The Regulation of Systemically Relevant Banks

Finanzwirtschaft, Banken und Bankmanagement I Finance, Banks and Bank Management

Sebastian C. Moenninghoff

The Regulation of Systemically Relevant Banks How Governments Should Manage Their Exposure to Banking System Risk

Finanzwirtschaft, Banken und Bankmanagement | Finance, Banks and Bank Management Reihe herausgegeben von A. Wieandt, Königstein, Deutschland S. C. Moenninghoff, Vallendar, Deutschland

Die Reihe präsentiert Forschungsbeiträge aus den Bereichen Finanzwirtschaft, Banken und Bankmanagement, die sich durch hohe wissenschaftliche Qualität und Praxisbezug auszeichnen. Sie richtet sich an Akademiker und Praktiker. The series presents research from the fields of finance, banking and bank management, which are characterized by high scientific quality and practical relevance. It is aimed at academics and practitioners.

More information about this series at http://www.springer.com/series/16023

Sebastian C. Moenninghoff

The Regulation of Systemically Relevant Banks How Governments Should Manage Their Exposure to Banking System Risk With a foreword by Hon.-Prof. Dr. Axel Wieandt

Sebastian C. Moenninghoff Vallendar, Germany Dissertation WHU – Otto Beisheim School of Management, Vallendar, Germany, 2017

ISSN 2524-6429 ISSN 2524-6437  (electronic) Finanzwirtschaft, Banken und Bankmanagement | Finance, Banks and Bank Management ISBN 978-3-658-23811-7  (eBook) ISBN 978-3-658-23810-0 https://doi.org/10.1007/978-3-658-23811-7 Library of Congress Control Number: 2018957569 Springer Gabler © Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2018 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer Gabler imprint is published by the registered company Springer Fachmedien Wiesbaden GmbH part of Springer Nature The registered company address is: Abraham-Lincoln-Str. 46, 65189 Wiesbaden, Germany

Foreword

Sebastian Moenninghoff´s thesis is dedicated to the perennial challenge that Too-Big-to-Fail (TBTF) in banking poses for financial, economic and political stability – a field of research that has received significant and increasing attention by both academics and regulators since the financial crisis. TBTF implicit guarantees became explicit in the great financial crisis of 2007-2009, with governments in the U.S. and Europe coming to the rescue of the respective banking systems with extensive guarantees, comprehensive bad-bank schemes and asset purchase programs, as well as significant recapitalizations of banks in trouble. In response to the crisis and under pressure from taxpayers and voters, politicians declared their intentions to abolish TBTF in banking. Since then, the competent regulatory bodies have developed and the respective legislative bodies have enacted a specific regulation targeted at systemically relevant banks and other financial institutions. The thesis consists of three significant contributions to the research on TBTF and the regulation of systemically relevant banks, • • •

a review of the empirical literature on TBTF in chapter 2, a conceptual/theoretical piece that allows for the discussion of regulatory policy options in the context of an overall sovereign portfolio perspective in chapter 3, and an empirical event-study examining the impact of the new regulation of systemically relevant banks on stock returns in chapter 4.

The conclusions drawn are clear and compelling: •

•

•

TBTF is a wide-spread phenomenon which creates significant (increasing in times of asset price volatility) sovereign (and therefore tax-payer) exposure and concurrent wealth distribution effects, can lead to competitive distortions in banking markets, and, might also impact bank behavior, notably risk taking behavior. There is no first-best regulatory solution to TBTF and the optimal policy design depends critically on the (ex ante) risk appetite of any given society and needs to take into account the implications for economic growth. Any kind of regulation that is specifically targeted at the largest, most systemically relevant institutions has to take into account potentially perverse incentives created by an official designation as systemically relevant institution.

Chapter 2, “Consequences of Government Guarantees for Banks – A Survey of the TBTF Doctrine”, provides a comprehensive literature review including the most recent contributions to the field. Starting point of the literature review is the concept of explicit or implicit

VI

Foreword

government guarantees for banks. The discussion is organized according to a comprehensive framework that distinguishes between the primary and secondary consequences of government guarantees. Primary consequences are the resulting government exposure to the banking sector and related wealth transfer / subsidies. Secondary consequences are competitive distortions and the influence on bank risk-taking, i.e. moral hazard. This comprehensive framework is new and original, and highlights the importance to distinguish between consequences and related different methodological approaches when interpreting findings on TBTF. This is an important contribution considering the extensive research body in this field and the variance of results between different publications. The framework forms the basis for a more granular discussion of how to measure the impact of TBTF on sovereigns and banks, and how to regulate systemically relevant banks going forward. The significant size of government exposure levels and competitive distortions revealed by several studies during times of heightened market volatility is also presented graphically. Measurement challenges are discussed in depth. A key conclusion from this chapter is that governments need to measure (and potentially) report their exposure to the banking system and adjust regulation accordingly. The chapter also identifies challenges in measuring the link between TBTF guarantees and bank (risk-taking) behavior. In summary, this chapter provides a state-of the art review of the existing literature in the rapidly evolving field of TBTF, provides a fundamental framework to discuss and categorize the extensive research body on TBTF going forward, and sets the agenda for future research in the field. Chapter 3, “Government Guarantees and Banking System Risk – A Regulatory Framework from an Exposure Perspective” develops a government exposure perspective on banking system risk and bank regulation. This perspective allows for a comprehensive evaluation of the full range of regulatory banking options. The approach to combine standard credit risk theory and portfolio theory in a single framework for bank regulation is novel and highly innovative and allows for a concise conceptual discussion and quantitative comparison of a broad range of regulatory options, ranging from free and narrow banking regimes (implying zero government exposure) over regulations such as minimum capital and liquidity requirements, taxes, bail-in regimes, etc. (implying limited exposure) to a fully nationalized banking system (implying full government exposure). The chapter first develops the fundamentals of credit risk exposure in bank regulation. Building on a macro-financial contingent claims approach of Bodie and Brière it then derives a stylized simulation model to highlight and discuss the major trade-offs in regulating (systemically) relevant banks. The optimal solution depends on the governments´ and henceforth societies´ risk preferences. The recent introduction of bank resolution regimes can be interpreted as a shift in focus of regulation from reducing the probability of default to also reducing the exposure at default in case of a banking crisis. All in all, this chapter elevates the discussion of bank regulation to the next level, from a mere “call for more regulation” to a more nuanced discussion of risk preferences and tradeoffs. It is important to highlight that the discussion in this chapter also includes a sub-chapter on limitations and avenues for future research.

Foreword

VII

Chapter 4, “Empirical Evidence from the New International Regulation Dealing with Global Systemically Important Banks” builds on the paper “The perennial challenge to counter Too-Big-to-Fail in banking: Empirical evidence from the new international regulation dealing with Globally Systemically Important Banks” co-authored with Steven Ongena, and myself, which has been published in the Journal of Banking and Finance. This paper estimates how the designation of specific banks as Globally Systemically Important Banks (G-SIBs) and announcements about the supervision and regulation of G-SIBs affected the relative stock returns of those banking organizations. The paper finds that, in general, announcements about new regulations depressed G-SIB returns relative to returns of other banks, but that announcements identifying specific banks as G-SIBs (or likely G-SIBs) caused the returns of those banks to increase relative to those of other banks. According to the paper, the empirical evidence indicates that any tendency of heightened supervision and regulation to reduce funding advantages for G-SIBs was at least partly offset by announcements that identify G-SIBs by name because those announcements reduced or eliminated any doubts about which banks regulators view as TBTF. The paper has been widely quoted in current research and policy papers, clearly demonstrating the uniqueness of the contribution, both in terms of the rigor of the empirical event study as well as the relevance of the policy conclusions. It has been a great privilege and a tremendous pleasure to supervise Sebastian Moenninghoff´s thesis on the regulation of systemically relevant banks. I am looking forward to having many more fruitful exchanges and discussions with him. Königstein i. Ts., March 2018 Axel Wieandt

Preface

This dissertation was written during my doctoral studies at WHU - Otto Beisheim School of Management. I wish to thank Professor Markus Rudolf, Professor Lutz Johanning and Professor Axel Wieandt for valuable comments and discussions. I am deeply indebted to Professor Axel Wieandt for continuous helpful guidance, important advice and constant encouragement and support during this work. I want to express my profound appreciation to Professor Steven Ongena for valuable guidance on the empirical analysis in Chapter 4. I would like to thank Professor Günter Franke, Professor Hans-Helmut Kotz and Professor Jan Pieter Krahnen for fruitful discussions at the beginning of my doctoral studies, and Professor Douglas Diamond and Professor Randy Kroszner for valuable discussions during my time at the University of Chicago Booth School of Business. I would like to thank my parents and my wife Annette for their support. Sebastian C. Moenninghoff

Table of Contents

1. Introduction .................................................................................................................... 1 2. Consequences of Government Guarantees for Banks – A Survey of the TBTF Doctrine ........................................................................................................................... 3 2.1 Introduction .............................................................................................................. 3 2.2 Surveys of Government Guarantees for Banks ........................................................ 5 2.3 TBTF as a Consequence of Government Guarantees ............................................... 6 2.3.1 Consequences of government guarantees ....................................................... 6 2.3.2 The logic of the TBTF doctrine ...................................................................... 8 2.3.3 Alternative theories in the context of government guarantees for banks ........ 8 2.3.4 Empirical approaches to measuring the prevalence of TBTF ....................... 10 2.4 Government Exposure and Subsidies ..................................................................... 11 2.4.1 Contingent claims approach and absolute subsidy estimates........................ 11 2.4.2 Funding advantages based on contingent claims approach and rating-implied spreads .................................................................................. 13 2.4.3 Costs of past rescue measures....................................................................... 15 2.4.4 Summary of empirical evidence of government exposure and subsidies ..... 16 2.5 Competitive Distortions from Government Guarantees ......................................... 17 2.5.1 Dimensions of competitive distortions ......................................................... 17 2.5.2 Empirical approaches to measure guarantee-return relationships ................. 19 2.5.3 Empirical evidence of competitive distortions ............................................. 21 2.5.3.1 Competitive distortions by individual institution systemic relevance .......................................................................................... 21 2.5.3.2 Competitive distortions by scope of activities covered by guarantees......................................................................................... 30 2.5.3.3 Competitive distortions by geography ............................................. 31 2.5.4 Summary of empirical evidence of competitive distortions ......................... 33 2.6 Government Guarantees and Risk Taking.............................................................. 34 2.6.1 The concept of moral hazard in banking ...................................................... 34 2.6.2 Empirical approaches based on guarantee-risk relationships ....................... 36 2.6.3 Empirical findings based on guarantee-risk relationships ............................ 36 2.6.4 Summary of empirical evidence of bank risk taking based on guarantee-risk relationships .......................................................................... 39 2.6.5 Empirical approaches based on risk-return relationships ............................. 41 2.6.6 Empirical findings based on risk-return relationships .................................. 42 2.6.7 Summary of empirical evidence of bank risk taking based on risk-return relationships .................................................................................................. 44 2.7 Conclusion ............................................................................................................. 45

XII

Table of Contents

3. Government Guarantees and Banking System Risk – A Regulatory Framework from an Exposure Perspective ........................................ 47 3.1 Introduction ............................................................................................................ 47 3.2 Banking System Exposure from a Credit Risk Perspective ................................... 49 3.2.1 Fundamentals of an exposure perspective for banking system risk .............. 49 3.2.2 Structural credit risk modeling in regulatory capital determination ............. 53 3.2.3 Application of structural credit risk models to government guarantees for banks ............................................................................................................. 54 3.2.4 Credit risk components of banking system exposure ................................... 56 3.2.4.1 Probability of distress ....................................................................... 57 3.2.4.2 Exposure at distress .......................................................................... 57 3.2.4.3 Loss given distress ........................................................................... 58 3.2.5 An exposure-based framework of principle policy choices .......................... 63 3.3 Banking System Exposure from a Sovereign Portfolio Perspective ...................... 65 3.3.1 Introduction to sovereign portfolio management .......................................... 65 3.3.2 Macroeconomic tradeoffs implied by regulatory policy choices .................. 66 3.3.3 Banking system exposure management from a portfolio perspective .......... 68 3.4 Regulatory Policy Options and Their Economic Tradeoffs ................................... 71 3.4.1 Management of probability of distress ......................................................... 71 3.4.1.1 Zero exposure: Narrow banking with all-equity financed banks...... 71 3.4.1.2 Limited exposure: Minimum capital and liquidity requirements ..... 73 3.4.2 Management of loss given distress ............................................................... 77 3.4.2.1 Zero exposure: Narrow banking with assets restricted to government securities....................................................................... 77 3.4.2.2 Limited exposure: Structural restrictions ......................................... 81 3.4.2.3 Limited exposure: Pigovian tax........................................................ 82 3.4.2.4 Limited exposure: Resolution powers and wind-down authorities .. 84 3.4.3 Management of exposure at distress ............................................................. 86 3.4.3.1 Zero exposure: Free banking ............................................................ 86 3.4.3.2 Limited exposure: Bail-inable claims............................................... 87 3.4.3.3 Full exposure: Nationalization of the banking system ..................... 90 3.4.4 Growth-stability tradeoff .............................................................................. 91 3.4.5 Exposure factor interaction and interconnections ......................................... 94 3.5 Discussion and Conclusion .................................................................................... 96 3.5.1 Institutional design, financial system structure and international policy ...... 96 3.5.2 The new regulation dealing with Global Systemically Important Banks ..... 99 3.5.3 Results, limitations and future research ...................................................... 100 3.5.4 Conclusion .................................................................................................. 102 4. Empirical Evidence from the New International Regulation Dealing with Global Systemically Important Banks ...................................................................... 105 4.1 Introduction .......................................................................................................... 105 4.2 G-SIB Regulation and Hypotheses ...................................................................... 107 4.2.1 Explicit and implicit government guarantees ............................................. 107 4.2.2 New G-SIB regulation ................................................................................ 108 4.2.3 Hypotheses ................................................................................................. 109

Table of Contents

XIII

4.3 Data and Methodology ......................................................................................... 110 4.3.1 Sample ........................................................................................................ 110 4.3.1.1 Sample compilation ........................................................................ 110 4.3.1.2 Sub-sample definitions ................................................................... 111 4.3.1.3 G-SIB designation .......................................................................... 111 4.3.2 Event dates.................................................................................................. 114 4.3.3 Methodology............................................................................................... 115 4.3.3.1 Abnormal return calculation........................................................... 115 4.3.3.2 Test statistics .................................................................................. 117 4.4 Empirical Results ................................................................................................. 118 4.4.1 Overall results ............................................................................................. 119 4.4.2 Regulatory announcements......................................................................... 120 4.4.3 Designation announcements ....................................................................... 123 4.4.4 Cross-sectional analysis of G-SIB returns .................................................. 125 4.5 Conclusion ........................................................................................................... 130 5. Conclusion ................................................................................................................... 133 Bibliography ..................................................................................................................... 137 List of Appendices ............................................................................................................ 151 Appendix ........................................................................................................................... 153

List of Abbreviations

α BCBS BHC BV β CDS DFX DLC DL EADGov ELB ELGov EFSF ESM EXTE FA FAB FANB FDICIA FIA FIBE FIL FL FSB G20 GB G-SIB G-SIFI GSS INTB INTE LGDGov LOLR LTCM MB MREL NFA NPV

Fraction of Sovereign Assets Dedicated to Financial Assets Basel Committee on Banking Supervision Bank Holding Company Book Value Total Sovereign Liabilities Dedicated to Foreign Debt Credit Default Swap Debt Issued in Foreign Currency Debt Issued in Local Currency Domestic Liabilities Government Exposure to a Bank at Time of Distress Bank Expected Loss Government Expected Loss from Exposure to a Bank European Financial Stability Facility European Stabilization Mechanism Negative Value of Externalities Due to Government Guarantees Financial Assets Bank Financial Assets Non-Bank Financial Assets Federal Deposit Insurance Corporation Improvement Act Fiscal Assets Financial Intermediation Benefits from Asset Transformation Fiscal Liabilities Foreign Liabilities Financial Stability Board Group of Twenty Guarantees to the Banking System Global Systemically Important Bank Global Systemically Important Financial Institution Global Sovereign Surplus Negative Value of Internalities From Government Ownership in Banks Negative Value of Internalities From Government Involvement in Financial Intermediation Loss Attributable to Government in Case of Distress Lender of Last Resort Long-Term Capital Management Monetary Base Minimum Requirements for Own Funds and Eligible Liabilities Net Fiscal Assets Net Present Value

XVI η OECD P ρ PDB PV r RecovMS RecovRS RecovWD Rit Rmt SIB SIFI SLB SLGov TARP TBTF TLAC ULB ULGov U γ

List of Abbreviations

Degree of Government Ownership in Equity of Domestic Banking System Organisation for Economic Co-operation and Development Probability Sovereign’s Relative Risk Aversion Probability of a Bank’s Distress Present Value Return Recoveries from Merger or Sale of a Bank Recoveries from Restructuring a Bank Recoveries from Winding Down a Bank Bank’s Period-t Return Market Portfolio Period-t Return Systemically Important Bank Systemically Important Financial Institution Bank Stress Loss Government Stress Loss Troubled Asset Relief Program Too-Big-to-Fail Total Loss Absorbing Capacity Bank Unexpected Loss Government Unexpected Loss Sovereign’s Utility Level of Government Guarantee for the Banking Sector

List of Figures

Figure 1: Figure 2: Figure 3: Figure 4: Figure 5: Figure 6: Figure 7:

Consequences of government guarantees .............................................................. 7 Empirical approaches to measure consequences of government guarantees ....... 10 Overview of estimated guarantee values in basis points over time ..................... 15 Implied guarantee policies based on industry and firm type covered.................. 18 Funding advantages for systemically relevant banks based on bonds ................. 23 Funding advantages for systemically relevant banks based on CDS ................... 24 Funding advantages for systemically relevant banks based on aggregate funding costs ....................................................................................................... 26 Figure 8: Funding advantages for systemically relevant banks based on deposits ............. 28 Figure 9: Funding advantages for banks versus non-banks and non-financial firms .......... 31 Figure 10: Expected, unexpected and stress losses of banks and the government ................ 51 Figure 11: Loss given distress, recovery value and exposure at distress .............................. 59 Figure 12: Banking system exposure within a sovereign’s overall portfolio ........................ 66 Figure 13: Capital requirements, economic value and risk-adjusted return .......................... 73 Figure 14: Asset-related restrictions and expected volatility levels ...................................... 79 Figure 15: Pigovian tax, taxation efficiency and capital requirements ................................. 84 Figure 16: Degree of government guarantees and relative volatility reduction compared to free banking system......................................................................................... 89 Figure 17: Risk-adjusted returns for policy choices and degree of risk aversion ................. 92 Figure 18: Risk-adjusted returns with high financial intermediation benefits and high government inefficiencies ................................................................................... 93 Figure 19: Exposure levers addressed by new G-SIB regulation........................................ 100 Figure 20: Stock prices of G-SIBs, Non-G-SIBs, and Other Banks, 2008 to 2011 ............ 118 Figure 21: Stock prices of G-SIBs, Non-G-SIBs, and Other Banks, 2003 to 2011 ............ 119

List of Tables

Table 1: Table 2: Table 3: Table 4: Table 5: Table 6: Table 7: Table 8: Table 9:

Empirical studies on guarantee values ................................................................ 13 Empirical approaches to measure the impact of guarantees on returns ............... 20 Empirical approaches to measure the impact of guarantees on risk .................... 36 Empirical approaches to measure risk-return distortions .................................... 42 Principle policy options to manage exposure to the banking system .................. 64 Overview of G-SIB sub-sample ........................................................................ 112 Overview of event dates .................................................................................... 114 Abnormal returns by event ................................................................................ 120 Regression results of cross-section of G-SIBs .................................................. 128

1.

Introduction

In response to the recent financial crisis, leading policymakers declared their intent to ‘take concrete steps to move forward with tough, new financial regulations so that crises like this can never happen again’, and ‘that banks can never again blackmail states and governments’. 1 Since then, the responsible international regulatory bodies have developed a new regulation specifically for systemically relevant banks to counter the Too-Big-to-Fail (TBTF) doctrine, a concept that suggests that government guarantees for banks produce incentives that can lead to competitive distortions, increased bank risk taking and, as a result, resource misallocation and an increased level of financial system risk. In light of these developments, this thesis attempts to provide answers to three overarching research questions: First, why should systemically relevant banks be regulated? Second, what are relevant policy options and how should systemically relevant banks be regulated? Third, does the new regulation for Global Systemically Important Banks (G-SIBs) succeed in limiting TBTF? The structure of this thesis follows these three questions. Chapter 2, “Consequences of government guarantees for banks – a survey of the TBTF doctrine” responds to the first question by examining the consequences of government guarantees in order to determine the need for regulating systemically relevant banks. We review the extensive body of literature on TBTF in banking and provide an overview of both empirical approaches and findings on the status of TBTF, with an emphasis on post-crisis contributions. We find that scholars have estimated guarantee values to range from more than a trillion US dollar annually based on contingent claims approaches to positive net present values based on past rescue measures, which emphasizes the need for governments to actively measure and manage their exposure to the banking system, considering risk and returns. Chapter 3, “Government guarantees and banking system risk – a regulatory framework from an exposure perspective” responds to the second question on relevant policy options and adequate regulation. We develop an exposure view on banking system risk and bank regulation, providing an analytic framework to evaluate regulatory policy options in order to achieve a desired level of government exposure to banking system risk. As part of this theoretical discussion, we present the results of a simplified model to demonstrate economic tradeoffs between different regulatory policy options and analyze the new regulation on systemically relevant banks through the lens of our framework. This contribution to our knowledge is the 1

White House press release, September 24, 2009; German federal government press release, September 25, 2009.

© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2018 S. C. Moenninghoff, The Regulation of Systemically Relevant Banks, Finanzwirtschaft, Banken und Bankmanagement  Finance, Banks and Bank Management, https://doi.org/10.1007/978-3-658-23811-7_1

2

1.

Introduction

first comprehensive approach to categorize and discuss the full range of major policy options for regulating banks within a single framework and to provide a quantitative methodology to compare the relative attractiveness of different policy choices. Overall, our stylized model suggests that the choice of an optimal regulatory policy mix depends on risk and return preferences of a society, and an economy’s institutional and cultural setting which impacts the tradeoffs implied by each regulation. Chapter 4, “Empirical Evidence from the New International Regulation Dealing with Global Systemically Important Banks”, responds to the third question, providing an empirical evaluation of the G-SIB regulation developed by the Financial Stability Board (FSB) and the Basel Committee on Banking Supervision (BCBS) based on an event study. We find that the new regulation negatively affects G-SIBs, yet that at the same time the official designation of banks as G-SIBs has a partly offsetting impact, suggesting that even though the individual components of the regulation have been effective, revealing the identities of G-SIBs eliminated ambiguity about the presence of government guarantees, and thereby may have run counter to the regulators’ intent to contain the effects of TBTF.

2.

Consequences of Government Guarantees for Banks – A Survey of the TBTF Doctrine

2.1

Introduction 2

The recent financial crisis and subsequent regulatory reform led to a broad discussion of the TBTF doctrine. Made popular by a 1984 statement by the U.S. Comptroller of the Currency in relation to the failure of U.S. bank Continental Illinois, the term TBTF originally referred to the size of large U.S. money center banks, which were considered to cause financial instability upon unorderly failure and thus mandated government intervention in case of crises. 3 Since then, the TBTF concept has been extended to include systemically relevant banks more broadly, and the academic discourse now reflects various streams of research including resulting government exposure, competition implications and moral hazard effects. Moral hazard, as we will further discuss in Section 2.6.1, refers to a change in bank risk taking behavior beyond the intended level of risk implied by a specific level of government guarantees and regulation, reflecting the fact that a bank’s systemic relevance can lead to expectations of government support in case of crisis. Moral hazard implies a lack or distortion of market discipline, a concept we will further introduce in Section 2.3.3 and which refers to a functioning market mechanism that influences economic decisions such as capital allocation or investments and thus determines economic outcomes. As part of our attempt to answer the overarching question of why systemically relevant banks should be regulated, several issues make it worth revisiting the underlying concept of TBTF and motivate this survey of empirical research. First, over the past years, academics, regulators and industry representatives have contributed extensive additional empirical research to advance the understanding of systemically relevant banks and TBTF and to derive relevant policy implications. Following early examinations of risk-return relationships based on individual bank risk factors since the 1970s, scholars increasingly have begun to examine guarantee-return and guarantee-risk relationships in the 2000s, with a growing variety of approaches since the financial crisis. Scholars have extended risk measures to include non2 3

This chapter is based on the unpublished working paper ”Consequences of Government Guarantees for Banks – A Survey of the TBTF Doctrine” co-authored by Axel Wieandt (see Moenninghoff and Wieandt (2017a)). Compare Committee on Banking, Finance and Urban Affairs (1984, p. 300). Responding to the Committee Chairman’s question whether he could ever foresee one of the 11 multinational money center banks fail, C. Todd Conover, then Comptroller of the Currency of the U.S. Treasury, responded: “I admit that we don't have a way right now. And so, since we don't have a way, your premise appears to be correct at the moment.”

© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2018 S. C. Moenninghoff, The Regulation of Systemically Relevant Banks, Finanzwirtschaft, Banken und Bankmanagement  Finance, Banks and Bank Management, https://doi.org/10.1007/978-3-658-23811-7_2

4

2.

Consequences of Government Guarantees for Banks

accounting indicators for individual bank risk and system-wide risk factors, while at the same time broadening the guarantee factors used, including institution-specific characteristics, official designations, government ownership in banks and political factors. Likewise, scholars have begun to apply the contingent-claims based approach in the context of government guarantees in the early 2000s. Developed in the 1980s and 1990s as part of an evolving credit risk debate, this approach had been used before to examine capital regulations and deposit insurance, and has been increasingly applied to quantify government exposure to banking system risk since the financial crisis. Second, this widespread discussion and the increasing variety of empirical approaches potentially imply room for ambiguity surrounding the concept of TBTF and how new empirical findings should be interpreted. Oftentimes, empirical studies necessarily focus on a specific aspect of the TBTF doctrine, while attempting to derive broader policy implications. Third, relatively few surveys examine empirical evidence of TBTF post-crisis. Recent post-crisis contributions mostly focus on a specific type of study (such as funding cost differentials) or embed select empirical evidence in the context of a more general account of TBTF. In contrast to existing surveys of empirical research on TBTF (which we review in Section 2.2), we structure our discussion along a comprehensive framework highlighting the three main consequences of government guarantees for banks: government exposure to banking system risk and implied wealth transfers, competitive distortions in the financial sector, and bank risk taking and potential moral hazard. We then survey the existing research to derive conclusions regarding each of these three consequences. In doing so, we categorize the large number of different empirical approaches used by researchers into four major categories: contingent claims-based approaches, examinations of banks’ risk-return characteristics, guarantee-return characteristics and guarantee-risk characteristics. 4 Our analysis emphasizes the need to differentiate among the major consequences of government guarantees – government risk exposure, competitive distortions and moral hazard. We find that studies on government exposure and related wealth transfers indicate significant exposure levels at least during times of heightened market volatility, suggesting that policymakers should take an active approach to managing this exposure, which we develop in Chapter 3. Studies attempting to examine the impact of government guarantees via competitive dynamics – or funding spreads – provide mixed evidence across refinancing instruments. However, most studies observe a recent spread compression after a significant spread widening during the crisis period, raising the question as to what factors are driving these developments, and, more generally, suggesting a more specific examination of the observed competitive dynamics. Studies focused on guarantee-risk and risk-return relationships, while providing partly mixed evidence, highlight the difficulty of empirically measuring the link between government guarantees and moral hazard, given that banks appear to trade off 4

Historical costs of rescue measures are an additional approach to quantify potential guarantees which we will briefly discuss.

2.2

Surveys of Government Guarantees for Banks

5

different types of risk and that the interaction between individual bank risk and banking system risk appears to be complex. Studies examining moral hazard should carefully select guarantee indicators to not only ensure predictive strength for actual guarantee levels but also correctly interpret potential underlying motivations for the provision of these guarantees. The remainder of this Chapter is organized as follows. In Section 2.2, we provide a brief overview of existing literature reviews on TBTF. In Section 2.3, we develop our analytical framework of consequences of government guarantees. We briefly revisit the underlying logic of the TBTF doctrine as well as alternate conceptual approaches on how government guarantees impact banks, and we identify the main empirical approaches used to examine TBTF. In Sections 2.4, 2.5 and 2.6, we review the existing empirical research on TBTF along our framework. In Section 2.7 we summarize our major conclusions. 2.2

Surveys of Government Guarantees for Banks

There have been a number of literature surveys of government guarantees for banks, TBTF and market discipline in banking more broadly, which we will briefly review in this section. Following the U.S. savings and loans crisis, Gilbert (1990) provides an overview of empirical studies on market discipline distinguishing between different refinancing instruments. While the author concludes that studies on subordinated debt and deposits have mixed results, most of the studies examining bank stock prices appear to confirm the prevalence of market discipline. Moyer and Lamy (1992) provide a general overview of TBTF including the history, underlying rationale and consequences, referencing evidence of bank risk-seeking and a lack of market discipline before and during the U.S. savings and loan crisis. Kwast et al. (1999) survey studies that examine market discipline in banking based on subordinated debt. While the authors find that most of the surveyed research suggests that subordinated debt prices do reflect bank risk, they conclude that investors were likely relying on implicit government guarantees during the periods that subordinated debt prices were unrelated to risk. De Ceuster and Masschelein (2003) survey literature to determine to which degree different refinancing instruments are suitable to exert market discipline. The authors conclude that there is some evidence of risk sensitivity in subordinated debt markets and that equity holders do not appear to exploit the option-features offered by bank equity in the context of guarantees. Flannery and Nikolova (2004) document progress made in analyzing market discipline, highlighting the expansion of refinancing instruments that have been examined by scholars and the consideration of funding quantities in addition to prices. Gropp (2004) surveys studies on market discipline for banks in the European Union and finds mixed evidence. Despite substantial potential for market discipline provided by the sizable subordinated debt market, the authors conclude that the effectiveness of market discipline remains weak. Rochet

6

2.

Consequences of Government Guarantees for Banks

(2004) discusses the use of market information in banking, concluding that complementing the information set of banking supervisors with market information is useful. Stern and Feldman (2004) provide a general account on TBTF and review empirical evidence in the context of policy actions, bank mergers, credit rating agency analyses, official designations, rescue measures and funding cost differentials. DeYoung et al. (2009) examine empirical studies on bank mergers and conclude that banks were motivated to enter mergers in order to obtain or strengthen a TBTF status. Post-crisis, Noss and Sowerbutts (2012) compare approaches and findings to quantify government exposure and subsidy estimations based on contingent claims approaches and rating-implied credit spreads, highlighting relative strengths and weaknesses of each approach. Strahan (2013) discusses empirical evidence of TBTF distortions such as leverage, risk taking, growth incentives and competitive advantages. Reviewing studies focused on debt funding cost differentials, Kroszner (2013) points out the importance of different methodologies, samples, time periods and refinancing instruments. 2.3

TBTF as a Consequence of Government Guarantees

2.3.1

Consequences of government guarantees

The impact of government guarantees on banks can be distinguished by their primary and secondary consequences for the government, the banking system and the wider economy. In this study, primary consequences refers to the immediate risk exposure a government assumes by guaranteeing bank liabilities and the implied wealth transfer from taxpayers to bank investors, which becomes visible once the guarantees are invoked in times of stress. The risk exposure and implied wealth transfer of the guarantee value take place the moment the guarantees are granted, whether implicitly or explicitly. The conceptual question of why governments may decide to accept exposure to the banking system and which level of exposure should be desirable has been largely excluded from the current discussion. The postcrisis mandate for regulators to eliminate TBTF implies a zero exposure target. While this conceptual aspect does not impact empirical exposure quantifications, it bears important implications for the interpretation of any exposure observation, as we will discuss further in Section 2.4.4. We understand secondary consequences to be changes in economic outcome due to the presence of government guarantees for banks. These consequences unfold via competitive distortions within the banking system and a change in banks’ risk-taking behavior. The distribution of guarantees across financial institutions, individual components of the financial system and different geographies as well as banks’ risk-taking behavior in response to government guarantees impact overall economic resource allocation, the level of systemic

2.3

TBTF as a Consequence of Government Guarantees

7

risk in the financial system and the financial system structure (compare Figure 1). 5 Different theories exist to explain how government guarantees influence competition and bank risk taking. The TBTF doctrine argues that government guarantees imply competitive distortions in favor of guaranteed banks and increased risk taking by these banks as a result of moral hazard. Other theories provide alternative predictions for competition and risk-taking behavior, such as the charter value theory or principal-agent arguments relating to the relationship between bank shareholders and bank managers. Scholars have used these different theories and concepts to empirically examine the impact of government guarantees. In the remainder of Section 2.3, we will briefly introduce these concepts and provide a classification of related empirical approaches, which we will use to review empirical findings in Sections 2.4, 2.5 and 2.6.

Figure 1:

5

Consequences of government guarantees

In a dynamic setting, secondary consequences and government exposure will interact. The economic outcome has implications for the government’s economic strength and implied guarantee values, which in turn will influence investor and bank behavior and related secondary consequences.

8

2.

2.3.2

Consequences of Government Guarantees for Banks

The logic of the TBTF doctrine

According to the TBTF doctrine, government guarantees for bank debt lower a bank’s refinancing costs and thus produce price and volume signals that induce a bank’s management to increase bank risk taking in order to enhance the bank’s return on equity. The increased level of risk can be achieved both via the risk profile of the assets as well as by altering the asset-liability structure in terms of leverage, and maturity transformation or liquidity transformation. In addition, incentives for banks to become systemically relevant can imply increased organizational complexity and a resulting higher operational risk. Bank managers acting as agents to bank shareholders increase risk taking to optimize shareholder returns. With limited liability, bank shareholders benefit from a greater upside while participating only with the amount of invested equity capital in the case of downside. 6 As a consequence, the risk-return relationship is fundamentally altered from the state without government guarantees. The consequences for the economic environment include competitive distortions due to refinancing advantages; a misallocation of resources due to increased risk taking and the financing of projects which in the absence of government guarantees and the resulting moral hazard would not have been financed; and, finally, an increased level of systemic risk (see illustration in Appendix A.1). 7 2.3.3

Alternative theories in the context of government guarantees for banks

Alternative theories and concepts in the context of government guarantees for banks include the charter value hypothesis, economies of scale and scope, market discipline and bank-internal principal-agent conflicts. The charter value hypothesis suggests that market entry restrictions alter bank risk taking due to the costs a loss of the bank charter implies for the bank’s shareholders in case of default. 8 The underlying reasoning is that once market entry restrictions are imposed, banks generate monopoly profits, the future value of which reduces shareholder incentives to risk losing the bank charter via bank default. From a different perspective, the granting of a bank charter can be considered an ex-ante wealth transfer to bank shareholders. The value of this conceptual ex-ante wealth transfer sets incentives similar to shareholder liability in excess of actual paid-in bank capital on the bank’s balance sheet. 9 Charter values alter the classic option-like setting in which a limited liability shareholder with unlimited upside can benefit from increased volatility. Instead, in the presence of charter values, the net present value of the bank’s equity depends on access 6 7 8 9

An exception to this setting are traditional private banks with unlimited partner liability. Compare Stern and Feldman (2004) for a general discussion of the TBTF doctrine. Compare Marcus (1984) for a model and discussion of charter values and bank risk taking. While these incentives may induce bank investors to cover actual losses in order to maintain the banking license, there is no legal liability to do so and shareholder action will depend on a variety of factors including an evaluation of future profits versus actual cost of covering current losses.

2.3

TBTF as a Consequence of Government Guarantees

9

to future period profits of the bank, which can be achieved by avoiding a default, or reducing volatility. 10 According to this logic, the impact of government guarantees on risk taking and equity returns is reduced as risk-incentives stemming from government guarantees are muted by incentives to maintain the value of the bank charter. At the same time, the riskreturn relationship for shareholders is altered in that lower risk coincides with higher equity returns (see illustration in Appendix A.2). Economies of scale and scope are important aspects in the discussion of TBTF as they provide alternative explanations as to why risk and return characteristics of banks could vary depending on financial institution attributes. In classic intermediation theory, scope economies can be explained by a variety of factors, such as diversification benefits, differing individual risk preferences, institutional liquidity risk management capabilities and information asymmetries. Likewise, economies of scale relate to fixed costs, liquidity insurance, and information asymmetries. 11 Unlike the charter value hypothesis, the concepts of economies of scale and scope do not explain how government guarantees impact banks, but instead are factors to consider when interpreting evidence of banks’ risk and return deviations in the context of government guarantees. The concept of market discipline refers to a functioning market mechanism that influences economic decisions such as capital allocation or investments and thus determines economic outcomes. In the context of banking, market discipline involves both the effective monitoring of banks by investors as well as investors’ ability to influence bank managers’ decisions. 12 Scholars generally agree that government guarantees interfere with effective market discipline and thus alter economic decision-making. Hence, the presence of market discipline conceptually implies the absence of distortions introduced via government guarantees, or the non-prevalence of the TBTF doctrine. Similar to the concept of economies of scale and scope, market discipline does not provide an explanation of how government guarantees affect banks, but has been applied as a means to provide evidence of whether market distortions exist or not. Information asymmetries and resulting principal agent conflicts can cause frictions, or internalities 13, within banking organizations, which may mitigate the effects of the TBTF doctrine. The underlying idea is that bank managers may not act only to maximize shareholder wealth, but will also consider their own agenda such as securing attractive follow-on employment opportunities, which may depend on the success of banks they have worked for in the past. This link between future job opportunity and bank performance can be established by markets, or by regulators not approving a manager of a 10 11 12 13

Compare Levonian (1991) and Gizycki and Levonian (1983) who incorporate the aspect of a multi-period setting it in an option pricing approach of government guarantees for banks. Compare Freixas and Rochet (2008). Compare Rochet (2004) and Bliss and Flannery (2002) for a discussion of market discipline in banking. Compare Spulber (1989) for a discussion of internalities.

10

2.

Consequences of Government Guarantees for Banks

failed bank for another post in the financial industry. In this setting, increased bank risk taking in the presence of a government guarantee, while potentially beneficial for shareholders, may not be optimal for a bank manager concerned about the implications a bank failure under his or her leadership would have on future employment opportunities. Conceptually, this phenomenon constitutes an internality potentially mitigating externalities arising under the TBTF doctrine. 2.3.4

Empirical approaches to measuring the prevalence of TBTF

Empirical approaches to measuring the prevalence of TBTF can be structured along the main consequences of government guarantees (see Figure 2). For primary consequences, scholars have attempted to measure government risk exposure and implied wealth transfers using contingent claims-based models, examinations of rating-implied funding advantages and observations of historical bailout costs.

Figure 2:

Empirical approaches to measure consequences of government guarantees

For secondary consequences, scholars have mostly focused on relationships between the three factors government guarantees, bank risk and bank returns. To analyze potential competitive distortions, scholars have conducted a variety of studies on the relationship of government guarantees and bank returns and, in addition, have drawn on findings from examinations of rating-implied funding advantages. To examine potential moral hazard, scholars have analyzed the impact of government guarantees on bank risk taking and on distortions of banks’ risk-return relationships. Generally, most of the empirical studies that examine the relationship between two of the three factors (guarantees, risk, and return) also explicit-

2.4

Government Exposure and Subsidies

11

ly control for or make assumptions about the remaining third factor. The alternative theories introduced in Section 2.3.3. have been used by scholars to provide counter-evidence or alternative interpretations of observed outcomes. 2.4

Government Exposure and Subsidies

2.4.1

Contingent claims approach and absolute subsidy estimates

The use of contingent claims approaches in the context of banking system risk has developed since the 1970s. Merton (1977) showed that government exposure can be conceptualized as an option, which reflects the bank manager’s right to sell the bank’s assets to the government at the price of the face value of insured liabilities once the bank defaults. The insured creditors are paid off, and the government receives the right to the future cash flows of the assets. A variety of studies attempt to determine the value and extent of implicit and explicit government subsidies for TBTF banks via guarantees. Scholars found that banks benefited from government guarantees during the 1990s, peaking during the Asian financial crisis of 1997. Kaplan-Appio (2002) observes guarantee values with an equivalent of a low double digit USD billion amount for Thai banks on average for the period from 1992 to 1996, and Lehar (2005) finds low double digit USD billion guarantee values per year on average for the period 1988 to 2002 for a sample of 149 international banks, peaking at approximately USD 100 billion during the Asian financial crisis. Likewise, individual large institutions were found to receive substantial guarantee values, such as the U.S. government-sponsored enterprises Freddie Mac and Fannie Mae or the two largest Swiss banks during the peak of the financial crisis. Lucas and McDonald (2006) derive a single digit billion USD value for the two U.S. government-sponsored enterprises. Haefeli and Jüttner (2011) observe low double digit billion Swiss Francs guarantee values for the two largest Swiss banks during the peak of the recent financial crisis, but notably a zero premium for the years 2004 and 2005, the years unaffected by the crisis and the run-up to the crisis. More generally, estimates for system-wide guarantee values during the recent financial crisis were found to be massive, ranging up to several hundreds of billions of US dollars annually. Oxera (2011) finds a high single digit billion GBP amount for the five largest UK banks, while, based on a different set of assumptions, Noss and Sowerbutts (2012) derive a high double digit to low triple digit billion GBP amount for the same sample. Haldane (2010) observes an average guarantee value of USD 169 billion for 37 global banks from 2007 to 2009, while a 2012 study by Haldane (2012) finds an average guarantee value for 26 large international banks of more than USD 1 trillion annually for the crisis period 2007 to 2010. As part of more recent attempts to determine the value of TBTF subsidies implied by government guarantees, scholars have extended the options-based approach to also take observed market spreads for bank debt into account, calculating the difference between

12

2.

Consequences of Government Guarantees for Banks

option-implied debt spreads and observed debt spreads, which provides an indication for funding advantages achieved due to government guarantees. 14 The aggregate funding advantages were found to be significant during the financial crisis, but decreased significantly in value after the crisis. For example, Tsesmelidakis and Merton (2013) find triple digit USD billion guarantee values for 27 large U.S. banks for the years 2008 and 2009, but only a very low single digit USD billion guarantee value for the year 2010. Li et al. (2011) observe triple digit USD billion guarantee values during the recent financial crisis for the 20 largest U.S. and European financial institutions respectively, based on the refinancing benefits achieved from a theoretical five-year zero coupon issuance. Lambert et al. (2014), for a sample of 100 large banks globally including officially designated G-SIBs, find implied subsidies via government guarantees worth to be several hundred of USD billion. Bijlsma et al. (2014) compare calculated credit default swap (CDS) spreads to those observed for 13 European G-SIBs and derive a funding advantage of 121 basis points on average for the years 2008 to 2011, which the authors translate into a yearly TBTF advantage per bank of approximately EUR10 billion (compare Table 1). This extremely broad range of results illustrates the importance of model assumptions regarding maturity and audit periods, equity cushions and volatility measures as well as distribution assumptions and discount factors. For example, regarding maturities, Haefeli and Jüttner (2010) show that guarantee values for a five-year maturity are approximately ten times higher than those for a one-year guarantee, elucidating the importance of maturity assumptions. Likewise, default probabilities are dependent on assumptions regarding the equity capitalization of banks. As Haefeli and Jüttner (2010) find, an increase in the assumed equity-to-assets ratio of the sample from 8% to 10% leads to a decrease in guarantee values of approximately 10%. The choice of volatility measures applied to a structural credit model is equally important. As Noss and Sowerbutts (2012) demonstrate, using option-implied volatilities (which are forward looking but may be biased in times of stress) instead of historical equity volatilities (which may be less indicative for future developments) can lead to a several times higher guarantee value in times of extreme stress. Also, distribution assumptions are critical. Noss and Sowerbutts (2012) find that considering occasional upward and downward jumps for price movements in the model (simulating socalled fat tails) can increase guarantee values by more than 40%. Finally, assumptions about the risk-free rate are impactful, too. Decreasing the discount rate from 5% used by Oxera (2011) to 1.2% (suggested by Noss and Sowerbutts to reflect the low interest rate environment in the UK in 2009 and 2010) increases the value of government guarantees by three to seven times for European and lookback options, respectively. Intuitively, with a lower upward drift of the asset value over time the probability of default increases.

14

At the same time, this approach considers the confidence level with which the market expects bank liabilities to be covered by government guarantees.

2.4

Government Exposure and Subsidies

Table 1:

2.4.2

13

Empirical studies on guarantee values

Funding advantages based on contingent claims approach and rating-implied spreads

As part of the more recent options-based approaches that take observed market spreads for bank debt into account, scholars have also expressed guarantee values in the form of basis points, which illustrates funding advantages resulting from government guarantees. When applied to the respective liability amounts of affected banks, the basis point estimate can be converted to an absolute dollar guarantee value. The general advantage of the basis points format is that results across studies are more comparable than absolute values given the deviation in underlying samples and reference to different portions of the liability structure. Also, results reported in basis points can be compared to findings from the second form of studies on exposure quantification, which is based on rating-implied spreads. This approach examines the rating uplift which rating agencies assign to banks, expressing expected government support based on rating agencies’ analysis. This rating uplift – the delta between the standalone and support rating – is then translated into a credit spread based on ratingspread relationships observed historically. While this methodology is based on subjective

14

2.

