An Introduction to Stress Testing
By Mike Tan, Park Sung Jae and Hazrul Akmal Hazarudin
What is a stress test?
Stress testing is a risk management tool that focuses on several key areas including credit, market, operational and liquidity risk. It is used to assess the strength of a bank’s balance sheet through its ability to survive during periods of economic stress, ensuring that banks have enough capital to weather worst-case scenarios. Bank stress testing involves creating hypothetical adverse macroeconomic and financial market scenarios to test the resilience of financial institutions to extreme negative conditions. It is performed under the direction of a single official entity (e.g. the central bank), in which the entire balance sheet of banks in the economy are simultaneously subjected to the same adverse scenarios.
Stress testing is also used to craft prudential policies designed at making sure that micro and macro-prudential policies are adequately resilient to economic slumps. For example, policymakers leverage stress test results to set capital requirements on banks, which are in place to ensure they fund themselves with sufficient loss-absorbing capital to avoid the likelihood of wide-spread banking failures.
How do stress tests work?
The stress testing process starts with the creation of an “adverse hypothetical scenario”. This is generally done using various economic and financial market variables and might include a standard set of scenarios, such as severe recessions (i.e. an increase in unemployment, falling GDP and sharp contractions in the property prices). Analysis of a scenario’s impact involves modelling the way in which the scenario would affect bank balance sheets. For example, rising unemployment might lead to higher loan default rates, potentially affecting the liquidity and solvency of banks.
The following section of this article discusses and compares the different stress testing approaches adopted by some major central banks.
A look at a few major central banks
1. United States Federal Reserve System
Following the aftermath of the 2007 financial crisis, the Federal Reserve launched the Supervisory Capital Assessment Programme (SCAP) in 2009. The SCAP takes a top-down approach of looking at financial institutions as integral parts of the market and in turn assesses systemic risk rather than operating risks of individual institutions in isolation. The participating firms are given macroeconomic situations: a baseline and adverse/severely adverse scenario, in which they calculate “potential losses on loans, securities, and trading positions, as well as pre-provision net revenue (PPNR) and the resources available from the allowance for loan and lease losses’’. Furthermore, they are asked to gauge their ability to internalize the shock with its resources over the two following years. By comparing both of these, the Fed is able to determine the ability of the institution to maintain its minimum regulatory capital standards given the adverse shock in the market.
In 2009, the following scenarios were used:
Baseline Scenario:
- Real GDP Growth and Unemployment Rate: Average of the projections provided by Consensus Forecasts, Blue Chip surveys, and the Survey of Professional Forecasters
- House Prices: Case-Shiller 10-City Composite index with reference to the average of the projections provided by the Blue Chip survey.
Adverse Scenario:
- Real GDP Growth and Unemployment Rate: 1.3 percent lower than the Baseline Scenario to “reflect conditions that are severe but plausible”, relying on the historical Blue Chip forecasts made since the 1970s.
- House Prices: 10 percent lower than estimated prices in the Baseline Scenario (based on historical data from 1900 about year-to-year variability of house prices.)
The conclusion of the test is used by the public to make informed decisions and have worked in recent years to build credibility and confidence for the nationally-essential banks.
In 2010, this was extended to the Comprehensive Capital Analysis and Review (CCAR) through which the Federal Reserve observes the capital adequacy and internal capital planning processes of the organisations participating in the SCAP. The main difference between the CCAR and the SCAP is that the CCAR looks at microprudential exercise. By using individual and industry-level data, it looks at the bank’s planned capital actions in relation to the bank’s baseline scenario rather than the capital action assumptions required in the SCAP and the Dodd-Frank Act, a replacement of the SCAP. In general, the Federal Reserve focuses on credit risk, interest rate risk (IRR), market and operational risk, and default risk.
Unlike the European Central Bank’s stress testing, where data and research are mainly used to raise bank’s awareness of its own capabilities, the CCAR’s research is directly used by the Fed to object to individual bank’s capital plans and to make further policy guidance as it has done in 2017 to create the Stress Capital Buffer.
