Simple Hazard Model.” Journal of Business 74
Washington, DC. (1): 101–24.
Kiyotaki, Nobuhiro, and John Moore. 1997. “Credit Cycles.” Song, Zheng Michael, and Wei Xiong. 2018. “Risks in China’s
Journal of Political Economy 105 (2): 211–48. Financial System.” NBER Working Paper 24230, National
Krishnamurthy, Arvind, and Tyler Muir. 2017. “How Credit Bureau of Economic Research, Cambridge, MA.
Cycles across a Financial Crisis.” NBER Working Paper 23850, Standard and Poor’s. 2013. “Corporate Methodology.” Stan-
National Bureau of Economic Research, Cambridge, MA. dard and Poor’s Rating Services, McGraw Hill Finan-
La Porta, Rafael, Florencio Lopez‐de‐Silanes, and Andrei cial, New York.
Shleifer. 2006. “What Works in Securities Laws?” Journal of Uluc, Arzu, and Tomasz Wieladek. 2017. “Capital Require-
Finance 61 (1): 1–32. ments, Risk Shifting and the Mortgage Market.” ECB Work-
Laeven, Luc, and Fabian Valencia. Forthcoming. “Systemic ing Paper 2061, European Central Bank, Frankfurt.
Banking Crises Database.” IMF Working Paper, International Vassalou, Maria, and Yuhang Xing. 2004. “Default Risk in
Monetary Fund, Washington, DC. Equity Returns.” Journal of Finance 59 (2): 831–68.

92 International Monetary Fund | April 2018


3
CHAPTER

HOUSE PRICE SYNCHRONIZATION: WHAT ROLE FOR FINANCIAL FACTORS?

Summary

R
ising house prices have been a feature of the economic recovery in many countries since the global finan-
cial crisis. But recent increases have also been occurring in an accommodative monetary policy environ-
ment in many advanced economies, raising the specter of financial instability should financial conditions
reverse and simultaneously lead to a decline in house prices.
This chapter analyzes whether and how house prices move in tandem across countries and major global cities;
that is, the synchronicity of global house prices. On the one hand, higher house price synchronization and deeper
global links in housing markets may be beneficial. On the other hand, higher synchronization may be the result of
global financial conditions influencing local house price dynamics and housing markets, thereby propagating local
economic and financial shocks. The analysis in this chapter aims to inform the views that policymakers ought to
take on the synchronicity in house prices.
Strikingly, the chapter finds an increase in house price synchronization, on balance, for 40 countries and 44
major cities in advanced and emerging market economies. The chapter’s analysis suggests that countries’ and cities’
exposure to global financial conditions may provide an explanation for the increase in house price synchronization.
Moreover, cities in advanced economies may be particularly exposed to global financial conditions, perhaps owing
to their integration with global financial markets or to their attractiveness for global investors searching for yield or
safe assets.
Thus, policymakers cannot ignore the possibility that shocks to house prices elsewhere may affect domestic
markets. While house price synchronization in and of itself may not warrant policy intervention, the evidence
presented in this chapter suggests that heightened synchronicity of house prices can signal a downside tail risk
to real economic activity, especially when taking place in a buoyant credit environment. The chapter finds that
macroprudential policies seem to retain some ability to influence local house price developments even in countries
with highly synchronized housing markets, and that macroprudential policy measures put in place to tame rising
vulnerabilities in a country’s financial sector may have the additional effect of reducing a country’s house price
synchronization with the rest of the world. These unintended effects are worth considering when evaluating the
trade-offs of implementing macroprudential and other policies.

International Monetary Fund | April 2018 93


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Introduction Figure 3.1. House Price Gains in Selected Cities and


Countries Have Been Widespread
Rising house prices have been a feature of the (Average annual real house price growth, 2013–17, percent)
economic recovery in many countries since the global
financial crisis (Figure 3.1). House price gains have City level Country level
been widespread and, in some markets, brisk. Indeed,
CHN: Shanghai
in recent years, the simultaneous growth in house NZL: Auckland
prices in many countries and cities located in advanced AUS: Sydney
HUN: Budapest
and emerging market economies parallels the coordi- TUR: Istanbul
nated run-up seen before the crisis (Figure 3.2). GBR: London
EST: Tallinn
House prices may comove across countries and cities NOR: Oslo
because economic activity has picked up at similar S E: Greater Stockholm
PHL: Manila
times. During 2017, there was a pickup in growth in DNK: Copenhagen
120 economies, accounting for three-quarters of world IRL: Dublin
GRC: Athens
GDP, which was the broadest synchronized global PER: Lima
NLD: Amsterdam
growth upsurge since 2010 (IMF 2018a). The wide- DEU: Berlin
spread boost to economic growth may support addi- HKG: Urban areas
CHL: South Santiago
tional housing demand across many countries, leading COL: Bogotá
to upward pressure on house prices. IND: Mumbai
ARE: Dubai
Global financial conditions—that is, those prevailing CZE: Prague
in major financial centers—and cross-border capital JPN: Tokyo
CAN: Toronto
flows may also explain the comovement in house prices AUT: Vienna
(see Rey 2015 and Chapter 3 of the April 2017 Global BEL: Brussels
USA: New York
Financial Stability Report [GFSR]).1 Recent increases in SRB: Belgrade
house prices have been occurring in an environment of PRT: Lisbon
ESP: Madrid
easy financial conditions in major advanced economies IDN: Jakarta
characterized by low policy rates, compressed spreads, KOR: Southern Seoul
BRA: São Paulo
and low volatility that has spread globally (Figure 3.2). T N: Taipei City
Moreover, in some housing markets, the motives of FIN: Metro area
MEX: Mexico City
global and institutional investors searching for yield FRA: Inner Paris
in a low-interest-rate environment have emerged as a SVN: Ljubljana
ARG: Buenos Aires
potential explanation for the brisk and synchronized SGP: Core central region
increases in house prices. In the past several years, ITA: Rome
RUS: Moscow
real estate investments—including in residential real
estate—by private equity firms, real estate investment –10 –5 0 5 10 15