Consequences of Government Guarantees for Banks

assessments by rating agencies, it has the advantage that the spread from the rating uplift versus the spread based on the standalone rating without rating uplift is clearly identified. The rating-implied approach, however, is based on the assumption that support ratings are indicators for market pricing. Critics of this approach argue that in practice bank credit spreads more closely follow standalone ratings than support ratings (compare Araten (2013) and Kroszner (2013)) and that even at a certain rating level non-default factors contribute to deviations (Bliss (2001)). Basis points results from contingent claims-based approaches include Li et al. (2011), who observe single digit and low double digit basis points funding advantages for large U.S. and European financial institutions before the crisis, which widen to approximately 380 basis points and 180 basis points, respectively, in 2008. Similarly, Tsesmelidakis and Merton (2013) observe slightly negative funding advantages for large U.S. banks before the crisis, which turn into approximately 700 basis points during the height of the crisis. Lambert et al. (2014), based on a sample of large banks in advanced economies, observes a middouble digit funding advantage pre-crisis, which extends to close to 100 basis points during the crisis. Notably, the authors observe elevated funding spreads in 2012 and 2013, ranging between 100 and 150 basis points. As one would expect, studies based on rating-implied spreads show more consistent spread levels over time, given that rating agencies typically adjust their assessments only every few months on average and take into consideration factors other than the immediate market volatility driving the results of the contingent claims-based approach. Ratingimplied results include Ueda and Weder di Mauro (2013), who – based on a large worldwide sample of banks – observe elevated support levels in 2009, increasing to 80 basis points from the 60 basis points pre-crisis levels. Likewise, Lambert et al. (2014) observe rating-implied spreads increasing from approximately 30 basis points in 2005 to approximately 90 basis points in 2009, and then decreasing to approximately 50 basis points in 2013 (compare Figure 3). Scholars have also estimated guarantee values by comparing the funding costs of large financial institutions to those of smaller ones. For example, Acharya et al. (2013) observe annual guarantee values for the largest 10% of U.S. financial institutions of close to zero during the 1990s, but observe an increase of the spread to 100 basis points in 2009, translating into an approximately USD 160 billion funding advantage. Likewise, Baker and MacArthur (2009) observe an increasing gap of funding costs between 18 large U.S. financial institutions with assets of more than USD 100 billion and smaller banks following the government rescue of Bear Stearns in early 2008. This observed increase of 49 basis points, according to the authors, implies a funding subsidy of approximately USD 34 billion. It is worth noting that studies using differences in funding costs to estimate guarantee values face practical challenges. With TBTF banks as defined in these studies contributing 70-85%

2.4

Government Exposure and Subsidies

Figure 3:

15

Overview of estimated guarantee values in basis points over time

of the total banking system’s assets 15, it is challenging to use the funding costs of the remaining 15-30% of the financial system as a representative reference point, considering supply and demand dynamics reflected in current market prices. In other words, with a given investor demand for senior bank bonds, eliminating an assumed guarantee for the vast majority of the supply would likely impact pricing for the remaining supply portion. Another aspect is that market-based funding cost differentials will be reflective of the bank’s home country’s sovereign rating. For banks in jurisdictions with weaker sovereign ratings, this implies lower funding advantages, potentially understating the actual exposure a government has and complicating cross-jurisdictional comparisons. 2.4.3

Costs of past rescue measures

Scholars have also referenced costs of past rescue measures in the context of discussing implicit guarantee values. Reinhart and Rogoff (2009) provide an overview of estimated bailout costs of past crises across countries, ranging from low single digit percentages of GDP for the U.S. savings and loan crisis in the 1980s and 1990s to an upper-range estimate of more than 50% of GDP for Argentinian bank rescue measures in 1984. On the other end of the spectrum, Moen et al. (2004) show that bailouts conducted by the Norwegian gov15

For example, U.S. BHCs with total consolidated assets of USD 50 billion or more (for which specific resolution requirements are put in place by the Dodd-Frank Wall Street Reform and Consumer Protection Act) contributed approximately 70% of U.S. banking assets in 2016.

16

2.

Consequences of Government Guarantees for Banks

ernment in response to the 1980 banking crisis had a positive present value considering increasing values of ownership stakes the government took in the banks. Likewise, Strongin et al. (2013) emphasize that the U.S. Troubled Asset Relief Program (TARP) implemented during the recent financial crisis made a 15% profit on its assistance to the six largest U.S. banks. The wide range of outcomes indicates that using past cost estimates as an approximation for guarantee values is challenging as each banking crisis is specific in nature and estimates for past crises show large discrepancies. 2.4.4

Summary of empirical evidence of government exposure and subsidies

In summary, scholars have found evidence of significant government exposure amounts to the financial system. However, the extremely broad range of results illustrates the importance of assumptions and data calibration when estimating government exposure amounts, especially if based on the contingent claims approach. The peaks observed by scholars during the crisis years 2008 and 2009 both for absolute guarantee value amounts and basis points funding advantage emphasize the importance of long-term empirical studies based on a broad set of determinants to derive estimates of representative guarantee values over time. Results based on rating-implied funding advantages exhibit less volatility over time, but are the outcome of the analysis of a singular institution (rating agency) as opposed to market prices, which drive contingent claims-based results. Analyses of cost of past rescue measures likewise show an extremely wide band of outcomes ranging from costs equivalent to significant percentage shares of GDP to actual gains from government intervention (which imply negative guarantee values). While this measure has the benefit of referring to actual costs incurred by governments – the reference governments should be most concerned about when assessing their exposure to the banking system – the specific nature and circumstances of past banking crises limit the transferability of past observations to expected future outcomes. Across all three methods, the significant variation in samples with regard to sample size and inclusion or exclusion of non-bank financial institutions not only makes comparisons across studies more difficult, but also leads to an important question regarding the scope of the underlying issue, i.e. whether government guarantees are assumed to affect primarily select TBTF banks (such as G-SIBs or SIBs), or also non-bank financial institutions (such as SIFIs), and if guarantees are assumed to support individual large institutions or the larger banking and financial system. This is an important definitional issue – which we will discuss from a competition point of view in Section 2.5 – and goes hand in hand with the broader question of how evidence of government exposure to the banking system should be interpreted – both from a political and economic perspective. From a political perspective, any observed exposure levels raise questions about wealth and income distribution. From an economic perspective, government exposure to the banking system raises a question about the role of a government in the context of a fractional reserve financial sys-

2.5

Competitive Distortions from Government Guarantees

17

tem. If the observed triple digit billions or even more than a trillion of US dollars of annual guarantee values are interpreted as an anomaly for a small set of large banks, the call for a removal of these unwanted effects through appropriate regulation may seem natural. This view is consistent with the post-crisis mandate for regulators to eliminate TBTF, which implies a zero exposure by governments to the banking system. However, if the observed exposure amounts are rather interpreted as the value creation of a fractional reserve banking system backed by government guarantees, the policy responses will depend on a government’s intended exposure level and exposure-related economic benefits, a view we will further examine in Chapter 3. 2.5

Competitive Distortions from Government Guarantees

2.5.1

Dimensions of competitive distortions

Government guarantees can cause competitive distortions if they are not provided equally to all participants in a particular market. The degree to which government guarantees within a certain economy are provided equally or less equally to market participants can be analyzed along multiple dimensions: individual bank characteristics, financial system components or financial system sub-sectors and geography. In an economy with a single policy framework, different guarantee policies can be conceptualized based on the degree of an individual firm’s systemic relevance that is required for an institution to be covered by government guarantees, and the scope of economic activities – or economic sectors – that are covered by government guarantees more broadly (see Figure 4). Government guarantees for only the most systemically important banks (G-SIBs or SIBs) imply a classic TBTF regime in banking. Abstracting from any potential regulatory costs associated with the TBTF status (such as higher capital requirements or other costs related to higher supervisory or regulatory requirements for those banks designated as TBTF), the funding cost advantages these banks enjoy can provide a competitive advantage compared to banks not covered by guarantees. At the other extreme, if the level of systemic importance of banks covered by government guarantees is very low – i.e. also the creditors of small banks are guaranteed – then a widespread banking system support prevails. The introduction of a blanket deposit guarantee for deposits of all banks in Germany by policymakers during the financial crisis is an example for a banking sector-wide guarantee (though limited to deposits as a specific refinancing instrument). 16

16

Other examples include Moyer and Lamy (1992) who observe that between 1979 and 1989 99.7% of all U.S. deposit liabilities for failed banks were fully protected. Li et al. (2011) caveat their findings about large banks subsidies as being “conservative, in that, smaller banks may also benefit from government support”.

18

2.

Figure 4:

Consequences of Government Guarantees for Banks

Implied guarantee policies based on industry and firm type covered

Similar financial functions can be performed by different subsectors within a financial system. 17 For example, banks and insurance companies may act as investors in similar projects, such as commercial real estate loans. To the degree one of these subsectors is covered by a government guarantee while the other is not (or to a lesser degree) there is a competitive distortion between the different financial system subsectors. Guarantees for the broader economy, i.e. also for non-financial industries, constitute a status of cross-industry TBTF, which, in the extreme case of guarantee coverage for all firms independent of a firm’s systemic importance would imply a fully state-guaranteed economy. 18 The resulting competitive distortions are a function not only of an uneven distribution of government guarantees, but also of the additional costs associated with guarantee benefits, such as regulations specifically aimed at systemically important banks. Extending the perspective beyond a single economy, competitive distortions between different economies with a different set of regulations are a third highly relevant dimension given the global nature of the

17 18

Compare Merton and Bodie (1995) for a functional perspective on the financial environment. Compare Gup (2004) for examples of government support for non-bank TBTF firms.

2.5

Competitive Distortions from Government Guarantees

19

financial system and differences in sovereign ratings, which refer to different degrees of support governments can provide to the respective home country financial institutions. In Section 2.5.3, we analyze existing empirical research along the three dimensions of competitive distortions: individual bank characteristics, financial system sub-sectors and geography. Most of the empirical contributions on competitive distortions focus on individual bank characteristics. We review these studies based on the refinancing instruments used to examine return distortions, a brief overview of which is included in the discussion of empirical approaches to measure guarantee-return relationships in Section 2.5.2. 2.5.2

Empirical approaches to measure guarantee-return relationships

When examining empirical studies on competitive distortions from government guarantees, the guarantee indicators used and the refinancing instruments examined are two important factors. This is because more explicit and more relevant guarantee indicators can be expected to provide greater predictability, and because government guarantees can be assumed to impact different refinancing instruments to different degrees depending on their seniority (e.g. debt vs. equity) and return participation (e.g. nominal interest for debt holders vs. profit and loss participation for equity holders). Scholars have applied a broad set of guarantee variables and a variety of bank’s refinancing instruments used by banks (compare Table 2). Studies examining explicit guarantees include O’Hara and Shaw (1990), which is based on the designation of U.S. money center banks as TBTF by the Comptroller of the Currency in 1984. Since then, scholars have reused this original 1984 sample, making the assumption that the explicit designation has not changed since the original statement decades ago (e.g. Balasubramnian and Cyree (2011)). In contrast, Moenninghoff et al. (2015) examine the recent explicit designation of banks as G-SIBs by the FSB and Group of Twenty (G20). Other studies examine government guarantees for portions of banks’ liabilities, such as deposit insurance, focusing on changes to the scope or amount of protection (e.g. Imai (2006)). An alternative guarantee increasingly applied by scholars post-crisis is the uplift rating agencies assign to bank ratings, expressing expected government support based on a rating agency’s assessment. More implicit guarantee measures used by scholars include a government’s willingness, regulatory ability or economic capacity to support stressed banks or a stressed banking system. Past rescue measures as an indication of a government’s willingness to step in have been applied on a country level (Cubillas (2014)), for specific financial institutions (see e.g. Kabir and Hassan (2005) who examine banks involved in the rescue of Long-Term Capital Management (LTCM)) and in the context of bank mergers (Molyneux et al. 2014). Likewise, scholars have examined formal antibailout policies such as the U.S. Federal Deposit Insurance Corporation Improvement Act (FDICIA) of 1991 as indicators of the prevalence of government guarantees (Angbazo and Saunders (1997)), which implies that regulatory or legal restrictions impact a government’s ability to spend taxpayer’s money in case of crisis. The significant increase of government

20 Table 2:

2.

Consequences of Government Guarantees for Banks

Empirical approaches to measure the impact of guarantees on returns

indebtedness of many countries following the recent financial crisis brought into question lower rated countries’ financial capacity to rescue banks, which led scholars to consider public finances and debt levels as indicators for the availability of government support (compare Demirguc-Kunt and Huizinga (2013)). Finally, scholars have used a variety of bank characteristics such as bank size – absolute or relative to the banking system –, sys-

2.5

Competitive Distortions from Government Guarantees

21

temic risk contribution or market share as proxies for potential government support. The application of these implicit guarantee measures assumes that these indicators of systemic relevance sufficiently approximate actual guarantee levels. The challenge of using implicit guarantee measures is that any associated returns will have to be carefully examined for possible alternative explanations (such as economies of scale in the context of size measures or economies of scope in the context of measures of complexity) as returns will potentially reflect sample-specific characteristics. To measure returns associated with the presence of government guarantees, scholars have used a variety of both debt and equity refinancing instruments. Return measures for debt include bond yields, subordinated debt yields and average funding costs across multiple refinancing instruments, which have been measured as actual yields, yield spreads to reference yields as well as rating-implied yield spreads based on rating agencies’ assessment of potential government support. Other empirical approaches include the use of CDS spreads to approximate the impact of guarantees on bank returns as well as the consideration of covenants in banks’ debt contracts, which can be considered a factor impacting returns from an investor point of view. Return measures used by scholars for equity instruments include stock returns at specific event dates, merger premiums paid as part of bank mergers and acquisitions, longer-term stock returns as well as market-to-book value measures expressing shareholders’ future return expectations. 2.5.3

Empirical evidence of competitive distortions

2.5.3.1 Competitive distortions by individual institution systemic relevance Evidence from bond spreads, subordinated bond spreads and interbank loan pricing Scholars have examined bonds, subordinated bonds and interbank loans in the context of government guarantees for banks in various settings. Using information on covenants in 415 bank debt contracts from 1974 to 1995, Goyal (2005) finds that bank charter values – as measured by Tobin’s q as a measure of market power, and banks’ demand deposit ratios – increase the likelihood of restrictive covenants being used in bank debt contracts, especially during the 1980s. Imai (2007) examines 279 subordinated debt issuances by Japanese banks from 1993 to 2004. The author observes that after the failure of Hokkaido Takushoku Bank, investors began requiring higher interest rates from weaker banks, indicating weaker guarantee expectations by investors. In the context of the failure of LTCM, Balasubramnian and Cyree (2011) examine 300 subordinated debt issuances by 70 U.S. bank holding companies (BHCs) during the period from 1994-1999 and observe an increasing size-discount after the Federal Reserve’s involvement in the 1998 bailout of LTCM, which the authors relate to TBTF perceptions. Examining close to eight hundred thousand federal funds transactions in 1998 between U.S. commercial banks, Furfine (2006) comes to a similar conclusion. The author observes that 32 commercial banks identified by a Federal Reserve publi-

22

2.

Consequences of Government Guarantees for Banks

cation as ‘large, complex banking organizations’ were able to receive funding at a rate of 4 basis points lower after the intervention by the Federal Reserve. Ongena and Penas (2009) examine bond prices in the context of European bank mergers analyzing 781 bond issues of 127 acquirers during the period from 1998 to 2002. The authors find that bonds of the acquiring bank react more positively to domestic than cross-border bank mergers and associate the larger domestic exposure of a bank with an increased likelihood of a swift government intervention in case of crisis, arguing that a single government within one legal and institutional framework can act faster than multiple governments in a coordinated action. Penas and Unal (2004) examine merger gains for 192 bank bonds in the context of 66 merger cases, observing gains as a result of mergers, which the authors attribute to diversification benefits, TBTF status and synergies. Using bank credit ratings, Morgan and Stiroh (2005) examine the rating-spread relationship for U.S. banks after the Continental Illinois rescue (1982-1986) and post-FDICIA (1993-1998). The authors observe a flattening relationship between spreads and rating for issues from banks named as TBTF compared to a steeper relationship for other issues after the 1984 statement by the Comptroller of the Currency. The authors find this relationship to persist in the post-FDICIA period. Rime (2005) examines ratings for banks from 21 industrialized countries from 1993 to 2003, observing that size-based explanatory factors such as total assets and market share have a significant, positive impact on bank issuer ratings. Post-financial crisis, Ueda and Mauro (2013) examine rating-implied spreads for 895 banks from 95 countries. While the authors observe an average increase in rating-implied spreads, various size variables are found to be insignificant in most countries. Schich and Kim (2012) examine rating uplifts for bonds of 229 banks in 20 countries as an indicator for implicit government guarantees and observe that a decrease in rating uplifts found across the sample post-crisis was mostly attributable to smaller banks. Acharya et al. (2013) examine bond spreads of 567 U.S. financial institutions from 1990 to 2010, differentiating their sample based on size and systemic risk measures such as CoVar and srisk. The authors observe a lack of risk-sensitivity of systemically important financial institutions’ bond spreads compared to other financial institutions bond spreads for most of the 1990s and 2000s, calculating that the 10% largest financial institutions have a 28 basis points funding cost advantage on average over time, peaking at more than 120 basis points in 2009. Kumar and Lester (2014a) use an analysis similar to the one by Acharya et al. (2013). While the authors agree with the observed higher bond spreads for large firms during the financial crisis, they observe a diminishing funding advantage post-crisis when extending the analysis until 2013. In contrast, Strongin et al. (2013) compare the six largest U.S. banks with more than USD 500 billion in assets to all other U.S. banks, observing only a very small funding cost advantage of 6 basis points for these largest U.S. banks throughout the crisis and concluding that these banks in 2013 had a negative funding cost spread. In the view of the authors, liquidity aspects of debt instruments can explain large bank funding cost advantages. Examining G-SIBs and domestically important banks (defined as the three largest

2.5

Competitive Distortions from Government Guarantees

23

banks in each sample jurisdiction), Lambert et al. (2014) observe a funding cost advantage for large banks compared to other banks of approximately 25 basis points for the period from 2003 to 2013, peaking at 250 basis points during the financial crisis. The authors, however, attribute a significant portion of this peak spread to emerging market outflows, which led to increasing corporate bond spreads, and to the fact that a significant portion of the sample banks are government owned, which positively impacts spreads, a phenomena referred to as ‘too public to fail’. In summary, as illustrated in Figure 5, most of the post-crisis contributions on funding advantages based on bonds have in common that they show elevated spreads for large banks during the crisis after a period of very low (or even negative) spreads pre-crisis. For the post-crisis period, while all studies show significantly lower spreads compared to the peak crisis years, there is disagreement across studies whether systemically relevant banks are close to zero, negative or positive.

Figure 5:

Funding advantages for systemically relevant banks based on bonds

Evidence from CDS-implied spreads Scholars have used CDS spreads in a variety of approaches to approximate how government guarantees impact funding costs. A general challenge in this context is the limited data availability of CDS spreads for smaller banks. Li et al. (2011) apply a structural contingent-claims model (as discussed in Section 2.4) and compare observed CDS spreads to spreads calculated based on equity returns for 83 U.S. financial institutions and 107 Euro-

24

2.

Consequences of Government Guarantees for Banks

pean financial institutions. Contrasting the funding costs for the largest 20 banks from the U.S. and Europe to those for the remaining sample firms, the authors observe that the largest U.S. financial institutions had an approximately 25 basis points funding advantage precrisis, which peaked amongst high volatility at approximately 90 basis points in 2009 and increased to more than 100 basis points in the aftermath of the crisis. For European banks, the authors observe only a marginal funding advantage before the crisis, a peak funding advantage of approximately 200 basis points during the financial crisis and an approximately 40 basis points funding advantage post-crisis. Bijlsma et al. (2014) calculate CDS spreads for 13 European G-SIBs based on regression parameters derived from calibrating a regression model for small banks’ CDS spreads to observed CDS spreads. The authors compare the calculated G-SIB CDS spreads to the observed ones and derive a funding advantage of 121 basis points for European G-SIBs on average for the years 2008 to 2011. As Figure 6 illustrates, the limited number of contributions on funding advantages of systemically relevant banks based on CDS spreads find positive spread levels during and post-crisis after only marginal funding advantages pre-crisis.

Figure 6:

Funding advantages for systemically relevant banks based on CDS

Using regression analysis, Völz and Wedow (2009) and Barth and Schnabel (2013) examine the relationship between international banks’ systemic relevance and CDS spreads. The authors generally confirm a negative impact of size or systemic relevance on CDS spreads. Völz and Wedow (2009), based on a sample of 91 banks across 24 countries, observe a 2 basis points reduction in CDS spread with every 1% increase of bank size relative to the bank’s home country GDP. Likewise, Barth and Schnabel (2013) observe an overall negative impact of systemic risk (as measured by a bank’s CoVar) on credit spreads for a sam-

2.5

Competitive Distortions from Government Guarantees

25

ple of international banks. However, both studies observe that after crossing a certain threshold of bank assets to home country GDP, increasing relative size in connection with increasing home country indebtedness can lead to a positive impact on CDS spreads, which the authors associate with investor concern about home country capacity to save very large banks. This observation is consistent with Demirguc-Kunt and Huizinga (2013), who observe that banks’ home countries’ deterioration of public finances positively impacted CDS spreads in 2008 compared to 2007. These results suggest that government guarantees for systemically relevant banks may provide a refinancing advantage as long as a bank is not too large to be rescued by its home country government. Evidence from aggregate funding costs Scholars have examined the aggregate funding costs of banks to analyze the impact of government guarantees on banks’ refinancing costs. While the advantage of this approach is that it takes all of a bank’s liabilities into account, it disregards structurally different funding compositions between large and small banks. As Kumar and Lester (2014b) illustrate based on a sample of U.S. G-SIBs and other banks, large banks are much more dependent on capital market funding than smaller ones. With that in mind, there are two views one can take from a competition perspective. One view is that the aggregate funding costs of a bank reflect the overall refinancing costs of a bank, which are the basis for competing for business such as commercial lending. This view essentially takes into account the fact that a bank’s liability structure has to grow via capital market instruments as the bank grows its asset base above a certain threshold that can be funded via a retail deposit franchise. An alternative view is that different liability structures reflect different business models (such as capital market activities large banks engage in) and, therefore, comparing aggregate funding costs across differently priced instruments does not adequately account for guarantee-induced funding advantages for individual refinancing instruments. The counterargument to this view is that smaller banks, which are often more reliant on deposits, benefit to a larger degree from an almost fully guaranteed instrument considering typical deposit insurance schemes and the associated fiscal backstops. These different views to some degree explain differing results observed by scholars when examining aggregate funding costs. Baker and MacArthur (2009), based on refinancing costs of U.S. depository institutions, observe a 29 basis points funding advantage on average for large banks with more than USD 100 billion in assets for the years 2000 to 2007. This funding advantage increased to 78 basis points during the crisis. However, data from depository institutions excludes a significant portion of capital market instruments, which are commonly issued at the BHC level. In contrast, Araten and Turner (2013), Kroszner (2013) and Kumar and Lester (2014b) examine aggregate funding advantages at the BHC level. This, according to the authors, not only more adequately reflects the complete liability structure (as it also includes capital market instruments issued at the BHC level), but also reflects a more risksensitive set of liabilities and considers those legal entities to which regulations are mostly

26

2.

Consequences of Government Guarantees for Banks

applied. Pre-crisis, all three studies observe significant negative funding spreads of up to 75 basis points for large banks. For U.S. G-SIBs, Kumar and Lester (2014b) and Kroszner (2013) observe a positive spread of approximately 45 basis points during the crisis, which completely diminished in the years following the crisis. Considering only U.S. BHCs with more than USD 500 billion in assets, Araten and Turner (2013) observe only a marginal positive funding advantage for large banks during the peak of the crisis and significant funding disadvantages of approximately 40 basis points following the crisis. In summary, as illustrated in Figure 7, scholars found differing results for funding advantages, partly driven by their decision to examine funding costs at the depository institution level or at the holding company level. During the crisis period, studies based on average funding costs find increasing positive spreads or decreasing negative spreads for systemically relevant banks. Post-crisis, scholars examining average funding costs at the holding company level found that funding advantages for systemically relevant banks approached zero or even dropped below zero.

Figure 7:

Funding advantages for systemically relevant banks based on aggregate funding costs

Evidence from deposits Early studies on the impact of government guarantees on deposit pricing and depositor risk sensitivity have focused on broad guarantees in the form of deposit insurance schemes. For example, Cook and Spellman (1996) observe that deposit rates for Federal Home Loan Bank member institutions in the U.S. reflected the risk of both the respective deposit-taking institution as well as the Federal Savings and Loan Insurance Corporation as guarantor.

2.5

Competitive Distortions from Government Guarantees

27

Moore (1997) examines deposit growth and asset quality for Mexican and Argentinian banks during the Mexican peso crisis in the mid-1990s, concluding that extensive government guarantees in Mexico muted depositor’s risk sensitivity while Argentinian depositors continued to be sensitive to risk. Similarly, Martinez-Peria and Schmukler (2001) examine the impact of changing deposit insurance schemes in Mexico, Chile and Argentina on insured and uninsured deposit prices and flows during the 1980s and 1990s, finding that deposits are sensitive to banks’ risk characteristics. Mondschean and Opiela (1999) observe that increased deposit insurance coverage implemented in Poland in 1995 decreased the importance of bank-specific factors for deposit pricing. Bartholdy et al. (2003) examine country-level deposit pricing across thirteen Organisation for Economic Co-operation and Development (OECD) membership countries and observe that – up to a certain point of maximum insurance coverage amount – deposits in countries without insurance show more than 40 basis points higher risk premiums. Examining bank-level data on deposit pricing and information on deposit insurance schemes across 51 countries, Demirguc-Kunt and Huizinga (2004) observe that explicit deposit insurance lowers deposit prices by 170 basis points. Davenport and McDill (2006) use account level data from Hamilton Bank during the time of the bank’s failure in 2001. The authors observe that withdrawals of insured deposits exceeded those of uninsured deposits and that certificates of deposits issued by Hamilton Bank reflected risk premiums, concluding that even insured depositors are sensitive to risk. Imai (2006) examines deposit pricing and flows across 50 Japanese banks in the context of newly introduced deposit insurance coverage limits in 2002 and observes that depositors became more sensitive to default risk as a consequence of the adjusted coverage limits. In summary, it appears that, while deposit insurance to a certain degree mutes depositor risk sensitivity, it does not fully eliminate depositor market discipline. Scholars have also used anti-TBTF regulations and costs of past rescue measures as potential indicators for the presence of government guarantees to examine the impact of guarantees on deposit pricing. Athavale (2000) compares the price of uninsured certificates of deposits to commercial paper yields and find that default risk premiums continued to decline throughout the 1980s and early 1990s despite the enactment of the anti-bailout regulation FDICIA. Cubillas et al. (2012) examine 2,593 banks across 66 countries and 71 banking crises from 1989 to 2007 and find that a post-crisis weakening of market discipline is positively related to accommodative policies applied during the preceding crisis. More recently, scholars have examined deposit funding spreads focusing on differences between large and small institutions. Beyhagi et al. (2014) examine deposit prices from 1990 to 2010 and observe a deposit funding advantage of 80 basis points on average for the six large Canadian banks compared to other Canadian banks. Araten and Turner (2013) examine domestic interest-bearing deposits for U.S. G-SIBs from 2002 to early 2011 and observe a funding advantage of 23 basis points. The authors argue that a funding advantage of this magnitude can be attributed to factors other than guarantees, such as a differentiated client service offering by larger banks (including nationwide branch and

28

2.

Consequences of Government Guarantees for Banks

automatic teller machine networks), higher cross-selling rates as well as access to noninterest bearing commercial deposits, which enables such banks to avoid attracting more expensive retail deposits. Jacewitz and Pogach (2014) analyze branch-level deposit rates for approximately 2,750 U.S. banks for the period from 2005 to 2010 and observe persisting pricing advantages for large banks compared to smaller ones. For the pre-crisis period, the authors observe a deposit funding advantage for banks with more than USD 200 billion in assets of approximately 50 basis points without controlling for risk and 15 basis points after controlling for common bank-specific risk factors. Notably, for the post-crisis period, the risk-controlled spread for large banks is close to zero. Kumar and Lester (2014b), while observing similar results for nominal spreads, observe only marginal deposit funding advantages of 4 to 6 basis points over time based on risk-adjusted spreads when using a sample of banks with more than USD 500 billion in assets. In summary, as illustrated in Figure 8, the few post-crisis contributions differ on the prevalence and extent of deposit-based funding advantages for systemically relevant banks during the crisis, with some observing a significant spread difference while others observe funding advantages close to zero. Post-crisis, both Jacewitz and Pogach (2014) and Kumar and Lester (2014) observe a decreasing funding advantage for systemically relevant banks, which is only a few basis points for risk-controlled spreads.

Figure 8:

Funding advantages for systemically relevant banks based on deposits

2.5

Competitive Distortions from Government Guarantees

29

Evidence from stock returns Early in the discussion of TBTF, scholars have begun using stock prices to analyze the impact of government guarantees on returns. In 1990, O’Hara and Shaw (1990) examined positive bank stock price reactions for eleven U.S. money center banks that were declared as TBTF by the U.S. Comptroller of the Currency in 1984. Angbazo and Saunders (1997) examine the FDICIA regulation and find that stock prices of banks with assets of more than USD 10 billion reacted negatively to narrower deposit coverage limits. In the late 1990s, Kabir and Hassan (2005) examine stock returns of 869 U.S. financial institutions in the context of the Federal Reserve’s involvement in the rescue of LTCM and observe that markets reacted negatively once investors learned that the Fed’s role was limited to facilitating the rescue, which the authors explain by the existence of TBTF perceptions prior to the rescue. Scholars have also examined bank stock returns in the context of merger announcements to identify TBTF perceptions. Kane (2000) examines a sample of 113 U.S. large bank mergers from 1991 to 1998 and observes gains on acquisitions of other large depository institutions, which the authors interpret as TBTF effects. In the context of European mergers, Molyneux et al. (2014) come to a similar conclusion. The authors examine 172 merger transactions between 1997 and 2008 in nine EU member countries and observe increasing merger premiums with increasing target size. A recent long-term study of bank stock returns by Ghandi and Lustig (2015) examines U.S. bank stock returns from 1970 to 2005, comparing portfolios of the 10% largest and 10% smallest bank stocks. The authors observe significant underperformance of the largest banks’ stocks compared to smaller banks’ stocks, which they attribute to an asymmetric government guarantee. Post financial crisis, Bongini et al. (2015) and Moenninghoff et al. (2015) examine the new regulation of G-SIBs, a regulatory approach which combines additional regulatory measures with an official designations of banks as G-SIBs. Bongini et al. (2015) find mixed evidence for the official 2011 and 2012 designation announcements of 28 and 29 G-SIBs, respectively. Taking also prior leakages of FSB lists into account, Moenninghoff et al. (2015) observe positive value effects from official designations, which partially offset the negative effects the announcements of additional regulations implied. Reflecting constrained public finances post-crisis, scholars have also examined bank stock price returns to explain the role of sovereign credit ratings on guarantee values. Demirguc-Kunt and Huizinga (2013), examining a sample of 717 banks from 34 countries from 1991 to 2008, observe a positive correlation between banks’ market-to-book ratio and bank assets and negative stock prices reactions to home country government debt levels and deficits. Likewise, examining 259 banks from 37 countries between 1995 and 2011, Correa et al. (2014) observe a negative impact of sovereign credit rating downgrades on bank stock returns for banks that rating agencies expect to be covered by government guarantees. In summary, there have been a number of post-crisis examinations of the prevalence of TBTF based on equity returns using a variety of approaches. It appears that even in the

30

2.

Consequences of Government Guarantees for Banks

context of an additional regulation specifically for G-SIBs, TBTF effects still exist due to the fact that regulators had to officially designate institutions as G-SIBs. Similar to CDS spreads, sovereign finances appear to influence potential guarantee benefits also in the context of equity returns. 2.5.3.2 Competitive distortions by scope of activities covered by guarantees Scholars have compared funding cost differentials between different sub-sectors of the financial system or between financial firms and non-financial firms. These analyses are important as they bring to attention the degree to which different financial sub-sectors or even non-financial industries are covered by government guarantees. For example, observations of large firm funding advantages not only for banks but for large firms from various industries could imply a cross-industry government support regime, as discussed in Section 2.5.1. Alternatively, the fact that funding advantages exist for large firms independent of their industry affiliation could be interpreted as an indication that factors other than guarantees may be relevant – such as economies of scale – potentially causing any observed spreads. Passmore et al. (2005) compare funding costs of the two U.S. government sponsored entities Fannie Mae and Freddie Mac to 68 other AAA- and AA-rated financial institutions during the period from 1997 to 2003, observing a 40 basis points funding advantage for the government-sponsored mortgage institutions. Lambert et al. (2014) examine funding costs based on a contingent claims approach for a global sample of 100 large banks during the period 2005 to 2013, distinguishing between investment banks and commercial banks. While the authors observe a broadly similar pattern for both groups for most of the observation period, they find that large investment banks’ funding advantage drops to zero directly after the failure of Lehman Brothers, while other large banks saw an increase of their funding advantage. Post-crisis, the funding advantage for large investment banks became similar to the funding advantage for other banks, which the authors attribute to the fact that U.S. investment banks converted into traditional bank holding companies regulated by the Federal Reserve during the crisis. Santos (2014) examines 8,399 U.S. bond issuances of banks, non-bank financial firms and non-financial firms during the period from 1985 to 2009, observing that the largest banks had a larger funding advantage than the largest non-bank financial firms and the largest non-financial firms. Tsesmelidakis and Merton (2013) examine yield spreads for large banks and large non-bank financial institutions based on a structured model. The authors find that both large banks and financial institutions exhibit only marginal funding advantages prior to the financial crisis. Beginning in 2007, the authors calculate increasing funding advantages, peaking at approximately 1,000 basis points for banks and 600 basis points for the broader financial institutions sample, which implies an approximately 400 basis points higher funding advantage for large banks compared to large financial institutions more broadly. Post-crisis, this stronger relative funding advantage for

2.5

Competitive Distortions from Government Guarantees

31

large banks versus other financial institutions persists at a 100 basis point level until 2010. In contrast, analyzing bond spreads for 44 U.S. banks from 1999 to 2013, Strongin et al. (2013) find that large banks have smaller funding advantages than non-bank financial firms and non-financial firms. According to the authors, these findings suggest that large firm’s funding advantages can be explained by size-related factors unrelated to government guarantees. Figure 9 summarizes these findings, showing significant spread volatility during the crisis years, but notably close to zero or even negative spreads for large banks compared to large non-bank financial institutions or non-financial firms pre- and post-crisis.

Figure 9:

Funding advantages for banks versus non-banks and non-financial firms

2.5.3.3 Competitive distortions by geography Competitive distortions between banks and banking systems of different economies are an important policy issue as they can have implications for the stability of the global financial system. Post-crisis, the world’s major economies coordinated their efforts to harmonize bank regulation through the G20 political format as well as international institutions such as the FSB and the BCBS. Differences between jurisdictions regarding regulations on bank resolution as well as governments’ willingness and ability to support banks can lead to regional competitive distortions. For example, cross-country studies examining deposit insurance show that the level of deposit insurance has an impact on the refinancing conditions for the respective banking system (see, for example, Bartholdy et al. (2003), MartinezPeria and Schmukler (2001) or Cubillas et al. (2012), which are discussed in Section

32

2.

Consequences of Government Guarantees for Banks

2.5.3.1). Likewise, Völz and Wedow (2009), Barth and Schnabel (2013) and DemirgucKunt and Huizinga (2013) show that a country’s rescue capacity – considering GDP and indebtedness – impact CDS spreads (compare Section 2.5.3.1). The effect of a government’s ability to rescue banks can result in a feedback loop of increasing financial system stress and deteriorating government finances, as evidenced by the recent financial crisis in the euro-area. 19 Recently established pan-European institutions such as the European Stability Mechanism in Europe that serves to protect the solvency of euro-area member states potentially lower the impact of individual governments’ financial capacity to support banks on cross-country competition (compare Moenninghoff et al. (2015)). Recent comparisons of large banks’ funding advantages between different geographies include Li et al. (2011) and Lambert et al. (2014). Li et al. (2011) examine observed CDS spreads to calculated spreads for large and small banks from the USA and Europe during the period 2001 to 2010, the results of which we discuss in Section 2.4.2. Post-crisis, the largest U.S. financial institutions exhibit a funding advantage compared to small ones of more than 100 basis points compared to approximately 40 basis points for the largest European financial institutions. These results contrast Lambert et al. (2014) who examine large banks’ funding spreads across different regions including the U.S., Japan, UK, the Eurozone and Switzerland. The authors conclude that large banks in advanced economies outside Europe have experienced decreasing funding advantages since the financial crisis, while large banks’ funding advantages in Europe remained elevated. There are several differences between these two studies with regard to sample institution type, sample size and methodology that are worth highlighting. While Lambert et al. (2014) focus their analysis on banks, Li et al. (2011) examine financial institutions more broadly, including, for example, insurance companies. Also, the large bank sample used by Lambert et al. (2014) includes 100 large banks globally, while Li et al. (2011) define large financial institutions as the 20 largest institutions in U.S. and Europe, respectively, comparing these institutions to the remaining 150 largest financial institutions in these geographies. Finally, the spreads referenced for Li et al. (2011) are the approximated difference between large banks’ funding advantage minus the remaining sample banks’ funding advantage, while the spreads referenced for Lambert et al. (2014) are just the sample bank funding advantages compared to the spread implied by a structured model as the authors do not examine a separate small bank example. 20

19 20

Compare Farhi and Tirole (2016) for a theory of feedback loops between sovereign and banking insolvency. Considering the large banks’ observed funding spreads versus the calculated ones for Li et al. (2011) does not alter the results: Post-crisis spread levels for large European banks on this basis are also lower than for large U.S. banks.

2.5

2.5.4

Competitive Distortions from Government Guarantees

33

Summary of empirical evidence of competitive distortions

Several findings are worth highlighting. First, a frequent observation across different refinancing instruments is that large banks exhibited increased funding spreads and heightened spread volatility during the crisis years. These observed spreads during the crisis period may be a crucial factor driving competitive dynamics, however, they do not appear to represent an average spread level over time or a longer-term trend. While a variety of studies find increased spread levels during the crisis, many studies also observe close-to-zero spreads before the crisis and diminishing funding spreads for large banks after the crisis. This pattern emphasizes the importance of long-term studies across economic cycles and periods of both financial stability and financial distress, respectively. More generally, most studies that examine bank spreads attempt to derive conclusions regarding the prevalence of TBTF. However, while spreads offer the advantage to compare results across studies independent of study samples, the exact implication of an observed level of spread differentials is largely unclear. As discussed in Section 2.4.2, using funding spreads between large and small banks to quantify TBTF guarantee values is difficult given that many large bank samples in these studies contribute most of the total banking system’s assets and liabilities and, as a consequence, observed small bank funding costs may not constitute a representative reference point considering supply and demand dynamics. The immediate effect of any observed spread differentials are competitive distortions, which, however, most of the studies do not focus on. Second, studies examining spreads during the post-crisis period mostly find mixed evidence across different refinancing instruments. Studies conducting repeat examinations using varying sample definitions or methodologies can be helpful in focusing the discussion on discrepancies in findings. For example, Kumar and Lester’s (2014a and 2014b) replication of studies by Acharya et al. (2013) and Jacewitz and Pogach (2014), or Noss and Sowerbutt’s (2012) replication of an analysis conducted by Oxera (2011) enable the discussion to be focused on specific definitional or methodological issues. Clearly defined data sets as well as a basic consensus on sample definitions across studies – i.e., which size thresholds to use for systemically relevant banks – are important factors in enabling the results to be compared. For the observed post-crisis decline in spreads, a better understanding of the factors driving this decline should be developed, i.e. whether these findings are a result of post-regulatory reform or due to a declining market volatility or other market-related factors. Third, recent examinations of large banks’ funding advantages compared to large non-bank and non-financial firms’ funding advantages more generally bring into focus how to adequately control for size-related factors. This aspect is important as many studies determine the samples of systemically relevant banks based on size. From a policy perspective this finding suggests to carefully differentiate between studies examining banks versus those including other financial firms when aiming to derive recommendations specifically for regulating banks. Fourth, studies examining

34

2.

Consequences of Government Guarantees for Banks

competitive distortions across economies indicate that local regulations with regard to deposit insurance and local government finances are important factors. A limited number of studies examining funding advantages across markets provide mixed evidence depending on sample definition and approach, and further cross-country examinations would be beneficial. 2.6

Government Guarantees and Risk Taking

2.6.1

The concept of moral hazard in banking

While the foundations for analyzing the impact of government guarantees on banks’ risktaking behavior were formalized by Merton in the 1970s (Merton (1977, 1978), scholars discussed moral hazard in the context of banking much earlier. For example, following the introduction of deposit insurance in four U.S. states in the wake of the 1907/1908 banking crisis in the U.S., Robb (1921) describes the fear that under the newly introduced system of deposit insurance “reckless straining after abnormal profits will tend to replace conservative banking”. 21 In the current discourse on TBTF, government guarantees for banks are regularly related to potential resulting moral hazard behavior by banks. In this context, the reference to moral hazard substantially follows the original meaning of the term in the area of risk insurance, which describes the alteration of an individual’s motive to prevent losses once insurance protection is provided. 22 Information asymmetries due to a lack of observability or too high costs for efficient monitoring of private action favor the emergence of moral hazard in the context of risk sharing. 23 Given that a bank’s overall risk profile depends on a variety of dimensions – including a bank’s assets, leverage and funding profile – any examination of bank risk taking needs to take a comprehensive view of risk, as we will further discuss in Section 2.6.3 in the context of banks trading off different risk types. It is important to note that the simple application of the moral hazard logic to bank behavior in a fractional reserve financial system which is deliberately and explicitly backed by government guarantees – the very purpose of which would be to allow for asset transformation, inherently involving the assumption of risk – falls short, because one would have to differentiate between a moral-hazard-induced increase in risk taking and guarantee-induced risk

21 22 23

Robb (1921), p. 189. Compare Shavell (1979), p.1, for moral hazard, Arrow (1963) for the concept of moral hazard, and Rowell and Connelly (2012) for a comprehensive discussion of the history of the term. Compare Arrow (1970), p. 11 and Spulber (1989), p. 61.

2.6

Government Guarantees and Risk Taking

35

taking. 24 For example, to identify moral hazard based on an observed increase in risk taking by banks following the introduction of explicit deposit insurance is difficult, because, from the point of view of the government’s motivation, any explicit guarantee for the banking system’s liability side would serve to stabilize a leveraged system to allow for fractional reserve financial intermediation, with the optimal level of asset risk of banks being dependent on the liability risk from an asset-liability-management perspective. With a government guaranteeing part or all of the liability side, an increase in risk assumed as part of banks performing asset transformation (for example, extended maturity transformation) does not necessarily imply an equivalent increase in a bank’s overall risk profile. In this setting, contrary to the standard moral hazard theory, higher levels of risk on behalf of the insured (the bank) may in fact maximize utility for the insurer (the government) given that the government may benefit from economic growth spurred by additional leverage. 25 Thus, differ24

25

To illustrate this point, we refer to worker’s compensation insurance in the U.S., a frequently used example in moral hazard literature, based on which scholars observed increased workrelated injuries with increasing insurance coverage, suggesting lacking care on behalf of employees or less pre-cautionary measures by employers due to the introduction of insurance coverage (compare Krueger (1988)). In this setting, the introduction of guarantees is assumed to have increased the insured’s risk taking compared to the ‘expected level of risk’ – which in this context would be best described as the employee’s and employer’s behavior without any insurance coverage. The change of behavior towards increased risk taking then can be described as moral hazard. In contrast, determining an ‘expected level of risk’ in the context of government guarantees for a fractional reserve banking system cannot refer to the same banking system without guarantees (as the existence of the current fractional reserve system in modern economies relies on some degree of government support), but would have to refer to other reference cases such as free banking without any lender of last resort, or narrow banking (such as all-equity financed banks or banks with asset restrictions to government-issued securities). However, using the risk levels of these alternatives as a benchmark in order to identify deviations indicating moral hazard is not suitable, because all these alternatives have very different implications for economic growth and thus for the government’s overall outcome (see following footnote). Summers (Financial Times, September 23, 2007) refers to the basis of this argument as moral hazard and confidence being opposite sides of the same coin. The complication in determining an adequate level of risk for a certain degree of explicit government insurance for a banking system stems from the role a financial system has for the overall economy and the resulting impact on welfare creation and government tax revenues. Referring to the worker’s compensation insurance example introduced in the previous footnote, the government’s interest in a leveraged financial system resembles a profit-sharing agreement between the insurer and the insured firm in addition to the existence of an insurance contract. In such a setting, it is easy to imagine situations, in which high-risk behavior by the insured conforms with the maximization of the insurer’s utility. The maximization of the insurer’s utility thus would be achieved by decreasing efforts to prevent injuries (or increase the risk of injuries) as long as the marginal increase of the insurer’s share in the firm’s profits is larger than the marginal cost increase due to increased injuries. Thus, in the context of a financial system, an economy’s wealth, which a government can be assumed to maximize, would be maximized by increasing the use of a fractional reserve financial system as long as the leverage-induced benefits outweigh the net costs resulting from the guarantees that are invoked in case of crises.

