Reverse Stress Testing in Banking

A Comprehensive Guide

Michael Eichhorn, Tiziano Bellini, Daniel Mayenberger, Michael Eichhorn, Tiziano Bellini, Daniel Mayenberger

• Editorial: De Gruyter
• Año: 2021
• ISBN:
  9783110644951
• Edición: 1
• Categoría: Economics
• Categoría: Banks & Banking

Reverse stress testing was introduced in risk management as a regulatory tool for financial institutions more than a decade ago. The recent Covid-19 crisis illustrates its relevance and highlights the need for a systematic re-thinking of tail risks in the banking sector. This book addresses the need for practical guidance describing the entire reverse stress testing process.

Reverse Stress Testing in Banking features contributions from a diverse range of established practitioners and academics. Organized in six parts, the book presents a series of contributions providing an in-depth understanding of:

• 
  
  

• Regulatory requirements and ways to address them

• 
  
  

  
  
  

• Quantitative and qualitative approaches to apply reverse stress testing at different levels – from investment portfolios and individual banks to the entire banking system

• 
  
  

  
  
  

• The use of artificial intelligence, machine learning and quantum computing to gain insights into and address banks' structural weaknesses

• 
  
  

  
  
  

• Opportunities to co-integrate reverse stress testing with recovery and resolution planning

• 
  
  

  
  
  

• Governance and processes for board members and C-suite executives

• 
  
  

Readers will benefit from the case studies, use cases from practitioners, discussion questions, recommendations and innovative practices provided in this insightful and pioneering book.

Part I: Fundamentals of Reverse Stress Testing

1 Reverse Stress Testing: A Versatile Thinking Tool

Definitions, Objectives, Challenges, Configurations and Organisation of the Book

Michael Eichhorn

Tiziano Bellini

Daniel Mayenberger

1.1 Motivation

To the best of our knowledge, this book is the first comprehensive collection of contributions dedicated to reverse stress testing (RST).

While we were finishing the book, the following transpired within a few weeks in June 2020:

• The Institute of Chartered Accountants in England and Wales (ICAEW) published an article with the title “Why all businesses should reverse stress test”. Together with a webcast from the ICAEW Audit and Assurance Faculty, it encourages entities of all sizes and industries – from a small coffee shop to a large bank – to consider using RST to enhance their going concern assessments and related disclosures. In essence, each business should answer three questions: What would it take for the entity to fail? What event or combination of events might lead to this outcome? What can we do now to prevent this from happening?(https://www.icaew.com/insights/viewpoints-on-the-news/2020/june-2020/why-all-­businesses-should-reverse-stress-test).
• On the back of the March Covid-19 market-wide stress, a Core College of regulators asked G-SIB banks to stress their revised 2020 funding plan using a combined scenario. The hypothetical scenario should start with a prolonged market-wide stress (i.e. slow L-shaped recovery from the Covid-19 crisis). It should then add a severe idiosyncratic stress with uncertainty about the survival of the bank. The combined outflows from the hypothetical scenario should push the bank significantly beyond the impacts seen in March 2020 and subsequent stress tests towards the highest severity levels of its contingency funding plan. The bank should land in a territory where it is not only in breach of its regulatory liquidity metrics requirements but is possibly in recovery and resolution. On this basis the bank should assess the remaining management actions, their feasibility, timing and execution risk.
• In a post-mortem of the Covid-19 crisis, a G-SIB bank reportedly conducted individual interviews with members of its Executive Board. The objective of the interviews was to understand how a function of the bank performed prior to, during and after the peak of the Covid-19 crisis in March 2020. One board member suggested to run more table-top simulations to confront the board with concrete stress situations and discuss tangible trade-offs. This, he said, would be more helpful than abstract discussions around risk appetite frameworks and standard updates.

In all of these discussions, RST was repeatedly referenced. This book will help readers to further develop their own thinking and RST frameworks.

We are of the view that RST should be first and foremost considered a versatile thinking tool – an understanding that holds all subsequent individual contributions of this book together. In our opinion, it represents the cornerstone for the next generation of risk management. Indeed, a big step has been accomplished by introducing regulatory stress tests. Nevertheless, RST should provide a more comprehensive view of potential unexpected sources of risks by focusing on each bank’s peculiarities. RST should be much more creative and extensively used by setting strategic and other targets. Using a spectrum of different configurations and test designs, the book is dedicated to demonstrating how RST can support deep and deliberate thinking. In fact, the main limitation of RST, in our opinion, is often not the method itself (which will also differ from case to case and be more or less difficult to overcome) but our own thinking.

The book is organised in six parts. Part I introduces RST as a versatile thinking tool with various configurations and test designs. Parts II, III and IV contain quantitative and qualitative case studies as well as applications based on artificial intelligence (AI), machine learning and quantum computing. Part V demonstrates the link between RST and recovery and resolution planning (RRP) by outlining how to best organise the infrastructure and construct the stress testing narrative and mitigating management actions. The book closes with Part VI, which provides an outlook on how the aforementioned objectives, challenges and choices will be covered in the remainder of the book.