trusts (REITs), and institutional investors appear Sources: Bank for International Settlements; CEIC Data Co. Ltd; Emerging Markets
to have grown (Figure 3.3), and anecdotes point to Economic Data Ltd; Global Financial Data Solutions; Haver Analytics; IMF, Research
increasing investor participation in select housing mar- Department house price data set; Organisation for Economic Co-operation and
Development; and IMF staff calculations.
kets, such as Amsterdam, Melbourne, Sydney, Toronto, Note: Data used in this figure comes from the sources listed above, and therefore
they could differ from data published by national authorities. Nominal house prices
are de ated by consumer price in ation, when real house prices are not readily
available in the sources above. Cities selected are the largest cities in each
Prepared by a staff team consisting of Jane Dokko (team economy in the sample based on population owing to data availability, and overlap
leader), Adrian Alter, Mitsuru Katagiri, Romain Lafarguette, with the top 50 cities for global investors identified by Cushman akefield
and Dulani Seneviratne, with contributions from Anil Ari, (2017). Labels in the figure use International Organi ation for Standardi ation (ISO)
Christian Bogmans, and Alan Xiaochen Feng, under the general codes. Latest available data as of 2017:Q2 for most economies; fewer than 15
guidance of Claudio Raddatz and Dong He. Claudia Cohen and economies have data through 2017:Q3.
Breanne Rajkumar provided editorial assistance.
1Moreover, global financial conditions may affect the comove-

ment in commercial real estate prices, but the chapter does not
analyze the synchronicity in commercial real estate prices because
high-quality, cross-country comparable data are limited to fewer than
10 countries.

94 International Monetary Fund | April 2018


CHAPTER 3 HOUSE PRICE SYNCHRONIZATION: WHAT ROLE FOR FINANCIAL FACTORS?

Figure 3.2. Widespread House Price Gains Have Accompanied Accommodative Financial Conditions (Diffusion Index of
House Price Growth and Global Financial Conditions)
(Left scale = percent; right scale = standard deviations)

Diffusion index S financial conditions index (right scale)


1. Advanced Economies: Country Level 2. Advanced Economies: City Level
100 8 100 8

80 6 80 6

60 4 60 4

40 2 40 2

20 0 20 0

0 –2 0 –2
1986:Q1
87:Q3
89:Q1
90:Q3
92:Q1
93:Q3
95:Q1
96:Q3
98:Q1
99:Q3
2001:Q1
02:Q3
04:Q1
05:Q3
07:Q1
08:Q3
10:Q1
11:Q3
13:Q1
14:Q3
16:Q1

1986:Q1
87:Q3
89:Q1
90:Q3
92:Q1
93:Q3
95:Q1
96:Q3
98:Q1
99:Q3
2001:Q1
02:Q3
04:Q1
05:Q3
07:Q1
08:Q3
10:Q1
11:Q3
13:Q1
14:Q3
16:Q1
Diffusion index orld financial conditions index (right scale)
3. Emerging Market Economies: Country Level 4. Emerging Market Economies: City Level
100 8 100 8

80 6 80 6

60 4 60 4

40 2 40 2

20 0 20 0

0 –2 0 –2
1996:Q1
97:Q1
98:Q1
99:Q1
2000:Q1
01:Q1
02:Q1
03:Q1
04:Q1
05:Q1
06:Q1
07:Q1
08:Q1
09:Q1
10:Q1
11:Q1
12:Q1
13:Q1
14:Q1
15:Q1
16:Q1

1996:Q1
97:Q1
98:Q1
99:Q1
2000:Q1
01:Q1
02:Q1
03:Q1
04:Q1
05:Q1
06:Q1
07:Q1
08:Q1
09:Q1
10:Q1
11:Q1
12:Q1
13:Q1
14:Q1
15:Q1
16:Q1
Source: IMF staff calculations.
Note: Diffusion index is based on year-over-year growth rates of real house prices. This index measures the share of positive house price growth observations in
each quarter.

and Vancouver (Zillow Research 2017; Bloomberg At the same time, however, the links across hous-
News 2018).2 ing markets may transmit or amplify financial and
Synchronicity, or the correlation, in house prices macroeconomic shocks, increasing the exposure of
should concern policymakers because it may signal local housing markets to global financial conditions
stronger transmission of external shocks to local hous- or to shocks affecting foreign investors active in local
ing markets. The global integration of housing markets markets. As a result, policymakers’ ability to address
may contribute to house price synchronization, as well imbalances in the housing market through national
as to more liquidity in housing and mortgage mar- or local policies may be constrained, particularly if
kets, higher capital flows from abroad, and enhanced house prices across many countries decline at once. In
risk-sharing opportunities for households and lenders. this case, a decline in external demand may exacerbate
the challenges of stabilizing household balance sheets,
financial markets, and economic activity. In this sense,
2Other factors, such as illicit capital flows, motives for tax evasion, or

the legal environment, may contribute to cross-border real estate pur-


a sharp reversal of the prevailing accommodative global
chases, but analyzing these issues is beyond the scope of this chapter. financial conditions could challenge how policymakers

International Monetary Fund | April 2018 95


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 3.3. Institutional Investor Participation Has Been on synchronization increased in recent years? Did it
the Rise increase before the global financial crisis?
1. Real Estate Investment Trusts
• What factors contribute to or dampen synchronic-
(Market capitalization of REITs normalized by the total market ity? Is there a role for financial factors, or is house
capitalization; index, 2005:Q1 = 100) price synchronization related mainly to the comove-
Developed markets: overall ment in economic activity? Do bilateral or two-way
Developed markets: residential
United States: overall links between country or city pairs matter for syn-
United States: residential chronicity or do only global factors matter?
Developed markets (excluding United States): overall • Should policymakers pay attention to house price
Developed markets (excluding United States): residential
300 synchronicity to gain a better understanding of
financial vulnerabilities and risks?
250
The chapter’s focus on house price synchronization
200 should not detract from the important task of moni-
toring house prices in individual markets. In fact, the
150
analysis in the chapter seeks to complement bilateral
100 surveillance efforts and country-level analysis that can
explore house price valuation and dynamics using
50 sophisticated models and rich data.
2005:Q1 07:Q1 09:Q1 11:Q1 13:Q1 15:Q1 17:Q1

The main findings are as follows:


2. Weighted Average Target Allocation to Real Estate: All Institutions • On balance, synchronization in house prices across
(Percent)
10.5 countries and major cities has increased over the
past several decades in advanced and emerging
10.0
market economies. This trend follows the rise in the
comovement of financial asset prices documented
elsewhere (see Chapter 2 of the April 2016 GFSR).
9.5
• The short-term comovement in house prices sharply
increases around the time of global recessions in
9.0
advanced economies. These spikes are much larger
among major cities than at the country level, sug-
8.5
2013 14 15 16 17 18 gesting that the ramifications of the global financial
cycle for cities may be particularly notable.
Sources: Jones and Weill (2017); Thomson Reuters Datastream; and IMF staff • Global financial conditions contribute to synchroni-
calculations.
Note: Developed markets correspond to the aggregate REITs series compiled by zation in house prices across pairs of countries and
Thomson Reuters Datastream for developed markets in line with the country cities even after accounting for the comovement in
classifications from Morgan Stanley Capital International and Dow Jones.
REITs = real estate investment trusts. economic activity and other fixed and time-varying
fundamentals. Their contribution is particularly
strong in major cities in advanced economies that
are usually more integrated with global financial
address financial and macroeconomic instability should markets but also where local supply constraints may
a simultaneous decline in house prices occur. be more binding. The presence of global investors
This chapter analyzes whether and how house prices searching for yield or safe assets in major cities may
move in tandem across countries and major global also be an explanation.
cities; that is, the synchronicity in global house prices • The dynamics of house prices are similar to those
and its determinants. Using quarterly data on house of other financial assets. For example, the expected
prices for countries and major cities (see Annex 3.1), return to investing in housing varies over time and
the chapter addresses the following questions: is predictable in the long term. In the financial
• What are the trends in the synchronization of house literature, this pattern is usually associated with vari-
prices across countries and across major cities? Has ations in the risk premium demanded by investors,

96 International Monetary Fund | April 2018


CHAPTER 3 HOUSE PRICE SYNCHRONIZATION: WHAT ROLE FOR FINANCIAL FACTORS?

indicating that the demand for housing may also be analyzing house price synchronization. Second, stylized
influenced by investors. facts are presented to document trends and heterogene-
• Higher house price synchronization corresponds to ity in house price synchronization across advanced and
increased downside risks to growth at horizons of up emerging market economies. Third, potential contrib-
to one year, controlling for other financial and mac- utors to house price synchronization are analyzed, as is
roeconomic conditions. This finding suggests that the importance of this measure for economic growth.
the comovement in house prices can help predict The final section concludes with a policy discussion.
the tail risk of an economic downturn.
House Price Synchronicity:
The policy discussion for this chapter centers around A Conceptual Framework
the following sets of issues: House prices may move in tandem across countries
• Policymakers may wish to monitor the synchroniza- and major cities because of synchronous supply and
tion of house prices with respect to other countries, demand factors (Figure 3.4).3 Supply-side consid-
in addition to the over- or undervaluation of house erations include the costs of construction and land
prices within a country. To that end, increasing acquisition. On the demand side, demographics, tax
the granularity, timeliness, and coverage of data on and other policy considerations, and depreciation and
house prices within countries would help provide maintenance play a role. Financial factors, such as
richer indicators for bilateral and multilateral sur- the mortgage interest rate, the risk premium on assets
veillance. In addition, more comprehensive data on with similar risk characteristics as housing, household
the participation of global and institutional inves- leverage, and the expected nominal house price appre-
tors in housing markets would strengthen surveil- ciation rate, also matter.4
lance efforts. The comovement in economic fundamentals may
• Macroprudential policies seem to retain some ability be a source of house price synchronization. Several
to influence local house price developments in of these factors, such as construction costs, taxes, and
countries with highly synchronized housing markets, demographics, tend to be slow moving and may lead
albeit to a lesser extent than in those that are less to synchronization only over long horizons. However,
synchronized. Consistently, macroprudential policy other economic fundamentals, such as rent, income,
measures put in place to tame rising vulnerabilities and inflation, may lead to comovement in housing
in a country’s financial sector are followed by a prices at shorter terms.5 Indeed, the coincidence of
decline in a country’s house price synchronization, recessions and housing downturns is well-documented,
suggesting that some of the drivers of synchroniza- with trade and financial links between countries
tion operate through local financial intermediaries. possibly playing a contributing role (Claessens, Kose,
Fiscal-based policies, such as ad valorem and buyers’ and Terrones 2011; Kose, Otrok, and Prasad 2012;
stamp duty taxes, may also lower house price syn- Kalemli-Özcan, Papaioannou, and Peydró 2013;
chronization, but less so than other measures, such Leamer 2015).
as limits on loan-to-value ratios. These unintended Simultaneous changes to financial factors can also
effects are an aspect to consider when evaluating the lead to greater house price synchronization. Changing
trade-offs of implementing macroprudential and interest rates, risk premiums, or expected capital gains
other policies (IMF 2013).
• Other policies that enhance resilience to global
3For an example of an asset pricing model that decomposes supply
financial shocks may also dampen house price
and demand factors for housing, see Poterba (1984) and Poterba,
synchronicity. This chapter presents evidence that Weil, and Shiller (1991).
exchange rate flexibility plays a role, but policies that 4Together, these nonfinancial and financial demand-side factors

deepen domestic real estate markets—or consumer determine the annual cost of homeownership, which is a function
of the user cost of housing. See US Department of Housing and
financial protections that discourage excessive or Urban Development (2000) for the precise details of how user costs
predatory lending to households—may also help. are calculated in the United States, and Poterba (1984) and Poterba,
Weil, and Shiller (1991) for a more general discussion.
5See the October 2013 World Economic Outlook (WEO) for a
The rest of this chapter covers four areas. First, discussion of the factors contributing to the synchronization in
the next section provides a conceptual framework for business cycles.

International Monetary Fund | April 2018 97


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 3.4. Global Financial Conditions, Portfolio Channels, and Expectations Contribute to House Price Synchronization,
as Do Supply Constraints and Local Policy

Expectations

Country A Country B

Global investors
Global lenders
Domestic Domestic
Supply Supply
financial financial
constraints constraints
conditions conditions

House Global financial House


prices conditions prices

Business Business
Local policy Local policy
cycle cycle
Trade, FDI

Source: IMF staff.