36

2.

Consequences of Government Guarantees for Banks

entiating between moral hazard and the more general consequence of resource misallocation appears theoretically as important as empirically challenging. We will discuss empirical evidence of how guarantees impact risk taking and distortions of banks’ risk-return relationship more broadly in Sections 2.6.2 through 2.6.4. 2.6.2

Empirical approaches based on guarantee-risk relationships

Scholars have examined the impact of government guarantees on bank risk using a variety of indicators for guarantees. Explicit guarantee indictors include the presence of deposit insurance and the public designation of banks as TBTF. Implicit guarantee indicators include political factors that are assumed to strengthen government guarantees or to facilitate potential future bank rescue measures, public ownership stakes in banks, guarantees implied by rating-uplifts and institution-specific measures such as size. To determine individual bank risk, scholars have examined bank leverage, loan size, non-performing loan levels and syndicated loan credit quality and pricing and have used macroeconomic indicators to estimate banking system-wide risk (compare Table 3). Table 3:

2.6.3

Empirical approaches to measure the impact of guarantees on risk

Empirical findings based on guarantee-risk relationships

Early studies examine the impact of guarantees on bank risk taking during the 1980s and early 1990s, a time period that includes the failure of U.S. bank Continental Illinois, the U.S. savings and loan crisis and subsequent regulatory changes in the form of FDICIA. Using bank size as an indicator for guarantees, Moyer and Lami (1992) observe declining capitalization ratios specifically for large U.S. banks from the 1960s to the 1980s. Likewise,

2.6

Government Guarantees and Risk Taking

37

Boyd and Gertler (1994) observe that large U.S. banks shifted their loan portfolios towards higher-risk asset classes and that large banks have lower capital ratios than other banks, attributing these findings to TBTF effects. Demsetz and Strahan (1997) also observe higher levels of risky loans and greater leverage at large U.S. BHCs from 1980 to 1993, however, the authors attribute this observation to economies of scale. Differentiating between banks’ systematic and idiosyncratic stock price variance, the authors examine bank diversification and conclude that large banks used their higher degree of portfolio diversification to generate riskier loans and assume greater leverage – an argument which points to a tradeoff between the riskiness of a bank’s funding sources, a bank’s leverage profile and its asset risk. A second stream of research uses the presence of deposit insurance as a guarantee indicator. Several studies observe a positive impact of deposit insurance on bank risk taking. Hovakimian and Kane (2000) apply contingent-claims models to examine equity return volatility of 123 U.S. chartered commercial banks from 1985 to 1994, observing that the presence of deposit insurance allowed banks to increase risk. Kwast and Passmore (2000), comparing large U.S. BHC capital ratios with those of non-bank firms from 1985 to 1997, find that large banks have lower capitalization even after the enactment of FDICIA, which the authors attribute to deposit insurance coverage. Ioannidou and Penas (2010) examine commercial loan originations by Bolivian banks after the introduction of deposit insurance and observe that banks with deposit insurance are more likely to initiate riskier loans with higher interest rates and worse ex-post performance, which according to the authors is facilitated by a lack of market discipline by large depositors. Contrasting these findings, a number of studies observe a decline in risk taking in response to explicit deposit insurance. Gropp and Vesala (2004), examining 128 European banks from 1992 to 1998, find that the introduction of explicit deposit insurance limits bank risk taking as expressed by leverage, asset risk and stock volatility. However, the authors argue that the introduction of explicit deposit insurance was a de facto reduction of more extensive implicit guarantees in place before. The authors also examine banks’ charter values based on banks’ margins and observe that higher charter values limit risk taking, independent of whether deposit insurance systems are in place. Gropp et al. (2011), observing credit risk and leverage measures for banks in OECD countries in 2004, find that banks not protected by deposit insurance had higher risk compared to banks protected by deposit insurance. The authors explain this increase in risk taking as a reaction to margin pressure caused by competing banks covered by deposit insurance, which led the non-covered banks to assume riskier and higheryielding loans. A number of studies on bank risk in the context of deposit insurance emphasize the importance of the contracting environment in determining how guarantees impact bank risk taking. Hovakimian et al. (2003) find that the political climate, economic freedom, and degree of government corruption are important determinants for risk-shifting under explicit deposit insurance. Based on an examination of banks from 56 countries from 1991 to 1999, the authors observe that explicit deposit insurance can lead to risk-shifting in poor contracting environments. In a cross-country setting, Demirguc-Kunt and Detragiache

38

2.

Consequences of Government Guarantees for Banks

(2002) examine panel data for 61 countries from 1980-1997 and observe that explicit deposit insurance had a negative impact on bank stability, especially in the context of deregulated interest rates and weak institutional environments. Other studies on bank risk in the context of deposit insurance point to a tradeoff between different types of individual bank risks or between individual bank risk and overall financial system risk. Forssbaeck (2011), using a model for market discipline based on variations in banks’ deposit insurance coverage and ownership structures, examines a global sample of several hundred banks from 1995 to 2005. The author observes that market discipline has only a limited effect on banks’ asset risk, but no effect on the banks’ z-score, which the author attributes to banks trading off asset risk and leverage. Conceptually, this view of risk tradeoffs within a bank is consistent with Demsetz and Strahan (1997), who observe that large banks used their greater portfolio diversification to generate riskier loans and assume more leverage, as discussed at the beginning of this section. DeLong and Saunders (2011) analyze the risk taking of 60 financial institutions after the introduction of the U.S. federal deposit insurance and find that banks and trusts increased their risk after introducing deposit insurance, while weaker banks benefited from increased depositor confidence, increasing the overall banking system stability. Anginer et al. (2013) observe a similar tradeoff between increased risk taking and overall banking system stability introduced by deposit insurance. Based on banks’ z-scores, the authors observe a positive correlation between explicit deposit insurance and individual bank risk 26 and at the same time an offsetting “stabilization effect” during times of financial system stress based on stock returns and marginal expected shortfall. Various studies have used ratings as indicators for guarantees. Soussa (2000), using a sample of 120 banks from five European countries and Japan finds that the relationship between TBTF banks and standalone bank credit ratings, loan loss provisions and asset quality is insignificant. Nier and Baumann (2006) find contrasting results. Based on a sample of 729 banks from 32 countries for the time period 1993-2000, the authors observe that capital buffers are lower for banks with higher government support expectations as measured by rating agency support ratings and share of insured funding. The authors relate these findings to moral hazard incentives from implicit government guarantees. Likewise, Gadanecz et al. (2012), based on a sample of 24,000 loan facilities during the period of 19932001, observe that banks with government support as expressed by rating uplifts have syndicated loan portfolios with lower spreads compared to unsupported peers, and Afonso et al. (2014), examining 612 banks from 92 countries, find that stronger sovereign support indicated by rating uplifts is associated with an increase in the ratio of impaired loans to total assets. Similar to findings in the context of deposit insurance, studies based on rating uplifts also found environmental factors to play an important role. Damar et al. (2012), 26

A limitation of these results, however, is that the z-score definition used is calculated based on pre-provision ROA and pre-provision ROA volatility (in addition to leverage), which excludes provisions as an important indicator of asset risk and resulting return volatility.

2.6

Government Guarantees and Risk Taking

39

examining z-scores and stock return volatilities for 64 banks worldwide following the introduction of a new rating methodology and respective new bank ratings published in 2006, observe increased risk taking by banks with high assumed government support during calm times, but a lower relative increase of risk by these banks during times of financial stress, which the authors relate to higher charter values due to funding advantages. Other implicit guarantee indicators scholars have used include political factors and rescue measures. Dam and Koetter (2012) use regional political factors in Germany as an indicator for bailout expectations and relate them to risk taking for a sample of 3,554 German banks during the period from 1995 to 2006, observing that higher bail-out expectations lead to an increase in bank risk taking. Using actual rescue measures as guarantee indicators, Black and Hazelwood (2013) examine how TARP provision for 441 U.S. banks from November 2007 to August 2009 impacted risk taking. The authors find that post TARP provision large banks originated higher-risk, higher-interest loans without increasing loan volume. Duchin and Sosyura (2014) examine risk-behavior by 278 U.S. banks post TARP and find risk-shifting within asset classes, leading to higher default risk and volatility for banks that accepted TARP funds. Scholars have also used explicit TBTF designations to examine bank risk taking in the context of government guarantees. Fischer et al. (2012) examine the removal of explicit government guarantees for German Landesbanken in 2001 and find that banks increase risk after guarantees are removed, an effect which the authors found to be most pronounced for banks with the highest expected franchise value. Likewise, Schnabel and Körner (2012) observe that German savings banks take higher risks (as measured by capitalization, risk provisions and bank liquidity) after their charter value decline when government guarantees for the associated Landesbanken extinguished in 2005 and lead to higher refinancing costs for savings banks. Gropp et al. (2014), examining the risk-taking behavior of 452 German savings banks during the five years before and after the removal of explicit government guarantees for savings banks in 2001, find opposing results. The authors observe a reduction in credit risk based on borrower-level z-scores, concluding that the previously existing public guarantees may have been associated with moral hazard effects. In contrast, examining explicit guarantees for money market funds in the U.S. during the financial crisis, Strahan and Tanyeri (2012) analyze approximately 450 money market funds from 2005 to 2010 and do not find an increase in risk taking as measured by the share of risky assets and the funds’ weighted average maturity. 2.6.4

Summary of empirical evidence of bank risk taking based on guarantee-risk relationships

In summary, while the studies discussed in Section 2.6.3 make partly opposing observations on how guarantees impact bank risk taking, several themes that are closely interrelated are worth highlighting. First, a number of studies indicate that comprehensive bank risk

40

2.

Consequences of Government Guarantees for Banks

measures are important as banks appear to trade off different risk types such as leverage versus diversification (Demsetz and Strahan (1997)) or leverage and asset risk (Forssbaeck (2011)). Taking into account the existence of risk tradeoffs in the context of government guarantees, however, complicates the empirical determination of moral hazard given that guarantees can be considered a means to lower a specific type of risk (i.e. funding risk) in response to which banks can be assumed to adjust other risk components without necessarily intending to alter their overall risk profile. Second, from a systemic risk perspective, cross-sectional competitive dynamics and varying effects over time and across the economic cycle appear to be relevant as studies found overall banking system stability to be higher despite increased risk taking of individual banks in response to guarantees (DeLong and Saunders (2011) and Aginer et al. (2013)). This observation in the context of systemic risk, again, is closely related to the aspect of individual bank risk tradeoffs. With higher funding stability as a result of a more stable financial system, individual financial institutions have a different set of risk choices than they would in a regime without any deposit insurance. This implies that – similar to our findings in the context of guarantee quantifications and funding advantage estimates – the observation period appears to play a crucial role. Scholars should emphasize long-term studies across cycles to understand how bank risk taking responds to government guarantees over time, and whether any observed change in risk taking during calm times is a reflection of a reduced likelihood of banking system stress due to the presence of guarantees intended to prevent bank runs. Third, most of the examined studies lack an explicit discussion of the guarantee indicators that are being used and how the choice impacts the interpretation of results. This discussion is important as there seems to be a conflict between explicitness of a guarantee – which can be assumed to provide stronger empirical signals of actual guarantee levels – and noise around a policymaker’s potential intent. Assuming that an explicit guarantee such as a deposit insurance scheme is meant to stabilize a financial system (in times of stress), it is hard to imagine that a government would provide this type of insurance without expecting an improved level of financial services to the economy across the cycle. In other words, the disadvantage of using an explicit guarantee indicator is that it is more likely to reflect a policymaker’s intent to foster through-the-cycle economic growth through increased bank intermediation. For example, in the context of explicit guarantees for German Landesbanken, studies concluding that the removal of explicit guarantees for publicly owned banks in Germany removed TBTF effects do not consider the particularities of the German savings banks system, which are legally obliged “to serve the common good” 27 and to provide households and local firms with easy access to credit. In this context, any level of explicit government guarantees can be assumed to be intended to increase bank intermediation of local savings banks, which ceteris paribus, may imply increased risk taking by those institutions. In return, the use of implicit guarantee indicators – while likely not directly expressing a policymaker’s intent to 27

Compare Gropp et al. (2014), p. 5.

2.6

Government Guarantees and Risk Taking

41

foster financial intermediation – presents the challenge of justifying the indicators’ relevance and avoiding potential sample bias if based on institution characteristics. In this context, the approach followed by Dam and Koetter (2012) of using regional political constellations in Germany as guarantee indicators is very promising in that it can be assumed that regional politicians do not have a mandate to determine bank bailout policies in Germany. As a result, any change in risk taking due to regional political constellations can be assumed to be unintended. While the authors’ finding may be pointing to an unsatisfactory status from a practical policy view, it is unclear from an economic theory perspective whether the transmission mechanism for risk taking in this context should be viewed as moral hazard, or whether the authors observe a dysfunctionality of the political system in which local politicians convince banks to stimulate local economies at the cost of the federal tax payers. These observations imply that a careful differentiation between guarantee measures is important. For example, the guarantee measure we will use in Chapter 4 – the FSB’s explicit designation of G-SIBs as part of the regulator’s attempt to eliminate TBTF – can be considered a particular case that combines an explicit guarantee with a credible motivation to not be intended to stimulate growth. If explicit guarantee measures are used, scholars should examine the institutional context of why these guarantees were provided and if a sufficient mandate was available for the institution providing the guarantees. 2.6.5

Empirical approaches based on risk-return relationships

A second approach scholars used to analyze banks’ risk taking in the context of government guarantees is based on risk-return relationships. Compared to the guarantee-risk approach we discussed in Sections 2.6.2 to 2.6.4., the risk-return relationship is less distinctly related to bank risk, but also takes return specifics into account. Scholars have used various risk measures in this context including individual bank accounting risk measures – such as capital ratios or leverage, loan loss rates or foreign debt exposure – or bank credit ratings in order to determine bank risk levels and risk behavior (compare, for example, Beighly et al. (1975), Pettway (1976), Flannery and Sorecscu (1996), Bliss (2001) and Hasan et al. (2013)). Examining banks’ asset mixes, Morgan and Stiroh (2001) analyze the proportion of riskier activities such as trading, credit card debt or commercial and industrial lending as a measure of bank risk. More recently, scholars have applied risk measures beyond accounting data. For example, Wu and Bowe (2012) use levels of information transparency – measured by reporting standards and disclosure frequency – to determine depositors’ perceptions about the likelihood that a bank could fail. Examining bank’s systemic vulnerability, Chava et al. (2014) use banks’ expected shortfall and marginal expected shortfall as risk indicators. In risk-return examinations, scholars have referred to funding instruments across the entire capital structure. The debt-related instruments studies have examined include deposits, bonds, CDS and subordinated debt, using yield spreads above reference debt in-

42

2.

Consequences of Government Guarantees for Banks

struments such as Treasuries or nominal rates. Similarly, stock prices have been used to measure the relationship between bank risk and equity returns (compare Table 4). Table 4:

2.6.6

Empirical approaches to measure risk-return distortions

Empirical findings based on risk-return relationships

Early studies examine the risk-return relationship of uninsured deposits, subordinate debt and equity returns based on accounting measures of bank risk. 28 Examining depositor behavior during the 1932 Chicago banking crisis, Calomiris and Mason (1994) compare the decline of deposit funding of 123 banks preceding the crisis and observe that changes in the composition of funding reflected banks’ relative weakness months before the actual crisis. Studies on depositor market discipline in the 1970s and 1980s find mixed results. For example, according to Gilbert et al. (1990), Herzig-Marx and Weaver (1979) find only limited explanatory power of risk variables examined. In contrast, Hannan and Hanweck (1988), examining large certificates of deposits for 300 U.S. banks in the first quarter of 1985, observe that deposit rates reflect leverage, return variability and bank capitalization. Likewise, Ellis and Flannery (1992) observe for six U.S. money center banks from 1982 to 1988 that certificate of deposits are sensitive to default risk based on banks’ equity returns. More recent studies examining the risk-return relationship based on deposits include DemirgucKunt et al. (2006). The authors examine deposit volumes for 257 banks from 35 countries during 36 banking crises and find declining deposit volumes for less profitable and weaker banks in times of crisis, concluding that depositors actively monitor banks. Likewise, Berger and Turk-Ariss (2015) observe generally strong market discipline for more than 2,000 banks from the U.S. and 22 European countries from 1997 to 2007. Recently, authors have 28

Compare Gilbert (1990) for a comprehensive overview of studies on market discipline in U.S. banking in the 1970s and 1980s.

2.6

Government Guarantees and Risk Taking

43

also examined how disclosure frequency and reporting standards or information not contained in financial statements impact market discipline. Wu and Bowe (2012) observe for 169 Chinese banks from 1998 to 2009 that a higher disclosure frequency and the application of international reporting standards impact the banks’ ability to attract deposits. Hasan et al. (2013) examine depositor behavior of 416 commercial banks in 11 Eastern European countries from 1994 to 2011 and find that negative news on the financial stability of foreign parent companies significantly impacted deposit flows of their Eastern European subsidiaries. Similar to early studies on depositor discipline, studies on subordinated debt holder market discipline during the 1970s and 1980s observe mixed results. For example, Pettway (1976) examines 77 bank subordinated debt issuances from 1971 to 1974 and concludes that investors were not very sensitive to capitalization ratios of large banks. In contrast, Gorton and Santomero (1990), using a contingent claims approach to derive a pricing model for subordinated debt, observe that accounting measures of risk only marginally predict banks’ asset volatility. In the context of the enactment of FDICIA in the U.S. in the early 1990s, Flannery and Sorescu (1996) examine 422 subordinated bond issues of 83 banks from 1983 to 1991 and find that bank risk factors correlated with subordinated debt yields, specifically in the time immediately prior to the FIDCIA enactment. Examining market discipline during the 1990s, Morgan and Stiroh (2001) analyze the relationship of banks’ asset composition and subordinated bond and bond spreads for 497 issuances by U.S. BHCs between 1993 and 1998. The authors find that, even when controlling for standard risk measures, bond and subordinated debt spreads reflect the risk implied by the different individual bank activities such as trading, commercial lending or industrial lending. Preceding the introduction of the Basel II accord in 2004 – which aimed to strengthen market discipline via enhanced disclosure as part of the accord’s third pillar – a number of studies assessed subordinated debt market discipline and the suitability of information from subordinated debt markets as a predictor of bank stability. For example, Evanoff and Wall (2001) examine 452 supervisory ratings assigned to U.S. banking organizations from 1991 to 1998 and find that subordinated debt yields are equally good or better predictors of supervisory ratings than capitalization ratios. In contrast, DeYoung et al. (2001), examining 1,079 U.S. banks from 1989 to 1995, observe that regulatory examinations produce additional valuerelevant information compared to subordinated debt spreads. Krishnan et al. (2005), analyzing secondary subordinated debt market data for 50 U.S. banks from 1994 to 1998, conclude that while credit spread levels provide information on bank-specific information, quarterly changes in bank risk exhibit only a weak relationship. Examining bond issuances of 96 U.S. financial institutions from 1990 to 2010, Chava et al. (2014) find that spreads reflect institutions’ expected shortfall, but do not respond to their systematic tail risk. Based on interbank loans, Furfine (2001) examines market discipline for 362 U.S. commercial banks between 1989 and 1997 and finds that interest rates reflect different levels of borrower credit risk, indicating that banks are effectively monitoring their peers.

44

2.

Consequences of Government Guarantees for Banks

Early studies examining market discipline based on equity returns focused on the correlation between accounting measures of individual bank risk and equity volatility and mostly found evidence for the prevalence of market discipline. 29 For example, Pettway (1980) finds that stock prices for seven large U.S. banks that failed between 1972 and 1976 reflected increasing bankruptcy potential. Likewise, Cornell and Shapiro (1986) observe that equity investors were able to differentiate between Latin American loan exposure of 43 U.S. banks during 1982 and 1983. Brewer and Mondschean (1994) observe a positive correlation between equity volatility and junk bond holdings of 74 savings and loan associations during 1985 to 1989 period. More recently, Gropp et al. (2006) observe for 86 European banks from 1991 to 2001 that distance to default measures reflect credit downgrades 6 to 18 months before the actual event. Stiroh (2006) examines equity returns from 1,227 U.S. BHCs from 1997 to 2004 and finds a close link between volatility and accounting measures of individual bank risk. 2.6.7

Summary of empirical evidence of bank risk taking based on risk-return relationships

A broad range of studies on market discipline uses various risk and return parameters. Early studies on market discipline for deposits find mixed results based on accounting-based risk measures, while a number of recent studies with greater variety of risk indicators mostly confirmed some level of market discipline. A variety of studies with differing results contribute to the discussion about the degree to which subordinate debt instruments should be mandated as part of Basel II in the early 2000s and, more recently, in the context of bailinable debt. Studies appear to suggest that investors in subordinated debt instruments can play a disciplining role for banks. Scholars examining equity returns found these returns to be mostly reflective of risk parameters. Generally, studies based on risk-return examinations are less suitable for explaining the exact transmission and consequences of government guarantees than studies analyzing guarantee-risk and guarantee-return characteristics. While the observation of market discipline suggests that there are no distortions (and consequently no primary and secondary consequences) from government guarantees, the observation of a lack of market discipline does not clearly indicate whether it is risk or return characteristics that have been affected by guarantees. Hence, making a statement on whether moral hazard takes place is more challenging given that neither absolute risk levels nor adjusted risk levels compared to postguarantee expected guarantee levels are the focus of the discussion. Likewise, studies discussing risk-return distortions without differentiating between factors indicating guarantees (such as large versus small banks) do not allow to draw clear competition implications.

29

Compare Gilbert (1990) for a comprehensive overview of studies on market discipline in U.S. banking in the 1970s and 1980s.

2.7

2.7

Conclusion

45

Conclusion

In this chapter we reviewed the extensive literature on the TBTF doctrine to answer the question of why systemically relevant banks should be regulated. An important observation is that many studies focus on a single aspect of the TBTF doctrine, but attempt to draw broader conclusions and policy recommendations. Our analysis, however, emphasizes the need to differentiate between the major consequences of government guarantees – government risk exposure, competitive distortions and moral hazard – in order to understand the need for regulating systemically relevant banks. Studies on government exposure and related wealth transfers suggest significant exposure levels at least during times of heightened market volatility, which implies that policymakers should actively engage in the question of how to regulate systemically relevant banks. The vast variance in findings – ranging from more than a trillion US dollar exposure levels based on contingent claims approaches to positive net present values of past rescue measures – does not alter this conclusion, but instead emphasizes that governments need to actively measure and manage their exposure to the banking system, considering returns and risk. In Chapter 3, we will develop an exposure perspective of banking risk and regulation a government should take and provide a framework of relevant policy options. Studies attempting to examine the impact of government guarantees via observed competitive distortions, as expressed by funding spreads between guaranteed and nonguaranteed institutions, provide mixed evidence across refinancing instruments. However, most studies observe a significant spread widening during the crisis period, followed by spread compression post-crisis, which emphasizes the importance of long-term studies and the need to further identify the driving factors behind recent spread contraction. More generally, scholars should attempt to further shed light on how observed spread differences impact competitive dynamics given that competitive distortions are conceptually the immediate consequences of spread differences. A higher degree of conventions regarding methodologies and samples would allow for more repeat studies with parameter variations and result in a more focused discussion of individual aspects. For example, Kumar and Lester’s (2014a and 2014b) replication of studies by Acharya et al. (2013) and Jacewitz and Pogach (2014) enable the discussion to be focused on specific definitional or methodological issues. Lambert et al. (2014) provide a valuable application of multiple empirical methodologies to measure government exposure and competitive distortions within a single paper. Likewise, more standardized and regularly updated empirical measurements of exposures and competitive distortions could be used for financial stability monitoring and regulatory purposes. Findings of cross-industry size-based funding advantages should be further examined in order to strengthen interpretations and policy conclusions of any observed large bank spreads. Studies focusing on guarantee-risk and risk-return relationships, while providing partly mixed evidence, highlight the difficulty in empirically measuring the link between govern-

46

2.

Consequences of Government Guarantees for Banks

ment guarantees and moral hazard. The apparent tradeoff of different risk types by banks – as observed by Demsetz and Strahan (1997) and Forssbaeck (2011) and seemingly complex interactions between individual bank risk and banking system risk – as found by DeLong and Saunders (2011) and Aginer et al. (2013) – require comprehensive risk measures and through-the-cycle observations. Studies should discuss guarantee indicators not only with regard to the strength as an indicator for actual guarantees, but also with regard to the potential underlying motivation of an institution providing the guarantee. For example, Dam and Koetter’s (2012) reference to regional political constellations as guarantee indicator is promising in that in the context of the study regional politicians can be assumed to not have any mandate to determine bank bailout policies. We believe that structuring the discussion of TBTF along the direct consequences of government guarantees – government risk exposure, competitive distortions and moral hazard – and likewise focusing conclusions on the directly observed effects will be beneficial for further identifying causes and effects, and for interpreting findings.

3.

Government Guarantees and Banking System Risk – A Regulatory Framework from an Exposure Perspective

3.1

Introduction 30

The recent financial crisis saw large-scale, globally coordinated bank bailouts by governments aiming to stabilize the financial system. The USA, EU and UK provided government and central bank support to their financial systems of approximately $15 trillion in total, or 73% of GDP for the USA, 88% of GDP for the UK and 18% of GDP for the European Union, respectively. 31 These significant rescue packages by governments around the world were aimed at limiting the economic distortions caused by the crisis, such as a negative impact on economic growth and employment. 32 These measures had distributive and fiscal consequences leading to wealth transfers from taxpayers to bank debt holders, increases in government indebtedness and oftentimes even to full or partial government ownership of banks and bank assets, which in many cases can still be observed years after the bailouts took place. 33 The extent of these rescue measures not only illustrates the degree to which governments were exposed to the banking system once at stress, but also brings into focus whether it was economically rational for governments to have entered and maintained this exposure ex ante and to have assumed the losses when the system was at stress. This translates into our second overarching question on what the relevant policy options are and how to adequately regulate systemically relevant banks. To answer this question, we develop an exposure perspective on banking system risk and bank regulation. Building on the fundamentals of credit risk, we develop a framework to evaluate the full range of major regulatory policy options from a sovereign’s portfolio optimization perspective. These options range from free banking and narrow banking regimes (implying zero government exposure) over regulations such as minimum capital and liquidity requirements, structural restrictions, Pigovian taxes, resolution powers, bail-in regimes, contingent 30 31 32 33

This chapter is based on the paper ”Government Guarantees and Banking System Risk – A Regulatory Framework from an Exposure Perspective” co-authored by Axel Wieandt (see Moenninghoff and Wieandt (2017b)). Compare Bank of England (2009), p. 20. Compare Panetta et al. (2009) p. 17; Despite the extensive rescue measures, per capita GDP for twelve affected countries declined by approximately 10% from peak to trough during the six years following the financial crisis (compare Reinhart (2014) p. 18). Advanced economies’ gross central government debt increased from approximately 60% precrisis to more than 80% post-crisis (compare Reinhart and Rogoff (2013) p. 7); three years after the financial crisis, governments continued to hold economic interest in 20 of the largest 50 banks (compare Moenninghoff et al. (2015) p. 236)).

© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2018 S. C. Moenninghoff, The Regulation of Systemically Relevant Banks, Finanzwirtschaft, Banken und Bankmanagement  Finance, Banks and Bank Management, https://doi.org/10.1007/978-3-658-23811-7_3

48

3.

Government Guarantees und Banking System Risk

capital and resolution funds (implying limited government exposure) to a nationalized banking system (implying full government exposure). Bank regulation and individual regulatory policies have been extensively discussed from a theoretical perspective and from a practical policy perspective. Freixas and Rochet (2008) provide a conceptual framework of bank regulation based on industrial organization theory. In their framework, public regulation or self-regulation is introduced to resolve market failures, potentially implying biased objectives of the regulators. The resulting postregulation market equilibrium not only depends on regulatory decisions, but also considers legal and institutional constraints, strategic behavior by regulated banks and competition effects. In the context of their framework, the authors discuss individual policies based on microeconomic theory. Barth et al. (2008) likewise distinguish between public interest versus private interest views of regulation. In contrast, Greenbaum and Thakor (2015) and Ugeux (2014) structure their discussion of regulatory policies along specific bank regulatory objectives. Contributions examining concrete regulatory policies include Gordy and Heitfield (2010) and Hull (2012) who discuss risk-based regulation, and Stern and Feldman’s (2004) discussion of bank regulation in the context of TBTF. In developing our exposure-based regulatory framework, we combine several streams of research. We apply standard credit risk theory developed amongst others by Merton (1974), Vasicek (2002) and Gordy (2003) and introduced by bank regulators for regulatory capital determination under Basel II. We combine this credit exposure view with a portfolio optimization perspective established by Bodie and Brière (2014a) in the context of sovereign wealth funds to derive our simple model of exposure based regulation. Similar to impact studies for capital and liquidity regulation conducted during the Basel III regulatory process, we define factors impacting risk-return trade-offs of individual policy choices. For the description and discussion of individual policy choices within our framework we survey literature on bank regulation. While regulators have used impact studies to examine risk-return tradeoffs inherent in individual policy choices such as capital and liquidity regulation, this contribution to our knowledge is the first comprehensive approach to categorize and discuss the full range of major policy options for regulating banks within a single framework and to provide a quantitative methodology to compare the relative attractiveness of different policy choices. Overall, our stylized model suggests that the choice of an optimal regulatory policy mix depends on risk and return preferences of a society, and an economy’s institutional and cultural setting which impacts the tradeoffs implied by each regulation. Our framework suggests that a society with higher risk tolerance will prefer limited exposure solutions such as bail-in regimes and capital regulations, while a highly risk-averse society may be more open to increased government involvement in the intermediation process, for example through structural restrictions. From a practical policy perspective, our framework can assist in making policy decisions and track and communicate a government’s resulting exposure outcomes. From a theoretical

3.2

Banking System Exposure from a Credit Risk Perspective

49

perspective, our framework allows to better structure the broad range of contributions on optimal regulatory policy for systemically relevant banks, and suggests further applications of portfolio theory in the context of economic policy examinations and bank regulation. Assessing the new regulation dealing with G-SIBs based on our framework, we demonstrate that after two decades of focusing on minimizing the likelihood of crisis, regulators increased their efforts to also reduce losses and government exposure in case of crisis by introducing international standards for resolution regimes including wind-down authorities and creditor bail-ins. In Section 3.2, we briefly present our exposure perspective of government guarantees for banks, describe the core components of credit risk as well as its theoretical underpinnings and empirical applications in the context of government guarantees, and develop a credit risk management approach for bank exposure. We then use this credit risk management approach as a basis to develop a simple framework of principle regulatory policy choices across exposure levels and primary exposure levers. In Section 3.3, we expand our view to an overall sovereign portfolio perspective, which, in addition to the level of banking system exposure, also takes into account economic growth and stability. We apply a macrofinancial contingent claims approach developed by Bodie and Brière (2014a) in the context of sovereign wealth funds to illustrate a formalized sovereign’s portfolio decision problem. In order to demonstrate the major economic tradeoffs implied by the government’s optimization problem, taking into account tradeoffs inherent in the individual policy choices, we develop a simple stylized model including economic frictions and tradeoffs emerging in the context of bank regulation. In Section 3.4, we discuss the individual regulatory options in keeping with our framework of policy choices considering the implied growth and stability tradeoffs derived from our stylized model. Section 3.5 discusses additional important factors for regulatory policy determination, summarizes the development of the new regulation dealing with G-SIBs, and draws conclusions. 3.2

Banking System Exposure from a Credit Risk Perspective

3.2.1

Fundamentals of an exposure perspective for banking system risk

Public exposure to the banking system became obvious during the recent crisis when the governments of the USA and various European countries guaranteed bank creditors by providing massive bailouts, amounting to several trillions US dollars globally. 34 While the different approaches to stabilizing the banking system in different jurisdictions and instances bore different implications for equity holders, debt holders of troubled institutions were generally made whole. Failing banks received liquidity injections, recapitalizations and asset guarantees. All three methods essentially result in the government assuming risks resulting from a (potential) decline in the value of the banks’ assets and, at the same time, 34

Compare Panetta et al. (2009) p. 17.

50

3.

Government Guarantees und Banking System Risk

guarantee bank creditors. In that sense, government guarantees to bank creditors – in an implicit or explicit way – effectively resemble credit risk. Generally, credit risk is the risk that a borrower will fail to meet her obligations under a credit contract, implying the borrower’s default and a potential loss of principal and interest for the lender. 35 As Merton (1974) shows, borrower default is triggered by the value of the borrower’s assets falling below the creditor’s claims. In case of a borrower’s default, the lender’s claims are converted into equity and it becomes the owner of the assets. Credit risk in the context of government exposure to the banking system has very similar features. If a bank fails, the government steps in, assumes creditors’ losses and essentially receives the right to the cash flows from the bank’s assets. As shown by Merton (1977), such a guarantee can be modeled as a put option, which allows a bank’s management to sell the assets to the government at the price equal to the amount of outstanding government guaranteed debt. Interpreting government guarantees for banks as credit risk enables credit risk management techniques to be used to analyze the individual risk components and to understand the range of available policy options. Because government guarantees for bank debt only become effective once the bank’s equity is depleted, the government’s expected losses do not equal expected and unexpected losses on the bank level, which are first covered by reserves or are absorbed by the bank’s capital, respectively. 36 Thus, a bank’s expected and unexpected losses only affect the bank’s equity holders, not the bank’s debt holders or the government as a potential guarantor. However, losses in excess of expected and unexpected losses on a bank level, so called stresslosses 37, lead to the failure of a bank and would contribute to losses the government would have to expect if it decided to guarantee part or all of the bank’s debt. The interconnectedness of individual banks – both within an economy and cross border – and the correlation of financial assets held by banks are important factors for the amount of combined stress losses in the banking system at a certain point in time and the resulting government losses. Consequently, the sum of the government’s expected, unexpected and stress losses from guarantees to the banking system are a direct result of the individual banks’ stress losses (see Figure 10). 38 35 36 37 38

In practice, a default will also involve significant transaction costs, as discussed in Section 3.2.4.3. Compare BCBS (2005). In this article, we will refer to losses above the confidence level as stress losses. It should be noted that a borrower’s stress losses generally translate into expected and potentially unexpected losses from the direct lender’s perspective, independent of the nature of the borrower or lender. In that sense, in a cascading loan relationship between a bank lending to a corporation, and the corporation providing sales financing to a client, each borrower’s stress losses will translate to expected and potentially unexpected losses for the direct lender. Equity cushions and collateral on each level will decide on how far the initial losses will be passed on or not. An example for stress losses being passed on provides the case of U.S. business jet manufacturer Eclipse Aviation. The company went bankrupt in November 2008, two months after its main customer DayJet (accounting for more than 50% of Eclipse’s total order volume, or 1,400 jets) filed for bankruptcy (Financial Times, November 26, 2008). As a result, DayJet’s stress losses translated into expected and potentially unexpected losses for Eclipse’s creditors.

3.2

Banking System Exposure from a Credit Risk Perspective

Figure 10: Expected, unexpected and stress losses of banks and the government

51

52

3.

Government Guarantees und Banking System Risk

Considering these guarantees to bank creditors, a consciously chosen and managed exposure provides a complementary view to a narrower externalities perspective on government rescue measures for banks. From a narrow externalities perspective, bank bailouts are considered unwanted negative externalities per se, resulting from private contracts between banks and investors that, in times of stress, impose losses on third parties such as the wider public, the impact of which is not considered by the pricing of the original contractual agreement. 39 This view, however, is somewhat irrational in a setting where the government knows about its exposure to the banking sector and market expectations of government guarantees. Our alternative view of an exposure perspective, which is developed throughout this chapter, is not generally inconsistent with the externalities perspective. 40 Similar to the externalities view, it emphasizes the potential macroeconomic costs and distributional consequences of banking crises. These costs have been one of the major concerns for policymakers and the broader public alike and a key motivation for new regulations being implemented in response to the crisis. In that sense, it also addresses the question why the government was – seemingly unexpectedly – exposed to these costs. However, in addition, it also raises an important question about the reasoning behind accepting any level of exposure. In contrast to the externalities perspective, which per se describes the costly fallouts from banking crises as unwanted and implicitly bears the normative implication that this type of exposure should be avoided, an exposure-based view raises the question about the appropriate level of government exposure and the return required in exchange for accepting a certain level of exposure. 41 One way to determine this return would be to take the perspective of a risk-return tradeoff as supported by traditional portfolio theory, according to which any level of risk should be accompanied by an appropriate return. In this view, a government has to define its desired exposure to the banking system based on the anticipated effects on its goal of maximizing a society’s welfare along the dimensions growth and stability. 42 From this perspective, analogous to a bank’s expected loss being part of the cost of its lending activity to clients, a government’s expected loss can be considered the cost of having a financial system fostering economic growth. Finally, an exposure-based approach defines the policy actions not as purely reactive to uncontrollable systemic risk events, but rather as pro-active in deciding about the range 39 40 41 42

See Coase (1960) for the concept of externalities and Spulber (1989) for a discussion of externalities in the context of regulation. In fact, we will also consider externalities as part of potential frictions in the context of a macrofinancial portfolio allocation. This does not imply that a government should per se accept exposure to the banking system. As we will show in Section 3.4, a zero exposure level is one of the principle policy choices available. In addition,‘distributional preferences’ (compare, for example, Mishan (1969)) can be considered an important welfare dimension as highlighted by the sustained criticism of wealth transfers through the recent government rescue measures for banks.

3.2

Banking System Exposure from a Credit Risk Perspective

53

of possible outcomes a society is willing to accept for a banking system as part of the society’s economic activity. 43 In that regard, it considers the coordinative role a government can have for an inherently fragile fractional reserve financial system both by establishing respective regulatory rules ex ante and in providing stability ex interim once a crisis looms. At the same time, an exposure-perspective of regulation allows the full range of major policy options to be structured and evaluated within a single framework, which – given the extent of regulatory proposals and options currently being discussed – provides a useful tool to select from the full range of major policy options from zero exposure such as free banking or narrow banking regimes over limited exposure regulations to full exposure in the form of a nationalized banking system. Using this exposure perspective, a government can determine the appropriate regulations considering economic growth and volatility implications as well as a society’s risk preferences. 3.2.2

Structural credit risk modeling in regulatory capital determination

The principal elements of structural credit risk modeling have become an integral part of bank capital determination with the introduction of Basel II by the BCBS in 2004. This development acknowledged a rapid evolvement of credit risk modeling over the past decades, from largely univariate and multivariate accounting-driven models to more complex approaches with stronger theoretical underpinnings, including option pricing or contingent claims approaches, capital market-based models and neural network analysis. 44 The advancements of credit risk management were driven to a significant degree by Black and Scholes (1973) and Merton (1974), who introduced option pricing models to value corporate debt, modeling equity as a call option on assets and establishing the link between equity volatility and asset volatility. Based on these theoretical foundations, Kealhofer, McQuown and Vasicek, or KMV, a provider of quantitative credit analysis, developed risk models that could be commercially applied by the financial industry. 45 These models became widely accepted in the financial services industry throughout the 1990s, and, based on an industry consultation, the BCBS decided to implement a similar credit risk management approach on a bank regulatory level, complementing preceding advancements in market risk regulation. 46 On the basis of Vasicek’s (2002) extension of Merton’s single asset model and Gordy’s (2003) proof of portfolio invariance of the Asymptotic Single Risk Factor 43 44 45 46

As Rudolf (2010), p. 819 shows, the relatively frequent occurrence of financial crises over the last 70 years (averaging one major financial crisis every 13 years) points towards the interpretation that these events may be less exceptional than suggested. Compare Altman and Saunders (1998) for an overview of credit risk developments from the 1970s to the 1990s. Compare Vasicek (1987) and KMV (1999). See Rehm and Rudolf (2000) for more information on KMV’s credit risk modeling approach and Johanning and Rudolph (2000) for an overview of the process of market and credit risk management for banks. Compare BCBS (1999, 2000).

54

3.

Government Guarantees und Banking System Risk

model, the BCBS decided to introduce the so-called internal ratings-based approach for regulatory capital determination under Basel II in 2004. 47 48 49 The essential risk components of the Basel II risk-weight functions to determine capital adequacy are the probability of default, exposure at default and loss given default, which will provide the basis for our framework of regulatory policy options. 50 3.2.3

Application of structural credit risk models to government guarantees for banks

In parallel to applying structural credit models in the context of credit risk management by the financial industry, these approaches also have been employed to measure the credit risk emanating from the financial industry. As early as 1977 and 1978, Merton formalized the value of government guarantees for banks by modeling deposit insurance as a governmentissued put option, which allows the bank’s management to sell the bank’s assets if the market value of the assets falls below the book value of the deposits, as discussed in Section 3.2.1. 51 52 Since then, Merton’s model has been applied by a large number of studies. The first applications focused on explicit guarantees for bank depositors and emerged in the context of the discussion around the U.S. fixed rate deposit insurance pricing in the 1980s and early 1990s, prior to the introduction FDICIA. The FDICIA regulation demanded a shift towards a risk-based deposit insurance regime, which was implemented in January 1993. Scholars found evidence in favor of risk-based deposit insurance pricing for the U.S. banking system, as banks had very different levels of risk, implying cross-subsidies under the fixed premium regime (Marcus and Shaked (1984)). 53 Solutions proposed to this problem based on structural credit model analysis included risk-adjusted deposit insurance pre47 48

49 50 51 52

53

See BCBS (2005), pp. 4 and 5. Numerous other structural and reduced-form models have evolved in parallel, weakening assumptions around time of default or dissolving the link between default and an underlying economic explanation, for example. For an overview of structural and reduced-form models compare Bohn (2000). Compare Adams et al. (2004) for an overview of the Basel II regulatory regime. Compare Danielson (2003) for a discussion of limitations of Basel II such as procyclicality and challenges in statistically measuring risk. In addition, the Basel risk weight functions take asset correlations and maturity adjustments into account. Compare Merton (1977 and 1978). Theoretically, debt holders (or guarantors) would incur no losses if a firm immediately defaults once the market value of assets equals the book value of debt. However, Crosbie and Bohn (2003) find that firms generally continue their operations beyond this point, i.e. have a much lower so-called distress-barrier than the nominal value of their total outstanding debt. Instead, losses at the time of default generally succeed the value of equity by a significant portion of long-term liabilities. Similarly, in the context of banking, supervisors may let banks continue to operate either because closure rules may depend on outdated book values, or because of regulatory forbearance (Duan et al. (1995)). Similar evidence regarding risk variances among banks was found for Canadian banks (Giammarino et al. (1989)) and Japanese banks (Sato et al. (1990)).

3.2

Banking System Exposure from a Credit Risk Perspective

55

miums (Ronn and Verma (1986)), flexible capital requirements according to the individual bank’s riskiness (Ronn and Verma (1989)) or a combination of both (Flannery (1991)). Despite the presence of mis-incentives for risk taking in this context, which, in theory, are introduced by a fixed rate deposit insurance, banks were not found to respond with riskshifting behavior in practice (Duan et al. (1995)). For single-period models, premiums were found to be too high overall (Marcus and Shaked (1984) and Pennacchi (1987b)). However, applying a multi-period approach, this observation reversed and premiums appeared to be too low (Pennacchi (1987b)), highlighting the importance of assumptions around regulatory powers and willingness to intervene early. It was also emphasized that the observed values for guarantees based on market prices only constitute a lower bound considering additional (partial) guarantees for uninsured depositors and even equity owners (Thomson (1987)). Also, it was suggested that an optimal closure policy should mandate to intervene early and use an assets-to-deposits threshold of larger than one (Acharya and Dreyfus (1989)), which effectively limits or avoids a loss once the guarantee is invoked. 54 The second wave of option-based models in the context of government guarantees for banks focused on evaluating supervisory processes and capital regulation throughout the 1990s. In order to monitor and control bank risk, risk-adjusted examination schedules based on structural credit models were suggested in order to establish a more formal examination policy and a more timely resolution process (King and O’Brien (1991)). Capital regulation was observed to be relatively effective. Evidence was found for a significant correlation between regulatory risk-based measures and market-based measures of capital. However, deviations were observed between market and regulatory assessment of specific riskweights for individual Basel I risk categories (Cordell and King (1995)). Applying optionbased models dynamically, large U.S. banks were found to have increased their capital in parallel to increasing their risk during the late 1980s (Levonian (1991)) and Australian banks were found to have increased their capital while keeping their risk profile constant in the 1980s and early 1990s (Gizycki and Levonian (1993)). In contrast, Hovakimian and Kane (2000) find that capital regulation was not able to avoid risk-shifting, independent of assumptions about the term-length of the deposit insurance in their model. Nickell and Perraudin (2001) find a mixed picture for UK banks, some being adequately capitalized and 54

Marcus and Shaked (1984) is widely cited as one of the first studies to apply Merton’s model in an empirical context. Subsequent adjustments to and extensions of their model include, for example, the consideration of regulatory forbearance and bankruptcy expectations (Ronn and Verma (1986 and 1989)), strike price adjustments due to the consideration of bankruptcy costs (Giammarino et al. (1989) and Duan et al. (1995)) and the incorporation of bank charter value (Levonian (1991) and Gizycki and Levonian (1993)). Pyle (1984 and 1986) extends Merton’s model for a random-time audit procedure and an unlimited term insurance. Pennacchi (1987a and 1987b) empirically demonstrates the difference in put option value of a limited term insurance (single period) versus an unlimited term insurance (multiple period). Cooperstein et al. (1995), also in an unlimited term insurance setting, include bank and regulatory behavior such as endogenous capital adjustments and regulatory forbearance.