1.2 Definition

Reverse stress testing (RST) is a concept that calls for a definition. It is useful to start by breaking down the acronym RST into its individual terms:

• The term stress comes originally from materials research where it denotes force over area. If the force is small, we are in the elastic regime: the medium may react ­linearly to external forces and tend to vibrate and revert to its original ­configuration. If the force is excessive, the medium may react non-linearly: firstly, it ­experiences plastic, irreversible deformations and then it fractures. A similar description also applies to finance: a small level of volatility and diffusion ­processes for risk factors can be hedged with no friction, but when shocks hit a critical threshold, friction develops and ultimately trading strategies undergo an abrupt blow up.
• The term reverse relates to the inversion of the cause-effect into an effect-cause relationship in the test design. In finance, traditional stress tests start with defining a (idiosyncratic and/or macro-financial) scenario and then assess its impact on their business, typically in terms of earnings, capital and liquidity. RST starts with defining the outcomes followed by reverse-engineering scenarios that, should they unfold, lead to the specified outcome. In this aspect, RST goes one step further than scenario-based stress testing, which examines synthetic future scenarios that have not already been experienced in the past and determines their financial impact. So scenario-based stress testing is forward-looking, but it first defines the cause (scenario) and then determines the effect (loss), while RST reverses this relationship.
• The term testing particularly relates to the scenarios being applied to a specific portfolio. In contrast, we know by historical experience what circumstances typically give rise to large trading book losses at financial institutions independently of their specific portfolio positions. RST instead is more specifically aimed at discovering stress scenarios that, when tested against the specific position at hand, give rise to a transient of severe friction and systematic losses or fracture. In RST, scenarios are often designed to be so severe that they “break the bank”, i.e. they render the business model unviable or cause the bank to fail.

The idea to reverse a cause-effect to an effect-cause relationship is not new. Comparable resilience tests are long established in material research. They are also common practice elsewhere, e.g. in IT safety management and the military. What is new is the regulatory requirement to apply RST to analyse the failure of a bank or the unviability of its business model.

1.3 Objectives

As a complementary method to sensitivity analysis and risk measures, RST can support a variety of different objectives:

• Capture high-severity events not covered by sensitivity analysis and scenario analysis.
• Back-test the existing scenario inventory for completeness, relevance and severity.
• Complement banks’ ICAAPs scenarios which were criticised for not being severe enough.
• Strengthen cross-linkage with other frameworks (ICAAP, ILAAP, recovery and resolution, risk appetite, strategy and financial planning).
• Mitigate cognitive biases such as availability bias, anchoring (to historical stress scenarios) or salience bias of traditional stress tests.
• Improve processes and methods used for regulatory stress testing.
• Support an integrated and holistic stress testing across risk types where traditional stress tests may fail to capture complex aspects from real situations.
• Identify hidden vulnerabilities that may otherwise remain undetected and are not considered during traditional stress tests.
• Identify specific vulnerabilities for individual business lines, geographies and institution size and complexity.
• Facilitate a better understanding of the dynamic interplay of risks over time and their impact on capital and liquidity – for example, between operational, conduct and reputational risk or macroeconomic, credit and structural interest rate risks.
• Understand the dynamic interplay of risks, capital and liquidity over time, including feedback loops and non-linear effects, and allow gathering insights into scenarios that involve combinations of these factors.
• Allow for a more narrative-orientated design to better capture the (inter)dependencies between risk types, financial resources and markets.
• Improve preparedness by enabling a constructive debate on the treatment of risks – decide whether they should they be accepted or managed through a combination of business activity reduction and additional controls. The broader the choice of scenarios and related mitigating actions, the better the response and the lower the actual losses in real stress occurrences.
• Provide an additional metric in risk-adjusted return quantification, comparing revenues of a business against potential losses.
• Overcome the limitation of risk models such as VaR, which break down in periods of crisis. Potentially help managers hedge against hidden scenarios.

Practitioners agree with most of these objectives. For example, a survey by Ernst & Young (2013) quotes an executive as follows:

“What does it take to break [a bank],” one executive said, “is a more interesting thing than just taking a random scenario and seeing what happens.”

However, at the same time, there are challenges to the implementation of RST.

1.4 Challenges

At least in parts typically stated, challenges directly relate to the aforementioned objectives:

• The search for scenarios that cause the business model to become unviable may mean that RST is by design constrained to rather remote (high severity, low probability) scenarios. At the same time, it would be difficult to set the stress scenarios with a plausible magnitude of the stress on the risk parameters, to determine the maximum level of stress the organisation can withstand.
• The infi...