Note: FDI = foreign direct investment.

may increase the comovement in house prices through • Portfolio channels: The presence of common lenders
the following mechanisms (Figure 3.4):6 or investors allows for the interdependence in house
• Changes in global financial conditions: The interna- prices in both crisis and normal times for reasons
tional transmission of financial conditions, such as potentially unrelated to economic fundamentals.8
those occurring because of a change in monetary For example, a shock in one country may lead
policy in one large country, usually occurs through global financial institutions to pull back on mort-
capital flows (Chapter 3 of the April 2017 GFSR).7 gage lending in many countries, perhaps to maintain
These flows do not need to go directly into housing capital requirements (Allen and Gale 2000; Ceto-
investments as long as they affect credit availability relli and Goldberg 2011). Alternatively, investors
and mortgage rates in the receiving country. In addi- experiencing distress in one market may liquidate
tion, an increase in the global demand for safe assets leveraged housing investments in other countries,
may compress the rates of sovereign bonds consid- possibly to meet margin calls or in anticipation of
ered as low risk, thereby holding down mortgage future redemptions, or may rebalance their portfo-
rates and supporting booming house prices across lios to follow predetermined investment mandates
many countries at once (Bernanke and others 2011). (Kodres and Pritsker 2002). Or shocks in one coun-
try can result in changes to investors’ risk appetite
and lead them to increase or withdraw their housing
6The user cost may also shift simultaneously across countries if
investments from many countries at once (Acharya
there is a coordinated tax reform that similarly changes tax rates or
aligns the tax deductibility of mortgage interest, but this chapter
and Pedersen 2005). In the housing market, recent
does not focus on these issues.
7Hirata and others (2012) find a role for a broader range of global

shocks, such as those to interest rates, productivity, credit, and 8See Chapter 2 of the April 2016 GFSR for a discussion of the

uncertainty. sources of financial market spillovers.

98 International Monetary Fund | April 2018


CHAPTER 3 HOUSE PRICE SYNCHRONIZATION: WHAT ROLE FOR FINANCIAL FACTORS?

developments point to the growing contribution flexibility may dampen the impact of global financial
of global and institutional investors to house price conditions because monetary policies may have more
dynamics in select major cities (Hekwolter of bite under such circumstances.
Hekhuis, Nijskens, and Heeringa 2017). Though Fluctuations in home values pose risks to house-
they are limited in number in the aggregate, the holds and financial institutions even if they occur in
geographic concentration of investors in certain only one country at a time. In a booming house price
cities may make house price synchronization more environment, households may engage in excessive risk
apparent among cities than among countries. These taking (Mian and Sufi 2009; Bhutta and Keys 2016),
channels may also contribute to house price syn- financial institutions may relax lending standards
chronization in normal times through arbitrage and (Demyanyk and Van Hemert 2009; Dell’Ariccia, Igan,
mortgage rates. and Laeven 2012; Chapter 2 of the April 2018 GFSR),
• Changes in expected capital gains: A coordinated and there may be overbuilding (Haughwout and
change in households’ or investors’ views of future others 2011). Thus, once the boom ends, a decline in
house prices across many countries can also result house prices may result in risks to macroeconomic and
in synchronicity. These changes in expectations can financial stability. Consumption may fall given that
be driven by rational views regarding future funda- housing is often the largest component of household
mentals (Himmelberg, Mayer, and Sinai 2005), but wealth in many countries, and household delever-
also by bouts of overoptimism, psychological factors, aging may be a further drag on growth (Chapter 2
and speculation (Shiller 2015). Rational or irrational of the October 2017 GFSR; Mian and Sufi 2009).
beliefs about house prices can propagate through Furthermore, banks’ exposures to house prices can
social networks, word of mouth, and other interper- cause them financial difficulties and may lead them
sonal links (Bailey and others 2016). If a wake-up to curtail many forms of lending, which, in turn, can
call leads to reassessment of these beliefs, perhaps lower employment (Berrospide, Black, and Keeton
in response to a shock in one country, a widespread 2016; Glancy 2017). Moreover, housing is a physical
realignment of house prices with fundamentals asset that requires maintenance and cannot be moved,
could occur (Goldstein 1998). There could even be so fire sales are often associated with blight and crime,
a systematic overcorrection if house prices exhibit which are destabilizing at the local level, because
momentum and excess variance relative to funda- distressed homes often sit vacant before they are sold
mentals (Case and Shiller 1990; Glaeser, Ponzetto, (Campbell, Giglio, and Pathak 2011; Anenberg and
and Shleifer 2016). Kung 2014). These costs are borne not just by the
households living in neighborhoods with distressed
As with many financial assets, institutional charac- sales but also by financial institutions if the legal sys-
teristics may influence whether financial factors lead tem is such that the ownership of foreclosed properties
to simultaneous changes in house prices across coun- is transferred to them.
tries. For example, financial integration can expose The challenges to macro-financial stability posed by
mortgage markets to global financial conditions and a house price decline in a given country can be larger
expose local financial markets to sudden stops in if the decline is synchronized with declines in other
capital flows (Chapter 3 of the April 2017 GFSR). countries. In this case, the pullback in consumption
Moreover, a country’s financial integration may and investment driven by balance sheet deleveraging
create a favorable environment for global investors to would coincide with a decline in external demand,
purchase housing directly, allowing global factors to
influence local house prices and local shocks to spread
arise through the specialization of production or because of how
more widely through a variety of mechanisms (see financially integrated banks differentially increase lending to coun-
earlier discussion).9 In contrast, greater exchange rate tries experiencing productivity shocks and contribute to divergent
output growth (Kalemli-Özcan, Papaioannou, and Peydró 2013).
9See Forbes (2012); Bekaert, Lundblad, and Siegel (2011); Finally, greater participation of foreign investors, especially those
Bekaert and Harvey (2000); Burger, Warnock, and Cacdac Warnock with long horizons, may be able to stabilize asset prices, including
(2012); and Miyajima, Mohanty, and Chan (2015) on equity and housing, if they behave countercyclically and take advantage of fire
bond market integration. Theoretically, greater financial integration sale opportunities. This would lead to a dampening of other drivers
may also correspond to less house price synchronization given that of synchronicity, although evidence on this countercyclical behavior
housing purchases are tied to business cycles. This relationship may may be limited (Chapter 2 of the April 2014 GFSR).