56

3.

Government Guarantees und Banking System Risk

some being undercapitalized based on derived default probabilities and implicit deposit insurance values. A third major application of option-based approaches aims to determine the value and extent of implicit and explicit government subsidies for TBTF banks via guarantees. Section 2.4.1 provides an overview of these studies, illustrating the high absolute guarantee values calculated and significant variation between studies and model calibrations. Many of these studies are motivated to examine and highlight the necessity for additional regulation – essentially arguing that positive implicit or explicit subsidy values for banks should be brought to zero – and to derive systemic risk measures, which can be applied by regulators to understand the exposure to the financial system at a specific point in time. 3.2.4

Credit risk components of banking system exposure

As implied by the previous literature’s approaches to determine government guarantees for individual banks or the overall banking system, the government’s exposure to a banking system – including its exposure via the central bank as a lender of last resort (LOLR) or via government-funded deposit insurance programs – ultimately depends on the potential losses that banks can incur in case of distress, the cushions banks have against distress in the form of capital and liquidity buffers, and the degree to which a government decides to guarantee the banking system’s creditors. 55 In other words, the government’s credit exposure can be expressed as a product of the probability of banking system distress, the banking system loss given distress, and the government’s exposure at distress, similar to the standard credit risk model. 56 Thus, we define the government’s expected loss from its exposure to an individual bank as ELGOV = PDB x LGDGov x EADGov with the government’s expected loss being the product of the probability of the bank’s distress (PDB), the loss in case of distress attributable to the government (expressed as the loss rate LGDGov) and the absolute amount of government exposure to the bank at the time of distress (EADGov). From an economic perspective, the government’s exposure includes both explicit guarantees such as official deposit insurance commitments as well as implicit guarantees for individual banks and the overall banking system to the extent they are

55

56

We use the broader term ‘distress’ in exchange for the term ‘default’ generally applied in the context of credit exposure, because governments regularly step in and rescue banks that are distressed (but have not formally defaulted yet) in order to avoid a formal default or legal bankruptcy, which potentially would trigger contagious effects. Compare BCBS (1999, 2000, 2005), KMV (1999) and Vasicek (2002).

3.2

Banking System Exposure from a Credit Risk Perspective

57

acknowledged by the government in case of distress. The government’s overall exposure to the economy’s banking system is the sum of the exposures to individual banks, as previously illustrated in Figure 10. 57 3.2.4.1 Probability of distress The probability of distress of an individual bank (PDB) defines the likelihood that a bank will become distressed in response to an idiosyncratic or systemic shock. Default events can occur due to both insolvency, i.e. the depletion of a bank’s equity basis following losses, or due to illiquidity, i.e. the banks lack of sufficient liquid assets and cash inflows to serve (unexpected) cash outflows within a certain period of time. The expected loss for a government regarding its exposure to a bank is derived based on the expected default rate under normal business conditions, or average PDB: PDB = P(Loss > Equity ∪ Cash outflows > (Available liquidity + Cash inflows)) 3.2.4.2 Exposure at distress EADGov is the amount of exposure the government has to a bank, i.e. the total amount the government will lose if the bank were to fail without any recoveries. 58 EADGov equates the sum of the government-guaranteed bank liabilities, which is equal to total liabilities minus equity and bail-inable claims. 59 EADGov = Total liabilities – Non-guaranteed liabilities = On-balance sheet liabilities + Off-balance sheet commitments – Equity – Bail-inable claims In order to capture shifts in funding prior to default, going-concern guaranteed liabilities have to be adjusted for increases in guaranteed liabilities prior to failure. Analogous to credit risk management for corporate loans, where adjusted exposure is calculated to reflect draw-downs of unused loan commitments or overdraft facilities just before failure, going-concern exposure to banks needs to be adjusted for shifts from uninsured to insured

57 58 59

Systemic (nonlinear) exposure factors are assumed to be reflected in the individual exposures. Because we will apply EADGov in the presentation of LGDGov, we discuss EADGov first, deviating from the order implied by the expected loss formula. With bail-inable claims we refer to all de facto non-guaranteed liabilities.

58

3.

Government Guarantees und Banking System Risk

sources, such as core deposits. 60 These funding shifts will be dynamic and dependent on the severity of a crisis. Depending on the degree of coverage of liabilities by the government, this could imply a shift from non-guaranteed liabilities to guaranteed liabilities, which would increase the government exposure. 61 Distress guaranteed liabilities = Going-concern guaranteed liabilities x (1 + Funding shift factor) Funding shift factor = Bail-inable claims replaced by guaranteed liabilities prior to distress / Going-concern guaranteed liabilities 3.2.4.3 Loss given distress LGDGov expresses the loss of a bank to which a government is exposed in case of distress. This includes both losses from idiosyncratic shocks and losses from systemic shocks transmitted through the financial system to a bank, such as by interbank contagion or macroeconomic shocks. Unlike a straightforward loan exposure in standard credit risk management, government banking exposure has to take into account lower seniority of equity and 60

61

A number of studies has observed this phenomena: Goldberg and Hudgins (1996 and 2002) observe decreasing levels and funding proportions of uninsured deposits for subsequently failing savings and loan associations compared to non-failing savings and loan associations during the 1980s in the USA. Failing savings and loan associations were found to attract relatively lower amounts of new uninsured deposits prior to their failure compared to non-failing savings and loan associations. Likewise, Park and Peristiani (1998) find a negative correlation between bank risk and ability to attract uninsured deposits for U.S. thrifts in the late 1980s, and Billet et al. (1998) find that riskier U.S. BHCs increase their use of insured deposits during the first half of the 1990s. Marino and Bennett (1999) find that uninsured deposits and unsecured liabilities tend to decrease in the months and years prior to failure for six large U.S. banks that failed during the 1980s and early 1990s. Jagtiani and Lemieux (2000) observe a significant decrease in uninsured deposits following a worsening financial condition after downgrades and prior to failure. Jordan (2000) observes uninsured depositors of New England commercial and savings banks starting to withdraw funds as early as two years prior to failure during the late 1980s and early 1990s. Cornett et al. (2011) observe a shift from wholesale deposits to insured deposits during the beginning of the recent financial crisis, and King (2008) finds a decreasing percentage of interbank borrowings prior to default for a sample of 1,182 failing U.S. banks during the period from 1986 to 2005. More recently, Icelandic banks Kaupthing Bank or Landsbanki shifted from more expensive wholesale funding to insured retail deposits in the UK and Continental European countries prior to default (Watterson et al. (2009) and Financial Times, February 23, 2010). Likewise, Bulgarian First Investment Bank and Portuguese Banco Espírito Santo started to raise insured retail deposits in Germany just before facing turbulences in mid-2014 (First Investment Bank experienced a bank run (Financial Times, June 27, 2014) and Banco Espírito Santo saw its parent company default on July 18, 2014 (Financial Times, July 18, 2014)). A lacking capability of a privately organized deposit insurance scheme to fulfill the ex ante promised coverage of deposits can also lead to a sudden increase in EADGov, potentially aggravating the course of a crisis by suggesting safety ex ante and increasing government exposure at distress ex interim.

3.2

Banking System Exposure from a Credit Risk Perspective

59

non-guaranteed claims of a bank. We therefore distinguish between the loss given distress for the overall bank (LGDB), which refers to the total liabilities of a bank, and the loss given distress for the government (LGDGov), which refers to the exposure amount of the government. The rate of LGDB, in line with the standard credit risk modeling of losses in the context of straightforward loan exposures, can be expressed as the total book value (BV) of assets prior to distress minus recoveries as a share of total BV of liabilities prior to distress: LGDB = (BV of assets – Recoveries) / BV of liabilities = 1 – Recovery rate In addition to recoveries, the rate for LGDGov has to consider equity and bail-inable claims, which fully assume losses prior to any losses becoming effective for the government: LGDGov = (BV of assets – Equity – Bail-inable claims – Recoveries) / EADGov = (BV of assets – Equity – Bail-inable claims – Recoveries) / (BV of assets – Equity – Bail-inable claims) = 1 – (Recoveries / EADGov) with LGDGov > LGDB for positive values of Equity and Bail-inable claims. For both loss rates, recoveries are an important component as troubled banks almost always own financial assets that can be sold or wound-down post failure (see Figure 11 for an illustration).

Figure 11: Loss given distress, recovery value and exposure at distress

Analogous to standard credit risk management approaches, loss given distress can be estimated based on current market values of the bank’s assets (i.e. market LGD) or based on

60

3.

Government Guarantees und Banking System Risk

discounted future cash flows from the assets (i.e. workout LGD). 62 However, unlike a single asset such as a corporate loan, a bank is a complex institution and, therefore, will likely always have at least some financial and non-financial assets that are not immediately marketable. In fact, a significant portion of a bank’s assets may not be immediately sellable if the overall market faces a liquidity run, as experienced during the recent financial crisis. Perfectly liquid assets on a distressed bank’s balance sheet that are not marked to market will likely entail losses if sold immediately, because otherwise the bank can be expected to already have attempted a disposal in order to avoid entering distress. As a consequence, a bank’s loss given distress will almost always have some workout components. Therefore, we will focus on workout options and consider liquidations via asset sales as part of the execution of these options. In principle, the government has three options in managing its workout recoveries. First, it can decide to wind down (WD) all of the assets in a dedicated government owned bad bank or public wind-down agency. Second, it can opt to restructure (RS) a failing bank and aim for a trade sale or a public listing via an IPO of all or part of the institution later on. Finally it can aim for an instant merger with or sale (MS) to another financial institution with sufficient capital to absorb the distressed bank’s losses and sufficient funding access to overcome any solvency and liquidity concerns. 63 The optimal choice will offer the maximum ultimate recovery value of these options for a given distress situation and will depend on the specific bank’s situation including the availability of wind-down plans and respective preparations, the market environment, the government’s risk preference as well as the government’s experience and ability to execute any of the available options. 64 Recoveries = Max(RecovWD, RecovRS, RecovMS) The recovery value from winding-down a bank (RecovWD) results from the sale or redemption of the failing bank’s financial assets within a (partially) government guaranteed institutional setup. 65 The net recovery value in the wind-down scenario depends on financial and 62

63

64 65

Compare Moody’s Analytics (2011) for a discussion of workout versus market or implied LGDs. In the context of corporate loan exposure, market LGDs can be derived by market prices of defaulting bonds or close-to defaulting bonds. We apply the term market LGD in the context of banks to the individual bank assets (for which tradable prices may or may not exist). Roland Berger Strategy Consultants (2012) uses probabilities for the individual recovery options in calculating the overall recovery value, which implies the same outcome when the decision maker is assumed to be an economically rational agent. The authors consider direct liquidation, merger/sale and wind-down as recovery options. There can be significant information costs for determining the optimal recovery option which are not explicitly included in our Recoveries formula. Different institutional and legal designs of bad bank setups have been used in history. Compare Rudolf (2012), pp. 34-41 for a comparison of different bad bank approaches chosen in Germany, Ireland, UK and the USA following the recent financial crisis and in Finland and Sweden after the Nordic banking crisis of the early 1990s.

3.2

Banking System Exposure from a Credit Risk Perspective

61

organizational considerations relating to both the bank’s asset and liability side. Market concentration and competition amongst potential buyers for financial assets that are being sold as part of the wind-down process will also affect the outcome. 66 Also, the sale of financial assets may negatively impact market prices and hence reduce the value of similar assets remaining on the bank’s balance sheet. Depending on the maturities of the individual assets, the wind-down process can take many years before a failed bank is completely dissolved, especially if a passive approach of managing assets is chosen. From a financial perspective, significant cost savings can be achieved if the maturities of the liabilities are shorter than those of the assets, which allows the government to refinance the bad bank at more favorable terms – supposing the run down agency is a publicly owned or guaranteed entity – and thus further enhance the value. While assuming bank assets and refinancing the liabilities increases the government’s debt burden and – in perfect markets – should increase the average cost of debt for the government, the impact in practice will depend on a multitude of different factors including the maturity structure of the outstanding government debt, the overall level of government indebtedness and resulting macroeconomic fundamentals. 67 68 In addition, as Rudolf (2012) demonstrates, a government owned bad bank or wind-down agency may have an advantage in waiting for the optimal point of selling assets as it faces less institutional pressures to immediately dispose assets at unfavorable prices, the value of which can be expressed as an option to wait. Given that in practice the optimal market timing will be difficult, the exact determination of potential refinancing advantages may be challenging ex ante. From an organizational perspective, the separate setup can involve additional costs especially if provided at a small scale for a single (i.e. non-recurring) instance, which oftentimes involves significant costs for external consultants and legal advisors. In return, however, regulatory costs can prove to be lower at the same time due to less strict regulatory requirements. 69 Furthermore, specific incentive problems can emerge in this context. Compared to a going-concern corporate setting where expansionary strategies oftentimes are incentive-consistent with increasing shareholder value, a wind-down scenario is a special case for bank employees, who essentially are tasked with eliminating their own jobs over time, with the employee’s success being (partly) defined by

66 67 68 69

Assets encumbered as part of collateralized funding may not be available for sale and – to the degree bank creditors not guaranteed by government guarantees have claims on these assets – may reduce the net present value of RecovWD. For example, Reinhart and Rogoff (2010) demonstrate that there is a threshold at around 90% debt-to-GDP, the exceeding of which can lead to a deterioration of macroeconomic fundamentals. The statement implies that the terms of the outstanding liabilities do not reflect a 100% government guarantee, i.e. that market participants did not perceive the bank as fully guaranteed by the government at the time the debt had been issued. Compare Rudolf (2012), p. 48.

62

3.

Government Guarantees und Banking System Risk

how quickly they resolve the remainders of the institution. 70 These aspects will be considered as additional net wind-down costs. The net present value of RecovWD, then, is the fair value of assets at distress minus the net present value (NPV) of the net wind-down costs plus the NPV of future refinancing advantages: RecovWD = Fair value of assets at distress – NPV of net wind-down costs + NPV of refinancing advantages The recovery value from restructuring (RecovRS) results from restructuring the bank, while keeping at least part of the entity intact. The degree of organization and legal separation can vary from simple guarantees for toxic assets to the establishment of a dedicated internal restructuring unit, a separate off-balance sheet special purpose vehicle or a completely separated bad bank. 71 The advantage of this option compared to a full wind-down scenario is that a potential franchise value of the entity beyond the value of the individual assets is maintained, and may be realized through a sale later on. In return, the refinancing will continue to be carried out by the bank as a privately chartered (though publicly owned) enterprise, and, while the direct government ownership will certainly improve refinancing conditions, the bank is likely not to achieve the same low level of refinancing costs a public wind-down agency would achieve. Also, more generally, the question is whether government ownership implies inefficiencies, or internalities, as lending decision may be influenced by political control, or because employee incentive systems may be influenced by politics. 72 RecovRS = NPV(Going-concern cash flows + refinancing advantages – internalities) The recovery value from a merger or sale of a distressed bank (RecovMS) to another financial institution allows to immediately realize the value of the financial assets and the goingconcern value of the entity. However, finding a suitable buyer for the failed bank or part of it over a very short time horizon (oftentimes a weekend) may prove difficult, depending on the characteristics of the troubled institution as well as the structure and state of the overall banking system. An outright sale or merger will be easier if the overall banking system is relatively stable and potential buyers are well capitalized and funded. The relative (strategic) attractiveness of the target will also play an important role as will available information 70

71 72

Compare Unger (2012), p. 271 for potential challenges in incentivizing employees of a bad bank and incentive differences compared to going-concern banks. The author suggests to apply nonmonetary incentives. To the degree that a wind-down agency acts as a servicer for (other) failing banks, the mis-incentives regarding implicit job eliminations over time may be less accentuated. Compare Clark et al. (2012) for a discussion of costs and benefits implied by these alternatives. Compare Wieandt and Moenninghoff (2011) for a case study on the restructuring of failed German lender Hypo Real Estate Holding. The issue of potential internalities arising in the context of public bank ownership will be further discussed in Section 3.4.3.3.

3.2

Banking System Exposure from a Credit Risk Perspective

63

about the possible extent of losses at the troubled bank. Both the purchase price offered (RecovMS) and the likelihood of finding a suitable buyer will decline as the scope of a crisis increases and, in turn, increase the more the bank’s distress is an isolated case of failure. RecovMS = Fair value of assets at distress + NPV going concern premium 3.2.5

An exposure-based framework of principle policy choices

Building on the credit exposure perspective developed in the previous sections, a government has a variety of options to manage its exposure to banks along the three main credit risk components, ranging from zero exposure in a free banking setup without any government intervention, to a full exposure setting in which the government runs a nationalized banking system. The different regulatory approaches emphasize different credit risk components as the primary lever to manage or limit the government exposure, while, of course, also affecting the overall risk profile of a bank. 73 Table 5 provides an overview of the principle regulatory options across the desired overall exposure levels and along the individual risk components. In order to achieve a zero exposure level to the banking system, a government has to make sure that one of the exposure risk factors equals zero. Technically, as long as not at least one of the three risk components equals zero, the government cannot claim a zero exposure. Different approaches have been suggested for bringing to zero each of the three risk components, i.e. a free banking environment or different forms of so-called narrow banking systems. On the other side of the spectrum, a government can nationalize its banking system and consequently would face full exposure (EADGov) both from an equity holder and a creditor perspective. For this policy choice, LGDGov and PDB are determined internally according to the government’s economic policy approach, for example by setting the respective credit policies and loan loss provisioning guidelines. In between these two extremes of zero and full exposure, multiple regulatory approaches exist to manage each of the three credit risk components. Minimum capital and liquidity requirements are means to lower PDB. Structural restrictions to bank’s investments in assets (e.g. by requiring a minimum borrower quality or minimum loan-to-value ratios) limit LGDGov by ensuring a higher 73

Inherent in the logic of the structural credit risk model are some interconnections between the different credit risk components. Generally, capital not only decreases the probability of distress, but also constitutes a deduction item for losses and for exposure from a government perspective. Likewise, volatility changes assumed due to structural restrictions not only lower losses given distress, but also the likelihood of distress. In contrast, government exposure limitations can be installed without affecting the overall risk parameters of a bank from an overall bank investor perspective, while of course lowering expected losses and the likelihood of loss materialization from a government perspective. Some particularities regarding interactions between different regulatory approaches are discussed in Section 3.4.5.

64

3.

Table 5:

Government Guarantees und Banking System Risk

Principle policy options to manage exposure to the banking system

recovery value of the bank’s assets. Likewise, a Pigovian tax can be viewed as lowering the net LGDGov after accounting for tax revenues received ex ante reflecting an individual bank’s riskiness. Furthermore, as discussed in Section 3.2.4.3, efficient resolution and wind-down authorities can enhance the recovery value in a wind-down scenario and thus also lower LGDGov. Finally, bail-inable debt, contingent convertible capital and pre-funded resolution or deposit insurance funds allow EADGov to be lowered. The advantage of an exposure-based framework is that it allows to categorize all individual regulatory policy options within a single framework and illustrates how they impact government exposure to the banking system. In other words, while there may be some interaction or hybrid regulatory approaches (which we will further discuss in Section 3.4.5), each regulatory proposal can be assessed based on its impact on the three risk components PDB, LGDGov and EADGov. 74 Thus, the framework allows to structure the regulatory discourse, which is currently characterized by fundamental reconsideration of regulatory approaches and a variety of different policy suggestions. Compared to alternative distinctions of regulatory approaches relating to quantity versus price regulation, rule-based regulations 74

In fact, the more complex and encompassing a regulatory proposal is, the more scrutiny should be applied regarding how the approach will alter government’s exposure to banking sector risk.

3.3

Banking System Exposure from a Sovereign Portfolio Perspective

65

versus built-in automatisms, or mere categorical comparisons, an exposure-based framework better addresses the issue of government exposure and associated risk, a question which has been the major impetus for the current regulatory overhaul. Likewise, a better understanding of the impact mechanisms of individual regulatory proposals on government guarantee levels and economic benefits addresses recent claims to increase cost-benefit analysis in financial regulation. 75 While the framework provides different regulatory approaches to achieve a specific level of exposure, it does not in itself provide an indication of what exposure level is most desirable for a government from an overall economic perspective. Also, when deciding between different regulatory options for a defined level of exposure, it does not provide guidance on the relative benefits of one or the other policy approach from a macroeconomic perspective. In order to develop a more complete view on the inherent tradeoffs of the regulatory options with regard to economic growth and stability, we extend in Section 3.3 the exposure perspective to the context of a sovereign portfolio view, within which a government has to choose its banking system exposure level and, more generally, decide about policy choices, from an portfolio optimization perspective, considering economic growth, stability and economic tradeoffs implied by market and government frictions. 3.3

Banking System Exposure from a Sovereign Portfolio Perspective

3.3.1

Introduction to sovereign portfolio management

A government’s credit risk exposure to a banking system can be viewed within the context of the sovereign’s overall portfolio, considering the individual sectors of an economy as interconnected portfolios of assets, liabilities and guarantees. 76 77 Extending the traditional focus of macroeconomics on macroeconomic flows, a national economy’s portfolio of assets, liabilities and guarantees can be brought into the form of a stylized accounting balance sheet, based on which typical portfolio properties can be identified (see Figure 12). 78 75 76

77 78

Compare Posner (2013) and Posner and Weyl (2014a and 2014b) for proposals and discussions of regulatory cost-benefit analysis for financial regulation. Compare Gray et al. (2007), Gray and Malone (2008) and Gapen et al. (2005) for a general discussion of macroeconomic portfolio management based on a contingent claims approach, and Das et al. (2013), pp. 18-19 for an overview of alternative sovereign asset and liability management approaches. We use the terms 'sovereign' and 'government' interchangeably in the context of our macroeconomic model discussion, always referring to the broad sense of the word including monetary institutions or central banks. Compare Buiter (1983) for the components of a sovereign balance sheet and Allen et al. (2002) for an overview of different approaches to balance sheet analysis for sovereign portfolio management. Following Bodie and Brière (2014a), we include the present value of target wealth for future generations, or global sovereign surplus, in Figure 12 and our subsequent formalization.

66

3.

Government Guarantees und Banking System Risk

Figure 12: Banking system exposure within a sovereign’s overall portfolio

A sovereign can apply this balance sheet perspective not only to understand the risks from and sensitivity to economic shocks or assess its ability to meet social and economic goals, but also to determine the risks transferred between economic sectors – such as risks from the financial industry to the government via implicit or explicit guarantees – in an overall economic context. 79 Government guarantees for banks enter the government’s balance sheet as a put option, which the banking sector is ‘long’ and the government is ‘short’. This perspective allows governments to evaluate the implications a banking sector has in terms of financial risk and potential resulting liabilities. 80 Following a formalization of the optimization problem developed by Bodie and Brière (2014a) in the context of sovereign wealth fund asset allocation, we will illustrate the economic tradeoffs a government faces when deciding on a level of banking system exposure and the regulatory means to achieve the desired exposure level. 81 We are not concerned with an optimal asset allocation for financial assets held by the government, for which the contingent claims approach presented here has been suggested. Instead we use our specification in Section 3.3.3 to illustrate the major tradeoffs implied by individual regulatory policies. These tradeoffs will be introduced briefly in the following section. 3.3.2

Macroeconomic tradeoffs implied by regulatory policy choices

For the following discussion of the different regulatory policy choices, we will assume fundamental tradeoffs in the context of market and regulatory benefits (or market versus regulatory failure), presuming that a society has to decide between different paths of opti79 80

81

Compare Bodie and Brière (2014b), p. 51 and Gray et al. (2007), p. 5. The quantification and inclusion on-balance sheet of guarantees raises the question about the implicit versus explicit nature of guarantees. A policy of constructive ambiguity favoring implicit guarantees that are not explicitly accounted for on a sovereign’s balance sheet would imply offbalance sheet commitments from an accounting perspective. Compare Bodie and Brière (2014b), pp. 13-15.

3.3

Banking System Exposure from a Sovereign Portfolio Perspective

67

mizing present and future wealth. 82 83Assuming that government guarantees for the banking system cannot fully be integrated into a contractual agreement – a setup most easily represented by implicit government guarantees, we distinguish between externalities and internalities, which we will further detail in the next two paragraphs of this Section. 84 Similar to impact studies conducted by regulators in the context capital and liquidity requirements, we define factors impacting the risk-return tradeoff implied by individual policy choices. Externalities can be of negative or positive nature. Negative externalities in the context of banking risk and government guarantees for banks are bailout costs borne by taxpayers without receiving any direct compensation. 85 Positive externalities in the context of a banking system are the economic benefits of having a financial system fostering economic growth, which could, for example, lead to higher tax income for the government (which could be translated into refunds for tax payers). In the context of impact studies on higher minimum capital requirements conducted as part of the Basel III process, regulators approximated the amount and cost of banks’ additional capital and estimated by how much investment in the broader economy would reduce due to higher borrowing costs, assuming that banks pass on higher costs of capital in the form of higher loan rates. 86 For the purpose of our model we consider these effects broadly as an economic contribution of a fractional reserve financial system that transforms short-term secure deposits into longer-term risky

82 83

84 85

86

The purpose is to highlight potential tradeoffs associated with the individual regulatory options, not to quantify these tradeoffs, which has to be left to further empirical analysis. In addition to a society’s risk return preferences, there may be other factors driving a government’s decision on banking risk exposure and respective policy choices, for example how important a society considers property rights and what the general understanding of a government’s role in the economy is. Compare Spulber (1989), pp. 46-48 and 54-55. Conceptually, the exposure perspective challenges the view of government guarantees as externalities, because these costs can be seen as not emerging in the full absence of related transactions, but rather as emerging in return for expected economic growth. In that sense, costs from government guarantees arise due to a consciously chosen level of exposure defined by the choice of regulations and are expected to be compensated by higher tax revenues for an increased exposure level. As part of the Basel III process, the FSB and BCBS examined the economic impact of increased capital and liquidity requirements and Total Loss Absorbing Capacity (TLAC) requirements through separate impact studies in 2010 and 2015, respectively, finding that a one percentage point increase in target ratio of tangible common equity leads to a maximum decline in the level of GDP by 0.17% from the baseline path, and a 25% increase in the holding of liquid assets relative to total assets is associated with a median decline in GDP in the order of 0.08% relative to the baseline trend (BCBS (2010) and BCBS (2015)). The Federal Reserve Bank of Minneapolis (2016) approximates bail-out probabilities and growth implications of higher capital requirements based on similar approach.

68

3.

Government Guarantees und Banking System Risk

assets. 87 The result is a tradeoff between the net effect of positive and negative externalities – i.e. a fractional banking system’s positive effects on economic growth minus the value of the guarantees provided by the government – and the level of economic volatility implied by the fragility of the fractional reserve financial system. Internalities refer to allocative inefficiencies arising from the costs of writing contingent contracts in the presence of risk, due to costs of monitoring and information processing under information asymmetries, and because of imperfectly observable behavior. 88 In the context of financial intermediation, we will consider inefficiencies resulting from political involvement in the capital allocation process. This is due to the fact that individual political goals may deviate from pure economic wealth maximization in the short- and mediumterm, or, more generally, that markets are assumed to be more efficient in allocating economic resources than governments. We consider these inefficiencies to manifest themselves both in economic growth rates through misallocation of financial resources, and in the net present value of a banking system’s equity due to a decreased bank profitability. The four variables used to capture these effects in our specification are the following: FIBE = Financial intermediation benefits, or, positive value of asset transformation performed by a fractional reserve banking system, such as increased economic growth, positive implication on employment, and additional tax income for the government EXTE = Negative value of externalities due to government guarantees for the banking system such as bank bailout costs in times of distress INTE = Negative value of internalities resulting from government involvement in the financial intermediation process, leading to lower economic growth, such as resource misallocation due to investment restrictions for banks or structural regulations limiting banks’ activities INTB = Negative value of internalities arising in the context of government ownership of banks, leading to a decrease in bank equity values due to resource misallocation by the banks owned by government 3.3.3

Banking system exposure management from a portfolio perspective

In order to model the different tradeoffs inherent in the individual policy choices, we begin with a stylized government accounting balance sheet equation based on a general contin87

88

A variety of scholarly contributions exist to explain benefits of a financial system. For example, Diamond and Dybvig (1983) demonstrate the value of banks in providing liquidity insurance. Diamond and Rajan (1999) show that guaranteed deposits issued by banks to depositors serve as a monitoring device, restricting a bank’s ability to absorb too much of the rents from its assets. We will discuss the potential benefits of a fractional reserve financial system further in Section 3.4.1.2 on minimum capital requirements. Compare Spulber (1989), p. 54.

3.3

Banking System Exposure from a Sovereign Portfolio Perspective

69

gent claims approach. Following Bodie and Brière (2014a), the government’s assets consist of financial assets the government owns (FA) and the present value of net future fiscal assets PV(NFA), which consists of the present value of future revenues minus the present value of nondiscretionary spending minus the present value of government guarantees to the banking system (following the general sovereign contingent claims approach, we rearrange to net out government guarantees from net present value of future fiscal assets). On the liability side, the government has foreign liabilities (FL) and domestic liabilities (DL). The adjusting item on the liability side is the global sovereign surplus (GSS), which is the present value of the wealth left to future generations. FA + PV(NFA) = FL + DL + GSS

(1)

The present value of the net fiscal assets is the sum of the present value of fiscal assets (FIA) and the present value of fiscal liabilities (FIL) minus the present value of guarantees to the banking system (GB). Fiscal assets include future tax income, revenues, fees and seigniorage. Fiscal liabilities are non-discretionary expenditures such as education, welfare and health care costs, military expenditures and core infrastructure. 89 Government guarantees for banks enter the equation to the degree they are provided depending on the regulatory regime the government decides for, expressed by the factor γ: PV(NFA) = PV(FIA) – PV(FIL) – PV(γGB) with 0 ≤ γ ≤ 1 the level of government guarantee for the banking sector For the purpose of considering different degrees of government ownership in banks in our simplified model, we distinguish between non-bank assets (FANB) and banking assets (FAB) with η being the degree to which the government owns equity of the domestic banking system. Distinguishing between non-bank and bank assets allows us to take into account the potential internalities INTB that emerge in the context of government ownership in banks. FA = FANB + ηFAB with 0 ≤ η ≤ 1 the share of government ownership in the banking sector Based on Bodie and Brière (2014a), domestic liabilities include the monetary base and debt issued in local currency. Foreign debt constitutes the debt issued in foreign currency. DL = MB + DLC FL = DFX Following Bodie and Brière (2014a), we formulate the return on the global sovereign surplus (GSS) and the optimization problem. 90 Solving equation (1) for the present value of 89 90

Compare Gray et al. (2007), pp. 10-11 and Gapen et al. (2005), p. 32. Equations (1) to (4) based on Bodie and Brière (2014a), pp. 48-50.

70

3.

Government Guarantees und Banking System Risk

future wealth, GSS equals the financial assets plus the present value of fiscal assets minus the guarantees for banks and foreign and local currency debt: GSS = FA + PV(NFA) – FL – DL

(2)

As the authors show, the return r on the global surplus then can be expressed as the return on financial assets plus the return on net fiscal assets minus interest payments on domestic and foreign currency debt: 91 rGSS = αrFA + (1 - α)rNFA – rβDFX – r(1- β)DL

(3)

with α the fraction of sovereign assets dedicated to financial assets and the remainder being the net fiscal surplus, and with β the fraction of total sovereign liabilities dedicated to foreign debt and 1-β the fraction of liabilities dedicated to domestic debt. Bodie and Brière (2014a) formulate the optimization problem. With constant risk aversion, the sovereign’s utility U on the GSS is maximized by: U(GSS) = 1/(1- ρ) x GSS1-ρ with ρ > 0 the relative risk aversion According to the authors, the sovereign then has to choose the regulatory option maximizing the expected return: Max (μGSS + ½(1- ρ) σ2GSS)

(4)

A formal differentiation of the optimization problem will require a large number of observed inputs due to the complex nature of the various regulatory options involved and would need to be subject to a separate impact study exercise with empirical observations. However, in order to demonstrate the relevant tradeoffs from a policy perspective we construct a simple stylized one-period model. In this simple model, we introduce the four variables presented in the preceding section, namely financial intermediation benefits (FIBE), the value of government guarantees for banks, or externalities (EXTE), internalities resulting from the government involvement in the financial intermediation process (INTE) and internalities resulting from government ownership in banks (INTB). We define the first three variables as factors altering the return on net fiscal assets – assuming that government taxes are a function of economic activity – and INTB as a factor lowering the return on financial holdings in banks by the government: rGSS = α(rFANB +(rηFAB x (1 – INTB))) + (1 - α)rNFA x (1 + FIBE) x (1 - EXTE) x (1 - INTE) – βrFL – (1- β)rDL

91

(5)

The authors’ formula reflects a portfolio of financial assets with different investment horizons.

3.4

Regulatory Policy Options and Their Economic Tradeoffs

71

The exact specifications of our simulation are presented in Appendices A.3 to A.9, including input variables and assumptions, government exposure calculations, quantification of externalities and internalities, the banking system and government balance sheet outcomes as well as the GSS risk-adjusted return outcomes for the individual policy options. With the primary purpose of the model being a demonstration of the interaction between the implied internalities and externalities, we will use the results in the form of graphical illustrations for our discussion in Section 3.4 to depict general tradeoffs pertaining to the individual regulatory options. 3.4

Regulatory Policy Options and Their Economic Tradeoffs

This section discusses the general regulatory policy options from our framework along the three main exposure factors PDB, LGDGov and EADGov and attempts to illustrate the related economic tradeoffs based on the outcomes of our simple sovereign portfolio model. 92 3.4.1

Management of probability of distress

3.4.1.1 Zero exposure: Narrow banking with all-equity financed banks Narrow banking with all-equity financed banks is an extreme approach to making the banking system less susceptible or even unsusceptible to bank runs. If all liabilities are equity claims provided by private investors, the probability of distress is de facto reduced to zero. Thus, the approach attempts to eliminate any government guarantees while installing full accountability for investors in banks. Initial proposals for all-equity financed narrow banks were made in conjunction with the second form of narrow banking, which restricts banks to investing in secure government-issued securities (see Section 3.4.2.1). In such a setting, allequity financed narrow banks would perform capital allocation within the economy, while the other part of the banking system with assets restricted to government securities would perform critical payment functions. 93 More recently, standalone all-equity narrow banking systems have been suggested as well, based on claims that payment transactions could soon be carried out without any deposit holdings due to technological advancements that will allow fractions to be deducted from a payer’s equity account and credited to a payee’s equity account. 94 Independent of how to provide a secure payment function, the fundamental assumption behind all-equity financed narrow banking is that the value of asset transformation as well as monitoring and information processing performed by a fractional reserve 92 93 94

The presentation of the individual regulatory option’s tradeoffs is based on a calibration of the input variables to yield the same risk-adjusted return for all policy choices at a relative risk aversion level of ρ = 2. We will change that assumption in Section 3.4.4. See also the Chicago Plan of 1933 and proposals by Kay (2010) and Kotlikoff (2010), which we will discuss in Section 3.4.2.1. Compare Cochrane (2014), pp. 199-200.

72

3.

Government Guarantees und Banking System Risk

banking system with short-term debt is outweighed by the cost of externalities resulting from government guarantees. We will further examine this claim in Section 3.4.1.2. on minimum capital requirements. One interesting natural experiment in this context is the recent emergence of peer-to-peer finance, allowing individual investors to perform loan investments and equity financing directly. This surge in direct investment activity could be regarded as preliminary evidence that alleged investor preferences for (short-term) secure debt are overrated. However as Moenninghoff and Wieandt (2013) discuss, funds invested via peer-to-peer lending platforms increasingly stem from financial institutions, rather than retail investors. This would imply a different conclusion, namely that investors in fact derive some value from asset transformation by involving financial institutions as intermediaries, even within a peer-to-peer financial setting. Furthermore, even when assuming that a large share of non-institutional investors accepts or even prefers to invest directly in equity and credit, giving up alleged advantages of short-term available deposits, the question remains whether all or part of the direct investments will definitely not be covered by the government if losses incur at a large scale to the broad public, and, as a result, the financing of the economy via direct investments would be at stake. 95 Conceptually, all-equity financed narrow banking is an extreme version of the regulatory approach establishing minimum capitalization requirements. In our illustration of the major tradeoff between financial intermediation benefits and externalities resulting from guarantees (see Figure 13), all-equity financed narrow banking is at the very right of the xaxis, with both financial intermediation benefits and externalities approaching zero. In our model, overall volatility is only partially reduced by increasing capital, explaining a remaining difference between raw return and risk-adjusted return even for the narrow banking setup. The important factor for a decision around this policy option is the slope of the risk-adjusted curve from its peak, which expresses how much lower the all-equity financed narrow banking returns are compared to the optimum capital level which yield the maximum risk-adjusted return. Aside from individual components of the financial system, such as, for example, mutual funds and closed-end funds, which have long complemented banks in a financial system, all-equity financed narrow banking is not currently implemented as an overall banking system configuration. While narrow banking proposals have been discussed and ultimately rejected by the UK Independent Banking Commission, 96 they did not become a notable part of the general Basel reform discussion, nor of the specific regulation for G-SIBs. 95

96

An example for government intervention into retail losses is the Hungarian government’s decision to install a borrower relief scheme at the end of 2013 covering more than 300,000 borrowers who took out Swiss Franc denominated mortgage loans and saw their interest burden rise due to exchange rate fluctuations. According to this scheme, borrowers can postpone 25% of the outstanding loan amount with the government covering half of the interest payments for the postponed amount (Financial Times, December 17, 2013). Compare Independent Commission on Banking (2011), pp. 44-45.

3.4

Regulatory Policy Options and Their Economic Tradeoffs

73

This figure shows the effects of capital ratio adjustments on intermediation benefits, cost of guarantees and global sovereign surplus based on our model. Increasing capital is assumed to reduce the amount of financial intermediation benefits, as asset transformation is reduced to a smaller share of liabilities (illustrated by the negative slope of FIBE). Also, increasing capital is assumed to reduce the cost of externalities as the probability of distress decreases (illustrated by the positive slope of EXTE). Further, increasing capital is assumed to reduce the volatility of economic output and related returns on net fiscal assets (illustrated by the decreasing distance between the expected return and the risk-adjusted return on the expected global sovereign surplus with increasing capital requirements).

Figure 13: Capital requirements, economic value and risk-adjusted return

3.4.1.2 Limited exposure: Minimum capital and liquidity requirements Minimum capital and liquidity requirements are means to decrease the likelihood of bank failure. In the context of exposure risk, higher capital requirements for a bank increase the bank’s loss absorbency and its distance to default, and decreases PDB respectively, which results in a decreasing government exposure ELGov. Capital for banks can be regulated on a risk-weighted basis, commonly expressed by capital ratios (referring to capital as a percentage of risk-weighted assets), and on an unweighted basis relating capital simply to overall assets, referred to as leverage ratio (calculated as the multiple of exposures to capital). From an exposure perspective, minimum capital requirements refer to going-concern capi-

74

3.

Government Guarantees und Banking System Risk

tal. 97 History shows a great range of bank capitalization ratios, with a decreasing tendency over the last 170 years. 98 The level of the appropriate minimum capital requirements was a point of contention even before the recent financial crisis. The question is whether the private market optimum deviates from the socially optimal outcome, and, if so, how the socially optimal outcome weighting economic costs and benefits can be determined. 99 While it is by and large accepted that there are economic benefits of financial intermediation, the exact benefits are hard to quantify. A variety of studies have examined the economic benefits of fractional reserve financial intermediation and found significant positive impact on growth. 100 In our model, the effects of capital ratio adjustments are threefold. First, increasing capital is assumed to reduce the amount of financial intermediation benefits, as asset transformation is reduced to a smaller share of liabilities (illustrated by the negative slope of the FIBE curve in Figure 13). Second, increasing capital is assumed to reduce the cost of externalities as the probability of distress decreases (illustrated by the positive slope of the EXTE curve in Figure 13). Finally, increasing capital is assumed to reduce the volatility of economic output and related returns on net fiscal assets (illustrated by the decreasing distance between the expected return and the risk-adjusted return on the GSS with increasing capital requirements in Figure 13). The optimum level of capital requirements thus depends on the intermediation benefits, cost of guarantees and the society’s level of risk aversion. Independent of the exact parameters determining social welfare, demands for unweighted

97

The Basel III capital requirements define different layers of capital, i.e. core equity tier 1 (which includes common shares and retained earnings), additional tier 1 (which includes preferred shares and high-trigger contingent convertible capital) and tier 2 capital, such as subordinated debt and low-trigger contingent convertible capital (compare BCBS (2011a), pp. 12-18). From an exposure perspective, it is important to recognize the difference between going-concern capital (i.e. most of the aforementioned components including high-trigger contingent convertible capital) and gone-concern capital, which consists of low-level trigger contingent convertible capital. While going-concern capital avoids bank failure by lowering PDB, gone-concern capital reduces government exposure in case of non-viability by lowering EADGov. Therefore, minimum capital requirements relate to going-concern capital. Gone-concern contingent convertible capital will be discussed as part of the EADGov-related measures. 98 For example, U.S. commercial banks’ capital ratios decreased from approximately 55% in 1840 to approximately 35% in 1870 to below 20% in 1900, further decreasing to below 10% in the second half of the 20th century (Berger et al. (1995)). 99 Diamond and Rajan (2000) provide a model showing that bank capital requirements are set considering bank safety, deposit refinancing costs and liquidation ability. Thakor (2014) and Admati et al. (2011) argue that private and social benefits deviate. 100 For example, Frenkel and Rudolf (2010) estimate that the introduction of a leverage ratio for the international banking system will have a significant economic impact including a reduction in lending and a slowdown in economic activity. The BCBS (2010) finds that one percentage point increase in banks’ capital ratios leads to a maximum decline in the level of GDP of about 0.17% from the baseline path, Ashcraft (2005) provides evidence that the failure of banks has negative effects on real economic activity.

3.4

Regulatory Policy Options and Their Economic Tradeoffs

75

capital ratios as high as 25% (i.e. a leverage ratio of 4) 101 would involve a massive reallocation of capital. To illustrate the consequences an unweighted 25% capital ratio implies, we briefly examine the two generic options banks have in response to changes in capital requirements, namely recapitalization and asset liquidation. 102 The first option, recapitalization, would require raising significant amounts of capital in order to continue to support the approximately USD 110 trillion in global banking assets. 103 Banks would have to quintuple their capital from currently approximately USD 5.5 trillion to approximately USD 27 trillion, which would equal additional required capital of USD 22 trillion. 104 Considering global financial and non-financial wealth of approximately USD 220 trillion as of 2011 and an average growth rate of global wealth between 4.1% (approximated 5-year average) and 7.7% (approximated 10-year average), the bank capital increase would absorb between 1.3 and 2.4 years of global wealth creation. 105 Alternatively, following the second option, asset liquidation, banks would only be able to support USD 22 trillion in assets with their current capital base of USD 5.5 trillion. 106 Given the current level of approximately USD 110 trillion in assets, this would imply a liquidation of assets of approximately USD 88 trillion, and a resulting capacity to only support deposits of approximately USD 16.5 trillion, less than one fourth of the current level of deposits of approximately USD 70 trillion intermediated by banks globally. 107 Assuming that bank liabilities in that setup would only consist of

101 Compare Admati and Hellwig (2013), p. 181. Likewise, the Federal Reserve Bank of Minneapolis (2016) suggests a risk-weighted capital ratio of 38% for systemically relevant banks, which – based on the risk-weighting multiplier of 1.75 assumed by the authors – translates into an unweighted capital ratio of approximately 22%. 102 Compare Admati et al. (2011), p. 10. The authors also suggest balance sheet expansion as a third option, which, for the purpose of our discussion, can be considered an even more extreme version of the recapitalization outcome. The authors argue that higher capital requirements lowers banks’ cost of equity reflecting lower default risk. 103 Global bank assets as of 2011 (IMF (2012), Statistical Appendix, p. 11). 104 Global bank tier 1 capital as of 2011 (McKinsey (2012)). The 2011 tier 1 capital figure already reflects an increase by USD 2 trillion since 2007. The additional capital needed results from the following calculation: USD 110 trillion assets x 25% capital/assets – USD 5.5 trillion existing tier 1 capital = USD 22 trillion additional required tier 1 capital. This calculation assumes that the 25% capital demand relates to tier 1 capital. 105 Global wealth data from Credit Suisse (2013), p. 18. The calculation for the number of years required to generate the capital gap is: USD 220 trillion x 1.041^(2.4) = USD 22 trillion, and USD 220 trillion x 1.077^(1.3) = USD 22 trillion. 106 Global bank tier 1 capital as of 2011 (McKinsey (2012)). The calculation for the supported amount of assets is USD 5.5 trillion existing tier 1 capital / 25% capital/assets = USD 22 trillion supported assets. 107 Bank asset as of 2011 (IMF (2012), Statistical Appendix, p. 11). Bank deposits and bank tier 1 capital as of 2011 (McKinsey & Company (2012). The calculation for the amount of assets to be liquidated is USD 110 trillion existing assets – (USD 5.5 trillion tier 1 capital / 25% capital/assets) = USD 110 trillion existing assets – USD 22 trillion supported assets = USD 88 trillion assets up for liquidation.

76

3.

Government Guarantees und Banking System Risk

deposits, bank balance sheets could support only USD 22 trillion deposits in total. 108 This means that investors would have to re-invest USD 48 trillion they previously held in secure deposits (or approximately one fifth of global wealth) directly into equity securities, debt securities, or other asset classes. 109 To summarize, demanding an increase in banks’ capitalto-assets ratios to 25% from current levels implies that, on average, households would have to either invest amounts equal to all of their increase in personal wealth over the next one to three years in bank stocks, or accept that they can only hold liquid and secure deposits of approximately one third of the amount they used to do. This simplified illustration shows how capital ratio determination fundamentally affects personal wealth and portfolio choices of the participants of an economy, and how significant the implications of seemingly small adjustments can be. Minimum capital requirements became a core tool for bank regulators beginning with the introduction of Basel I in 1988. 110 Since then, the scope and calculation of riskweighted assets have been further refined (Basel II), and, in response to the recent financial crisis, the quality and amount of capital requirements have been increased (Basel III). Likewise, additional capital requirements have become an integral part of the new regulation dealing with G-SIBs, with progressive capital surcharges amounting to 1.0% to 2.5% core tier 1 capital based on an indicator approach measuring the individual bank’s systemic relevance in comparison to its peers. 111 Unlike minimum capital requirements, formal regulatory liquidity requirements were not introduced internationally until after the recent financial crisis. Prior to the Basel III changes, liquidity management was left largely to bank managers, with supervisors focus-

108 The calculation for the amount of deposits that can be supported by USD 5.5 trillion in capital is USD 5.5 trillion tier 1 capital / 25% capital/assets - USD 5.5 trillion tier 1 capital = USD 16.5 trillion deposit capacity. With a current ratio of deposits to non-equity liabilities significantly below 1, the actual amount of deposits that could be supported would be even lower. 109 We will discuss the possibility of investments in secure government securities in Section 3.4.2.1. Expressing the deposits to be re-invested as a share of global household wealth disregards the fact that a certain share of deposits is held by corporations and other institutions, which are ultimately owned by private households. The shift of non-household deposits to direct investments thus would potentially increase the volatility of household direct investments in corporations and other institutions, but not provide a shift to be performed by households directly. 110 The formal use of bank capital regulation emerged in the U.S. in the 1980s. Prior to that, bank supervisors applied informal and subjective factors focusing on other aspects than capital. A measure requiring minimum capital to the population served by a bank was already enacted in the 1860s. Capital-to-deposits ratios and capital-to-assets ratios were examined after the Great Depression, but not applied (Burhouse et al. (2003)). 111 In its current reform proposal, the Federal Reserve Bank of Minneapolis (2016) also suggests tiered risk-weighted capital requirements ranging from 10% for community banks over 23% for banks not deemed systemically important by the U.S. Treasury and 38% for banks deemed systemically important by the U.S. Treasury.

3.4

Regulatory Policy Options and Their Economic Tradeoffs

77

ing on ensuring that appropriate risk management practices were in place. 112Although the structured credit risk model does not explicitly measure liquidity risk, it has been shown that liquidity and capital can be seen as (partial) substitutes. 113 Formal liquidity rules for the overall banking system were introduced as part of Basel III and will become fully effective after an observation period of several years. 114 The Liquidity Coverage Ratio is used to regulate short-term liquidity over 30 days, requiring banks to hold sufficient stock of high quality liquid assets to meet expected net cash outflows in stress. The Net Stable Funding Ratio requires banks to have a sufficient amount of longer-term stable funding to cover the required stable funding over a one-year period. For G-SIBs, additional regulatory liquidity requirements were only considered in the initial stages. While policy makers, recognizing that in some circumstances liquidity surcharges could reduce risks posed by G-SIBs, left the door open to also consider liquidity related measures at the G20 meeting in Seoul in November 2010, no liquidity related additional regulation was included in their final proposal. 3.4.2

Management of loss given distress

3.4.2.1 Zero exposure: Narrow banking with assets restricted to government securities In managing or limiting exposure to the banking system, a government can also aim to reduce the loss given distress. For zero exposure to the banking system, a government can attempt to bring LGDGov to zero by limiting bank assets to risk-free securities, such as government issued bonds. In case of rescuing and taking over a distressed bank, a government would effectively receive its own liabilities, which it could then redeem at par independent of the current market value (contracting its own balance sheet). 115 There are a variety of proposals for narrow banking with government assets. 116 The so-called Chicago Plan de112 For example, internal liquidity risk management and stress testing was supervised as part of the national implementation of the EU capital requirements directive (ECB (2007), p. 21-22.). A variety of informal rules for internal liquidity risk management exist for a long time. For example, the Louisiana Banking Act of 1842 required banks to cover their deposits and notes issued by gold or bills of exchange and promissory notes with a maturity of equal to or less than 90 days (Pennacchi, 2012). Another example is the so-called golden banking rule developed by German economist Otto Hübner in 1854, suggesting that the maturities of a bank’s assets should not exceed the maturity of its liabilities in order to avoid the risk of illiquidity. 113 For example, Diamond and Rajan (2000) show that the bank capital structure reflects effects of bank capital on liquidity creation. Compare Schanz (2012) for an approach to jointly calibrate capital and liquidity requirements. 114 Banks have to disclose the Liquidity Coverage Ratio beginning in 2015, with a phase-in period until full implementation in 2019. The Net Stable Funding Ratio is set to be introduced as of 2018. 115 This logic abstracts from the concept of debt valuation adjustments on a bank level. 116 Compare Pennacchi (2012) for a comprehensive overview of different narrow banking policy proposals.

78

3.

Government Guarantees und Banking System Risk

veloped by a group of economists in Chicago after the Great Depression suggests a 100% reserve requirement for deposits. 117 Likewise, during the U.S. savings and loan crisis, publicly administered ‘deposited currency accounts’ marketed via the governmental postal office system (Tobin (1987)) or requirements to hold U.S. Treasury bills as collateral equal to the amount of deposits (Merton and Bodie (1993)) were suggested. More recently, when the regulatory agenda was being set in the UK following the financial crisis, Kay (2010) proposed a so-called utility banking restricted to investments in government securities. Further proposals recommend the government to be involved in loan rating and auctioning (Kotlikoff (2010)) or put forward that a country’s monetary authority, via a public private partnership model, should calibrate banks’ asset quality requirements over time, optimizing credit supply (Ricks (2011)). While, in principal, following these proposals in a strict manner would allow default-free bank liabilities with government guarantees for banks to be reduced to essentially zero, the scope of application is a critical issue. Several of the aforementioned concepts rely on the existence of a separate all-equity financed banking system, which would perform the allocation of capital across the economy, i.e. distributing capital from surplus units to deficit units. 118 In contrast, variants of the Chicago Plan established to solve credit provision by banks that issue debt to the government. 119 Taken together with a 100% reserve requirement, this puts the government effectively in the position of the financial intermediary. Likewise, Kotlikoff’s (2010) and Ricks’ (2011) proposals require allocative decisions to be made by the government. From a perspective of economic tradeoffs, the question of how broadly this policy choice is applied has important implications. A pure narrow banking system with assets restricted to government securities relies on the government’s involvement in credit allocation, which results in a tradeoff between government inefficiencies and lower volatility (assuming financial intermediation benefits from the creation of debt are maintained and externalities are reduced to zero). In contrast, in the banking system resulting from a combination of the two narrow banking approaches, government inefficiencies would play a smaller role, and the tradeoff would include lower financial intermediation benefits and lower externalities as discussed in Section 3.4.1.1. In our model, the role of the government in carrying out intermediation of financial resources can be observed in the significantly increased amount of government’s holdings of financial assets (Appendix A.7), representing its involvement in providing credit. Respectively, we assume inefficiencies relating to economic growth resulting from the government’s role in intermediation. In return for the inefficiencies arising in this context, more stability would 117 See Phillips (1995) for a history of the Chicago Plan. 118 For example, Tobin (1987) suggests investment trusts and low-risk commercial banks next to the proposed deposit currency accounts, and Kay (2010) proposes all-equity financed banks to complement the suggested utility banks. 119 Other versions of the Chicago Plan also consider investment trusts similar to all-equity financed banks as complements to narrow banks with assets restricted to government securities (compare Benes and Kumhof (2012)).

3.4

Regulatory Policy Options and Their Economic Tradeoffs

79

be expected and, in order to make the option attractive from a risk-return perspective, even required. The tradeoff is displayed in Figure 14, showing that for strong reductions in volatility, narrow banking with assets restricted to government securities achieves the desired return (at the very left of the graph), however, that, as soon as volatility is assumed to be higher and approaches the standard levels used in our model, this policy choice becomes increasingly unattractive from a risk-return perspective.

This figure shows the tradeoff between inefficiencies from government involvement in financial intermediation and reduction in economic volatility. For a situation where the government involvement in financial intermediation leads to a strong reduction in volatility, narrow banking with assets restricted to government securities or structural restrictions for the banking sector can achieve a desired positive return (at the very left of the graph). In contrast, assuming only a low volatility reduction is achieved by government intervention (with resulting volatility levels at standard levels modeled) this policy choice becomes increasingly unattractive from a risk-return perspective.

Figure 14: Asset-related restrictions and expected volatility levels

Some proponents of narrow banking with assets limited to government securities have suggested that while the government could act as a provider of secure and liquid securities, it would not necessarily have to get involved in allocation decisions, given that there is a genuine refinancing demand by the government and sufficient capacity implied by current levels of government indebtedness. In essence, this argument relates to the logic discussed in Section 3.4.1.2 in the context of a 25% capital requirement. With approximately USD 43 trillion in global public debt outstanding, a narrow banking setup relying on government

80

3.

Government Guarantees und Banking System Risk

securities without any government involvement in the financial intermediation process would require approximately USD 27 trillion of deposits to be shifted into direct investments. 120 In addition, of course, all of the current USD 43 trillion investments in government debt would have to shift into non-government debt or other asset classes to allow the depositors to invest in government securities. Thus, suggesting a narrow banking setup with government securities as bank assets either implies strong assumptions about investor liquidity and security preferences, or will involve some of the inefficiencies discussed earlier. A second critical aspect in the context of banks holding government securities has become obvious during the recent financial crisis, which evolved to a sovereign debt crisis in Europe. In times of stress, the nexus between sovereigns and banks can result in a negative feedback loop, within which lower sovereign debt mark-to-market values can lead to deteriorating bank balance sheets. As a consequence of these devaluations and due to the potential erosion of the government’s support for banks, the banking system’s funding costs can increase, negatively impacting profitability. The deterioration in the banking system’s solvency again increases the potential contingent liability for the government. This feedback mechanism can be similarly applied to connections of the local banking system to foreign sovereigns, and to connections of the local banking system to foreign banks that are connected to foreign sovereigns. 121 While this issue extends the context of narrow banking, its potential volatility implications in times of stress have to be considered when examining the risk-return tradeoff of a narrow banking policy option with assets restricted to government securities. In practice, money market mutual funds, a deposit-like product, which emerged in the USA as a response to Regulation Q – a regulation that limited interest rates on bank deposits – can be considered a form of narrow banking with assets restricted to government securities, certificates of deposit, and other highly liquid and low-risk securities. However, the recent financial crisis showed, first, that the unquestioned quality of the securities held by money market mutual funds is crucial in times of stress, and, second, that a regulator in deciding to apply narrow banking elements has to be very clear about the degree of coverage of different products and institutions. Money market mutual funds were considered extremely safe, though officially not guaranteed by the government. However, quite contrary, they turned out not to be entirely safe and, consequently, banks issued guarantees to those funds they administered, essentially making them part of the guaranteed banking system.

120 Global public debt level as of 2013 (World Bank Public Sector Debt Statistics, Q4 2013). The amount of deposits to be shifted is calculated as USD 70 trillion deposits – USD 43 trillion government debt = USD 27 trillion. The amount is only an approximation because the amount of global deposits is based on 2011 figures (McKinsey (2012)). 121 Compare Merton et al. (2013), p. 26 and Jobst and Gray (2013), p. 34.

3.4

Regulatory Policy Options and Their Economic Tradeoffs

81

In the context of post-crisis regulatory debate, narrow banking with assets restricted to government securities was not seriously considered. As mentioned in Section 3.4.1.1, related proposals were discussed but ultimately rejected by the UK Independent Banking Commission, and, on an international level, have not found their way into the Basel reform discussion neither for the general banking system nor for G-SIBs. 3.4.2.2 Limited exposure: Structural restrictions Structural restrictions in the context of bank regulation refer to a variety of approaches that attempt to restrict banking activities or bank business models to limit risk while still allowing banks to generate financial intermediation benefits and issue run-prone debt. The concept of structural restrictions is broad and can range from specific individual exposure restrictions that apply to the overall banking system, such as large exposure restrictions defined by Basel III, to fundamental constraints regarding entire business models, such as the elimination of the universal banking model by the U.S. Glass-Steagall Act of 1933. Suggested structural restrictions include proposals to limit loan provision by financial holding companies to separate subsidiaries financed by uninsured deposits (Litan (1987)), to restrict so-called monetary service companies to investing only in low risk assets (Pierce (1991)) or to require safe collateralized assets equal to the amount of deposits accepted (Pollock (1992)). In the wider sense, the regulatory design of the market (infra-) structure and corresponding transparency requirements constitute structural interferences with implications on bank’s business models as well, in which the government’s economic value as a standard setter may be relatively high. Structural restrictions constitute a partial government intervention in the capital allocation process, leading to an alteration of the market equilibrium and implying effects on economies of scale and scope. 122 Regulators, in return, have argued with market failure and claimed to increase social benefits with structural measures. 123 In our model, consistent with our modeling of government inefficiencies in the context of narrow banking with asset restrictions, we apply the inefficiency to only part of the banking system’s balance sheet (Appendix A.4). Bank assets, however, are assumed to remain off balance sheet from a government’s perspective (Appendix A.7). Similar to the narrow banking setup, which limits banks to investing in government securities, the fundamental tradeoff for structural restrictions is inefficiencies versus the increase in stability in the banking system (Figure 14). However, unlike narrow banking, the required volatility reductions compared to the standard assumptions in order to achieve a specific risk-return level are lower, and sensi122 Compare, for example, Diamond (1984) for economies of scale and Schmid and Walter (2009)) for economies of scope. 123 For example, UK regulators declared to improve the social benefits of the banking system in that they correct returns from investment banking activities which in their view strained value creation in commercial banking due to bank internal cross-subsidization (compare Haldane (2012), p. 9).

82

3.

Government Guarantees und Banking System Risk

tivity to volatility changes is higher, reflecting only partial application of inefficiencies from government involvement in the intermediation process on the one side, and potential losses in case of distress due to a non-zero LGDGov on the other side. The recent application of structural regulatory measures on a national and regional level constitutes a shift in paradigm, breaking with a tendency to deregulation and less interfering regulatory measures over the past decades. 124 Current regulatory reform proposals focus on limiting asset-related risk by restricting or excluding proprietary trading, hedge fund and private equity activities from regulated banks’ balance sheets (U.S. Volcker Rule), by ring-fencing retail operations, consumer and business lending and trade finance from certain investment activities (UK Vickers Commission), or by ring-fencing significant proprietary trading and market making activities from the remainder of a bank (EU Liikanen Expert Group). The Basel III specific regulations for G-SIBs do not include any elements of structural regulation, with regulators concluding at the G20 meeting in Toronto that diverse business models and organizational structures are themselves sources of systemic resilience. 3.4.2.3 Limited exposure: Pigovian tax The concept of a Pigovian tax traces back to the observation of positive and negative externalities – or deviations between marginal social net product and marginal private net product – and the conclusion that taxes imposed by the government could limit these externalities and thus increase economic welfare. 125 However, while price regulations like a Pigovian tax were considered a compelling response to market failure in theory, it was pointed out that government intervention potentially leads to unintended consequences due to a limited understanding and knowledge of the involved market participants’ utilities and the actual post-intervention outcomes. 126 127 Furthermore, due to a political process involving different interest groups, agents implementing the government intervention may not intend to achieve the social optimum. 128 In the context of bank regulation, the costs caused by systemically relevant institutions during the financial crisis sparked a renewed interest in Pigovian taxes by academics. Following the observation that systemic interconnections 124 Compare Gambacorta and van Rixel (2013), p.1. 125 Compare Pigou (1924), pp. 183-185, 224-225. 126 Compare Pigou (1954), p. 6 and Coase (1960), pp. 15-16 for unintended consequences in the context of a Pigovian tax, and Merton (1936) more generally for unintended consequences. Likewise, Buchanan argues that because a social optimum principally is determined by market forces, the external determination of an optimum is impossible as the market optimum is not observable (Buchanan (1969a), pp. 174-175 and Buchanan (1969b), p. 74). 127 In addition to unintended consequences and regulatory failure, there are settings (industry structures) in which quantity restrictions generally can achieve superior outcomes compared to price restrictions (compare Weitzman (1974)). 128 Compare Pigou (1954), p. 6: “ 'Fancy' finance, …, whatever its theoretical attractions, has, at all events in a democracy, dim practical prospects.”

3.4

Regulatory Policy Options and Their Economic Tradeoffs

83

between financial institutions were a key contributing factor to the recent crisis, the taxation of banks based on their individual systemic risk contribution was proposed. Suggested approaches for systemic risk measures include, for example, how much the value at risk of the overall banking system changes once a specific bank moves from a median state to a distress state (CoVar), or, from the opposite perspective, by how much an individual institution is likely to be undercapitalized once the overall banking system approaches an undercapitalized state (Systemic Expected Shortfall). 129 Referencing to the fundamental instability resulting from run-prone debt in a fractional reserve financial system, taxes on shortterm debt also were suggested. 130 In practice, post-crisis implementation of a Pigovian tax has been relatively limited. On national levels, temporary taxes were introduced to compensate ex post to cover losses from failed banks, as for example in the UK. On a global level, regulators discussed the introduction of levies at the G20 meeting in Toronto in June 2010, however, they were dismissed as potentially reducing loss absorbency of banks at a critical point in time, and eventually were not further considered within the Basel framework or as part of the international FSB guidelines. More generally, elements of Pigovian taxes have been present in a variety of regulatory approaches attempting to account for different risk and cost contributions, such as the progressive nature of G-SIB surcharges or risk-based deposit insurance premiums discussed in Section 3.2.3. In our model we reflect the important aspect of implementation efficiency by using a factor between zero and 1, which is applied to externalities. With a factor of one, all externalities stemming from the banking system are covered by the tax and the net LGDGov is zero after consideration of the ex ante collected taxes. 131 The risk-adjusted return under a Pigovian tax policy is a linear function of the implementation efficiency given a certain capitalization level (Figure 15). The Pigovian tax value is highest when potential externalities are largest. In that sense, Pigovian tax and capital requirements can be partial substitutes in lowering the government’s exposure. 132 While the degree of implementation efficiency is important at low capitalization levels, higher levels of capital requirements and the related lower PDB decrease the exposure (and the amount of externalities) so that the different levels of tax implementation efficiency converge. The black line in the right graph of Figure 15 (low implementation efficiency) resembles the risk-adjusted return of mini129 For CoVaR compare Brunnermeier and Adrian (2011) and for systemic expected shortfall Acharya et al. (2016). 130 Compare Perotti and Suarez (2011) and Cochrane (2014), p. 217-218. 131 In our model, we abstract from the discussed unintended consequences and second-round effects. 132 For example, the Federal Reserve Bank of Minneapolis (2016) suggests to implement a Pigovian tax for shadow banking institutions in connection with higher capital requirements for the regulated banking system in order to avoid regulatory arbitrage between these two financial sectors. Notably, the authors favor a Pigovian tax over capital requirements citing difficulties to calibrate and implement capital requirements for shadow banks.

84

3.

Government Guarantees und Banking System Risk

mum capital requirements in Figure 13 (which also does not consider any Pigovian taxes). The (theoretical) optimum of a highly implementation efficient Pigovian tax is superior to the optimum offered by capital requirements, because it essentially allows to generate financial intermediation benefits without accepting any externalities (disregarding secondround wealth effects).

The left chart exhibits the risk-adjusted return under a Pigovian tax policy as a linear function of implementation efficiency at a given capitalization level. A low implementation efficiency (due to unintended consequences, a limited understanding of market participants’ utilities or agency conflicts) implies low economic growth, while higher degrees of implementation efficiencies imply higher economic growth. The right chart shows the risk-adjusted return under a Pigovian tax depending on implementation efficiencies and capital requirements. The value of a Pigovian tax is highest when potential externalities are largest (i.e. capitalization levels are low). The degree of implementation efficiency is important at low capitalization levels. Higher levels of capital requirements and the related lower PDB decrease the exposure (and the amount of externalities) so that the different levels of tax implementation efficiency converge. The black line in the right graph of Figure 15 (low implementation efficiency) resembles the risk-adjusted return of minimum capital requirements in Figure 13 (which also does not consider any Pigovian taxes). The (theoretical) optimum of a highly implementation-efficient Pigovian tax is superior to the optimum offered by capital requirements, because it essentially allows to generate financial intermediation benefits without accepting any externalities.

Figure 15: Pigovian tax, taxation efficiency and capital requirements

3.4.2.4 Limited exposure: Resolution powers and wind-down authorities Resolution powers and wind-down authorities are legal, procedural and institutional preparations to allow the government to actively manage its exposure to banks in distress. The government’s principle recovery alternatives are an orderly wind-down of the bank’s assets, a restructuring and subsequent privatization of the bank, and an immediate merger with or sale to another bank (compare Section 3.2.4.3). Crisis management preparations, supervisory coordination as well as bank organizational factors contribute to the government’s ability to perform adequate crisis management and to choose the most value enhancing recovery alternative. First, preparations on behalf of the supervisory authority include a dedicated resolution authority, appropriate laws and powers, and the development of crisis manage-

3.4

Regulatory Policy Options and Their Economic Tradeoffs

85

ment capabilities and routines. Ex ante procedural preparations and transparency on behalf of banks include the development of recovery and resolution plans (so-called living wills) and respective resolvability assessments by supervisors. 133 Second, global coordination aims at international harmonization of insolvency and resolution laws as well as the establishment of crisis management groups and institution-specific cross-border cooperation agreements. 134 From a credit exposure perspective, eliminating cross-border legal restrictions and international political frictions in resolving banks can be seen as approaches to lower settlement risk of a government’s exposure to a banking system with crossjurisdictional activities. 135 Third, in order to support wind-down planning and immediate resolution, the organizational setup of bank liabilities has also been discussed. Proposals range from a single point of entry for all liabilities at a single entity or holding level, to socalled subsidiarization, which suggests supporting largely separate local entities with local funding. 136 Principally, the ex interim and ex post enhancement of the recovery value by resolution powers and wind-down authorities reduces government exposure without necessarily negatively impacting financial intermediation benefits, aside from legal, organizational and administrative costs. In our model, the ex post net recovery rate enhancement as a percentage of EADGov is assumed such that the risk-adjusted return of this policy option is calibrated to the outcomes of the other policy choices at the chosen level of risk aversion. The level of ex post net recovery enhancement reflects the benefit of having readily available functioning resolution powers and authorities in case of distress and determines the overall return improvement. In practice, the focus on this option by regulators has increased significantly. Following the un-orderly failure of Lehman Brothers in September 2008, which caused systemic spillovers to large parts of the global financial system, regulators worked on the aforementioned components of resolution powers and wind-down authorities within two streams of the Basel and FSB regulatory process for a G-SIB specific regulation. First, the establishment of supervisory colleges as part of enhanced supervision was suggested at the G20 summit in London in April 2009, and enacted by the G20 members in November 2011. Second, the FSB was tasked to support contingency planning in April 2009, the core out133 Compare Financial Stability Board (2011a). For bank resolution plans in the context of internationally harmonized legal resolution regimes compare Avgouleas et al. (2013). 134 Earlier attempts to harmonize international bank insolvency and liquidation rules exist. For example, after the failure of Pakistani bank BCCI in 1991, the BCBS suggested principles for cross-border bank resolution including timing, communication, supervisory interaction and coordination (compare BCBS (1992)). The 2001 EU Winding-Up Directive mandated a single entity insolvency regime for banks incorporated in the European Economic Area. 135 We will further elaborate on this issue in Section 3.5.1. 136 Tradeoffs in the context of these structural regulations of liabilities include bank expansion efficiency, tax efficiency and business focus on the one hand, and resolvability on the other (compare Fiechter et al. (2011)).

86

3.

Government Guarantees und Banking System Risk

comes of which were endorsed in November 2011, including elements of resolution tools and frameworks, resolution plans, crisis management groups and information sharing. While these changes constitute significant advancements in resolving large and complex banks, the effectiveness in practice under system-wide stress has yet to be proven. 3.4.3

Management of exposure at distress

3.4.3.1 Zero exposure: Free banking A third approach for a government to establish a zero exposure level to the banking system is to bring to zero EADGov by denying any LOLR function in case of distress. Such a setting prevails in free banking, which can be defined as an environment with no or barely any specific monetary and banking-related regulations, and, most importantly, with no LOLR role carried out by a public entity. 137 Instead, bank debt holders have to either expect losses in case of bank failure and would price these losses into their investment decisions ex ante, or, alternatively, private LOLR arrangements provide guarantees for a bank and ensure at least some stability in case of crises. Historical private arrangements replacing a public LOLR function include unlimited liability of bank stockholders, industry clearing house associations and coordination by dominating industry participants. 138 However, while these private solutions can moderate some shocks, they lack the coordinative power and financial capacity of a sovereign with the ability to impose taxes. Historical variants of free banking with different degrees of regulatory freedom existed in several dozens of countries around the world throughout the 17th to 19th centuries, including the Scottish free banking era from 1716 to 1845 or the antebellum U.S. free banking period from the 1830s to the 1860s. 139 In free banking, PDB and LGDGov are to a large degree determined by market forces. 140 Scholars controversially discussed to which degree free banking would nowadays results in a fractional reserve banking system or not. 141 While some scholars argue that a lack of 137 Compare Bordo (1990), p. 19 for alternative views on the LOLR function. 138 Friedman and Schwartz (1986), p. 302 describe the stabilizing effect of unlimited stockholder liability for banks during the Scottish free banking era. Banks were owned by established members of the community with substantial wealth and reputation at stake. Bordo (1990), p. 22 presents the argument that clearing house associations can offset information asymmetries potentially leading to bank runs. An example for a dominant actor in the industry coordinating in times of crisis and thus providing stability is J.P. Morgan who intervened during the 1907 Bankers’ Panic in the USA (see Rudolf (2010), pp. 821-822 for a description of the crisis). Certain Swiss banks’ partners face unlimited liability (compare Dermine (2009), p. 93). 139 See Schuler (1992), pp. 40-45 and Hickson and Turner (2004), p. 906 for comprehensive overviews of different free banking regimes. 140 To the degree restrictions exist these will also determine the equilibrium outcome. 141 Compare Bordo (1990), p. 21. Bodenhorn (2003), p. 10 evidences a comparably low maturity transformation of only 68 days for U.S. banks in the year 1855, indicating a moderating impact of a relatively low maturity transformation on PDB.

3.4

Regulatory Policy Options and Their Economic Tradeoffs

87

unintended consequences resulting from too much regulation would introduce stability to the system (compare Bordo (1990)), others argue that, because of bank’s asset transformation function, a public LOLR is a prerequisite for a stable banking system (Goodhart (1987) and Friedman and Schwartz (1986)). To the degree that banks continue to perform asset transformation in a free banking environment, volatility can be assumed to increase as a result of the lack of government guarantees providing stability in times of stress. 142 In our model we assume that a free banking system would continue to operate as a fractional reserve system, resulting in financial intermediation benefits, but no externalities. As an offsetting factor, we assume a strong increase in volatility to reflect the lack of stabilization measures provided due to the absence of government guarantees. In post-crisis regulatory discourse, Wieandt’s (2014) suggestion of enacting a constitutional bailout prohibition similar to constitutional balanced budget amendments already in place in several European countries and in the U.S on the state level would allow for a regulatory regime with a zero exposure level similar to free banking. 143 While the aim of abandoning any government guarantees was high on the agenda, free banking proposals including drastically less regulations were not actively discussed by international standard setters. 3.4.3.2 Limited exposure: Bail-inable claims Bail-inable claims such as bail-inable debt, contingent convertible capital and pre-funded resolution funds are gone-concern instruments to reduce the government’s exposure at distress when a bank is no longer viable or considered by supervisors to be likely no longer viable. While all three instruments achieve an ex ante reduction in EADGov, they apply different mechanisms in doing so. Bail-inable debt allows a regulator to write-down and convert into equity unsecured and uninsured creditor’s claims. The bail-in of unsecured creditors follows a ‘liability cascade’ considering the seniority of outstanding debt and equity instruments. 144 On a global level, the concept of TLAC has been developed by the FSB, which requires G-SIBs to have loss absorbing instruments of 16-20% of risk-weighted assets and 6% of total in excess of capital requirements. The equivalent instrument developed by European authorities is the minimum requirement for own funds and eligible liabil142 For example, Kupiec and Ramirez (2013) find evidence for a long-lasting negative effect on economic growth caused by bank failures. 143 Similar to Wieandt’s (2014) suggestion of a constitutional prohibition of bank bailouts, the Federal Reserve Bank of Minneapolis (2016) proposes that the U.S. Treasury Secretary certifies that individual banks are no longer systemically relevant, which would exempt these banks from a higher capital requirement of 38% on a risk-weighted basis. 144 Conlon and Cotter (2014) find that, even if bail-in regimes for unsecured creditors had existed for European banks during the recent financial crisis, the aggregate impairment charges of EUR 534 billion would still have been borne mostly by equity holders and subordinate debt holders. Senior debt holders, due to the seniority of their claims, would only have experienced losses in Austria, Greece and Italy.

88

3.

Government Guarantees und Banking System Risk

ities (MREL), which requires European G-SIBs to hold loss-absorbing instruments of at least 8% of their liabilities. Contingent convertible capital enables a regulator to convert or write-down instruments based on pre-defined conversion triggers or regulatory judgment in order to support the gone-concern resolution process. 145 The current outstanding contingent convertible capital instruments still vary greatly with regard to their characteristics, such as conversion triggers and conversion or write-down mechanisms. Notably, these instruments have not yet been tested during a financial crisis and – if held by other systemically relevant financial institutions – could also be a source of contagion and financial instability. Privately financed pre-funded resolution funds constitute an additional funding source a resolution authority can access to cover losses amidst the resolution process of a bank. Pre-funded resolution funds may allow for Pigovian tax elements in their contributions, and, in the case of the Eurozone, are currently also considered to contribute to a cross-border resolution mechanism and potential burden sharing. Generally, pre-funded resolution funds allow for distributing financial burdens over time, avoiding peak funding requirements during a crisis as implied by ex-post funding of resolution funds. Scholars have controversially discussed bail-inable debt and contingent convertible capital, with some regarding them as a useful means to impose losses on bank investors (French et al. (2010) on contingent convertible capital) and others criticizing the complexity and the implied tax-driven motivation of hybrid instruments (Admati et al. (2010)). The question of complexity and practicality of contingent convertible capital points towards a more general challenge in ensuring a zero EADGov. From a government’s perspective, it may prove difficult to credibly commit ex ante to not intervene in case of crises, because the short-term benefits of intervention may appear to outweigh long-term costs such as the creation of moral hazard, especially in the context of short-term political election cycles. 146 This time consistency problem emerges as severe financial crises typically occur less frequent in a certain jurisdiction than political election cycles. In our model, we assume that 50% of the non-equity liabilities are bailed in at distress, reflecting first legal implementations of the FSB guidelines that indicate exceptions and caps for the degree to which unsecured creditors will be bailed-in. From a government perspective, the resulting reduction in EADGov significantly lowers the amount of externalities. Because some guarantees remain for non-bail-inable claims, bail-in solutions would naturally require some degree of economic volatility reduction in order to meet a similar risk-adjusted return compared to a free banking system with a zero EADGov. This required 145 Compare Avdjiev et al. (2013) for an overview on contingent convertible capital. We refer in the context of limiting EADGov to gone-concern contingent convertible capital with low triggers. Going-concern contingent convertible capital (i.e. instruments with high triggers) are, from an exposure perspective, means to increase capital and thus lower PDB, and therefore are included in the discussion about minimum capital requirements. 146 Compare Moenninghoff and Wieandt (2011) for a discussion of the TBTF doctrine and moral hazard arising in this context.

3.4

Regulatory Policy Options and Their Economic Tradeoffs

89

relative reduction in volatility, based on our model assumptions, is comparably low as shown in Figure 16, which compares the risk-adjusted returns of the three EADGov-oriented policy approaches depending on the corresponding relative volatility decrease.

This chart compares the risk-adjusted returns for the three regulatory regimes free banking, bail-inable debt/contingent claims and a nationalized banking system depending on the volatility reduction achieved by the provision of government guarantees in each of the three policy options. The chart illustrates that each of the three policy options requires a different level of volatility reduction by government guarantees for it to be attractive. In other words, the chart shows that the two regulatory options relying on government guarantees require a certain minimum reduction in volatility to be attractive compared to the free banking system. The returns in a free banking system are independent of any assumptions around volatility reduction by government guarantees as no government guarantees are provided. A regulatory regime with bail-inable debt or contingent claims is more attractive than a free banking system if the inherent partial government guarantees achieve to reduce economic volatility to some extent. This point is the intersection of the return graph for free banking and the return graph for bail-inable debt/contingent claims. If the reduction in volatility from government guarantees is not large enough to compensate for the cost of guarantees (to the left of the intersection of the return graph for free banking and the return graph for bail-inable debt/contingent claims), free banking is more attractive than a regime with bail-inable debt or contingent claims. If the reduction in volatility by the partial government guarantees is higher (to the right of the intersection of the return graph for free banking and the return graph for bail-inable debt/contingent claims), a regime with bail-inable debt/contingent claims is more attractive than a free banking system. A regime with a nationalized banking system results in high inefficiencies because of the assumed internalities from the government’s involvement in the financial intermediation process and from government ownership in banks. In order to achieve risk-adjusted returns comparable to free banking and bail-inable debt/contingent claims regimes, the government guarantees provided in this regulatory regime need to lower economic volatility significantly. If the volatility reduction by government guarantees in a nationalized banking system is very high, a nationalized banking system achieves higher risk-adjusted returns than a free banking system, as shown at the very right of the chart.

Figure 16: Degree of government guarantees and relative volatility reduction compared to free banking system

90

3.

Government Guarantees und Banking System Risk

Following the G20 leaders’ calls to never let banks blackmail governments and states again, bail-inable debt, contingent convertible capital and pre-funded resolution funds have moved to the center stage of regulatory attention. As part of the key attributes of effective resolution regimes developed by the FSB, all three bail-in instruments are becoming core elements of the currently implemented resolution regimes. 3.4.3.3 Full exposure: Nationalization of the banking system A banking system owned and fully guaranteed by the public would result in full exposure of the government to the banking system, leading to an EADGov equal to all banking system liabilities. In this setting, the government would control banks and, in its role as a principal, would determine bank safety and soundness via internal arrangements. PDB and LGDGov thus would be direct results of government decisions and – given the government’s fixed EADGov – the major determinants of the government’s overall exposure and resulting expected loss. In this setting, government ownership replaces contracts between different parties and, as a result, eliminates any externalities from public guarantees. Instead, potential friction and agency costs would arise in the form of internalities. In practice, government owned banks constitute a significant part of the global banking system with their equity summing up to approximately 42% in 1995 compared to 59% in 1970. 147 In Germany, publicly owned Landesbanken and savings banks constitute a large share of the banking system. 148 Scholars have interpreted the impact of government ownership in banks in different ways. Political incentives were found to interfere with economic growth. For example, La Porta et al. (2002) observes that government ownership of banks is associated with lower per capita income growth and lower productivity growth. Sapienza (2002) observes that state-owned banks are a mechanism for supplying political patronage and Dinç (2005) finds evidence for political motivations influencing actions taken by government-owned banks. 149 Likewise, agency problems such as corruption and misallocation have been seen as disrupting the implementation of well-intended political proposals, as demonstrated by Banerjee (1997). In contrast, other authors emphasized the fact that publicly owned banks address market failures (Stiglitz (1993)). 150 While empirical evidence indicates that public ownership is associated with lower economic growth (e.g. La Porta et al. (2002)), this view has been partly challenged once more weight is given to institutional factors (Körner and Schnabel (2010)). In our model, because of assuming internalities from government involvement in the financial intermediation process and from government ownership in banks, the nationaliza147 Compare La Porta et al. (2002). 148 Compare Sinn (1999) for a discussion of German state banks. 149 Compare Tanzi (2011) for a general discussion of the role of a government and implications for economic growth. 150 Compare Sapienza (2004) for different views on how government ownership affects economic outcomes.

3.4

Regulatory Policy Options and Their Economic Tradeoffs

91

tion of the banking system results in comparably high inefficiencies. Externalities, in contrast, are considered to be zero in this setting. In order to achieve risk-adjusted returns comparable to other policy options despite the assumed inefficiencies, the volatility would have to significantly decrease compared to the other two policy choices that focus on EADGov, which is illustrated by the fact that the risk-adjusted return curve for the nationalized banking system intersects with the risk-adjusted-return curve for the free banking system at the very right in Figure 16. Despite the temporary nationalizations of banking institutions during the recent financial crisis – with some of these assets still being held by governments five years after the bailouts took place – claims for permanent nationalization of banks were limited in the mainstream regulatory discourse. Some scholars suggested the nationalization of banks in connection with a reduction of fractional reserve asset transformation and changes in lending policies (Moseley (2011)) or demanded a nationalization of banks to eliminate externalities and promote public economic goals (Epstein (2010)). It has been argued that permanent public ownership of depository institutions is needed because of implicit government guarantees for these banks (Buiter (2008)) and because systemically relevant banks would be too difficult to regulate with other policy measures (Alperovitz (2012)). Government ownership as a regulatory policy option was not included in the international Basel and FSB reform agenda nor in any major official national regulatory effort. Quite to the contrary, incentive mechanisms of the recent bailout packages encouraged banks to redeem the government as a shareholder sooner rather than later both in the USA and the EU, and EU competition laws require governments to re-privatize stakes in banks acquired during the recent financial crisis within a defined time frame. 3.4.4

Growth-stability tradeoff

The various regulatory policy options laid out allow the government to define and manage its exposure to the banking system focusing on the individual exposure components, or levers. Different tradeoffs are implied by the individual policy choices, which determine financial intermediation benefits as well as inefficiencies due to market frictions and government involvement in capital allocation. Altogether, these factors determine economic growth. The major offsetting factor used to balance different growth outcomes from these tradeoffs is volatility, making the simplifying assumption that all regulatory options yield a similar risk-adjusted return for a defined level of risk aversion of ρ = 2. As shown in Figure 17, varying assumptions around the society’s risk preferences alter the attractiveness of individual policy options because of their implied volatility profile. Policy choices with higher levels of guarantees and correspondingly assumed lower levels of volatility perform relatively better if a society’s assumed level of risk aversion is high. In contrast, regulatory settings that imply only partial government guarantees, such as

92

3.

Government Guarantees und Banking System Risk

This figure compares the different policy choices in a risk-return framework. Different policy choices are more or less attractive based on a society’s level of risk aversion. Our model makes the simplifying assumption that all regulatory options yield a similar risk-adjusted return for a defined level of risk aversion of ρ = 2. At lower levels of risk aversion, a free banking regime or bail-in regimes are comparably more attractive. At higher levels of riskaversion, asset restrictions, banking system nationalization, structural regulations and capital regulations are comparably more attractive.

Figure 17: Risk-adjusted returns for policy choices and degree of risk aversion

bail-in regimes, or no guarantees at all, like in a free banking setup, yield the highest riskadjusted returns for relatively low levels of assumed societal risk aversion. Until now, we assumed in our discussion that, for a given level of risk aversion, the regulatory policy choices had similar risk-adjusted return outcomes. In Figure 17, this calibration is expressed by the fact that all risk-adjusted return curves intersect for the relative risk preference of ρ = 2. Abandoning the assumption of risk-return-neutrality of the individual policy choices for the modeled level of risk aversion lets us recognize how different the involved tradeoffs regarding economic growth can turn out. As one would expect, assuming a higher level of financial intermediation benefits and increasing government inefficiencies leads to a dominance (upward shift) of more market-oriented regulatory choices over policy options with higher government involvement (downward shift), as shown in

3.4

Regulatory Policy Options and Their Economic Tradeoffs

93

Figure 18. 151 With stronger financial intermediation benefits, narrow banking proposals become unattractive regardless of the level of risk aversion. Likewise, the nationalization of the banking system is dominated by limited exposure solutions. Aside from the attractiveness of free banking for very low levels of risk aversion – reflecting that we did not adjust volatility assumptions – bail-in regimes, resolution and wind-down authorities and a Pigovian tax are the economically dominant policy choices, because we assumed that they allow for financial intermediation benefits while limiting government involvement in the capital allocation process.

This figure compares the different policy choices in a risk-return framework assuming a higher level of financial intermediation benefits than in figure 17. As a result, the assumed risk-return neutrality assumed before (at ρ = 2) no longer holds. Instead, bail-in regimes, free banking and capital regulations are the dominant choices depending on the society’s level of risk aversion.

Figure 18: Risk-adjusted returns with high financial intermediation benefits and high government inefficiencies

151 Figure 18 represents the results after adjusting the assumptions about financial intermediation benefits and government inefficiencies (as laid out in Appendix A.3 to A.9) by a factor of 3.

94

3.

Government Guarantees und Banking System Risk

Accordingly, lowering assumptions about intermediation benefits and considering government involvement in capital allocation to bear less internalities would shift the curves of regulatory choices with high government involvement upwards, and more market-oriented regulatory options downwards, leading to a different set of optimal choices. In addition to the level of risk aversion, the level of existing government debt bears important implications for the exposure a government can accept from its banking system. With a given amount of government-owned assets and a determined present value of target wealth for future generations, the amount of debt determines a government’s maximum capacity to extend guarantees to the banking system (compare Figure 12). In summary, our simple, stylized model demonstrates that individual regulatory policy choices can be assessed based on their contribution to or elimination of government exposure to the banking system and their impact on economic growth and volatility. In practice, our approach will require adequate empirical data to draw exposure, return and volatility implications, and a pragmatic tradeoff between model complexity and the consideration of all relevant inputs. Modeling the economic tradeoffs inherent in individual policy choices has the benefit to explicitly assess the resulting government exposure to the banking system while considering the risk-return outcome from these policy choices. 3.4.5

Exposure factor interaction and interconnections

We demonstrated that the individual regulatory policy options each affect primarily one of the three exposure factors, while, as implied by the logic of structural credit risk modeling, factors partly also interact with other exposure factors. Generally, capital not only decreases the probability of distress, but also constitutes a deduction item for losses and for exposure from a government perspective. Likewise, volatility changes assumed due to structural restrictions not only lower losses given distress, but also the likelihood of distress. In contrast, government exposure limitations can be installed without affecting the overall risk parameters of a bank from an average liability perspective, though still affecting expected losses and the likelihood of loss materialization from a government perspective. In addition, from the perspective of individual policies, regulatory options can also affect other exposure levers in more particular ways. Narrow banking with restrictions to government securities, or similar structural approaches can imply liquidity regulation in the sense that more secure government securities held by the banks may also be more liquid, effectively reducing liquidity risk and thus PDB in addition to bringing to zero LGDGov. Also, narrow banking with assets restricted to government securities very much resembles the nationalization option once considering that, in order to finance the economy, the government has to provide and allocate capital. Therefore, while narrow banking with assets restricted to government securities provides a zero exposure to a privately run banking system, it strongly exposes the government to the direct credit risk from the economy once capital is allocated. Likewise, wind-down regimes, depending on the exact implementation,

3.4

Regulatory Policy Options and Their Economic Tradeoffs

95

can imply structural restrictions to the extent that requirements relating to the structure of the liability side affect the overall business model. 152 Similarly, the progressive capital surcharges for G-SIBs potentially imply structural reactions by banks (as intended) such as reducing complexity or size. At the same time, as mentioned earlier, the progressive nature of the capital surcharges also resembles some Pigovian tax element. Furthermore, the impact on the exposure factor may not always be fully visible ex ante. Whether contingent convertible capital is considered a measure reducing EADGov (i.e. gone-concern contingent convertible capital) or a measure reducing PDB (i.e. going-concern contingent convertible capital) largely depends on the definition of a bank’s point of nonviability or the conversion trigger level. Especially for instruments triggered by supervisory judgment, the distinction between gone- and going-concern instruments close to the point of non-viability may be hard to determine objectively ex ante. In addition, pre-funded resolution funds conceptually could be considered an additional economy-wide equity layer reducing PDB, however, from an exposure perspective, because resolution funds are goneconcern contributions to equity, they in fact reduce EADGov. Furthermore, several options interact with other regulatory measures. Resolution powers and wind-down authorities effectively go hand in hand with bail-in claims, as illustrated by the fact that both were developed within the single resolution regime work stream of the Basel III G-SIB regulation. From a legal perspective, bail-ins of creditors are part of the overall insolvency and resolution framework, although from an exposure perspective bailinable claims affect a different lever (EADGov) than resolution powers and wind-down authorities (LGDGov). Finally, a variety of forms of hybrid regulatory approaches are conceivable in addition to the principal choices presented here. For example, Cochrane (2014) principally suggests all-equity narrow banking, however, allowing some short-term debt, which would only constitute a small share of total debt and be subject to a Pigovian tax. Likewise, macro-prudential regulation – a regulatory approach focused on system-wide risk – has gained increasing attention post-crisis and is oftentimes mentioned in the same breath with other policy options. From an exposure perspective it is, however, a tool allowing regulators to calibrate a variety of policy options such as capital requirements or asset restrictions based on the state of the economy or banking system risk levels. While these interconnected, interacting or hybrid regulatory proposals may in principal offer desirable policy choices, examining their exact exposure levers and their impact on growth and volatility is mandated to properly evaluate these proposals and compare them to the available set of alternatives.