International Monetary Fund | April 2018 99


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 3.5. Synchronization Has Steadily Increased across leaving little room for the current account to offset the
Countries and Cities contraction in domestic demand. Indeed, in the past,
(Median synchronization; closer to zero denotes higher synchronization)
large and widespread house price swings have been
Country pairs Linear trend (country pairs) Major city pairs associated with periods of financial instability across
1. Within Advanced Economies many countries at once (Claessens, Kose, and Terrones
–0.05 2008, 2011; Reinhart and Rogoff 2008). These risks
would be compounded if a pullback among global
Higher synchronization

–0.10 investors were to lead to fire sales across asset classes,


capital flight, and tighter mortgage market conditions
–0.15 (Kaminsky and Reinhart 2000; Campbell, Giglio, and
Pathak 2011; Bekaert and others 2014; Chinco and
–0.20 Mayer 2015).10

–0.25
House Price Synchronization in
1991:Q1
92:Q3
94:Q1
95:Q3
97:Q1
98:Q3
2000:Q1
01:Q3
03:Q1
04:Q3
06:Q1
07:Q3
09:Q1
10:Q3
12:Q1
13:Q3
15:Q1
16:Q3 Countries and Cities
Different measures of house price synchronization
2. Within Emerging Market Economies
capture distinct dimensions of this phenomenon.
0.00 Synchronization can be measured in different ways and
at different frequencies. To capture these distinctions,
this chapter uses a broad set of measures applied to
Higher synchronization

–0.05 the comovement in house prices across countries and


cities. All measures focus on either the cyclical compo-
nent of real house prices—henceforth, the house price
–0.10
gap—or the quarterly growth rate in real house prices.
The former removes the medium-term trend in these
prices and allows for comparisons of housing markets
–0.15
with different medium-term cycles. The latter provides
1995:Q1
96:Q2
97:Q3
98:Q4
2000:Q1
01:Q2
02:Q3
03:Q4
05:Q1
06:Q2
07:Q3
08:Q4
10:Q1
11:Q2
12:Q3
13:Q4
15:Q1
16:Q2

a higher-frequency measure of house price growth that


can be analyzed at long horizons. Annex 3.2 provides
details of these measures.
3. Between Advanced and Emerging Market Economies
–0.05
Synchronicity in housing markets has markedly
increased over time.
• On balance, the house price gap has become more
Higher synchronization

–0.10
synchronized in countries and cities in advanced
and emerging market economies (Figure 3.5).11
–0.15
The synchronization in the house price gap reflects
medium-term changes to how shocks propagate
–0.20

10More specifically, such a pullback could directly cause or


–0.25 accompany instability. Previous GFSR research analyzing the
1995:Q1
96:Q2
97:Q3
98:Q4
2000:Q1
01:Q2
02:Q3
03:Q4
05:Q1
06:Q2
07:Q3
08:Q4
10:Q1
11:Q2
12:Q3
13:Q4
15:Q1
16:Q2

financial stability implications of rising equity and bond price


correlations has found that financial factors explain cross-country
spillovers and investor retrenchment during crises (see Chapter 3
of the April 2014 GFSR and Chapter 2 of the April 2016 GFSR).
Source: IMF staff estimates.
Moreover, as with financial assets, at times of fire sales, global
Note: The Synch1 measure capturing the negative of the absolute difference
between two countries’ house price gaps is used. See Annex 3.2 for methodology. investors may base their decision to sell on how liquid a partic-
Shaded areas correspond to US recessions. ular housing market is rather than on the fundamentals of the
housing market.
11The period for which data are available for each group starts

in 1973 for advanced economies and 1995 for emerging mar-


ket economies.

100 International Monetary Fund | April 2018


CHAPTER 3 HOUSE PRICE SYNCHRONIZATION: WHAT ROLE FOR FINANCIAL FACTORS?

across countries or cities (see Annexes 3.2 and Figure 3.6. The Relative Contribution of the Global Factor
3.3).12 Between 1991 and 2016, synchronicity is Has Grown
(Window = 15 years; percent)
lower among major cities in advanced economies
than among the countries where they are located, 60 Upper and lower bound (75th and 25th percentiles)
but it has gradually moved closer to country-level Median
synchronicity. This pattern is intriguing because
50
synchronicity should be lower among cities that are
affected by idiosyncratic shocks that average out at
the country level, and it indicates that the factors 40
driving house price synchronicity have become
disproportionately more important for cities. This 30
finding motivates a closer look at the house price
dynamics of major cities. Among emerging mar-
20
kets, synchronicity between countries and between
major cities is similar, perhaps for purely statistical
reasons (the major city often represents the bulk of 10
the national house price index) or because of more
integrated internal housing markets. 0
• In many advanced economies, moreover, the
1986:Q1
88:Q1
90:Q1
92:Q1
94:Q1
96:Q1
98:Q1
2000:Q1
02:Q1
04:Q1
06:Q1
08:Q1
10:Q1
12:Q1
14:Q1
16:Q1
increase in synchronization is evident in the ris-
ing share of the variation in house price growth
explained by a common global factor (Figure 3.6). Source: IMF staff estimates.
Note: The figure shows the rolling estimation with a 15-year window for the share
A dynamic factor model estimates that the share of of the variation in house price growth explained by a common global factor in the
the variance explained by the estimated global factor dynamic factor model. See nnex 3.3.
increases from about 10 percent to 30 percent over
the period from 1971 to 2016.13 This common
global factor summarizes the long-term contribu- economies.14 For example, the housing boom of the
tion of many sources of house price synchronicity, 2000s extended to many advanced economies, and
including the role of global financial developments simultaneous declines in house prices triggered large
and the tightening of financial links, among others financial sector losses worldwide during the global
(see Annex 3.3). financial crisis. Common shocks appear to affect
emerging market economies differently, as evidenced
The short-term comovement in house prices by the fact that the comovement in house prices is less
increases sharply around the time of global recessions likely to shoot up around the time of global recessions.
in advanced economies. This can be seen in Figure 3.7, Among advanced economies, the increase in short-term
which depicts the instantaneous quasi correlation, a synchronicity before recessions is much larger between
measure of short-term comovement, in house price major cities than between countries. This again
gaps. The sharp increases around global economic suggests that the factors driving this synchronous
downturns are noticeable and may reflect common movement may particularly affect major cities in
shocks affecting housing markets in many advanced advanced economies.
Countries and cities differ in how synchronized
they are. Their exposure to the common global factor
varies, with a larger contribution of this factor to
12A similar pattern is found when using seven-year rolling correla-
house prices in countries and cities in Europe than
tions in house price gaps.
13The factor loadings and vector autoregression parameters are in other regions (Figure 3.8). In addition, advanced
simultaneously estimated by the two-step procedure proposed in economies are more exposed than emerging market
Koop and Korobilis (2013) using data for 19 advanced economies
from the second quarter of 1971 to the fourth quarter of 2016.
This procedure requires long time series, so it cannot be adequately 14The instantaneous quasi correlation is constructed not to have a