152 Compare also Independent Banking Commission (2011), p. 66, suggesting that ringfencing activities lets the market decide whether common ownership of ringfenced and non-ringfenced activities remains efficient.

96

3.

Government Guarantees und Banking System Risk

3.5

Discussion and Conclusion

3.5.1

Institutional design, financial system structure and international policy

In addition to considering the discussed economic tradeoffs, a government has to take general economic, institutional and international policy factors into account when defining and implementing regulatory policy choices. General economic factors play an important role in choosing the appropriate regulatory regime in a variety of ways. First, a country’s capacity to deal with a failing banking system, i.e. its ability to withstand expected and unexpected losses, bears important regulatory implications. A small country with a large banking system as, for example, measured by banking assets to GDP, has to choose exposure levels reflecting its lower capacity to cover banking system stress losses. 153 Within a sovereign portfolio perspective, this would be reflected by a large contribution of the banking system to the overall portfolio volatility. Choosing regulations in line with a country’s capacity to cover the banking system stress loss can be challenging to the degree that banks may compete globally and may find themselves at a competitive disadvantage due to stricter regulations in their home country. Second, a country’s economic and financial system structure also determines the optimal policy mix. A country with primarily bank-driven financial intermediation – as opposed to financial intermediation performed by capital markets or non-bank intermediaries – could find itself with a comparably large exposure relative to its capacity. However, at the same time prudential measures restricting bank intermediation may hurt the local economy relatively more, highlighting the importance of careful calibration of regulations to the economic environment. 154 Also, a diverse financial system with a variety of non-bank financial institutions raises the question about horizontal regulatory consistency. If a regulator decides to use different exposure factors and to target different exposure levels for different parts of the financial system, such as a free banking style policy for a so-called shadow banking sector versus comprehensive regulatory choices limiting exposure to the official banking system, regulatory arbitrage may take place between those two parts of the financial system. 155 Such horizontal regulatory inconsistency may be problematic if the different parts of the financial system with different regulations are interconnected – as for example large investment banks had exposure to LTCM – because a government may underestimate its de facto exposure. Third, regulatory consistency over time in defining exposure levels is also an important aspect, for which different strategies have emerged over time. A consciously chosen strategy of time inconsistency is constructive 153 For example, the additional regulations for large banks recently installed in Switzerland ascribe to this issue. 154 Compare Johanning (2009) for the importance of calibrating strategy and risk management in a bank setting. 155 Compare the Federal Reserve Bank of Minneapolis’ (2016) proposal to implement a Pigovian tax for the shadow banking system to avoid regulatory arbitrage between regulated banks (which face capital requirements) and non-bank financial institutions not facing capital requirements.

3.5

Discussion and Conclusion

97

ambiguity, which implies a lack of explicit guarantees ex ante to avoid moral hazard, and the provision of rescue measures ex post to provide stability once a crisis occurs. However, if market participants expect government interventions despite a lack of explicit guarantees, the value of constructive ambiguity disappears. 156 But even time-consistent strategies may pose challenges, as mentioned in Section 3.4.2.4 in the context of bail-inable claims. In order to overcome the fundamental lack of credibility a government faces in committing to a no-bailout policy ex ante, an explicit constitutional prohibition of bank bailouts would be the strongest instrument (Wieandt (2014)). Finally, fundamental questions about a government’s leverage may determine the extent of additional leverage via a banking system. Considering both government debt and bank debt as ultimately being leverage to an economic system, both types of debt can be viewed as substitutes. 157 In addition to the (assumed) relatively higher economic benefits a privately organized banking system can provide compared to a government, other institutional or behavioral factors could play a role as well. If, for example, government debt holders would not fully consider contingent liabilities by the government to its banking system in evaluating and pricing the government’s debt, the government would have an incentive to transfer part of the leverage to a private – though government backed – banking system, similar to the idea of constructive ambiguity. In that sense, the interaction between banking system stability and government solvency we discussed in Section 3.4.2.1 is also important for government debt investors. Institutional, cultural and demographic factors may impact a society’s risk tolerance and impact policy implementation, making these factors important aspects to consider when making regulatory policy decisions. 158 Culture in itself and a society’s demographic profile may influence relative risk aversion and therefore alter risk-adjusted returns for policy choices. Furthermore, the implementation of different approaches may differ. If, for example, a country has a strong market economic foundation, temporary government support in times of distress may prove to be easier than bailouts in a country where the government can be assumed to maintain invested over the long-term once having gained access to decisions on credit allocation (again, assuming political motivations leading to inefficiencies in the context of government ownership in banks). Also, constitutional aspects have to be considered in choosing regulatory policies. Extreme measures such as nationalization, even if they would be considered economically desirable, may not be constitutionally viable, and

156 Compare also Cochrane (2014), p. 237, describing ex ante bail-out expectations by market participants that are not fulfilled by government action later on as an undesirable state. From a time consistency perspective, this choice bears the danger to evoke both moral hazard ex ante and economic volatility ex post. 157 Compare Jordà et al. (2016) for an examination of the co-evolution and the interaction of public and private sector debt. 158 Compare Rieger et al. (2014) for a survey of individual risk preferences across 53 countries and Chen et al. (2015) for an examination of corporate risk behavior depending on cultural factors across 43 countries.

98

3.

Government Guarantees und Banking System Risk

therefore not a practical choice for a specific country. 159 Finally, regulatory approaches should reflect a government’s and a society’s capacity and willingness to engage in risk management. In the context of bank credit risk management, this aspect is referred to as ‘risk culture’. The risk culture of a government and regulator should reflect the necessary skills to identify and evaluate risks associated with banking system exposure and to transparently communicate these risks. Quantifying exposure levels and expected losses as well as the anticipated economic benefits of a banking system can help to rationally decide upon and manage exposure. Such a risk culture can also be reflected in choosing appropriate crisis management tools and timing. From a behavioral finance perspective, it can be argued that governments will often intervene at irrationally low market value levels, reflecting the acute financial system distress. From this perspective, rescue measures for banks, despite appearing costly at first sight, can turn out to yield positive returns considering the overall stabilizing effect a government can have on the system at the moment it becomes invested. Finally, international policy has considerable implications for national exposure management. In addition to mere economic considerations, political factors may contribute to risk arising in the context of cross-border activities, an aspect referred to as settlement risk in credit risk management. This settlement risk affects all three factors of government exposure. First, regarding PDB, a cross-border group may from an economic perspective have sufficient capital in aggregate, but, for political reasons, may not be able to fully access subsidiary capital or liquidity, increasing the home country’s government exposure value. 160 Likewise, regarding LGDGov, the value of foreign subsidiary bank assets may be affected by foreign government financial policy actions, such as suspending foreign exchange convertibility or restricting capital flows. Finally, regarding EADGov, a government has to consider whether the unwritten rule that a bank’s home country government is responsible for rescuing a troubled bank including its foreign subsidiaries holds true in the future as well. A foreign government’s refusal to bail out a bank’s subsidiary despite its economic capability to do so would constitute settlement risk the host country government would have to consider when assessing its overall exposure. 161 More general, international competition and 159 For example, German financial market stabilization laws enacted during the recent financial crisis included the temporary possibility of bank ownership expropriation as a last resort in order to avoid a systemic collapse. Although not applied, the possibility of expropriation was criticized by leading opposition politicians with reference to historical expropriations under dictatorship regimes in Germany. 160 For example, Italian bank UniCredit expressed expectations that a change from a national to a European banking supervision regime would allow it to access capital held by its German subsidiary HypoVereinsbank which it could not access before (Financial Times, November 14, 2013). 161 For example, the proposed regulation to require U.S. subsidiaries of foreign banks to also apply enhanced capital regulation required for large local banks (compare Financial Times, February 19, 2014) can be considered a move to limit settlement risk.

3.5

Discussion and Conclusion

99

burden sharing in banking is a complex issue with repercussions on exposure management, because positive and negative externalities may occur asymmetrically across jurisdictions and time. Due to the interconnectedness of the global financial system, risk spillovers may induce costs to countries that have not been immediately involved in the ex ante regulatory decisions about the source risk. From a competition perspective, competitive regulatory settings may be conceived as subsidies to a local financial system, the cost of which potentially are partially borne by other countries. While this led to calls for an organization equivalent to the World Trade Organization, the regulatory policy institutions responsible for the development of the international G-SIB regulations – the FSB and the BCBS – can be seen as a significant step towards harmonizing global rules and leveling playing fields. 162 3.5.2

The new regulation dealing with Global Systemically Important Banks

The revision of the Basel accord after the recent financial crisis and the FSB’s development of additional guidelines for regulating G-SIBs resulted in a significant extension of exposure factors covered by international bank regulations. After focusing almost exclusively on limiting PDB for more than two decades, regulators established international regulations to also limit LGDGov and EADGov by mandating resolution powers and wind-down authorities as well as bail-inable claims for G-SIBs as part of comprehensive resolution regimes (see Figure 19). At the same time, PDB-related capital regulation was strengthened by increased capital requirements for the overall banking system and additional progressive capital surcharges for G-SIBs. Further, capital requirements were complemented by new liquidity requirements for the overall banking system. The extension of the exposure factor coverage by the Basel regulatory framework acknowledges that the sole focus on PDB – unless dealt with in the form of an all-equity financed banking system – does not fully eliminate potential banking sector risk and resulting government exposure. Wind-down authorities and bail-in regimes aim at limiting losses given distress and government exposure at distress. Also, if the marginal cost of decreasing a single exposure factor increases, a certain level of limited exposure can be achieved in a less costly way by simultaneously lowering all three exposure factors through a combination of regulatory measures as exemplified by the extended exposure factor coverage of the new G-SIB regulation. Notably, unless the bail-in regimes achieve a full bail-in of bank creditors, the government will retain some exposure to the banking system even under the currently proposed regime. The designation of the most systemically important banks worldwide as G-SIBs, considered a necessary step to apply additional regulation to these banks, confirms this view by implying that these banks cannot be allowed to fail. 163 The current proposals for bail-inable claims do not imply a full bail-in 162 Compare Boone and Johnson (2010) for proposals of a World Trade Organization-style institution for financial system regulation. 163 Compare Moenninghoff et al. (2015).

100

3.

Government Guarantees und Banking System Risk

Figure 19: Exposure levers addressed by new G-SIB regulation

and a corresponding reduction of government exposure to zero, despite vocal calls for a zero exposure level by politicians during the recent financial crisis. In defining an appropriate regulatory response, policy makers chose to maintain a limited – though likely strongly reduced – exposure by governments to the banking system, considering the economic tradeoff between growth and stability. 3.5.3

Results, limitations and future research

We present a framework to quantify the economic impact of different policy choices for regulating systemically relevant banks and related economic tradeoffs. While regulators have used impact studies to examine risk-return tradeoffs inherent in individual policy choices such as capital and liquidity regulation, this contribution to our knowledge is the first comprehensive approach to categorize and discuss the full range of major policy options for regulating banks within a single framework and to provide a quantitative methodology to compare the relative attractiveness of different policy choices. Overall, our stylized model shows that the choice of an optimal regulatory policy mix depends on risk and return preferences of a society, and an economy’s institutional and cultural setting which impacts the tradeoffs implied by each regulation. Our framework suggests that a society with higher risk tolerance will prefer limited exposure solutions such as bail-in regimes and capital regulations, while a highly risk-averse society may be more open to increased government involvement in the intermediation process, for example through structural restrictions.

3.5

Discussion and Conclusion

101

Our framework takes a credit risk view of government guarantees for banks to illustrate the tradeoffs inherent in individual policy choices, structuring the individual regulations along the three components of credit risk – probability of distress, loss given distress and exposure at distress. Managing the probability of distress of banks through capital and liquidity requirements implies a tradeoff between growth, externalities and financial intermediation benefits. Managing the loss given distress of a government through narrow banking or structural restrictions implies a tradeoff between inefficiencies from government involvement in capital allocation versus economic volatility, while the benefit of a Pigovian tax to limit a government’s net loss given default depends on its implementation efficiency. Finally, managing government exposure at distress to the banking system can range from zero exposure in a free banking regime to limited exposure in the context of bail-inable debt and full exposure for a nationalized banking system. An important tradeoff inherent in these exposure-at-distress-related policy choices is the implied economic volatility versus the government’s resulting risk exposure. From a practical policy perspective, our framework can assist in making policy decisions and track and communicate a government’s resulting exposure outcomes. Adopting an exposure-based approach to regulating the banking system extends the narrow interpretation of government guarantees as externalities. Our model demonstrates that the consequent next step for regulators is to further apply credit risk management logic and tools across the full range of major regulatory options to determine the desired exposure levels and define appropriate regulatory action. Governments should proactively decide on the level of banking system risk they are willing to cover by choosing from a variety of regulatory options ranging from zero to full exposure. From a theoretical perspective, our framework allows to better structure the broad range of contributions on optimal regulatory policy for systemically relevant banks, and suggests further applications of portfolio theory in the context of economic policy examinations and bank regulation. Our framework has several limitations. The practical application of our model depends on the availability of appropriate data on growth and volatility implications of individual regulatory policies. Hybrid policies and multiple exposure levers triggered by individual regulations will require careful empirical calibration and impact validation. Likewise, implementing regulatory policies may imply significant costs in reality while those costs may be hard to adequately reflect in a model. Further, determining risk preferences of a society and translating them into desired economic volatility levels and acceptable banking risk exposure levels may be challenging in theory and practice. Also, the global nature of banking and the need for regulatory harmonization appears to contradict local risk preferences based on local demographics, cultural factors and budget constraints. For example, the implied growth and volatility implications by a globally harmonized minimum capital requirement may not be suited to simultaneously meet the risk-return preferences of a young, innovative and growing economy and those of an aging, risk-averse economy. Regulatory competition between countries can further complicate the calibration of desired exposure

102

3.

Government Guarantees und Banking System Risk

and appropriate policies given that in a global interconnected financial system other countries’ regulatory decisions may impact a sovereign’s ultimate exposure. 164 Finally, the explicit communication of exposure levels as part of a government’s exposure management and communication can lead to secondary effects – as demonstrated in Chapter 4 in the context of communications around G-SIB regulation – which are not reflected in our simplified model. Future research could focus on conducting impact studies across regulations, to quantify the factors driving the tradeoffs for each regulation and to better understand individual regulations’ effectiveness and consequences. Also, a framework for determining a society’s risk preferences in the context of financial stability would allow to translate cultural risk attributes into risk exposure preferences. Moreover, attempts to further standardize the measurement of government exposure will allow to derive more consistent and relevant exposure quantifications that can be applied to determine preferred regulatory policies. 3.5.4

Conclusion

The recent financial crisis highlighted more than ever how much governments were exposed to banking system risk. The subsequent regulatory debate posed the question what are relevant policy options and how to adequately regulate systemically relevant banks. To answer this question, we present a comprehensive framework to categorize and discuss the full range of major policy options for regulating banks and to provide a quantitative methodology to compare the relative attractiveness of different policy choices from a sovereign’s portfolio optimization perspective. These options range from free banking and narrow banking regimes (implying zero government exposure) over regulations such as minimum capital and liquidity requirements, structural restrictions, Pigovian taxes, resolution powers, bail-in regimes, contingent capital and resolution funds (implying limited government exposure) to a nationalized banking system (implying full government exposure). Using a sovereign’s portfolio optimization approach by Bodie and Brière (2014a), we develop a stylized model suggesting that the choice of an optimal regulatory policy mix depends on risk and return preferences of a society, and an economy’s institutional and cultural setting which impacts the tradeoffs implied by each regulation. In our model, a society with higher risk tolerance will prefer limited exposure solutions such as bail-in regimes and capital regulations, while a highly risk-averse society may be more open to increased government involvement in the intermediation process, for example through structural restrictions. 164 Hence, any regulatory policy tailored to the risk return preferences of a certain country will have to consider the implications other countries’ regulatory policies have on its own exposure as a result of global financial system interconnectedness and potential cross-border risk contagion, whether in the context of globally harmonized regulation or varying regulatory approaches across countries.

3.5

Discussion and Conclusion

103

Examining the new international regulation for G-SIBs developed by the BCBS and FSB through the lens of our framework, we find that it considers an expanded set of exposure factors to manage banking system risk. After two decades of focusing on minimizing the probability of banking system distress, regulators proceeded to now also reduce loss given distress and exposure given distress by setting international standards for resolution regimes including wind-down authorities and creditor bail-ins. It appears that regulators refined and extended their approach to a limited exposure, instead of deciding to fully eliminate exposure, considering the inherent economic tradeoff between growth and stability. Time and experience with the new regulations will prove how limited the remaining banking system exposure will be and what the resulting risk-return levels will be. Our model demonstrates that the consequent next step for regulators is to further apply credit risk management logic and tools across the full range of regulatory options to determine the desired exposure levels and define appropriate regulatory action. Governments have to proactively decide on the level of banking system risk they are willing to cover by choosing between a variety of regulatory options ranging from zero to full exposure. An exposure perspective on government guarantees and bank regulatory policy options, as presented in this article, extends the narrow interpretation of government guarantees as externalities, and provides an analytic framework to manage banking system risk considering tradeoffs regarding growth and stability from a sovereign’s overall portfolio perspective.

4.

Empirical Evidence from the New International Regulation Dealing with Global Systemically Important Banks

4.1

Introduction 165

In response to the financial crisis of 2008, the G20 leaders declared their intention to ‘take concrete steps to move forward with tough, new financial regulations so that crises like this can never happen again’ and ‘[so] that banks can never again blackmail states and governments’. 166 Since then, the responsible international regulatory bodies have developed new regulation measures consisting of enhanced supervision, capital surcharges, and the establishment of resolution regimes specifically for banks that would pose high risks to the financial system if they were to fail. 167 In this context, the concept of the “Global Systemically Important Bank”, in short “G-SIB”, 168 has emerged, characterizing the banks that are subject to the new additional regulation and ultimately resulting in an official list of 29 global banks deemed too-systemically-relevant to fail. In an attempt to answer our third question on whether the new regulation succeeds in mitigating the cost and risks of TBTF 169, we conduct an event study. We observe that, despite the progress made in developing these new regulatory measures, the new regulation turns out to be a double-edged sword. While the new regulation per se appears to impact the affected banks significantly, the audible welcoming of G-SIB status by several senior bank

165 Reprinted from Journal of Banking & Finance, Volume 61, Moenninghoff, S.C., Ongena, S. and Wieandt, A., The perennial challenge to counter Too-Big-to-Fail in banking: Empirical evidence from the new regulation dealing with Global Systemically Important banks, Pages 221-236, Copyright (2015), with permission from Elsevier. 166 White House press release, September 24, 2009; German federal government press release, September 25, 2009. 167 The affected banks have to apply this regulation in addition to the general Basel III standards, which include increased capital and liquidity requirements. 168 The term “Global Systemically Important Financial Institution” or “G-SIFI” has been used as an equivalent to the term “G-SIB”, a description referring more broadly to systemically significant financial institutions including non-bank financial intermediaries such as insurance companies and market infrastructure providers. As our analysis refers exclusively to banks, we will use the term G-SIB throughout this chapter. 169 The degree to which the new regulation of G-SIBs is intended to fully abolish TBTF varies between individual regulatory communications, ranging from the goal to reduce the probability and impact of failure of G-SIBs or the purpose to reduce moral hazard risks of G-SIBs to the declared intention to end TBTF (see for example Financial Stability Board (2013)). © Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2018 S. C. Moenninghoff, The Regulation of Systemically Relevant Banks, Finanzwirtschaft, Banken und Bankmanagement  Finance, Banks and Bank Management, https://doi.org/10.1007/978-3-658-23811-7_4

106

4.

Empirical Evidence from the New International Regulation

executives upon the designation of their bank potentially already heralds one of the regulation’s unintended consequences. 170 As discussed in Chapter 2, designating banks as too big or too systemically relevant can undermine market discipline and allow banks to refinance themselves below market cost, potentially creating incentives for these banks to misallocate resources and assume too much risk. Low refinancing costs increase the banks’ shareholder value at the cost of the taxpayer, and in times of financial crisis bank bailouts are essentially direct transfers of taxpayer wealth to bank creditors. Todd Conover, the then Comptroller of the Currency of the U.S. Treasury, acknowledged this phenomenon by stating in his testimony before the U.S. House of Representatives on September 19, 1984, that regulators would not allow any of the eleven multinational U.S. money center banks to fail. Since then, various studies have examined the presence of the TBTF doctrine as well as the effectiveness of regulatory reform to abolish it. TBTF effects can result from an explicit TBTF policy for banks or from implicit expectations (for example based on observations of past government action) by bank managers and investors of government intervention in case of a financial crisis. In contrast to the extant literature (which we review later), our study does not focus on the explicit or implicit creation of TBTF effects by government announcements or bank rescues, but rather on the TBTF effects that arise in the context of a new regulation specifically designed to mitigate the costs and risks of the TBTF doctrine. Our study first examines the likely unintended consequences of the new regulation for G-SIBs, which (almost unavoidably) designates individual banks as G-SIBs – that is to say, as too systemically important to fail, thereby reinforcing existing TBTF perceptions in the market. Second, our study contributes to the wider discussion of TBTF by examining specific factors such as government ownership and home country rating characteristics that may determine the potency of the TBTF doctrine. Third, our study aims to examine the ultimate net effectiveness, from a policy perspective, of the current G-SIB regulation, weighing the regulation’s impact on G-SIBs against the strengthening of their TBTF designation. We find that the new regulation negatively affects G-SIBs, yet that at the same time the official designation of banks as G-SIBs has a partly offsetting impact, suggesting that investors did not believe that governments would allow those banks to fail. Our cross170 For example, the CEO of Deutsche Bank reportedly stated that, despite the additional regulatory burden associated with G-SIB status, he was glad that Deutsche Bank would likely be on the list of systemically important banks as G-SIB status would allow for benefits in the refinancing and depository business (Handelsblatt, June 27, 2011). Likewise, the CEO of JP Morgan was reported as saying that G-SIB status could imply more business for the bank because customers would be looking for strong counterparties (Wall Street Journal MarketWatch, November 4, 2011). Similarly, the Bank of China released a statement upon its designation saying that the bank’s G-SIB status would reflect ‘the comparative advantages […] of Bank of China versus our domestic peers’ (Financial Times, November 4, 2011) and that ‘the opportunity outweighs the challenge’ (People’s Daily Online, November 6, 2011).

4.2

G-SIB Regulation and Hypotheses

107

sectional analysis of the valuation effects with respect to, for example, government ownership of banks supports the view that the positive reaction to these designations can be attributed to a TBTF perception. These results suggest that even though the individual components of the regulation have been effective, revealing the identities of G-SIBs eliminated ambiguity about the presence of government guarantees, and thereby may have run counter to the regulators’ intent to contain the effects of TBTF. The remainder of this chapter is organized as follows: Section 4.2 provides a brief review of relevant empirical literature on explicit and implicit government guarantees and TBTF in the context of regulations (a more comprehensive overview of which is provided in Chapter 2). This section also introduces the new G-SIB regulation, and develops our hypotheses. Section 4.3 describes our sample, the relevant event dates, and the methodology that we follow. Section 4.4 demonstrates and discusses our empirical results and Section 4.5 contains a conclusion. 4.2

G-SIB Regulation and Hypotheses

4.2.1

Explicit and implicit government guarantees

With regard to explicit government guarantees, scholars have found that the TBTF doctrine has strengthened as a result of official government announcements or the establishment of deposit insurance schemes. O’Hara and Shaw (1990), for example, observe positive shareholder wealth effects for large money center banks following the U.S. Comptroller of the Currency’s announcement of a TBTF policy after the bailout of large U.S. banks in the 1980s. Based on observations of long-term credit rating actions, Billett et al. (1998) likewise conclude that insured deposit financing shielded U.S. BHCs from market discipline in the early 1990s. Chernykh and Cole (2011) observe increased moral hazard in banks’ risk taking following the introduction of a deposit insurance scheme in Russia, and Imai (2006) observes a strengthening of the TBTF doctrine by the Japanese deposit insurance reform of 2002. 171 With regard to implicit government guarantees, research indicates that concrete rescue measures can contribute to the emergence of the TBTF doctrine. Relating to the Federal Reserve Bank’s coordination of the rescue of LTCM, Kabir and Hassan (2005) find evidence of TBTF effects based on bank stock returns. Likewise, based on an examination of subordinated debt yield spreads, Balasubramnian and Cyree (2011) conclude that the Federal Reserve Bank’s intervention in the LTCM rescue has diminished market discipline especially for large banks. In examining subordinated debt yield spreads for large European banks, Sironi (2003) similarly observes implicit government guarantees for large banks in the early 1990s, but a weakening of those guarantees during the late 1990s. 171 For an overview of approaches to and challenges in measuring funding cost differentials as an indication for TBTF effects, see Kroszner (2013).

108

4.

Empirical Evidence from the New International Regulation

Based on long-term bank stock returns, Ghandi and Lustig (2015) provide evidence of implicit government guarantees for large but not for small U.S. banks. Based on credit ratings for a global sample of banks, Ueda and Weder di Mauro (2013) observe substantial funding advantages for large banks before and during the recent financial crisis. On the contrary, Ellis and Flannery (1992) find that rates paid by large banks for uninsured certificates of deposit included significant default risk premiums in the late 1980s and conclude that market discipline prevailed despite the government bailouts of large U.S. banks throughout the 1980s. Examining balance sheet structures, Nier and Baumann (2006) find that market discipline is likely to be weaker for banks benefiting from implicit government guarantees, and Dam and Koetter (2012) conclude that heightened bailout expectations increase risk taking by banks. Several authors even go beyond describing bank behavior as a reaction to an already obtained TBTF status, and observe banks acting in order to obtain the desired TBTF label. Penas and Unal (2004), for example, associate increasing returns for bonds of midsize banks following a merger with the achievement of TBTF status by the combined institution due to that merger. Likewise, Deng et al. (2007) attribute significant bond yield spread tightening for midsize banks following diversification efforts to TBTF effects. Differentiating between domestic and cross-border mergers, Ongena and Penas (2009) find TBTF effects for bondholders of banks conducting domestic mergers due to the possibility of quicker and more effective government intervention in case of financial distress. Gropp et al. (2011) examine competitive conduct in the context of government guarantees and conclude that guarantees for large banks increase the level of risk competitor banks are willing to take. Finally, regarding the regulator’s ability to limit TBTF, Flannery and Sorescu (1996) examine whether the large U.S. bank bailouts in the 1980s impacted the relationship between bond yield and bank risk in the early 1990s and find that the introduction of the FDICIA, a regulation specifically aiming for an anti-bailout policy, proved to be effective. According to the authors, the regulation reduced discretionary government support by mandating a least-cost resolution method and an immediate resolution of failing banks, allowing exceptions only in cases of systemic risk and requiring a complex decision-making process and respective governance. However, while relating their findings to an anti-TBTF regulation, the authors examine TBTF stemming from a preceding implicit or explicit government bailout policy (see also Schäfer et al. (2016), as do the other two strands of research already mentioned. 4.2.2

New G-SIB regulation

In April 2009, the G20 established the FSB in order to develop guidelines for additional regulation and oversight of G-SIBs, and, in September 2009, tasked the FSB to develop concrete measures by October 2012 to reduce the moral hazard posed by these banks. Over the next two and a half years, in cooperation with the BCBS, the FSB made concrete rec-

4.2

G-SIB Regulation and Hypotheses

109

ommendations for additional supervision, regulation, and resolution of G-SIBs. 172 In order to enhance supervision, the FSB recommended more supervisory powers, an improved set of standards and methods, and stricter assessment regimes, as well as unambiguous mandates, independence, and appropriate resources. In the area of regulation, the FSB proposed that G-SIBs hold additional capital of 1.0‒2.5% Common Equity Tier 1, depending on the individual institution’s systemic relevance as measured by an indicator-based approach, in order to increase the loss-absorbency of G-SIBs. 173 174 Finally, regarding resolution regimes, the FSB recommended strengthening national resolution regimes, cross-border cooperation, and resolution planning, and the enhanced resolvability of G-SIBs. 4.2.3

Hypotheses

While, generally speaking, the net wealth effect of a new regulation ultimately depends on the resulting equilibrium and could also be beneficial for the affected firms, 175 the additional regulatory measures for G-SIBs are widely considered to involve costs for the affected institutions. First, regarding regulation, the capital surcharges imply higher average funding costs as bank equity is perceived to be more costly than debt from the banks’ perspective. This is because of the tax deductibility of interest paid on debt as opposed to dividends paid to common equity holders. In addition, potential funding advantages for large institutions resulting from the expectation of public sector support should a financial crisis occur could further lower the net cost of debt from the banks’ perspective. 176 Furthermore, potential information asymmetries between bank management and bank investors could lead to an underpricing of newly issued equity especially in times of high uncertainty about the value of banks’ assets, and therefore make debt refinancing at least temporarily economically 172 For an overview of the complete process of G-SIB regulation see Appendix 4.13. The table summarizes the individual regulatory announcements made by the G20, the FSB and the BIS throughout the development of the new G-SIB regulation, distinguishing between general regulatory announcements, announcements relating to specific regulatory measures such as supervision, resolution, loss absorbency and levies or structural measures, and announcements to define and identify G-SIBs. 173 Initial considerations of contingent capital and bail-in options were not included in the final recommendations. Likewise, in its final proposal, the FSB did not include liquidity surcharges, tighter large exposure restrictions, structural constraints, or bank levies, which were discussed later in the process. Although the FSB proposals do not officially contain merger restrictions, statements by high-level supervisors involved in the process suggest that supervisory authorities have an implicit understanding to prevent further consolidation amongst G-SIBs (Banking Day, November 12, 2010). 174 Notwithstanding the G20 leaders’ endorsement of the final proposal, individual jurisdictions continue to debate additional regulatory measures for G-SIBs on a national level. 175 See Spulber (1989), p. 32‒36. 176 The BCBS explicitly mentions its intention to ‘level the playing field in the banking sector by reducing the funding advantages of G-SIBs that arise from expectations of public sector support’ in explaining its motivation for focusing on Common Tier 1 Equity (BCBS (2011b), p. 17).

110

4.

Empirical Evidence from the New International Regulation

more attractive. 177 Second, the additional supervisory requirements will likely entail additional administrative and operational costs for G-SIBs. Third, the establishment of resolution regimes is also assumed to result in significant administrative, legal, and operational costs for G-SIBs due to their efforts to demonstrate a sufficient level of resolution planning and ensure resolvability. Furthermore, resolvability as such, if credibly installed, would be an important instrument for decreasing the likelihood of a government bailout should a crisis occur, and thus would lower implicit government support, which in turn would increase funding costs for G-SIBs. In contrast, the value of an explicit government guarantee for large banks’ shareholders should be positive, as has been discussed in Section 4.1. The additional value received by a bank upon designation would be highest for banks without any perceived government support before being designated as globally systemically important. However, to the extent that an explicit guarantee provides stronger assurance of government support than an implicit one, a transition from implicit government guarantees to explicit ones should also provide value to the affected banks’ shareholders. 4.3

Data and Methodology

4.3.1

Sample

4.3.1.1 Sample compilation We created a sample of the 300 largest publicly listed banks worldwide. The sample is divided into three sub-samples: first, banks that were officially designated as G-SIBs throughout the entire regulatory process; second, banks that were discussed for designation but eventually were not considered to be G-SIBs; and, third, banks that were never considered for designation at any point in time. Choosing a global sample reflects the BCBS’s initial consideration of banks from 17 countries and five continents as G-SIB candidates and the BCBS’s intention to establish globally applicable regulatory guidelines. It also enables us to maximize the number of G-SIBs within the sample. The construction of our sample based on banks’ size resembles the approach chosen by the BCBS in its initial procedure to identify G-SIB candidates. The sample was generated from Datastream based on our primary selection criterion — asset size as of the end of 2009 — which we applied to the “financial institutions” subcategory within the database. In addition to commercial banks, the sample includes 34 investment banks and one bancassurance company. 178

177 See Admati et al. (2011) for a discussion of whether or not bank equity is more expensive than bank debt in theory and practice. 178 Life and casualty insurance companies, investment holding companies, real estate investment vehicles, and publicly listed central banks were not included in the sample.

4.3

Data and Methodology

111

4.3.1.2 Sub-sample definitions As already indicated the sample contains three relevant sub-samples (see Appendix A.10 for an overview of the individual sub-samples). The first sub-sample consists of those banks categorized as G-SIBs at a certain stage in the regulatory process and these banks will be referred to as “G-SIBs”. These banks will indeed be subject to the new additional regulation and, after the first designation event took place, are expected to benefit from obtaining G-SIB status. Technically, this sub-sample comprises 24 to 30 banks depending on the designation stage in focus. The second sub-sample comprises banks not designated as G-SIBs at a specific designation stage in the regulatory process even though their size would in principle merit G-SIB categorization. We refer to this sub-sample as “Non-G-SIBs”. For example, if, at the first designation event, a bank is among the largest banks worldwide, but was not included in the leaked list of GSIBs, this bank would be considered a Non-G-SIB for this event. We determine this sub-sample as those banks that are among the 50 largest banks but are not in the G-SIB sub-sample at that respective stage. The third sub-sample comprises banks not considered G-SIBs at any point in the regulatory process. These banks are neither subject to the additional regulation nor were they considered potential G-SIBs during the regulatory process. 179 The third sub-sample will be named “Other Banks” and comprises the approximately 250 remaining banks from our sample. 4.3.1.3 G-SIB designation We dynamically classify the banks into our three sub-samples G-SIBs, Non-G-SIBs, and Other Banks throughout the observation period according to the three subsequent stages of G-SIB designation, an overview of which can be found in Table 6. The first stage constitutes the beginning of the regulatory process during which the involved regulators ruled out any form of official G-SIB designation. 180 Although there were indications that the number

179 Three banks from the third group will be temporarily included in the group of G-SIBs following the designation events, reflecting that factors beyond size were considered in the G-SIB designation process. However, we assume that markets do not in principle expect banks in the third group to be identified as G-SIBs. 180 For example, Jaime Caruana, Chairman of the Bank for International Settlements, stated that he did not ‘think at least at the beginning [there was] the intention to publish these names of the banks’ (Dow Jones Newswires, October 23, 2010). Likewise, Mario Draghi, the then Chairman of the FSB, reportedly said that a list of the institutions would not be published (Reuters, June 26, 2011). Similarly, referring to potential unintended certification effects, Lord Adair Turner, Chairman of the British Financial Services Authority and Director at the Bank of England, stated that ‘what is not useful is having an absolute definitive list’, but that he did not ‘think we should be terrified if somebody manages to produce a list that's not far from what the authorities consider to be the 25 most important [banks]’ (Dow Jones Newswires, January 6, 2011). By pursuing the goal of constructive ambiguity as opposed to full supervisory transparency, policy makers attempted to avoid the moral hazard potentially resulting from the official designation of G-SIBs (see Thomson (2009) for a discussion of designation transparency and Goodhart and Huang (2005) for a model of the value of constructive ambiguity for central banks as lenders of last resort).

112 Table 6:

4.

Empirical Evidence from the New International Regulation

Overview of G-SIB sub-sample

4.3

Data and Methodology

113

of banks affected by the new regulation would be in the low double digits, there was no concrete information available about which banks would be affected in the initial phase of the regulation. 181 Thus, we argue that, initially, market participants expected additional regulation for the most systemically relevant banks, but did not expect a fully transparent official designation of these banks as G-SIBs. Consequently, in assigning our sub-samples in the initial phase, we categorize strictly by size and define the 30 largest banks as perceived G-SIBs, the 31st to 50th largest banks as Non-G-SIBs, and the remaining 250 banks as Other Banks. In a robustness test, we vary the number of G-SIBs for these first three events and find that our results are mostly unaffected (compare Appendices 4.11 and 4.12). 182 The second stage is marked by the leak of a preliminary list of 24 G-SIBs compiled by the FSB and the publication of the list in a front page article in the Financial Times on November 10, 2009. 183 From that date on, not only were these 24 banks perceived as GSIBs by the market, the mere fact that such a list existed and had been leaked to the press constituted an important change. We argue that the leak resulted in the market considering regulators to be significantly less likely to maintain their intended constructive ambiguity and no longer committed to refraining from officially designating the affected banks as GSIBs once finally identified. The same list was re-published by the Financial Times in another front page article approximately one year later on November 30, 2010, now explicitly relating it to the regulation of G-SIBs. 184 This increased the confidence level that the 181 For example, the FSB reported in September 2009 that supervisory colleges had been set up for more than 30 large, complex, cross-border financial institutions (Reuters, September 25, 2009). Former Federal Reserve Chairman Paul Volcker estimated in September 2009 that ‘a range from about 5 to 25’ financial institutions in the world should be considered systemically significant and protected from failure outside of commercial banking and insurance organizations (United States Congress, September 24, 2009). In July 2009, Federal Reserve Chairman Ben Bernanke referred to ‘roughly 25 financial firms’ that would be considered systemically important as part of legislation giving the central bank authority to oversee those institutions (Wall Street Journal MarketWatch, July 4, 2009). Senator Tester, in a hearing by the Senate Committee on Banking, Housing and Urban Affairs in March 2009, referred to ‘all 17 that are too big to fail’ (Financial Market Regulatory Wire, March 24, 2009). 182 In our robustness test, we define the G-SIB sample as the 20 largest and 40 largest banks, respectively. Both variations show significant negative returns for G-SIBs at the first regulatory event. However, while the sample with 40 G-SIBs exhibits negative return differentials compared to Non-G-SIBs across all three designation events — similar to our results with a sample of 30 G-SIBs — the smaller sample of 20 G-SIBs shows negative return differentials compared to Non-G-SIBs for the first and the third event, but a positive return differential for the second event. Within the TBTF logic this could imply that the market strongly expected some of the banks ranking from 21 to 30 to be designated as G-SIBs (Appendices A.11 and A.12). 183 See the Financial Times, November 30, 2009. The list published by the Financial Times also contains six insurance companies, which are not considered in our study. 184 See the Financial Times, November 10, 2010. While the Financial Times reprinted exactly the same list, the article mentioned the possible exclusion of most big Asian banks, which was immediately denied by sources involved in the process (Dow Jones Business News, November 11, 2010).

114

4.

Empirical Evidence from the New International Regulation

leaked list was relevant to the regulation of G-SIBs, which we would expect to amplify the market reaction to the first leaked list. To our knowledge, these two list publications were the only complete lists of G-SIBs published in the media before the announcement of the official list. 185 The third stage is marked by the official publication of the list of G-SIBs, which is in line with the view that an official designation had become more likely after the preliminary list was leaked. On November 4, 2011, paralleling the G20 leaders’ final approval of the new regulation, the FSB published its official list of 29 banks. 4.3.2

Event dates

In order to capture all relevant stock returns, we identify 12 event dates, including nine that are associated with regulatory announcements and three that are associated with GSIB identification. An overview of the event dates and their description (i.e., the significant news they entailed) can be found in Table 7. A comprehensive illustration of all relevant Table 7:

Overview of event dates

185 Several articles referred to individual G-SIBs that were included in the leaked list published by the Financial Times. A list compiled in a research note by Morgan Stanley around July 19, 2011, was discussed in the Financial Times. However, only eight banks from that list were mentioned in the article.

4.3

Data and Methodology

115

event dates and the regulatory process is contained in Appendix A.13. From the whole range of dates of announcements made by the three international institutions involved in the development of the regulation — the G20, the FSB, and the BCBS — we consider as important all six G20 summits that took place until the regulation was officially endorsed in November 2011. In addition, we consider the publication of the concrete policy proposals for each of the three components of the new regulation (i.e., supervision, capital requirements, and resolution regimes and tools) as significant news. In contrast, all other regulatory publications by the FSB and BCBS regarding G-SIBs constitute progress reports mainly prepared for the meetings of G20 finance ministers and central bank governors and these do not contain any significant new information compared to documents published around preceding G20 summits. In examining these regulatory event dates, we distinguish between the broad, general regulatory announcements made by the G20 leaders and policy-specific announcements by the FSB and the BCBS. 186 187 Regarding events identifying banks as G-SIBs, the three relevant events are the two Financial Times publications of the leaked preliminary list of GSIBs and the official list published by the FSB in November 2011. 4.3.3

Methodology

4.3.3.1 Abnormal return calculation In order to determine the impact of the identified regulatory announcements and designations we measure the time-zone adjusted, average daily abnormal returns on a trade-to-trade basis for each of the three sub-samples of banks at the respective event date. The value of the average daily abnormal return thus reflects the event’s impact on the market value of 186 Regarding general regulatory announcements, the G20 leaders declared at their first three summits their strong intention to develop regulatory measures for systemically significant banks. At the first summit, in Washington, after having previously focused on government support for large banks around the world in order to stabilize the financial system, the focus shifted toward calling for increased regulation and oversight of these banks. At the second summit, in London, it was decided to establish the FSB in order to set guidelines for G-SIB regulation and oversight. And at the third summit, in Pittsburgh, the G20 leaders tasked the FSB with developing measures including enhanced supervision, resolution regimes, and tools, and increased prudential requirements commensurate with the cost of G-SIB failure. The fourth and fifth summits, in Toronto and Seoul, did not yield significant new developments within the three areas of regulation compared to the broad suggestions laid out earlier. Essentially, these two summits recognized the progress made by the FSB and endorsed the core elements discussed at earlier summits. The major news of the sixth summit, in Cannes, was the publication of the official list of G-SIBs. Therefore, we will treat this event as a designation event. 187 Regarding specific policy announcements, the FSB published its recommendations to enhance supervision in November 2010, and its recommendations on resolution regimes in July 2011. The BCBS announced and approved the additional capital requirements in June 2011 and September 2011 respectively.

116

4.

Empirical Evidence from the New International Regulation

the banks included in the respective sub-sample. As a measure of abnormal return we use the market model, relating the individual bank equity securities’ returns to the returns of broad-based stock indices of the respective banks’ home countries: 188 Rit = αi + βiRmt + ϵit with E[ϵit] = 0 and Var[ϵit] = σ² For any bank security i, we calculate the abnormal return ϵit based on the bank’s period-t returns Rit, the market portfolio Rmt , and the parameters of the market model αi, βi and σ².189 Negative values for ϵit imply that the bank’s market value “abnormally” decreases following an event, and positive values for ϵit indicate that the market value increases. In addition to the absolute average abnormal return, we consider the return differential between the different sub-samples to be especially revealing as the news should impact the groups asymmetrically. In that sense, additional regulation harming G-SIBs should leave the other two groups of our sample unaffected or should even benefit these banks in terms of resulting competitive advantages. We expect the Non-G-SIBs sub-sample to exhibit the strongest asymmetric reaction compared to G-SIBs especially around the designation events. This is because the ex ante expectation for individual Non-G-SIBs might have been that they would be designated as G-SIBs as well (as opposed to the average of Other Banks, which the market never expected to be included) and thus we expect a stronger sign of relief or disappointment upon being included in or excluded from the list. In addition, considering not only the absolute return but also the return spread between the three sample groups allows us to further control for confounding effects impacting the whole financial sector. Several of our model’s specifications merit discussion. First, as we are investigating a global sample, significant news occurring later on a business day in Europe or the U.S. (e.g., afternoon GMT) would affect Asian stocks only on the next calendar day because the local stock exchanges would already be closed. Therefore, we adjust the non-weekend event returns for time zones using the local stock exchange closing times as provided by MSCI and event-specific estimates about the time the information was released. 190 Second, we use broad local market indices in order to calculate the abnormal returns as suggested by Campbell, Cowan, and Salotti (2010). These authors conclude that local indices are well 188 See Appendix A.14 for an overview of home countries and respective indices used to calculate the abnormal returns for the sample banks. 189 See Campbell, Lo, and MacKinlay (1997), p. 155. 190 For the G20 summits taking place on a weekday (events 2, 3, 8, and 12) we use the Reuters U.S. News Archive in order to determine the first mention of a (draft) communiqué, a summit statement, or an official list of G-SIBs respectively. For the first two designation events (events 4 and 7) we use the online publication date of the Financial Times (5:00 a.m. GMT) as the information release date. For the three weekend event dates (events 1, 5, and 9) we use the next calendar day without any time zone adjustments. For the three regulatory report publication or decision dates (events 6, 10, and 11) we assume that these reports were released at noon GMT based on the fact that the responsible regulatory bodies are based in Basel, Switzerland and some of these reports were observed to be released around midday.

4.3

Data and Methodology

117

suited for capturing abnormal returns and, specifically, that non-parametric tests, which we also apply in our study, adequately capture abnormal returns in this context. A potential pitfall of local indices is that large individual banks or a group of local banks contained in a sample may impact the local index and therefore lower the observed abnormal returns. However, we expect this effect to be limited for our sample. 191 Third, we perform our analysis with an estimation window of 150 days for the days -1 to -151, excluding the event date and the days -1 and +1 around the event date. In determining the most applicable estimation window there is a tradeoff between ensuring that market participants’ risk perceptions (and respective correlations) are relevant to the respective event date and, therefore, not too far away from the actual event, and, on the other hand, ensuring that the estimation window is not contaminated by prior events (and other significant developments around the event date). Finally, regarding the event window length, we use a one-day sampling interval as the event dates are precisely defined and we attempt to exclude additional noise from other events occurring during the financial crisis. In robustness checks, we vary our analysis along all four dimensions. We use returns from the respective calendar days instead of time zone-adjusted event returns, accepting that return reactions of Asian stocks may not reflect news occurring later on the event day. We apply a pre-crisis “fixed” estimation window that consists of the first 150 days in 2008 instead of a rolling estimation window preceding the event dates. We use the MSCI World as a global market index instead of using local market indices to estimate the betas and abnormal returns. This method implies a potential inconsistency by relating closing prices of Asian stocks to the closing price of the MSCI index, which may reflect news occurring later on the event day. Finally, we use a five day (+2/-2) event window instead of a one day event window, which increases the likelihood that additional (non-related) news is captured by the stock returns, especially during the height of the crisis. All of these variations, despite adding noise and consequently lowering statistical significance, confirm our overall results regarding the return differentials of G-SIBs to Non-G-SIBs (see Appendix A.15). 4.3.3.2 Test statistics We test our results with both parametric and non-parametric tests of significance. As a parametric test, we employ the t-test. As non-parametric tests, we use the generalized sign 191 We approximated the index weights of our G-SIB and Non-G-SIB samples for all event dates based on Datastream data of actual historical index weightings and market values (data available for approx. 76% of our sample). First, across all events, the average G-SIB constitutes 3.9% of an index (median 4.1%) and the average Non-G-SIB 3.3% (median 3.0%). As a group of index constituents, G-SIBs constitute on average 10.0% of an index (median 10.0%) and Non-G-SIBs 6.0% (median 5.4%). We consider these figures sufficiently small to overcome concerns that the majority of the abnormal returns would be diminished, especially when considering that we focus on return differentials and the fact that the differences of index weights between G-SIBs and Non-GSIBs are even less significant.

118

4.

Empirical Evidence from the New International Regulation

test, taking into account the proportional distribution of positive versus negative abnormal performance, and the Wilcoxon signed-rank test, analyzing both the sign and the magnitude of the abnormal performance. 192 4.4

Empirical Results

Before presenting our results, and in order to provide context to our event study, we briefly examine the long-term returns of G-SIBs throughout the development phase of the new GSIB regulation. From the beginning of the regulatory process on November 17, 2008, to the endorsement of the regulation on November 4, 2011, G-SIBs exhibited a strong underperformance of -21.2%, compared to +42.0% and +63.2% for Non-G-SIBs and Other Banks, respectively (see Figure 20). Likewise, at the end of the process when the new G-SIB regulation was endorsed, G-SIBs traded -76.7% below their pre-crisis high compared to losses of only -31.6% for Non-G-SIBs and -21.9% for Other Banks (see Figure 21). 193 The following event study analysis aims to contribute to an explanation of the observed relative underperformance of G-SIBs during this period.

Indexes for G-SIBs, Non-G-SIBs, and Other Banks contain the sample banks according to the designation stages as discussed in Sections 4.3.1.2 and 4.3.1.3. a

Figure 20: Stock prices of G-SIBs, Non-G-SIBs, and Other Banks, 2008 to 2011

192 See Brown and Warner (1985), p. 217‒218. 193 Based on an indexation starting on January 1, 2003. The percentage decline expresses the stock price decline from the individual group’s pre-crisis high to November 4, 2011. G-SIBs reached their pre-crisis high on May 18, 2007, Non-G-SIBs on July 18, 2007, and Other Banks on October 31, 2007.

4.4

Empirical Results

119

a Indexes for G-SIBs, Non-G-SIBs, and Other Banks contain the sample banks according to the designation stages as discussed in Sections 4.3.1.2 and 4.3.1.3.

Figure 21: Stock prices of G-SIBs, Non-G-SIBs, and Other Banks, 2003 to 2011

4.4.1

Overall results

In this section, we examine how the regulatory announcements impact stock returns of GSIBs. Table 8 provides an overview of the abnormal returns by event and sub-sample and Appendix A.16 displays the individual banks’ returns over all 12 event dates. We distinguish between general regulatory events – namely, G20 summits referencing multiple regulatory measures, specific regulatory events, which further detail a specific regulatory component (such as capital regulation, supervision, and resolution regimes), and designation events. Based on a simple aggregation exercise, G-SIBs show negative returns of -4.1% across all nine regulatory announcement dates and negative aggregate return differentials compared to Non-G-SIBs and Other Banks of -5.6% and -5.4%, respectively. Considering only the statistically significant event returns, the aggregate return for G-SIBs is -3.8% (events 1 and 5) and the aggregate return differentials compared to Non-G-SIBs and Other Banks -4.1% (events 1, 9, 10, and 11) and -5.4% (events 1, 3 5, and 11), respectively. These negative returns for G-SIBs contrast with the positive abnormal returns of +1.9% and the positive return differential to Non-G-SIBs of +4.3% G-SIBs experience on aggregate across all three designation events. Considering only statistically significant returns, the aggregate abnormal return for G-SIBs upon designation is +2.5% (events 4 and 7) and the aggregate return differential to Non-G-SIBs is +3.4% (events 4 and 7).

120 Table 8:

4.4.2

4.

Empirical Evidence from the New International Regulation

Abnormal returns by event

Regulatory announcements

Across the five general regulatory events, G-SIBs experienced aggregate negative returns of -3.4%, while Non-G-SIBs and Other Banks showed positive returns of +1.2% and +1.5%, respectively (considering only statistically significant returns, G-SIBs experienced an aggregate abnormal return of -3.8% (events 1 and 5) while Non-G-SIBs and Other

4.4

Empirical Results

121

Banks showed aggregate abnormal returns of 0.0% (events 3 and 5) and +1.2% (events 3, 5, and 8), respectively. These returns result in a total abnormal return differential between GSIBs and Non-G-SIBs across all five events equal to -4.6% (-2.9% if considering only the statistically significant return differential from event 1). All five individual events also show a negative abnormal return differential between G-SIBs and Non-G-SIBs. We observe the biggest abnormal return differential of -2.9%, which is statistically significant at the 5% level, occurring after the first G20 summit, in Washington, where the government switched from ensuring support for large banks toward calling for increased regulation and oversight. In that sense, the first G20 summit seems to have led market participants to anticipate much of the regulatory announcements made at the second and third G20 summits, which show negative abnormal return differentials of “only” -0.5% and -0.7%, respectively. The comparably weaker negative return differentials of only -0.4% and -0.2% on the fourth and fifth G20 summits, in Toronto and Seoul, could be attributed to two factors. First, these two summits resulted in less significant news compared to the initial three summits. Second, these two summits both took place after the occurrence of the first designation event, which, as we will argue in the next section, may have partially undermined the effect of announcing additional regulation. On aggregate, the general announcements made by the G20 leaders involve negative wealth effects for the world’s largest banks, caused by the market’s anticipation of the costs of the new additional regulation. The four FSB and BCBS announcements of specific policy proposals also show negative returns for G-SIBs and positive returns for Non-G-SIBs. Based on a simple aggregation across all four events, G-SIBs had negative returns of -0.7% and Non-G-SIBs of +0.3%, while Other Banks showed slightly negative returns of -0.1%; considering only statistically significant aggregate abnormal returns, Non-G-SIBs showed a return of +0.3% (events 9 and 10) and Other Banks of -0.1% (events 6, 10, and 11). This leads to an aggregate return differential of G-SIBs compared to Non-G-SIBs of -1.0% and compared to Other Banks of -0.5%; again considering only statistically significant abnormal returns G-SIBs show an aggregate return differential of -1.2% (events 9, 10, and 11) compared to Non-G-SIBs and 0.7% (event 11) compared to Other Banks. The specific policy announcements require a differentiated discussion. For the decision on capital surcharges, G-SIBs exhibit an abnormal return of -0.4% and a return differential to Non-G-SIBs of -0.9%, which is statistically significant according to the t-test at the 5% level. It is noteworthy that we observe the opposite pattern on the preceding announcement date in June 2011, which yields a positive return differential of +0.7% compared to Non-G-SIBs, with a statistical significance at the 10% level. This is despite the significant new information contained in the announcement of June 2011. First, a decision had been made regarding the instrument by focusing on Core Tier 1 Equity.

122

4.

Empirical Evidence from the New International Regulation

Consequently, contingent capital was no longer considered by the BCBS. 194 Second, the absolute height of the additional capital surcharges of up to 2.5% was announced. Third, the design of the capital surcharges as a progressive component was introduced, ranging from 1.0% to 2.5% depending on the systemic relevance of the individual institution. The positive abnormal return reaction by G-SIBs to the announcement could be attributed to two factors. First, the magnitude of the proposal could have been below market participants’ expectations given the initial announcements made by the G20 leaders and expectations would already have been priced in as a reaction to the initial announcements. Likewise, the fact that the announcement included news across several dimensions — type of instrument, amount of additional capital, and progressive nature — could also have led to opposing effects, with some of the proposal’s dimensions involving higher than and other dimensions involving lower than expected costs. Second, considering the negative abnormal return differentials on the decision date, uncertainty about whether these proposals would eventually be endorsed by the BCBS could have limited the market’s reaction to the June announcement. Regarding resolution regimes, we find a marginally negative abnormal return for GSIBs and a negative return differential of -1.1% compared to Non-G-SIBs. Notably, the -1.1% negative return differential at a significance level of 5% indicates that markets, on a relative basis, expect G-SIBs to face comparably higher costs from these measures. Hence, the fact that we observe positive returns for Non-G-SIBs could indicate that markets expect the funding advantages of G-SIBs over Non-G-SIBs to be lowered by the increased likelihood of a credible resolution regime being installed. This observation is important as the resolution regime was published after the first designation event occurred, and thus raised the relevance of the resolution regime as a tool for undoing exactly the effects of explicit government guarantees attributed to G-SIBs. Finally, regarding enhanced supervision, we observe only a minor negative abnormal return of G-SIBs of -0.2%, which results in a slightly positive return differential for G-SIBs of +0.2% and close to zero compared to Non-G-SIBs and Other Banks, respectively. These results could be explained by market expectations — based on the initial high-level announcements — being largely in line with the newly published proposals. In addition, despite describing core elements of the new supervisory framework, the resulting costs will, in part, depend on concrete implementation by the respective national and regional legislators. In sum, these negative returns of G-SIBs reflect the negative wealth effects the new regulation implies for the shareholders of these banks. To the extent that these anticipated costs represent a reduction in implicit government guarantees, these results would confirm the effectiveness of the broadly announced reform proposals in limiting TBTF. 194 In its announcement, the BCBS mentioned that it would continue reviewing contingent capital and support the use of contingent capital to meet higher national loss absorbency requirements.

4.4

Empirical Results

123

However, as some of the later regulatory announcements indicate, and as we will examine more closely in the next section, the designation of G-SIBs may have partly undermined the new regulation. 4.4.3

Designation announcements

Across the three designation events, G-SIBs exhibit positive abnormal returns of +1.9% on aggregate, in contrast to -2.4% for Non-G-SIBs, which results in a positive return differential of +4.3% compared to Non-G-SIBs; again considering only statistically significant returns, G-SIBs show an aggregate abnormal return upon designation of +2.5% (events 4 and 7) and a positive return differential of +3.4% (events 4 and 7) compared to Non-G-SIBs. Notably, the asymmetric returns of G-SIBs and Non-G-SIBs we observe for regulatory announcements have reversed for the designation events. G-SIBs now exhibit positive abnormal return differentials across all designation events. The fact that G-SIBs exhibit a relatively higher aggregate return differential versus Non-G-SIBs than versus Other Banks (+1.1%) is in line with our expectation that investors would be disappointed about the eventual exclusion of Non-G-SIBs from the list and that Non-G-SIBs and Other Banks are affected to a different degree by competition from G-SIBs. Recalling our discussion of designation events in Section 4.3.1.3, we consider the significant new information in the first designation event to be not only the specification of the banks that are considered G-SIBs, but the fact that there is an official list and the related increased probability that this list will be officially published at some point in time. In that sense, the designation constitutes significant news for all of the 24 banks included in the list and yields a positive abnormal return of +1.5% for G-SIBs and an abnormal return differential of +1.5% compared to Non-G-SIBs, which is statistically significant at the 5% level. The second designation event, which is the republication of this leaked list, explicitly relates this list to the new G-SIB regulation. Thus, it increases the confidence level of the leaked list being relevant for the regulation of G-SIBs, leading to a positive abnormal return for G-SIBs of +1.1% and an abnormal return differential compared to Non-G-SIBs of +1.9% with a statistical significance at the 1% level. As one additional step in our analysis, we examine how similar the stock price reactions of G-SIBs and Non-G-SIBs at the second designation event are compared to the reactions at the first designation event, given that the sample of G-SIBs at the second designation event remains unchanged from the one at the first designation event. We partition the sample into G-SIBs and Non-G-SIBs with a positive versus negative abnormal return at the first designation event and use these samples to analyze the banks’ reactions at the second event. Consistent with our findings based on the distinction of G-SIBs and Non-G-SIBs, we observe for the second designation event a positive abnormal return differential of +0.7% for banks with positive returns at the first desig-

124

4.

Empirical Evidence from the New International Regulation

nation event versus banks with negative returns at the first designation event. These findings are statistically significant at the 1% level. 195 The third designation event, the official publication of the list of G-SIBs by the FSB in November 2011, again constitutes significant news. First, it extends the list from 24 to 29 G-SIBs. Second, it excludes four banks previously included in the leaked list and thus identifies in total nine banks as G-SIBs for the first time. In order to measure the impact of the new information at this event, we focus for this designation event on the banks that were newly included in or excluded from the list of G-SIBs. Newly announced G-SIBs yield an abnormal return differential of +0.9% based on an abnormal return of -0.6% for new G-SIBs compared to an abnormal return of -1.5% for Non-G-SIBs. 196 197 We repeat the analysis for this designation event including all G-SIBs and Non-G-SIBs (i.e., also those that are not newly designated or removed from the list) and find that the overall GSIB sample only shows a marginally positive abnormal return differential of +0.02% compared to Non-G-SIBs and a small negative abnormal return differential of -0.3% compared to Other Banks, which could be explained by market anticipation of the repeat designation of those banks that were already included in the previously leaked lists (see Appendix A.18). These results show that G-SIBs benefited from obtaining the official status of being “too globally systemically important to fail” and the explicit government guarantee associated with this status. In that sense, the official designation of G-SIBs may have counteracted and in this sense undermined the proposed goal of the regulation to diminish the potency of the TBTF doctrine. This indicates that designation as a G-SIB creates value for the affected banks’ shareholders, consistent with our second hypothesis. These findings of positive return differentials are also consistent with O’Hara and Shaw’s (1990) observation of positive returns upon the announcement of explicit government guarantees. However, the 195 We perform a similar analysis for the third designation event. However, given that the G-SIB sample changes at the third designation event, we would not necessarily expect a certain return pattern. The abnormal return differential between G-SIBs and Non-G-SIBs with positive returns at the first designation event versus those with negative returns at the first designation event is -0.6% at the third designation event, lacking statistical significance (compare Appendix A.17). 196 The relatively low number of newly declared banks for the third event could have affected the levels of statistical significance for the third designation event. 197 We also compare the abnormal returns of the nine newly designated G-SIBs at the third event with the abnormal returns for these banks at the first designation event (0.39% with a p-value of 0.6004) and second designation event (-0.14% with a p-value of 0.6923). The observed lack of statistical significance for the abnormal returns for these nine banks at the first two events may generally relate to the small sample size and to the fact that the market potentially did not anticipate two of these banks (State Street and Bank of New York Mellon) to be included at the first two designation events, given that their balance sheets are comparably smaller and complexity and non-substitutability had not been officially discussed as factors of systemic importance at that point in time (compare Appendix A.19).

4.4

Empirical Results

125

fact that in the case of G-SIBs these explicit government guarantees are part of a regulation designed to mitigate the costs and risks resulting from government guarantees unveils the partially paradoxical nature of the new regulation. 4.4.4

Cross-sectional analysis of G-SIB returns

In this section, we analyze three cross-sectional aspects of G-SIBs in order to better understand the positive abnormal returns of G-SIBs upon designation. In a first step, we examine the aggregate abnormal returns across the three designation events for G-SIBs on an individual bank level (see Appendix A.16). 198 For banks with G-SIB status at all three designation events we observe that 14 out of 19 show positive aggregate returns and only two show negative returns exceeding -1%. The three Japanese G-SIBs show the strongest aggregate returns followed by 11 U.S. and European G-SIBs that show single digit aggregate abnormal returns. In order to further examine the impact of home country affiliation we repeat our analysis separating U.S. and European banks (see Appendix A.20). The results generally support our overall results with a statistically significant abnormal return differential of +5.8% for U.S. G-SIBs versus U.S. Non-G-SIBs at the first designation event and a (nonsignificant) abnormal return differential of +0.8% at the second designation event. Likewise, European G-SIBs show positive (non-significant) abnormal return differentials versus European Non-G-SIBs for the first two designation events, of +0.7% and +1.8%, respectively, but a negative return differential of -0.7% for the third designation event. It is important to note that the separated analysis significantly decreases the individual sample sizes (in several cases to less than three banks per sub-sample), which we expect to reduce statistical significance. In addition to the examination of abnormal return differentials by region, we include home country ratings in our regression analysis in order to understand the impact of sovereign rating differences on the value of G-SIB designation. We consider the sovereign rating of the home country of the individual G-SIBs to be an indication of financial strength and resources to provide government support in the form of bailouts in times of crisis. Intuitively, assuming national government guarantees, a higher sovereign rating of a bank’s home country should imply a higher value of G-SIB status, as the home country’s ability to provide sufficient funds for bailout increases. Although, in addition to the sovereign rating, the size of the bank relative to the home country’s GDP should also play an important role, the recent European sovereign debt crisis highlighted the importance of a country’s rating and ability to access capital markets in order to provide support to its financial sector. 198 For banks that change their status over the three designation events, the returns are calculated as an index expressing a bank’s stock price reaction to designation announcements in order to capture both the effect of designations and of removals from the list of G-SIBs. The returns are calculated as the sum of the stock return(s) upon designation plus (-1) times the stock return(s) upon the loss of G-SIB status.

126

4.

Empirical Evidence from the New International Regulation

Furthermore, we examine banks with a G-SIB status in fewer than three designation events and find that four out of six banks with negative returns are partially government owned at some point during the designation process, including the three banks with the most negative returns. For the banks with positive returns, only two out of seven banks are owned by the government during the regulatory process. Building on these indicative observations, we examine how government ownership influences the value of the G-SIB label for a bank by including government ownership as a variable in our regression analysis. Government ownership could imply existing government guarantees and thus lower the value resulting from an additional designation of a government owned bank as GSIB. This phenomenon, referred to as “Too-Public-to-Fail”, 199 implies that banks owned by governments are more likely to be bailed out should failure occur than are banks without any government ownership. Various studies empirically confirm this view. Faccio et al. (2006) find that, in general, politically connected firms are more likely to receive government bailouts. Likewise, Borisova and Megginson (2011) find increasing government ownership to be associated with lower cost of debt. Consequently, in the context of G-SIB regulation, we would expect the value of the G-SIB label to be highest for banks without any prior government support component implied by government ownership, and lowest for banks that are largely government owned. Thus, more positive stock return reactions of non-government owned G-SIBs upon designation would support a TBTF hypothesis. Finally, differences in aggregate abnormal returns of G-SIBs could reflect the market’s assessment of the banks’ systemic relevance and potential related higher costs for more systemically significant institutions. We therefore consider the level of additional capital requirements resulting from the G-SIB designation as defined by the regulators in late 2012 as well as total reported bank assets at the respective designation events. Generally, we would expect the inclusion of a G-SIB in a higher capital surcharge bucket to result in a less positive return reaction compared to its inclusion in a lower capital surcharge bucket, as the net effect of the designation’s benefits and costs of additional regulation should be lower. However, it is important to note that no official classification for individual G-SIBs into different capital surcharge buckets existed during the three observed designation events. Likewise, neither the fact that there is a progressive component in applying the capital surcharge nor the reference to all five systemic risk indicators took place until June 2011, and the specification of the assessment methodology was not released before July 2011. Therefore, this analysis presumes that markets were able to anticipate the progressive component for capital surcharges and to sufficiently differentiate between the systemic relevance of G-SIBs as defined later on in the process. We regress the observed abnormal returns on government ownership, sovereign risk rating, required capital surcharges and total bank assets across the three designation 199 See Seelig (2004), p. 220.

4.4

Empirical Results

127

events. The results are presented in Table 9. For government ownership, we first perform a regression with the actual ownership percentage. 200 In a second step, we perform a regression using a dummy variable, which equals 1 for G-SIBs in which the government owns a blocking minority stake of 25% or more, and 0 for G-SIBs in which the government does not own a blocking minority stake, resembling the previous literature’s definition of a “majority shareholder”. 201 For the sovereign risk rating analysis we assign numerical values to the sovereign ratings of the home countries of G-SIBs. 202 As a robustness test, we also conduct this analysis based on indicator variables of rating quality, distinguishing between G-SIBs based in countries with a AAA rating and G-SIBs based in countries with a rating below AAA at the respective designation event. For capital surcharges, we use the assigned required capital surcharge of 1.0% to 2.5% as assigned by regulators in November 2012. 203 The regression results confirm the importance of government ownership for the value of the G-SIB label. With a negative coefficient of -0.0493 and a p-value of 0.0019 (differentiating by minority blocking stake results in a coefficient of -0.0316 and a p-value of 0.0004), G-SIBs with higher government ownership react less positively to designation announcements compared to G-SIBs with low government ownership or no government ownership at all. These observations are in line with our general findings of positive designation event returns for G-SIBs. In contrast to government ownership, the G-SIB home country sovereign risk rating does not seem to significantly affect the value of the G-SIB label for a bank, as the low coefficient of -0.0004 with a high p-value of 0.8603 implies. Our robustness test with indicator variables for rating quality yields similar results. One factor which could explain this is the policy environment many G-SIBs in the eurozone faced over previous years. With the establishment of the European Financial Stability Facility (EFSF) and the European Stabilization Mechanism (ESM) in June 2010 and September 2012, respectively, market participants may have anticipated G-SIB status early on involving supranational guarantees for banks, at least in the eurozone. In fact, already in July 2010 the ability of European countries to draw on resources from the EFSF to bail out their banking system was officially confirmed — long before official policy changes allowed direct support for eurozone banks 200 See Appendix A.21 for an overview of economic stakes held by governments in G-SIBs. 201 See Faccio et al. (2006). 202 See Ueda and Weder di Mauro (2013) for the translation of ratings into numerical values. See Appendix A.22 for an overview of sovereign risk ratings for G-SIB home countries across the designation events. 203 According to the November 2012 publication of the FSB, we use 2.5% for HSBC, Deutsche Bank, JP Morgan Chase, and Citigroup; 2% for BNP Paribas and Barclays; 1.5% for Royal Bank of Scotland, Bank of America, Mitsubishi UFJ, UBS, Credit Suisse, Goldman Sachs, Morgan Stanley, and Bank of New York Mellon; and 1% for Crédit Agricole, ING Group, Mizuho Financial, Banco Santander, Société Générale, UniCredit, Bank of China, Sumitomo Mitsui, Wells Fargo, BBVA, Nordea, Standard Chartered, and State Street.

128 Table 9:

4.

Empirical Evidence from the New International Regulation

Regression results of cross-section of G-SIBs

4.4

Empirical Results

129

from rescue funds. 204 In this sense, market participants’ anticipation of a supranational guarantee could have at least partially leveled out return reactions of higher- and lowerrated home countries. The funds provided by the ESM and EFSF for the bailout of Spanish banks in June 2012, for example, confirm that such anticipation would have been justified. Likewise, the joint rescue of Dexia by France, Belgium, and Luxembourg demonstrates that banks with large cross-border exposure will potentially be supported by multiple countries involved, reflecting the complexity and cross-jurisdictional activity of G-SIBs. Also, recent requests by U.S. regulatory authorities for separate capital requirements for the U.S. investment banking entities of large European banks indicate that the recent crisis’s implicit agreement of home countries to bail out their own banks worldwide may no longer be considered unconditionally viable in the future. The level of capital surcharge has a negative coefficient of -0.3182, indicating that the additional costs of relatively higher capital requirements for more systemically significant G-SIBs compared to less systemically significant G-SIBs could have a muting effect on the designation event stock return, consistent with the TBTF hypothesis. However, with a pvalue of 0.6657 this result lacks statistical significance, which could be explained by the fact that the classification into buckets was conducted after the designation events took place, as discussed at the beginning of this section. 205 Finally, in conjunction with the before mentioned variables, the level of total assets does not appear to be a significant contributing factor, resulting a coefficient of 0.000. This may be due to the fact that market participants anticipated factors other than size to be used to identify G-SIBs. These were discussed throughout the regulatory process and eventually – in addition to size – defined as interconnectedness, complexity, cross-jurisdictional activity and non-substitutability. Furthermore, all G-SIBs have balance sheets greater than 100 billion Euros. As a result, our cross-sectional analysis would not have captured any size effects G-SIBs may have compared to smaller banks. Lastly, considering the global nature of our sample, differences in accounting regimes between different jurisdictions – such as the degree to which netting of derivative positions is permitted – may have limited the representativeness of reported total bank assets for de facto bank size. Summarizing our cross-sectional analysis, we find that the value of obtaining G-SIB status is lower for G-SIBs with a higher government ownership stake. These findings support our more general results regarding the positive value of G-SIB designation. Regarding the sovereign rating of the home countries of G-SIBs, we do not observe a clear relationship between home country rating and value of G-SIB designation. In part, this may be due to the fact that supranational bank bailouts — as exemplified by rescue measures employed 204 Reuters, July 6, 2010. 205 Our regression analysis of systemic relevance as expressed by the classification of banks into buckets reduces the number of observations across the three designation events to 46, because banks that were only temporarily included in our G-SIB sample were never classified by regulators. Therefore, we include the variable in a separate regression.

130

4.

Empirical Evidence from the New International Regulation

by the ESM and EFSF — may have been anticipated by markets and thus offset the importance of home country ratings. The degree of systemic relevance as expressed by the required level of capital surcharge appears to have a dampening effect on the returns on designation. 4.5

Conclusion

In a reaction to the comprehensive bailouts of large banks during the financial crisis, the G20’s leaders decided to introduce a new regulation specifically aimed at G-SIBs. Over the subsequent three years, international regulatory authorities developed concrete proposals for the regulation, supervision, and resolution of G-SIBs. As a consequence of these additional regulatory measures, we observe negative absolute as well as negative relative abnormal returns on stocks of G-SIBs compared to “Non-G-SIBs” and “Other Banks” upon announcements related to the new regulatory measures. This confirms the hypothesis that the new regulation for G-SIBs involves costs for the affected banks. However, the new regulation also comes with an official designation of particular banks as G-SIBs. This designation leads to positive aggregate absolute and relative abnormal returns for G-SIBs across all three designation events, which suggests that official designation as a G-SIB creates value for the affected banks’ shareholders. The level of ownership of a bank by domestic government determines the value of G-SIB designation. Yet, an analysis of the ratings of G-SIBs’ home countries suggest that, to a certain degree, G-SIB status also has a supranational support component. Furthermore, the additional costs associated for G-SIBs with a higher systemic relevance compared to less systemically relevant G-SIBs seems to have a decreasing effect on the returns on the designation event. Overall these results confirm the hypothesis that G-SIB designation has positive value effects for the affected banks. These results illustrate the potentially unintended consequences of the new regulation. At the same time, they demonstrate that TBTF effects not only stem from government announcements or bank rescue measures, but can also be created by a regulation specifically designed to mitigate the costs and risks of TBTF — a somewhat paradoxical aspect of the new regulatory proposals. Finally, in evaluating the new G-SIB regulation from a policy perspective, to the extent that the observed future costs from the new regulation represent a reduction in implicit government guarantees, our results confirm the effectiveness of the announced reform proposals to limit TBTF. However, the fact that the official designation in part contradicts these effects raises an important issue. Apparently, revealing the identities of G-SIBs eliminated ambiguity about the presence of government guarantees, and thereby may have run counter to the regulators’ intent to contain the effects of TBTF. This suggests the need for a

4.5

Conclusion

131

conscious decision about the level of supervisory transparency versus constructive ambiguity when future policy reevaluations are carried out and recommendations are made. Furthermore, our findings bring into focus the importance of credible resolution regimes, as this may be the right conceptual tool to undo the effects we observe as a result of designating banks as G-SIBs.

5.

Conclusion

This thesis discusses the regulation of systemically relevant banks along three subsequent questions: Why should systemically relevant banks be regulated? What are relevant policy options and how should systemically relevant banks be regulated? Does the new regulation for Global Systemically Important Banks succeed in limiting TBTF? The structure of this thesis follows these three questions. Chapter 2, “Consequences of government guarantees for banks – a survey of the TBTF doctrine” responds to the first question by examining the consequences of government guarantees in order to determine the need for regulating systemically relevant banks. We review the extensive body of literature on TBTF in banking and provide an overview of both empirical approaches and findings on the status of TBTF, with an emphasis on post-crisis contributions. Our analysis emphasizes the need to differentiate between the major consequences of government guarantees – government risk exposure, competitive distortions and moral hazard. We find that scholars have estimated guarantee values to range from more than a trillion US dollar annually based on contingent claims approaches to positive net present values based on past rescue measures. This emphasizes the need for governments to actively measure and manage their exposure to the banking system, considering risk and returns, a view we develop in Chapter 3. Studies attempting to examine the impact of government guarantees via competitive dynamics – or funding spreads – provide mixed evidence across refinancing instruments. However, most studies observe a recent spread compression after a significant spread widening during the crisis period, raising the question as to what factors are driving these developments, and, more generally, suggesting a more specific examination of the observed competitive dynamics. Studies focused on guarantee-risk and risk-return relationships, while providing partly mixed evidence, highlight the difficulty of empirically measuring the link between government guarantees and moral hazard, given that banks appear to trade off different types of risk and that the interaction between individual bank risk and banking system risk appears to be complex. Studies examining moral hazard should carefully select guarantee indicators to not only ensure predictive strength for actual guarantee levels but also correctly interpret potential underlying motivations for the provision of these guarantees. Chapter 3, “Government guarantees and banking system risk – a regulatory framework from an exposure perspective” responds to the second question on relevant policy options and adequate regulation. We develop an exposure view on banking system risk and bank regulation, providing an analytic framework to evaluate regulatory policy options in order to achieve a desired level of government exposure to banking system risk. This contribution to our knowledge is the first comprehensive approach to categorize and discuss the full range

© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2018 S. C. Moenninghoff, The Regulation of Systemically Relevant Banks, Finanzwirtschaft, Banken und Bankmanagement  Finance, Banks and Bank Management, https://doi.org/10.1007/978-3-658-23811-7_5

134

5.

Conclusion

of major policy options for regulating banks within a single framework and to provide a quantitative methodology to compare the relative attractiveness of different policy choices. Overall, our stylized model suggests that the choice of an optimal regulatory policy mix depends on risk and return preferences of a society, and an economy’s institutional and cultural setting which impacts the tradeoffs implied by each policy option. These options range from free banking and narrow banking regimes (implying zero government exposure) over regulations such as minimum capital and liquidity requirements, structural restrictions, Pigovian taxes, resolution powers, bail-in regimes, contingent capital and resolution funds (implying limited government exposure) to a nationalized banking system (implying full government exposure). Our stylized model suggests that a society with higher risk tolerance will prefer limited exposure solutions such as bail-in regimes and capital regulations, while a highly risk-averse society may be more open to increased government involvement in the intermediation process, for example through structural restrictions. Assessing the new regulation dealing with G-SIBs based on our framework, we demonstrate that after two decades of focusing on minimizing the likelihood of crisis, regulators increased their efforts to also reduce losses and government exposure in case of crisis by introducing international standards for resolution regimes including wind-down authorities and creditor bail-ins. Our framework enables a broad range of regulatory policy options to be categorized and evaluated within a single framework based on exposure implications, which allows the postcrisis regulatory debate to be structured better. Also, it makes explicit the tradeoffs implied by different regulatory policy options, allowing for a more comprehensive and differentiated discussion of benefits and costs of policy decisions, a step in the policy decision making which had been mostly skipped during the recent crisis management and reform efforts. Finally, compared to the traditional externalities view in which bank bailouts are considered unwanted negative externalities, it raises the question of what the appropriate level of government exposure to the banking system should be and thus takes a pro-active stance to potential banking system risk acceptance by the government. Chapter 4, “Empirical Evidence from the New International Regulation Dealing with Global Systemically Important Banks”, responds to the third question, providing an empirical evaluation of the G-SIB regulation developed by the FSB and the BCBS based on an event study. We find that the new regulation negatively affects G-SIBs, yet that at the same time the official designation of banks as G-SIBs has a partly offsetting impact, suggesting that investors did not believe that governments would allow those banks to fail. Our cross-sectional analysis of the valuation effects with respect to, for example, government ownership of banks supports the view that the positive reaction to these designations can be attributed to a TBTF perception. These results suggest that even though the individual components of the regulation have been effective, revealing the identities of G-SIBs eliminated ambiguity about the presence of government guarantees, and thereby may have run counter to the regulators’ intent to contain the effects of TBTF.

5.

Conclusion

135

Overall, our findings suggest that policymakers should take an active view of government exposure to banking system risk by considering it part of a broader policy decision. While empirical evidence shows that government exposure to the banking system is significant, our theoretical exposure framework offers the full range of major policy options to calibrate exposure based on economic risk-return optimization and a society’s risk preferences. The new regulation of G-SIBs, while extending the set of policies to manage government risk exposure, does not appear to be aiming to fully eliminate government exposure from a conceptual point of view, which is in-line with our empirical observation of TBTF effects in the context of the regulation’s implementation.

Bibliography

Acharya, S., Dreyfus, J.F., 1989. Optimal Bank Reorganization Policies and the Pricing of Federal Deposit Insurance. Journal of Finance, 44, 1313-1333. Acharya, V.V., Pedersen, L., Philippon, T., Richardson, M.P., 2016. Measuring Systemic Risk. Review of Financial Studies, Forthcoming. Acharya, V.V., Anginer, D., V.V., Warburton, A.J., 2013. The end of market discipline? Investor expectations of implicit state guarantees. New York University Leonard N. Stern School of Business Research Paper. Adams, M., Muck, M., Rudolf, M., 2004. Basel II – A Guarantee for a Stable Banking System? Financial Markets and Portfolio Management, 18, 306-311. Admati, A.R., and Hellwig, M.F., 2013. The Banker’s New Clothes. Princeton University Press. Admati, A.R., DeMarzo, P.M., Hellwig, M.F., Pfleiderer, P., 2011. Fallacies, Irrelevant Facts, and Myths in the Discussion of Capital Regulation: Why Bank Equity is Not Expensive. Stanford GSB Research Paper 2063. Afonso, G., Santos, J.A.C., Traina, J., 2015. Do “too-big-to-fail” banks take on more risk? Journal of Financial Perspectives, 3, 129-143. Allen, M., Rosenberg, C., Keller, C., Setser, B., Roubini, N., 2002. A Balance Sheet Approach to Financial Crisis. International Monetary Fund Working Paper 210. Alperovitz, G., 2012. Wall Street Is Too Big to Regulate. New York Times, July 22, 2012. Altman, E.I., Saunders, A., 1998. Credit Risk Measurement: Developments Over the Last 20 Years. Journal of Banking & Finance, 21, 1721-1742. Angbazo, L., Saunders, A., 1997. The effect of TBTF deregulation on bank cost of funds. New York University Leonard N. Stern School of Business Research Paper. Anginer, D., Demirguc-Kunt, A., Zhu, M., 2013. How does deposit insurance affect bank risk? Evidence from the recent crisis. Journal of Banking & Finance, 48, 312-321. Araten, M., 2013. Credit Ratings as Indicators of Implicit Government Support for Global Systemically Important Banks. Journal of Risk Management in Financial Institutions, 7, 345-352. Araten, M., Turner, C., 2013. Understanding the funding cost differences between global systemically important banks (GSIBs) and non-G-SIBs in the USA. Journal of Risk Management in Financial Institutions, 6, 387-410. Arrow, K.J., 1963. Uncertainty and the welfare economics of medical care. American Economic Review, 53, 941-973. Arrow, K.J., 1970. Political and Economic Evaluation of Social Effects and Externalities. In: Margolis, J. (Ed.), The Analysis of Public Output. National Bureau of Economic Research, 1-30. Ashcraft, A.B., 2005. Are Banks Really Special? New Evidence From the FDIC-Induced Failure of Healthy Banks. American Economic Review, 95, 1712-1730. Athavale, M., 2000. Uninsured deposits and the too-big-to-fail policy in 1984 and 1991. American Business Review, 18, 123-128. Avdjiev, S., Kartasheva, A., Bogdanova, B., 2013. CoCos: a Primer, BIS Quarterly Review. Bank for International Settlements. Avgouleas, E., Goodhart, C., Schoenmaker, D., 2013. Bank Resolution Plans as a Catalyst for Global Financial Reform. Journal of Financial Stability, 9, 210-218. Baker, D., McArthur, T., 2009. The value of the “too big to fail” big bank subsidy. Center for Economic and Policy Research Issue Brief, September.

© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2018 S. C. Moenninghoff, The Regulation of Systemically Relevant Banks, Finanzwirtschaft, Banken und Bankmanagement  Finance, Banks and Bank Management, https://doi.org/10.1007/978-3-658-23811-7

138

Bibliography

Balasubramnian, B., Cyree, K.B., 2011. Market discipline of banks: Why are yield spreads on bankissued subordinated notes and debentures not sensitive to bank risks? Journal of Banking & Finance, 35, 21-35. Banerjee, A.V., 1997. A Theory of Misgovernance. Quarterly Journal of Economics, 112, 1289-1332. Bank of England, 2009. Financial Stability Report, June, 25. Barth, A., Schnabel, I., 2013. Why banks are not too big to fail – evidence from the CDS market. Economic Policy, 28, 335-369. Barth, J.R., Caprio, G., Levine, R., 2008. Rethinking bank regulation: Till angels govern. Cambridge University Press. Bartholdy, J., Boyle, G.W., Stover, R.D., 2003. Deposit insurance and the risk premium in bank deposit rates. Journal of Banking & Finance, 27, 699-717.BCBS, 1992. The Insolvency Liquidation of a Multinational Bank. Bank for International Settlements. BCBS, 1999. Credit Risk Modelling: Current Practices and Applications. Bank for International Settlements. BCBS, 2000. Range of Practice in Banks’ Internal Rating Systems. Bank for International Settlements. BCBS, 2005. An Explanatory Note on the Basel II IRB Risk Weight Functions. Bank for International Settlements. BCBS, 2010. Final Report: Assessing the macroeconomic impact of the transition to stronger capital and liquidity requirements. Bank for International Settlements. BCBS, 2011a. Basel III: A Global Regulatory Framework for More Resilient Banks and Banking Systems. Bank for International Settlements. BCBS, 2011b. Global systemically important banks: assessment methodology and the additional loss absorbency requirement. Bank for International Settlements. BCBS, 2015. Assessing the economic costs and benefits of TLAC implementation. Bank for International Settlements. Beighly, H.P., Boyd, J.H., Jacobs, D.P., 1975. Bank Equities and Investor Risk Perceptions: Some Entailments to Capital Adequacy Regulation. Journal of Bank Research, Autumn 1975, 190-221. Benes, J., Kumhof, M., 2012. The Chicago Plan Revisited. International Monetary Fund Working Paper 202. Berger, A.N., Turk-Ariss, R., 2015. Do depositors discipline banks and did government actions during the recent crisis reduce this discipline? An international perspective. Journal of Financial Services Research, 48, 103-126. Berger, A.N., Herring, R.J., Szegö, G.P., 1995. The Role of Capital in Financial Institutions. Journal of Banking & Finance, 19, 393-430. Berger, A.N., Imbierowicz, B., Rauch, C., 2016. The roles of corporate governance in bank failures during the recent financial crisis. Journal of Money, Credit and Banking, 48, 729-770. Beyhaghi, M., D’Souza, C., Roberts, G.S., 2014. Funding advantage and market discipline in the Canadian banking sector. Journal of Banking & Finance, 48, 396-410. Bijlsma, M.J., Lukkezen, J., Marinova, K.H., 2014. Measuring Too-Big-To-Fail Funding Advantages from Small Banks’ CDS Spreads. TILEC Discussion Paper 2014-012. Billett, M.T., Garfinkel, J.A., O’Neal, E.S., 1998. The cost of market versus regulatory discipline in banking. Journal of Financial Economics, 48, 333-358. Black, F., Scholes, M., 1973. The Pricing of Options and Corporate Liabilities. Journal of Political Economy, 81, 637-654. Black, L.K., Hazelwood, L.N., 2013. The effect of TARP on bank risk-taking. Journal of Financial Stability, 9, 790-803. Bliss, R.R., 2001. Market discipline and subordinated debt: A review of some salient issues. Federal Reserve Bank of Chicago Economic Perspectives, 24-45. Bliss, R.R., Flannery, M.J., 2002. Market discipline in the governance of US bank holding companies: Monitoring vs. influencing. European Finance Review, 6, 361-396.

Bibliography

139

Bodenhorn, H., 2003. Short-Term Loans and Long-Term Relationships: Relationship Lending in Early America. Journal of Money, Credit and Banking, 35, 485-505. Bodie, Z., Briere, M., 2014a. Sovereign Wealth and Risk Management: A Framework for Optimal Asset Allocation of Sovereign Wealth. Journal Of Investment Management, First Quarter 2014. Bodie, Z., Briere, M., 2014b. Optimal Asset Allocation for Sovereign Wealth Funds: Theory and Practice. Bankers, Markets and Investors, 128, 49-54. Bohn, J.R., 2000. A Survey of Contingent-Claims Approaches to Risky Debt Valuation. Journal of Risk Finance, 1, 53-70. Bongini, P., Nieri, L., Pelagatti, M., 2015. The importance of being systemically important financial institutions. Journal of Banking & Finance, 50, 562-574. Boone, P., Johnson, S., 2010. Will the Politics of Moral Hazard Sink Us Again? In: Turner, A., Haldane, A., Woolley, P., Wadhwani, S., Goodhart, C., Smithers, A., Large, A., Kay, J., Wolf, M., Boone, P., Johnson, S., Layard, R. (Ed.): The Future of Finance – The LSE Report, 238-273. Bordo, M.D., 1990. The Lender of Last Resort: Alternative Views and Historical Experience. Federal Reserve Bank of Richmond Economic Review, 76, 18-29. Borisova, G., Megginson, W.L., 2011. Does government ownership affect the cost of debt? Evidence from privatization. Review of Financial Studies, 24, 2693-2737. Boyd, J.H., Gertler, M., 1994. Are banks dead? Or are the reports greatly exaggerated? Federal Reserve Bank of Minneapolis Quarterly Review, 18. Brewer, E., Mondschean, T.H., 1994. An empirical test of the incentive effects of deposit insurance: The case of junk bonds at savings and loan associations. Journal of Money, Credit and Banking, 26, 146-164. Brown, S.J., Warner, J.B., 1985. Using daily stock returns: The case of event studies. Journal of Financial Economics, 14, 3-31. Brownlees, C.T., Engle, R., 2010. Volatility, correlation and tails for systemic risk measurement. New York University Leonard N. Stern School of Business Research Paper.Brunnermeier, M.K., and Adrian, T., 2011. Covar. National Bureau of Economic Research Working Paper 17454. Buchanan, J.M., 1969a. External Diseconomies, Corrective Taxes, and Market Structure. American Economic Review, 59, 174-177. Buchanan, J.M., 1969b. Cost and Choice: An Inquiry in Economic Theory. University of Chicago Press. Buiter, W., 1983. Measurement of the Public Sector Deficit and its Implications for Policy Evaluation and Design. In: Blejer, M.I, Cheastey, A. (Ed.), How to Measure the Fiscal Deficit. International Monetary Fund, 297-344. Buiter, W., 2008. The End of American Capitalism as We Knew It. Financial Times, September 17. Burhouse, S., Feid, J., French, G., Ligon, K., 2003. Basel and the Evolution of Capital: Moving Forward, Looking Back. FDIC Research Paper. Calomiris, C.W., Mason, J.R., 1994. Contagion and bank failures during the Great Depression: The June 1932 Chicago banking panic. American Economic Review, 87, 863-883. Campbell, J.Y., Lo, A.W., MacKinlay, A.C., 1997. The Econometrics of Financial Markets. Princeton University Press. Campbell, C.J., Cowan, A.R., Salotti, V., 2010. Multi-country event-study methods. Journal of Banking & Finance, 34, 3078-3090. Cannella Jr, A.A., Fraser, D.R., Lee, D.S., 1995. Firm failure and managerial labor markets evidence from Texas banking. Journal of Financial Economics, 38, 185-210. Chava, S., Ganduri, R., Yerramilli, V., 2014. Do Bond Investors Price Tail Risk Exposures of Financial Institutions? Working Paper. Chen, Y., Dou, P.Y., Rhee, S.G., Truong, C., Veeraraghavan, M., 2015. National culture and corporate cash holdings around the world, Journal of Banking & Finance, 50, 1-18. Chernykh, L., Cole, R., 2011. Does deposit insurance improve financial intermediation? Evidence from the Russian experiment. Journal of Banking & Finance, 35, 388-402.