applied to most emerging market economies. trend (see Annexes 3.2 and 3.3).

International Monetary Fund | April 2018 101


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 3.7. Instantaneous Quasi Correlation of House Price Gaps Shows Financial Cycle Properties (Median Shown)
1. Country Pairs: Within Advanced Economies 2. Major City Pairs: Within Advanced Economies
1.4 1.4
1.2 1.2
1.0 1.0
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
–0.2 –0.2
1995:Q1
96:Q2
97:Q3
98:Q4
2000:Q1
01:Q2
02:Q3
03:Q4
05:Q1
06:Q2
07:Q3
08:Q4
10:Q1
11:Q2
12:Q3
13:Q4
15:Q1
16:Q2

1995:Q1
96:Q2
97:Q3
98:Q4
2000:Q1
01:Q2
02:Q3
03:Q4
05:Q1
06:Q2
07:Q3
08:Q4
10:Q1
11:Q2
12:Q3
13:Q4
15:Q1
16:Q2
3. Country Pairs: Within Emerging Market Economies 4. Major City Pairs: Within Emerging Market Economies
1.0 1.0
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
–0.2 –0.2
–0.4 –0.4
1995:Q1
96:Q2
97:Q3
98:Q4
2000:Q1
01:Q2
02:Q3
03:Q4
05:Q1
06:Q2
07:Q3
08:Q4
10:Q1
11:Q2
12:Q3
13:Q4
15:Q1
16:Q2

1995:Q1
96:Q2
97:Q3
98:Q4
2000:Q1
01:Q2
02:Q3
03:Q4
05:Q1
06:Q2
07:Q3
08:Q4
10:Q1
11:Q2
12:Q3
13:Q4
15:Q1
16:Q2
5. Country Pairs: Between Advanced and Emerging Market Economies 6. Major City Pairs: Between Advanced and Emerging Market Economies
0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0

–0.2 –0.2

–0.4 –0.4

–0.6 –0.6
1995:Q1
96:Q2
97:Q3
98:Q4
2000:Q1
01:Q2
02:Q3
03:Q4
05:Q1
06:Q2
07:Q3
08:Q4
10:Q1
11:Q2
12:Q3
13:Q4
15:Q1
16:Q2

1995:Q1
96:Q2
97:Q3
98:Q4
2000:Q1
01:Q2
02:Q3
03:Q4
05:Q1
06:Q2
07:Q3
08:Q4
10:Q1
11:Q2
12:Q3
13:Q4
15:Q1
16:Q2

Source: IMF staff estimates.


Note: Higher quasi correlation values imply that the house price gaps of both countries (cities) are simultaneously above or below their respective historical averages.
See Annex 3.2 for methodology for quasi-correlation computation. Shaded areas correspond to US recessions.

102 International Monetary Fund | April 2018


CHAPTER 3 HOUSE PRICE SYNCHRONIZATION: WHAT ROLE FOR FINANCIAL FACTORS?

economies to global factors. Over time, the relative Figure 3.8. Relative Contribution of the Global Factor Varies
importance of the global factor has increased, but not across Regions
(Percent)
uniformly across advanced economies.15
Countries and cities also differ in how intercon- 1. Country Level
nected they are. The approach in Diebold and Yilmaz 40
(2014) offers one way to measure the interconnected-
ness in housing markets via an examination of quar- 30
terly house price growth correlations.16 This approach
shows that, after controlling for various global factors, 20
countries’ housing markets account differentially for
house price developments in other countries. More- 10
over, countries differ in the degree to which their
house prices can be attributed to other countries’ house 0
prices. For example, many large advanced economies’ dvanced Emerging Europe and sia North and
economies market other South
housing markets are closely interconnected, as sug- economies merica
gested by their central location and proximity to other
economies in a network map representing the links 2. City Level
40
in housing markets (Figure 3.9). In contrast, many
emerging market economies show weaker connectivity
with other countries. 30

Cities may have housing markets that are highly


interconnected even if their countries do not have 20

strong connectivity (Figure 3.10). Some cities lie


more at the core of the network, possibly reflecting 10
the deviation of house price dynamics in these cities
from the rest of their respective countries’ experi- 0
Europe and other sia North and South
ences. For instance, while at the country level Japan merica
is on the periphery of the network, at the city level,
Tokyo is more centrally located, closer to cities such Source: IMF staff estimates.
Note: The figure shows the share of the variation in house price growth from 2002
as London and Stockholm, perhaps reflecting the to 2016 explained by a common global factor in the dynamic factor model.
relative attractiveness of Tokyo to global investors “Europe and other” category comprises European countries, South frica, and
Israel. See nnexes 3.2 and 3.3 for methodology.
over other cities in Japan. Moreover, looking at cities,
it is apparent that many financial centers are more
centrally positioned and influential, suggesting that
city-level house price dynamics may also be transmit-
ted across borders. equities (see Chapter 2 of the April 2016 GFSR).17
The interconnectedness of housing markets has also Spillovers are particularly strong among advanced
increased over time (Figure 3.11). Consistent with economies, but the proportional increase is the largest
the rising trend in synchronicity discussed earlier, the for spillovers from advanced economies to emerging
network analysis shows that, on average, the share market economies and then from emerging market
of the house price variance in a country that can be economies to advanced economies, with average
accounted for by changes in another single country— interconnectedness increasing by about 60 percent and
henceforth, “spillovers”—increased from 1.4 percent 40 percent, respectively.
in 1990–2006 to 2.1 percent in 2007–16, which is
a notable increase and comparable to that seen for

15These results are available on request.


16Chapter 2 of the April 2016 GFSR explains the Diebold and 17Data limitations preclude omitting the global financial crisis

Yilmaz (2014) methodology that is applied here. period in this comparison.

International Monetary Fund | April 2018 103


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 3.9. Economies Differ in Their House Price Interconnectedness

ISR

AUS
FIN CHE
ZAF
SWE
DNK
BEL CAN
GBR COL

USA NZL
FRA
IRL
ESP
ITA HKG
PRT SGP
CHL
KOR HUN
RUS
JPN
NLD
IDN CHN
DEU MYS

AUT

THA

Source: IMF staff estimates.


Note: The figure is based on a vector autoregression of house price growth rates (quarter over quarter), controlling for global factors, from a sample covering 1990:Q1
to 2016:Q4. For methodology details, see Chapter 2 of the pril 2016 Global Financial Stability Report. Node si e is based on an economy s total outward spillovers.
Pink nodes represent advanced economies, and blue nodes represent emerging market economies. rrow thickness is based on link distribution. Only links above the
50th percentile are considered. The figure layout is based on the algorithm by Fruchterman and Reingold (1991) and plotted using the “qgraph” R package. Node
labels used in the figure are International Organi ation for Standardi ation (ISO) codes. Following Morgan Stanley Capital International markets classification criteria
and the IMF s World Economic Outlook country classification in 1990, the beginning of our sample, orea is classified as an emerging market economy.

Analyzing Contributors to House Price Countries with deeper financial links, as captured by
Synchronization their bilateral banking linkages, exhibit more synchroni-
What are the factors behind house price synchro- zation (Figure 3.12). This result, which is independent
nization? What is the role of financial factors? As of the comovement in output and other economic fun-
discussed earlier, the comovement in house prices damentals, is consistent with financial factors propagat-
may arise from synchronous business cycles or other ing local economic or financial developments between
nonfinancial economic fundamentals. To distinguish
among potential factors, the econometric framework
these issues. The analyses are performed at the country-pair level using
analyzes house price synchronization within country quarterly data from 1990 through 2016 for 40 countries (as well as
and city pairs over time.18 for major city pairs) using two synchronicity measures: (1) negative
value of the absolute difference in house price gaps (synch1), in which
a value closer to zero suggests that the differences in house price gaps
18The bilateral panel data approach removes hard-to-observe between two countries have declined; and (2) instantaneous quasi
country characteristics influencing synchronicity in house prices across correlation of house price gaps (QCORR), in which a higher value
countries or cities, such as strong cultural ties, similar mortgage market implies that the house price gaps of both countries are simultaneously
design, or similar tax treatment of housing capital gains. Thus, the above or below their respective historical averages. See Annex 3.2 for a
results discussed in this section are less likely to be confounded by technical discussion of the econometric model.

104 International Monetary Fund | April 2018


CHAPTER 3 HOUSE PRICE SYNCHRONIZATION: WHAT ROLE FOR FINANCIAL FACTORS?

Figure 3.10. Interconnectedness among Cities’ House Prices Varies

Hls

Syd HKG
Snt
Osl
Stc
Mmb
Lnd
Msc Mnl
Vnn

Tky Brl
Ack SGP
Trn Prs

Rom
Lim
Ams Brs
NYC
Jkr MxC

Dbl Dub
Sel
Mdr

Bgt

Shn

Source: IMF staff estimates.


Note: The figure is based on a vector autoregression of city-level house price growth rates (quarter over quarter), controlling for global factors, spanning 2004:Q1 to
2017:Q2. For methodology details, see Chapter 2 of the pril 2016 Global Financial Stability Report. See nnex Table 3.1.2, note 1, for city selection criteria,
conditional on data availability. Node si e is based on the city s total outward spillovers. Pink nodes represent advanced economies, and blue nodes represent
emerging market economies. rrow thickness is based on link distribution. Only links above the 66th percentile are considered. The figure layout is based on the
algorithm by Fruchterman and Reingold (1991) and plotted using the “qgraph” R package. ck = uckland; ms = msterdam; Bgt = Bogot ; Brl = Berlin;
Brs = Brussels; Dbl = Dublin; Dub = Dubai; H G = Hong ong S R; Hls = Helsinki; Jkr = Jakarta; Lim = Lima; Lnd = London; Mdr = Madrid; Mmb = Mumbai;
Mnl = Manila; Msc = Moscow; MxC = Mexico City; N C = New ork City; Osl = Oslo; Prs = Paris; Rom = Rome; Sel = Seoul; SGP = Singapore; Shn = Shanghai;
Snt = Santiago; Stc = Stockholm; Syd = Sydney; Tky = Tokyo; Trn = Toronto; nn = ienna. Following Morgan Stanley Capital International markets classification
criteria and the IMF s World Economic Outlook country classification in 1990, the beginning of our sample, orea (and thus Seoul) is classified as an emerging
market economy.

two countries.19 Moreover, the magnitude of the in transmitting shocks across countries (for example,
relationship is nearly as large as that between business see Allen and Gale 2000; Calvo and Mendoza 2000;
cycle synchronization and house price synchronization, Perri and Quadrini 2011). For instance, when a negative
suggesting that financial frictions, such as contagion and shock affects a country (or a set of countries), banks
sudden capital flow stops, may play an important role may retrench from activity abroad, triggering a credit
crunch in other countries, which might lead to deeper
19These conclusions are robust to the inclusion of monetary policy recessions and lower asset prices.20
synchronization and bilateral trade linkages as controls. While a
causal link from bilateral banking linkages to house price synchro- 20For example, during the global financial crisis, subsidiaries

nicity cannot be directly established from this analysis, the inclusion of foreign banks had to reduce their operations in eastern Europe
of country-pair fixed effects and multiple time-varying bilateral because of the subprime crisis and the new regulatory environment
determinants reduces the possibility of confounding factors. Also, (Chapter 2 of the April 2015 GFSR). However, the results discussed
reverse causality, in which house price synchronicity increases bilat- here and in the literature are limited in identifying the mechanisms
eral banking linkages, is difficult since diversification motives should by which bank retrenchment may occur, as data on bilateral banking
lead to a negative correlation between these two variables. flows do not differentiate by their intended use.