140

Bibliography

Cihak, M., Maechler, A.M., Schaeck, K., Stolz, S.M., 2012. Who disciplines bank managers? Review of Finance, 16, 197-243. Clark, J., Gerken, A., Guse, F., Härle, P., 2012. Lösungsansätze zur Übertragung von Portfolios und nicht strategienotwendigen Geschäftsbereichen auf eine “Bad Bank”.
In: Bolder, M., Wargers, M. (Ed.): Modell‚ Bad Bank: Hintergrund – Konzept - Erfahrungen, Gabler, 79-100. Coase, R.H., 1960. The Problem of Social Cost. Journal of Law and Economics, 3, 1-44. Cochrane, J.H., 2014. Toward a Run-free Financial System. In: Taylor, J.B., Baily, M.N. (Ed.), Across the Great Divide: New Perspectives on the Financial Crisis, Hoover, 197-249. Committee on Banking, Finance and Urban Affairs, 1984. Hearings Before the Subcommittee on Financial Institutions Supervision, Regulation and Insurance, September 18, 19 and October 4, 98-111. Conlon, T., Cotter, J., 2014. Anatomy of a Bail-in. Journal of Financial Stability, 15, 257-263. Cook, D.O., Spellman, L.J., 1996. Firm and guarantor risk, risk contagion, and the interfirm spread among insured deposits. Journal of Financial and Quantitative Analysis, 31, 265-281. Cooperstein, R.L., Pennacchi, G.G., Redburn, F.S., 1995. The Aggregate Cost of Deposit Insurance: A Multiperiod Analysis. Journal of Financial Intermediation, 4, 242-271. Cordell, L.R., King, K.K., 1995. A Market Evaluation of the Risk-based Capital Standards for the US Financial System. Journal of Banking & Finance, 19, 531-562. Cornell, B., Shapiro, A.C., 1986. The reaction of bank stock prices to the international debt crisis. Journal of Banking & Finance, 10, 55-73. Cornett, M.M., McNutt, J.J., Strahan, P.E., Tehranian, H., 2011. Liquidity Risk Management and Credit Supply in the Financial Crisis. Journal of Financial Economics, 101, 297-312. Correa, R., Lee, K.H., Sapriza, H., Suarez, G.A., 2014. Sovereign credit risk, banks' government support, and bank stock returns around the world. Journal of Money, Credit and Banking, 46, 93121. Credit Suisse, 2013. Global Wealth Databook 2013. Crosbie, P., Bohn, J., 2003. Modeling Default Risk. KMV. Cubillas, E., Fonseca, A.R., González, F., 2012. Banking crises and market discipline: International evidence. Journal of Banking & Finance, 36, 2285-2298. Cubillas, E., González, F., 2014. Financial liberalization and bank risk-taking: International evidence. Journal of Financial Stability, 11, 32-48. Dam, L., Koetter, M., 2012. Bank bailouts and moral hazard: Evidence from Germany. Review of Financial Studies, 25, 2343-2380. Damar, E., Gropp, R., Mordel, A., 2013. The Ex Ante versus Ex Post Effect of Public Guarantees. In: Evanoff, D.D., Holthausen, C., Kaufman, G.G., Kremer, M. (Ed.), The Role of Central Banks in Financial Stability: How Has It Changed? World Scientific, 347-364. Daníelson, J., 2003. On the feasibility of risk based regulation. CESifo Economic Studies 49.2 (2003): 157-179. Davenport, A.M., McDill, K.M., 2006. The depositor behind the discipline: A micro-level case study of Hamilton Bank. Journal of Financial Services Research, 30, 93-109. DeLong, G., Saunders, A., 2011. Did the introduction of fixed-rate federal deposit insurance increase long-term bank risk-taking? Journal of Financial Stability, 7, 19-25. Demirgüç-Kunt, A., Detragiache, E., 2002. Does deposit insurance increase banking system stability? An empirical investigation. Journal of Monetary Economics, 49, 1373-1406. Demirgüç-Kunt, A., Detragiache, E., Gupta, P., 2006. Inside the crisis: An empirical analysis of banking systems in distress. Journal of International Money and Finance, 25, 702-718. Demsetz, R.S., Strahan, P.E., 1997. Diversification, size, and risk at bank holding companies. Journal of Money, Credit, and Banking, 29, 300-313. Dermine, J., 2009. Bank Valuation and Value-Based Management. McGraw Hill. Das, U.S., Lu, Y., Papaioannou, M.G., Petrova, I., 2013. Sovereign Risk and Asset and Liability Management - Conceptual Issues. Journal of Reviews on Global Economics, 2, 330-355.

Bibliography

141

De Ceuster, M.J.K., Masschelein, N., 2003. Regulating Banks through Market Discipline: A Survey of the Issues. Journal of Economic Surveys, 17, 749-766. Demirgüç-Kunt, A., Huizinga, H., 2004. Market discipline and deposit insurance. Journal of Monetary Economics, 51, 375-399. Demirgüç-Kunt, A., Huizinga, H., 2013. Are banks too big to fail or too big to save? International evidence from equity prices and CDS spreads. Journal of Banking & Finance, 37, 875-894. Deng, S.E., Elyasiani, E., Mao, C.X., 2007. Diversification and the cost of debt of bank holding companies. Journal of Banking & Finance, 31, 2453-2473. DeYoung, R., Flannery, M.J., Lang, W.W., Sorescu, S.M., 2001. The Information Content of Bank Exam Ratings and Subordinated Debt Prices. Journal of Money, Credit and Banking, 33, 900925. DeYoung, R., Evanoff, D.D., Molyneux, P., 2009. Mergers and Acquisitions of Financial Institutions: A Review of the Post-2000 Literature. Journal of Financial Services Research, 36, 87-110. Diamond, D.W., Dybvig, P.H., 1983. Bank Runs, Deposit Insurance, and Liquidity. Journal of Political Economy, 91, 401-419. Diamond, D.W., 1984. Financial Intermediation and Delegated Monitoring. Review of Economic Studies, 51, 393-414. Diamond, D.W., Rajan, R.G., 1999. Liquidity Risk, Liquidity Creation and Financial Fragility: A Theory of Banking. Journal of Political Economy, 109, 287-327. Diamond, D.W., Rajan, R.G., 2000. A Theory of Bank Capital. Journal of Finance, 55, 2431-2465. Dinç, I.S., 2005. Politicians and Banks: Political influences on Government-owned Banks in Emerging Markets. Journal of Financial Economics, 77, 453-479. Duan, J.C., Moreau, A.F., Sealey, C.W., 1995. Deposit Insurance and Bank Interest Rate Risk: Pricing and Regulatory Implications. Journal of Banking & Finance, 19, 1091-1108. Duchin, R., Sosyura, D., 2014. Safer Ratios, Riskier Portfolios: Banks’ Response to Government Aid. Journal of Financial Economics, 113, 1-28. ECB, 2007. Liquidity Risk Management of Cross-border Banking Groups in the EU. EU Banking Structures, October. Ellis, D.M., Flannery, M.J., 1992. Does the debt market assess large banks' risk? Time series evidence from money center CDs. Journal of Monetary Economics, 30, 481-502. Epstein, G., 2010. Finance without Financiers: Prospects for Radical Change in Financial Governance: David Gordon Memorial Lecture. Review of Radical Political Economics, 42, 293-306. Evanoff, D.D., Wall, L.D., 2001. Sub-debt yield spreads as bank risk measures. Journal of Financial Services Research, 20, 121-145. Faccio, M., Masulis, R.W., McConnell, J.J., 2006. Political connections and corporate bailouts. Journal of Finance, 61, 2597-2635. Farhi, E., Tirole, J., 2016. Deadly embrace: Sovereign and financial balance sheets doom loops. National Bureau of Economic Research Paper 21843. Federal Reserve Bank of Minneapolis, 2016. The Minneapolis Plan to End Too Big to Fail, November. Financial Stability Board, 2011a. Key Attributes of Effective Resolution Regimes for Financial Institutions, October. Financial Stability Board 2011b. Policy Measures to Address Systemically Important Financial Institutions, November 4. Financial Stability Board, 2013. Progress and Next Steps Towards Ending “Too-Big-To-Fail” (TBTF), September 2. Fiechter, J., Ötker, M.I., Ilyina, A., Hsu, M., Santos, M.A., Surti, J., 2011. Subsidiaries or Branches: Does One Size Fit All? International Monetary Fund Discussion Note 1104. Fischer, M., Hainz, C., Rocholl, J., Steffen, S., 2012. Government guarantees and bank risk taking incentives. AFA 2012 Chicago Meetings Paper.

142

Bibliography

Flannery, M.J., Sorescu, S.M., 1996. Evidence of bank market discipline in subordinated debenture yields: 1983-1991. Journal of Finance, 51, 1347-1377. Flannery, M.J., Nikolova, S., 2004. Market Discipline of U.S. Financial Firms: Recent Evidence and Research Issues. In: Borio, C., Hunter, W.C., Kaufman, G., Tsatsaronis, K. (Ed.): Market Discipline Across Countries and Industries, MIT Press, 87-100. Forssbæck, J., 2011. Ownership structure, market discipline, and banks’ risk-taking incentives under deposit insurance. Journal of Banking & Finance, 35, 2666-2678. French, K., Baily, M., Campbell, J., Cochrane, J., Diamond, D., Duffie, D., Kashyap, A.K., Mishkin, F.S., Rajan, R.G., Scharfstein, D.S., Shiller, R.J., Shin, Hyun Song, Slaughter, M.J., Stein, J.C., Stulz, R., 2010. The Squam Lake Report: Fixing the Financial System. Journal of Applied Corporate Finance, 22, 8-21. Frenkel, M., Rudolf, M., 2010. The Implications of Introducing an Additional Regulatory Constraint on Banks' Business Activities in the form of a Leverage Ratio. Association of German Banks, March 1. Freixas, X., Rochet, J.-C., 2008. Microeconomics of Banking. MIT Press. Friedman, M., Schwartz, A.J., 1986. Has Government Any Role in Money? Journal of Monetary Economics, 17, 37-62. Furfine, C.H., 2001. Banks as Monitors of Other Banks: Evidence from the Overnight Federal Funds Market. Journal of Business, 74, 33-57. Furfine, C., 2006. The Costs and Benefits of Moral Suasion: Evidence from the Rescue of Long‐Term Capital Management. Journal of Business, 79, 593-622. Gadanecz, B., Tsatsaronis K., Altunbasb, Y., 2012. Spoilt and Lazy: The Impact of State Support on Bank Behavior in the International Loan Market. International Journal of Central Banking, 8, 121-173. Gambacorta, L., Rixtel, A.V., 2013. Structural Bank Regulation Initiatives: Approaches and Implications. Bank for International Settlements Working Paper 412. Gapen, M.T., Xiao, Y., Gray, D.F., Lim, C.H., 2005. Measuring and Analyzing Sovereign Risk with Contingent Claims. International Monetary Fund Working Paper 01-155. Ghandi, P., Lustig, H., 2015. Size anomalies in U.S. bank stock returns. Journal of Finance, 70, 733768. Giammarino, R., Schwartz, E., Zechner, J., 1989. Market Valuation of Bank Assets and Deposit Insurance in Canada. Canadian Journal of Economics, 22, 109-127. Gilbert, R.A., 1990. Market discipline of bank risk: Theory and evidence. Federal Reserve Bank of St. Louis Review, 72, 3-18. Gizycki, M.C., Levonian, M., 1993. A Decade of Australian Banking Risk: Evidence from Share Prices. Reserve Bank of Australia Research Dicussion Paper 9302. Goldberg, L.G., Hudgins, S.C., 1996. Response of Uninsured Depositors to Impending S&L Failures: Evidence of Depositor Discipline. Quarterly Review of Economics and Finance, 36, 311-325. Goldberg, L.G., Hudgins, S.C., 2002. Depositor Discipline and Changing Strategies for Regulating Thrift Institutions. Journal of Financial Economics, 63, 263-274. Goodhart, C.A., 1987. Why Do Banks need a Central Bank? Oxford Economic Papers, 39, 75-89. Goodhart, C.A., Huang, H., 2005. The lender of last resort. Journal of Banking & Finance, 29, 10591082. Gordy, M.B., 2003. A Risk-factor Model Foundation for Ratings-based Bank Capital Rules. Journal of Financial Intermediation, 12, 199-232. Gordy, M.B., Heitfield, E.A., 2010. Risk-based Regulatory Capital and Basel II. In: Berger, A.N., Molyneux, P., Wilson, J.O.S., The Oxford Handbook of Banking, Oxford University Press, 357376. Gorton, Gary, and Santomero, A.M., 1990. Market discipline and bank subordinated debt: Note. Journal of Money, Credit and Banking, 22, 119-128.

Bibliography

143

Goyal, V.K., 2005. Market discipline of bank risk: Evidence from subordinated debt contracts. Journal of Financial Intermediation, 14, 318-350. Gray, D.F., Merton, R.C., Bodie, Z., 2007. Contingent Claims Approach to Measuring and Managing Sovereign Credit Risk. Journal of Investment Management, 5, 5-28. Gray, D., Malone, S., 2008. Macrofinancial Risk Analysis. John Wiley and Sons. Greenbaum, S.I., Thakor, A.V., 2008. Contemporary Financial Intermediation. Academic Press. Gropp, R., Gruendl, C., Guettler, A., 2014. The Impact of Public Guarantees on Bank Risk-Taking: Evidence from a Natural Experiment. Review of Finance, 18, 457-488. Gropp, R., Hakenes, H., Schnabel, I., 2011. Competition, risk-shifting, and public bail-out policies. Review of Financial Studies, 24, 2084-2120. Gropp, R., Vesala, J., 2004. Deposit insurance, moral hazard and market monitoring. Review of Finance, 8, 571-602. Gropp, R., 2004. Bank Market Discipline and Indicators of Banking System Risk: The European Evidence. In: Borio, C., Hunter, W.C., Kaufman, G., Tsatsaronis, K. (Ed.): Market Discipline Across Countries and Industries, MIT Press, 55-68. Gropp, R., Vesala, J., Vulpes, G., 2006. Equity and bond market signals as leading indicators of bank fragility. Journal of Money, Credit and Banking, 38, 399-428. Gup, B.E. (Ed.), 2003. Too Big to Fail. Praeger. Haefeli, M., Jüttner, M.P., 2012. The Value of the Liability Insurance for Credit Suisse and UBS. Journal of Institutional and Theoretical Economics, 168, 612-635. Haldane, A., 2010. The $100 billion question. Speech at the Institute of Regulation and Risk, Hong Kong, March. Haldane, A.G., 2012. Control rights (and wrongs). Economic Affairs, 32, 47-58. Hannan, T.H., and Hanweck, G.A., 1988. Bank insolvency risk and the market for large certificates of deposit. Journal of Money, Credit and Banking, 20, 203-211. Hasan, I., Jackowicz, K., Kowalewski, O., Kozłowski, Ł., 2013. Market discipline during crisis: Evidence from bank depositors in transition countries. Journal of Banking & Finance, 37, 54365451. Herzig-Marx, C., Weaver, A.S., 1979. Bank soundness and the market for large negotiable certificates of deposit. Federal Reserve Bank of Chicago. Hovakimian, A., Kane, E.J., 2000. Effectiveness of Capital Regulation at US Commercial Banks, 1985 to 1994. Journal of Finance, 55, 451-468. Hovakimian, A., Kane, E.J., Laeven, L., 2003. How country and safety-net characteristics affect bank risk-shifting. Journal of Financial Services Research, 23, 177-204. Hickson, C.R., Turner, J.D., 2004. Free Banking and the Stability of Early Joint-stock Banking. Cambridge Journal of Economics, 28, 903-919. Houston, J.F., James, C., 1995. CEO compensation and bank risk Is compensation in banking structured to promote risk taking? Journal of Monetary Economics, 36, 405-431. Hull, J.C., 2012. Risk Management and Financial Institutions. John Wiley and Sons. Imai, M., 2006. Market discipline and deposit insurance reform in Japan. Journal of Banking & Finance, 30, 3433-3452. Imai, M., 2007. The emergence of market monitoring in Japanese banks: Evidence from the subordinated debt market. Journal of Banking & Finance, 31, 1441-1460. Independent Banking Commission on Banking, 2011. Final Report: Recommendations. The Stationery Office,. Ioannidou, V.P., Penas, M.F., 2010. Deposit insurance and bank risk-taking: Evidence from internal loan ratings. Journal of Financial Intermediation, 19, 95-115.Jacewitz, S., Pogach, J., 2014. Deposit Rate Advantages at the Largest Banks. Federal Reserve Board Staff Working Papers 201402.

144

Bibliography

Jagtiani, J., Lemieux, C., 2000. Stumbling Blocks to Increasing Market Discipline in the Banking sector: A Note on Bond Pricing and Funding Strategy Prior to Failure. Emerging Issues Series, Supervision and Regulation Department, Federal Reserve Bank of Chicago, April. Jobst, M.A.A., Gray, M.D.F., 2013. Systemic Contingent Claims Analysis – Estimating MarketImplied Systemic Risk. International Monetary Fund Working Paper 1354.Johanning, L., 2009. Strategisches Risikomanagement. In: Schäfer, Burghof, Johanning, Wagner, Rodt (Ed.): Risikomanagement und kapitalmarktorientierte Finanzierung: Festschrift zum 65. Geburtstag von Bernd Rudolph, 459-471. Johanning, L., Rudolph, B., 2000. Entwicklungslinien im Risikomanagement. In: Johanning, L./Rudolph, B. (Ed.): Handbuch Risikomanagement, Uhlenbruch, 15-52. Jordà, Ò., Schularick, M.H., Taylor, A.M., 2016. Sovereigns versus Banks: Credit, Crises, and Consequences. Journal of the European Economic Association, 14, 45-79. Jordan, J.S., 2000. Depositor discipline at failing banks. New England Economic Review, March/April, 15-28.Kabir, M.H., Hassan, M.K., 2005. The near-collapse of LTCM, US financial stock returns, and the fed. Journal of Banking & Finance, 29, 441-460. Kane, E.J., 2000. Incentives for banking megamergers: what motives might regulators infer from event-study evidence? Journal of Money, Credit and Banking, 32, 671-701. Kaplan‐Appio, I., 2002. Estimating the value of implicit government guarantees to Thai banks. Review of International Economics, 10, 26-35. Kaufman, G.G., 2014. Too big to fail in banking: What does it mean? Journal of Financial Stability, 13, 214-223. Kay, J., 2010. Should We Have Narrow Banking? In: Turner, A., Haldane, A., Woolley, P., Wadhwani, S., Goodhart, C., Smithers, A., Large, A., Kay, J., Wolf, M., Boone, P., Johnson, S., Layard, R. (Ed.): The Future of Finance – The LSE Report, 238-273. King, K.K., O'Brien, J.M., 1991. Market-based, Risk-adjusted Examination Schedules for Depository Institutions. Journal of Banking & Finance, 15, 955-974. King, T.B., 2008. Discipline and Liquidity in the Interbank Market. Journal of Money, Credit and Banking, 40, 295-317. KMV Corporation, 1999. Modeling Default Risk. Moody’s KMV. Körner, T., Schnabel, I., 2011. Public Ownership of Banks and Economic growth. Economics of Transition, 19, 407-441. Kotlikoff, L.J., 2010. Jimmy Stewart is Dead: Ending the World's Ongoing Financial Plague with Limited Purpose Banking. John Wiley and Sons. Krishnan, C.N.V., Ritchken, P.H., Thomson, J.B., 2005. Monitoring and controlling bank risk: Does risky debt help? Journal of Finance, 60, 343-378. Kroszner, R.S. A Review of Bank Funding Cost Differentials. University of Chicago Booth School of Business Working Paper. Krueger, A.B., 1988. Moral Hazard in Workers’ Compensation Insurance. Princeton University Working Paper. Kumar, A., Lester, J., 2014a. Do Bond Spreads Show Evidence of Too Big To Fail Effects? Evidence from 2009–2013 Among US Bank Holding Companies. Oliver Wyman. Kumar, A., Lester, J., 2014b. Do Deposit Rates Show Evidence of Too Big to Fail Effects? Evidence from 2009–2013 Among US Bank Holding Companies. Oliver Wyman. Kupiec, P.H., Ramirez, C.D., 2013. Bank Failures and the Cost of Systemic Risk: Evidence from 1900 to 1930. Journal of Financial Intermediation, 22, 285-307. Kwast, M.L., Covitz, D.M., Hancock, D., Houpt, J.V., Adkins, D.P., Barger, N., Bouchard, B., Connolly, J.F, Brady, T.F., English, W.B., Evanoff, D.D., Evanoff, Wall, L.D., 1999. Using subordinated debt as an instrument of market discipline. Federal Reserve System Staff Study 172. Kwast, M.L., Passmore, S.W., 1999. The subsidy provided by the federal safety net: Theory and evidence. Journal of Financial Services Research, 16, 125-145.

Bibliography

145

La Porta, R., Lope de Silanes, F., Shleifer, A., 2002. Government Ownership of Banks. Journal of Finance, 57, 265-301. Lambert, F.J., Ueda, K., Deb, P., Gray, D. F., Grippa, P., 2014. How Big Is the Implicit Subsidy for Banks Considered Too Important to Fail? Global Financial Stability Report, April. Lehar, A., 2005. Measuring systemic risk: A risk management approach. Journal of Banking & Finance, 29, 2577-2603. Levonian, M.E., 1991. Have large Banks Become Riskier? Recent Evidence from Option Markets. Federal Reserve Bank of San Francisco Economic Review, Fall, 3-17. Li, Z., Qu, S., Zhang, J., 2011. Quantifying the value of implicit government guarantees for large financial institutions. Moody’s Analytics. Litan, R.E., 1987. What Should Banks Do? Brookings Institution Press. Lucas, D., McDonald, R.L., 2006. An options-based approach to evaluating the risk of Fannie Mae and Freddie Mac. Journal of Monetary Economics, 53, 155-176. Marcus, A.J., and Shaked, I., 1984. The Valuation of FDIC Deposit Insurance Using Option-pricing Estimates. Journal of Money, Credit and Banking, 16, 446-460. Marino, J.A., and Bennett, R.L., 1999. The Consequences of National Depositor Preference. FDIC Banking Review, 12, 19-38. Martinez Peria, M.S., Schmukler, S.L., 2001. Do depositors punish banks for bad behavior? Market discipline, deposit insurance, and banking crises. Journal of Finance, 56, 1029-1051. McKinsey & Company, 2012. The Triple Transformation: Achieving a sustainable Business Model, October. Merton, R.C., 1974. On the Pricing of Corporate Debt: The Risk Structure of Interest Rates. Journal of Finance, 29, 449-470. Merton, R.C., 1977. An analytic derivation of the cost of deposit insurance and loan guarantees an application of modern option pricing theory. Journal of Banking & Finance, 1, 3-11. Merton, R.C., 1978. On the Cost of Deposit Insurance When There are Surveillance Costs. Journal of Business, 51, 439-452. Merton, R.C., Bodie, Z., 1995. A Conceptual Framework for Analyzing the Financial Environment, in Crane, D.B., Froot, K.A., Mason, S.P., Perold, A.F., Merton, R.C., Bodie, Z., Sirri, E.R., Tufano, P. (Ed.) The Global Financial System: A Functional Perspective, Harvard Business School Press, 3-31. Merton, R.C., Bodie, Z., 1993. Deposit Insurance Reform: A Functional Approach. CarnegieRochester Conference Series on Public Policy, 38, 1-34. Merton, R.C., Billio, M., Getmansky, M., Gray, D., Lo, A., Pelizzon, L., 2013. On a New Approach for Analyzing and Managing Macrofinancial Risks. Financial Analysts Journal, 69, 22-33. Merton, R.K., 1936. The Unanticipated Consequences of Purposive Social Action. American Sociological Review, 1, 894-904. Mishan, Edward J., 1969. Welfare Economics: Ten Introductory Essays. Random House. Molyneux, P., Schaeck, K., Zhou, T.M., 2014. ‘Too systemically important to fail’ in banking – Evidence from bank mergers and acquisitions. Journal of International Money and Finance, 49, 258-282. Moen, H., 2004. The present value of central government investments in and support to Norwegian banks. In: Moe, T.G., Solheim, J.A., Vale, B. (Ed.): The Norwegian Banking Crisis, Norges Banks Skriftsserie, 33, 225-249. Moenninghoff, S.C., Wieandt, A., 2011. Too big to fail?! Leçons de la crise financière. Revue d'économie financière, 101, 231-260. Moenninghoff, S.C., Wieandt, A., 2013. The Future of Peer-to-Peer Finance. Zeitschrift für betriebswirtschaftliche Forschung, 65, 466-487. Moenninghoff, S.C., Wieandt, A., 2017a. Consequences of Government Guarantees for Banks – A Survey of the TBTF Doctrine, unpublished working paper.

146

Bibliography

Moenninghoff, S.C., Wieandt, A., 2017b. Government Guarantees and Banking System Risk – A Regulatory Framework from an Exposure Perspective, unpublished working paper. Moenninghoff, S.C., Ongena, S., Wieandt, A., 2015. The Perennial Challenge to Abolish Too-BigTo-Fail in Banking: Empirical Evidence from the New International Regulation Dealing with Global Systemically Important Banks. Journal of Banking & Finance, 61, 221-236. Mondschean, T.S., Opiela, T.P., 1999. Bank time deposit rates and market discipline in Poland: the impact of state ownership and deposit insurance reform. Journal of Financial Services Research, 15, 179-196. Moody’s Analytics, 2011. Overcoming challenges in PD/LGD Modeling in the Absence of High Quality and Rich Data. February. Moore, R.E., 1997. Government Guarantees and Banking: Evidence from the Mexican Peso Crisis. Federal Reserve Bank of Dallas Financial Industry Studies, 13-21. Morgan, D.P., Stiroh, K.J., 2001. Market discipline of banks: The asset test. Journal of Financial Services Research, 20, 195-208. Morgan, D.P., Stiroh, K.J., 2005. Too big to fail after all these years. Federal Reserve Bank of New York Staff Report 220. Moseley, F., 2011. The US Economic Crisis. International Journal of Political Economy, 40, 59-71. Moyer, R.C., Lamy, R.E., 1992. ‘Too Big to Fail’: Rationale, Consequences, and Alternatives. Business Economics, 27, 19-24. Nickell, P., Perraudin, W., 2001. How Much Bank Capital is Needed to Maintain Financial Stability? Bank of England Working Paper. Nier, E., Baumann, U., 2006. Market discipline, disclosure and moral hazard in banking. Journal of Financial Intermediation, 15, 332-361. Noss, J., Sowerbutts, R., 2012. The implicit subsidy of banks. Bank of England Financial Stability Paper 15. O'Hara, M., Shaw, W., 1990. Deposit insurance and wealth effects: The value of being "Too Big to Fail”. Journal of Finance, 45, 1587-1600. Ongena, S., Penas, M.F., 2009. Bondholders' wealth effects in domestic and cross-border bank mergers. Journal of Financial Stability, 5, 256-271. Oxera, 2011. Assessing state support to the UK banking sector, March. Panetta, F., Faeh, T., Grande, G., Ho, C., King, M., Levy, A., Signoretti, F.M., Taboga, M., Zaghini, A., 2009. An assessment of financial sector rescue programmes. Banca d’Italia Occasional Papers, July. Park, S., Peristiani, S., 1998. Market Discipline by Thrift Depositors. Journal of Money, Credit and Banking, 347-364. Passmore, W., Sherlund, S.M., Burgess, G., 2005. The Effect of Housing Government‐Sponsored Enterprises on Mortgage Rates. Real Estate Economics, 33, 427-463. Penas, M.F., Unal, H., 2004. Gains in bank mergers: Evidence from the bond markets. Journal of Financial Economics 74, 149-179. Pennacchi, G., 2012. Narrow banking. Annual Review of Financial Economics, 4, 141-159. Pennacchi, G.G., 1987a. Alternative Forms of Deposit Insurance: Pricing and Bank Incentive Issues. Journal of Banking & Finance, 11, 291-312. Pennacchi, G.G., 1987b. A Reexamination of the Over-(or Under-) Pricing of Deposit Insurance. Journal of Money, Credit and Banking, 19, 340-360. Perotti, E., Suarez, J., 2011. A Pigovian Approach to Liquidity Regulation. International Journal of Central Banking, 7, 3-41. Pettway, R.H., 1976. Market tests of capital adequacy of large commercial banks. Journal of Finance, 31, 865-875. Pettway, R.H., 1980. Potential insolvency, market efficiency, and bank regulation of large commercial banks. Journal of Financial and Quantitative Analysis, 15, 219-236. Phillips, R.J., 1995. The Chicago Plan and The New Deal Banking Reform. M.E. Sharpe, 1994.

Bibliography

147

Pierce, J.L., 1991. The Future of Banking. Yale University Press. Pigou, A., 1954. Some Aspects of the Welfare State. Diogenes, 7, 1-11. Pigou, A.C., 1924. The Economics of Welfare. Transaction Publishers. Pollock, A.J., 1992. Collateralized Money: An Idea Whose Time Has Come Again. Challenge, 35, 62-64. Posner, E., 2013. Benefit-Cost Analysis for Financial Regulation. American Economic Review, 103, 393-397. Posner, E., Weyl, E.G., 2014a. Benefit-Cost Paradigms in Financial Regulation. Journal of Legal Studies, 43, 1-34. Posner, E.A., Weyl, E.G., 2014b. Cost-Benefit Analysis of Financial Regulations: A Response to Criticisms. University of Chicago Coase-Sandor Institute for Law and Economics Research Paper 683. Pyle, D., 1984. Deregulation and Deposit Insurance Reform. Federal Reserve Board of San Francisco Economic Review, Spring, 5-15. Pyle, D.H., 1986. Capital Regulation and Deposit Insurance. Journal of Banking & Finance, 10, 189201. Rehm, F., Rudolf, M., 2001. KMV Credit Risk Modeling. In: Frenkel, M., Hommel, U., Rudolf, M. (Ed.): Risk Management - Challenge and Opportunity, Springer, 141-154. Reinhart, C.M., Rogoff, K., 2009. This Time is Different: Eight Centuries of Financial Folly. Princeton University Press. Reinhart, C.M., Rogoff, K., 2010. Growth in a Time of Debt. National Bureau of Economic Research Paper 15639. Reinhart, C.M., Rogoff, K., 2013. Financial and Sovereign Debt Crises: Some Lessons Learned and Those Forgotten. International Monetary Fund Working Paper 13266. Reinhart, C.M., 2014. The Global Financial Crisis of 2008-2009 and the Developing World: A Historical Perspective. Annual Bank Conference on Development Economics. Ricks, M., 2011. A Regulatory Design for Monetary Stability. Vanderbuilt Law Review, 65, 1287. Rieger, M.O., Wang, M., Hens, T., 2014. Risk Preferences Around the World, Management Science, 61, 637-648. Rime, B., 2005. Do Too-Big-To-Fail Expectations Boost Large Banks Issuer Ratings. Swiss National Bank Working Paper. Rochet, J.C., 2004. Market discipline in banking: Where do we stand. In: Borio, C., Hunter, W.C., Kaufman, G., Tsatsaronis, K. (Ed.): Market Discipline Across Countries and Industries, MIT Press, 55-68. Roland Berger Strategy Consultants, 2012. Stress Testing Spanish Banks – Final Report. June 21. Ronn, E.I., Verma, A.K., 1986. Pricing Risk‐Adjusted Deposit Insurance: An Option‐Based Model. Journal of Finance, 41, 871-896. Ronn, E.I., Verma, A.K., 1989. Risk-based Capital Adequacy Standards for a Sample of 43 Major Banks. Journal of Banking & Finance, 13, 21-29. Rowell, D., Connelly, L.B., 2012. A history of the term “moral hazard”. Journal of Risk and Insurance, 79, 1051-1075. Robb, T.B., 1921. The guaranty of bank deposits. Houghton Mifflin Company, Boston and New York. Rudolf, M., 2010. Eine Historie von Finanzmarktkrisen.
In: Joecks, W., Ostendorf, H., Rönnau, T. (Ed.) Festschrift für Erich Samson, Hüthig Jehle Rehm, 819-834. Rudolf, M., 2012. Alternative Konzepte zur Rettung angeschlagener Banken. In: Bolder, M., Wargers, M. (Ed.): Modell‚ Bad Bank: Hintergrund – Konzept - Erfahrungen, Gabler, 33-59. Santos, J.A., 2014. Evidence from the Bond Market on Banks''Too-Big-To-Fail' Subsidy. Federal Reserve Board of New York Economic Policy Review, December, 29-39. Sapienza, P., 2004. The Effects of Government Ownership on Bank Lending. Journal of Financial Economics, 72, 357-384.

148

Bibliography

Sato, R., Ramachandran, R.V., Kang, B., 1990. Risk Adjusted Deposit Insurance for Japanese Banks. National Bureau of Economic Research Working Paper 3314. Schäfer, A., Schnabel I., Weder di Mauro B., 2016. Financial sector reform after the subprime crisis: Has anything happened? Review of Finance, 20, 77-125. Schanz, J., 2012. A Joint Calibration of Bank Capital and Liquidity Ratios. Discussion Paper, Bank of England. Schich, S., Kim, B.H., 2012. Developments in the value of implicit guarantees for bank debt: The role of resolution regimes and practices. OECD Journal: Financial Market Trends, 2, 35-65. Schmid, M.M., Walter, I., 2009. Do Financial Conglomerates Create or Destroy Economic Value? Journal of Financial Intermediation, 18, 193-216. Schnabel, I., Körner, T., 2012. Abolishing public guarantees in the absence of market discipline. Ruhr Economic Papers 437. Schuler, K., 1992. The World History of Free Banking: an Overview. In: Dowd (Ed.), The Experience of Free Banking, 4-47. Schweikhard, F., Tsesmelidakis, Z., 2011. The impact of government interventions on CDS and equity markets. AFA 2012 Chicago Meetings Paper 2011. Seelig, S.A., 2004. Too Big to Fail: A Taxononomic Analysis. In: Gup, B.E. (Ed.), Too Big to Fail, Praeger, 219-230. Shavell, S., 1979. On Moral Hazard and Insurance. Quarterly Journal of Economics, 93, 541-562. Sinn, H.-W., 1999. The German State Banks: Global Players in the International Financial Markets, Edward Elgar. Sironi, A., 2003. Testing for market discipline in the European banking industry: Evidence from subordinated debt issues. Journal of Money, Credit and Banking, 35, 443-472. Soussa, F., 2000. Too big to fail: moral hazard and unfair competition. Financial Stability and Central Banks, In: Centre for Central Banking Studies (Ed.): Selected Issues for Financial Safety Nets and Market Discipline, 5-31. Spulber, D.F., 1989. Regulation and markets. MIT Press. Stern, G.H., Feldman, R.J., 2004. Too big to fail: The hazards of bank bailouts. Brookings Institution Press.Stiglitz, J.E., 1993. The Role of the State in Financial Markets. The World Bank Economic Review, 7, 19-52. Stiroh, K.J., 2006. New evidence on the determinants of bank risk. Journal of Financial Services Research, 30, 237-263. Strahan, P.E., 2013. Too Big to Fail: Causes, Consequences, and Policy Responses. Annual Revue of Financial Economics, 5, 43-61. Strahan, P.E., Tanyeri, B., 2015. Once burned, twice shy: Money market fund responses to a systemic liquidity shock. Journal of Financial and Quantitative Analysis, 50, 119-144. Strongin, S., Hindlian, A., Lawson, S., Murillo, J., Sadan, K., Balakrishna, S., 2013. Measuring the TBTF Effects on Bond Pricing. Goldman Sach Global Markets Institute. Summers, L., 2007. Beware moral hazard fundamentalists. Financial Times, September 23. Tanzi, V., 2011. Government versus Markets: The Changing Economic Role of the State, Cambridge University Press. Thakor, A.V., 2014. Bank Capital and Financial Stability: An Economic Tradeoff or a Faustian Bargain? Annual Review of Financial Economics, 6, 185-223. Thomson, J.B., 1987. The Use of Market Information in Pricing Deposit Insurance. Journal of Money, Credit and Banking, 19, 528-537. Thomson, J.B., 2009. On Systemically Important Financial Institutions and Progressive Systemic Mitigation. Federal Reserve Bank of Cleveland Policy Discussion Paper 7. Tobin, J., 1987. A Case for Preserving Regulatory Distinctions. Challenge, November/December, 1017. Tsesmelidakis, Z., Merton, R.C., 2013. The value of implicit guarantees. Massachusetts Institute of Technology Working Paper.

Bibliography

149

Watterson, J.P., Lowey, E.G., Houghton, J.R., 2010. First (Interim) Report of the Select Committee on Kaupthing, Singer and Friedlander (Isle of Man) Limited, June. Ugeux, G., 2014. International Finance Regulation: The Quest for Financial Stability. John Wiley and Sons. Unger, M., 2012. Berufsbild “De-Investitionsbanker” – Personalpolitische Erfahrungen beim Aufbau der Ersten Abwicklungsanstalt,
in: Bolder, M., Wargers, M. (Ed.): Modell‚ Bad Bank: Hintergrund – Konzept - Erfahrungen, Gabler, 267-277. Ueda, K., Weder di Mauro, B., 2013. Quantifying structural subsidy values for systemically important financial institutions. Journal of Banking & Finance, 37, 3830-3842. Vasicek, O., 1987. Probability of Loss on Loan Portfolio. KMV Corporation. Vasicek, O., 2002. Loan Portfolio Value. RISK, December 2002, 160-162. Völz, M., Wedow, M., 2009. Does banks size distort market prices? Evidence for too-big-to-fail in the CDS market, Deutsche Bundesbank Discussion Paper Series 2: Banking and Financial Studies. Weitzman, M.L., 1974. Prices vs. Quantities. Review of Economic Studies, 41, 477-491. Wu, Y., Bowe, M., 2012. Information disclosure and depositor discipline in the Chinese banking sector. Journal of International Financial Markets, Institutions and Money, 22, 855-878. Wieandt, A., 2011. Moenninghoff, S.C., The Financial Crisis: Observations and Implications. Zeitschrift für betriebswirtschaftliche Forschung, 63, 508-530. Wieandt, A., 2014. Nach 5 Jahren Finanzkrise: Hat der Wettbewerb als Ordnungsprinzip auf den Finanzmärkten ausgedient? In: Körber, T., (Ed.): Wettbewerbsbeschränkungen auf staatlich gelenkten Märkten, Referate der 4. Göttinger Kartellrechtsgespräche vom 13. Juni 2014 anlässlich des 80. Geburtstags von Prof. Dr. Dr. h.c. Ulrich Immenga, Nomos, 107-112.

List of Appendices

Appendix A.1 Appendix A.2 Appendix A.3 Appendix A.4 Appendix A.5 Appendix A.6 Appendix A.7 Appendix A.8 Appendix A.9 Appendix A.10 Appendix A.11 Appendix A.12 Appendix A.13 Appendix A.14 Appendix A.15 Appendix A.16 Appendix A.17 Appendix A.18 Appendix A.19 Appendix A.20 Appendix A.21 Appendix A.22

Stylized logic of the TBTF doctrine ....................................................... 153 Stylized logic of the charter value theory............................................... 153 Introduction to risk-return model for regulatory policy choices ............ 154 Input variables and exposure factors for regulatory policy choices ....... 155 Internalities, externalities and volatility implications ............................ 156 Banking system balance sheet ................................................................ 157 Sovereign balance sheet at t = 0 ............................................................. 158 Sovereign balance sheet at t = 1 ............................................................. 159 Global sovereign surplus risk-adjusted returns ...................................... 160 Overview of event study sub-samples .................................................... 161 Abnormal returns for event 1-3 assuming 40 largest banks are G-SIBs ... 161 Abnormal returns for event 1-3 assuming 20 largest banks are G-SIBs. .. 161 Overview of regulatory process for G-SIBs ........................................... 162 National market indices used for market model ..................................... 163 Overview of abnormal returns with methodology variation .................. 164 Individual G-SIB / Non-G-SIB abnormal returns at designation events ..................................................................................................... 165 G-SIB / Non-G-SIB returns differentiating between positive and negative first designation return ............................................................. 166 Event 12 abnormal returns for all G-SIBs (incl. repeat designations).... 166 Abnormal returns of newly declared G-SIBs at 3rd designation event .. 167 Abnormal returns by event for U.S. and European banks ...................... 168 Economic interest held by governments ................................................ 169 Sovereign risk ratings for G-SIB home countries .................................. 170

© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2018 S. C. Moenninghoff, The Regulation of Systemically Relevant Banks, Finanzwirtschaft, Banken und Bankmanagement  Finance, Banks and Bank Management, https://doi.org/10.1007/978-3-658-23811-7

Appendix

Appendix A.1:

Stylized logic of the TBTF doctrine

Appendix A.2:

Stylized logic of the charter value theory

© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2018 S. C. Moenninghoff, The Regulation of Systemically Relevant Banks, Finanzwirtschaft, Banken und Bankmanagement  Finance, Banks and Bank Management, https://doi.org/10.1007/978-3-658-23811-7

154 Appendix A.3:

Appendix

Introduction to risk-return model for regulatory policy choices

In order to illustrate interaction mechanisms and general tradeoffs inherent in individual policy choices, we develop a simple simulation of the range of regulatory policy choices introduced in Section 3.3. All figures are exemplary only and do not imply a specific policy preference. Instead, in the standard setting all figures are chosen to yield the same expected return on the global sovereign surplus for a risk aversion factor of ρ = 2.0. While changes in the assumptions alter the resulting risk-weighted returns for the individual policy choices, the general tradeoffs remain largely unchanged as our sensitivity analysis shows. We model the balance sheet of a government over a one-year period and the implications different regulatory options bear for economic growth and stability. We also model a banking system as part of the economy with assumptions regarding the capitalization level, government ownership, inefficiencies introduced via the government’s involvement in the intermediation process as well as benefits of the financial intermediation on overall economic growth. Based on these assumptions, government exposure to the financial system is calculated, and internalities, externalities and volatilities are quantified. On the basis of the adjusted year-end balance sheet of the government, wealth changes are calculated and adjusted for risk. Based on these risk-adjusted returns, major tradeoffs inherent in the individual policy options regarding one or several input factors are derived. Appendix A.4 introduces the chosen input variables and resulting exposure factors. Externalities, internalities and volatilities relating to the input variables are shown in Appendix A.5. The bank and government balance sheets are shown in Appendix A.6 to A.8. The resulting risk-weighted returns are displayed in Appendix A.9.

155

Appendix

Appendix A.4:

Input variables and exposure factors for regulatory policy choices

156 Appendix A.5:

Appendix

Internalities, externalities and volatility implications

157

Appendix

Appendix A.6:

Banking system balance sheet

158 Appendix A.7:

Appendix

Sovereign balance sheet at t = 0

159

Appendix

Appendix A.8:

Sovereign balance sheet at t = 1

160 Appendix A.9:

Appendix

Global sovereign surplus risk-adjusted returns

Appendix

Appendix A.10: Overview event study sub-samples

Appendix A.11: Abnormal returns for event 1-3 assuming 40 largest banks are G-SIBs

Appendix A.12: Abnormal returns for event 1-3 assuming 20 largest banks are G-SIBs

161

162 Appendix A.13: Overview of regulatory process for G-SIBs

Appendix

Appendix

Appendix A.14: National market indices used for market model

163

164 Appendix A.15: Overview of abnormal returns with methodology variation

Appendix

Appendix

Appendix A.16: Individual G-SIB / Non-G-SIB abnormal returns at designation events

165

166

Appendix

Appendix A.17: G-SIB / Non-G-SIB returns differentiating between positive and negative first designation return

Appendix A.18: Event 12 abnormal returns for all G-SIBs (incl. repeat designations)

Appendix

Appendix A.19: Abnormal Returns of newly declared G-SIBs at 3rd designation event

167

168 Appendix A.20: Abnormal returns by event for U.S. and European banks

Appendix

Appendix

Appendix A.21: Economic interest held by governments

169

170 Appendix A.22: Sovereign risk ratings for G-SIB home countries

Appendix