International Monetary Fund | April 2018 105


GLOBAL FINANCIAL STABILITY REPORT: A BUMPY ROAD AHEAD

Figure 3.11. Average Country-Level Housing Market Spillovers Figure 3.12. Bilateral Links between Countries Are
Have Increased Associated with House Price Synchronization
(Percentage points)
0.09
3.0 1990–2006 2007–16
0.08

2.5 0.07

0.06

Standard deviations
2.0
0.05

1.5 0.04

0.03
1.0
0.02

0.5 0.01

0.00
0.0 Business cycle Bilateral bank
Total Es to Es EMEs to Es to EMEs EMEs to Es synchroni ation of ij integration of ij
EMEs
Source: IMF staff estimates.
Source: IMF staff estimates. Note: Synchronicity is measured by the synch1 of gaps measure; see Annex 3.2
Note: The figure is based on a vector autoregression of country-level house price for computation methodology. Figure shows statistically significant standardi ed
growth rates (quarter over quarter), controlling for global factors, for two sample coefficients that are calculated using the coefficients in specification 4 in nnex
periods (1990–2006 and 2007–16). Spillovers are defined as the share of the Table 3.2.1 and their respective standard deviations, and presented in terms of
house price variance in a country that can be accounted for by changes in another standard deviations of the dependent variable; this specification also controls for
single country. For methodology, see Chapter 2 of the pril 2016 Global Financial global financial conditions (proxied through global liquidity) in addition to
Stability Report. Es = advanced economies; EMEs = emerging market economies. country-pair fixed effects and quadratic and linear time trends (standard errors are
clustered at multiway at time, country i, and country j ). The standard deviation for
business cycle synchroni ation is 0.0124 and 1.040 for bilateral bank integration.
See Country-Pair Analysis section of Annex 3.2 for further details. i = country 1
and j = country 2 in the country pair.
A country’s financial openness contributes to house
price synchronicity. Among advanced and emerging
market economies, countries with greater capital
account openness, as proxied by the Chinn-Ito index, Moreover, global financial factors play a role even after
are more exposed to global factors (Figure 3.13). accounting for the comovement in business cycles,
Moreover, among the advanced economies that can which points to an independent role for global factors
be observed for a longer period, the rise in exposure in accounting for house price synchronization.
to the global factor is observed in parallel with the Greater exchange rate flexibility appears to
increase in the comovement in equities documented dampen the importance of global financial condi-
here, in previous GFSRs, and elsewhere (Figure 3.14; tions (Figure 3.15). The impact of global liquidity is
Jordà and others 2017). Taken together, these results lower in countries with high exchange rate flexibility,
suggest that house price synchronization can be perhaps because countries with this feature may have
understood in the broad context of the asset price tools for dealing with imbalances resulting from
synchronization spurred by the evolution of finan- exposure to global financial conditions (Obstfeld,
cial openness. Ostry, and Qureshi 2017).21 For instance, in coun-
Past increases in global liquidity, as well as good tries where local currency loans prevail and exchange
market sentiment and loose global financial condi- rates are flexible, central banks may have a stronger
tions, are strongly associated with a higher short-term
comovement in house prices. These relationships apply
to the instantaneous quasi correlation in house price 21Nominal rigidities may be less relevant in countries with
gaps when looking within country pairs in advanced exchange rate flexibility, dampening the role for global finan-
and emerging market economies (Figure 3.15). cial conditions.

106 International Monetary Fund | April 2018


CHAPTER 3 HOUSE PRICE SYNCHRONIZATION: WHAT ROLE FOR FINANCIAL FACTORS?

Figure 3.13. Greater Financial Openness Is Associated with Figure 3.14. On Average, the Global Factor for House Prices
Higher House Price Synchronization Has Increased along with That for Equities
(Window = 15 years; percent)
50
80
45 House prices Equity prices

40 70
House price synchronicity (percent)

35 60

30
50
25
40
20

15 30

10 20

5
10
0
Less than 0 0 to 2 More than 2 0

1986:Q1
88:Q1
90:Q1
92:Q1
94:Q1
96:Q1
98:Q1
2000:Q1
02:Q1
04:Q1
06:Q1
08:Q1
10:Q1
12:Q1
14:Q1
16:Q1
Capital account openness

Sources: Chinn and Ito (2006); and IMF staff estimates.


Note: Synchronicity is measured by the share of the variation in house price
growth from 2002 to 2016 explained by a common global factor in the dynamic Source: IMF staff estimates.
factor model. See Annexes 3.2 and 3.3 for more details on methodologies. Note: The figure shows the rolling estimation with a 15-year window for the share
of the variation in house and equity price growth explained by a common global
factor in the dynamic factor model. See nnexes 3.2 and 3.3 for methodology.

influence on short-term interest rates and thus on


financing conditions.22 analysis of housing market segments within the United
The contribution of global financial conditions States suggests that higher-priced homes are more
to house price synchronization in cities is somewhat responsive to changes in house prices of non-US cities.
larger than for countries (Figure 3.15). If large cities In particular, house prices in non-US cities charac-
attract global investors, house price comovement in terized as destinations for global investors (such as
cities may be particularly responsive to global financial London) exert more influence on higher-priced homes
conditions. This seems to be the case. Notably, cities in the largest US cities, as would be the case if demand
in advanced economies show greater responsiveness to from global investors were exerting upward pressure on
global financial conditions, using global liquidity as house prices in US and non-US markets.
a proxy. These cities are, on average, more exposed to The participation of global investors in local real
the global factor (Figure 3.8), but also may face con- estate markets may contribute to the behavior of
strained housing supply such that changes in housing housing returns, in addition to contributing to syn-
demand driven by global liquidity conditions may have chronized house prices across countries. As Box 3.2
a more pronounced and coordinated impact. discusses, as wi