2. European Central Bank
The European Central Bank (ECB) confronts financial shocks like no other region in the world as the EU has many countries involved in its single market framework. This means that they are at greater risk from both internal and external shocks. For this reason, they run many different stress tests. Like other central banks, it does this on an annual basis. During this process, the bank approaches a bottom-up methodology by analyzing data from individual banks and then creating top-down comparative measurements such as benchmark scores. Like the Fed, in this test, it focuses on credit risk, IRR, market and operational risk, and default risk.
The Bank also conducts an EU-wide stress test which is run every two years. Considered to be thorough in assessing different types of risks, it judges credit risk, net interest income risk, market risk, fees and commission risk, operational risk, conduct risk and liquidity risk. Unlike the Federal Reserve which has significant historical data to predict the probability of the level of stress in the market, due to the nature of the European Single Market, the EU-wide stress test does not show the probability of its scenarios.
The ECB has two scenarios: baseline and adverse scenarios. Although the tests were cancelled due to COVID-19, the 2020 adverse scenario was set as the “lower for longer” adverse scenario. The conditions were given as the following:
Other Factors Include: 10-year government bond yield rates, euro swap rates, and real estate prices
Furthermore, as business models significantly differ between the European nations, the risk of the banks are determined in risk parameters rather than scenario itself. However, this methodology is not applicable throughout the EU, meaning that legislature and policymaking may be difficult. Due to this factor, the results of the EU-wide stress test rarely bring about new policies directly.
The ECB also runs comprehensive assessments upon request from its member states. Thematic tests are also conducted for specific types of shocks such as liquidity risk in 2019 and interest risk in 2017.
3. Bank of England
The Bank of England (BoE) started its annual concurrent stress tests in 2013 upon recommendation from the Monetary Policy Committee. Unlike the previous two central banks who run extensive testing on a large number of banks, the UK runs stress tests on only seven major banks upon which its financial activity is heavily dependent (i.e. Barclays, HSBC, Lloyds Banking Group, Nationwide, The Royal Bank of Scotland Group, Santander UK Group Holdings plc and Standard Chartered). The BoE also employs an annual cyclical scenario and a biennial exploratory scenario at times in addition to the commonly utilized baseline scenario. As its name suggests, the cyclical scenario focuses on economic performance and financial imbalances whereas the biennial exploratory scenario seeks to understand a wider range of indicators not related to historical context. The objective here is to understand the unpredictable and to prepare banks to be resilient to covertly approaching shock.
One of the biggest characteristics of the Bank of England’s stress tests is that its results are directly translated into policies and capital planning. The figure below outlines this process:
How has stress testing adapted over the years?
Big data and methodology
In 2018, 2.5 quintillion bytes of data were created each day and is projected to reach 463 exabytes by 2025. As more data is created and concomitantly collected by financial institutions, models adopting more granular approaches become more feasible. This allows central banks to shift away from coarse frameworks like VAR models which provide good estimates of overall losses but are limited in revealing losses of individual banks or that of particular loan types.
Cyberattacks
By 2025, the global fintech market is projected to be worth approximately 460 billion USD. The increasing digitalisation of financial services is accompanied by heightened risks of cyberattacks, which can be particularly insidious and devastating, evident from prominent attacks like WannaCry, NotPetya and more recently, the SolarWinds hack. Given the potential for attacks on systemically important financial institutions (SIFI), more central banks are moving beyond the microprudential approach that they have traditionally taken towards cyberattacks to one which recognises the financial instabilities they can generate.
These instabilities are generated via the following channels: risk concentration, risk contagion and erosion of confidence. Risk concentration entails the disruption of particular entities which the financial system is heavily reliant on such as payment and settlement systems, engendering liquidity and solvency woes. Erosion of confidence can result in a self-fulfilling prophecy in which a bank run results from a cyberattack. Risk contagion occurs when a cyberattack on a particular entity transmits solvency and liquidity risks to other financial institutions which have links to it.
The Bank of England has recognised the far-reaching implications of cyberattacks and piloted a cyber stress test of lenders in 2019, with future stress tests plausibly including more scenarios like corruption of data. While the European Central Bank has implemented the TIBER-EU framework to boost the cyber resilience of financial institutions, it has not taken significant steps to incorporate cyber resilience into stress testing. The same can be said of the Federal Reserve.
To incorporate the systemic risks stemming from cyberattacks into stress test models, cybersecurity data from financial institutions and central bank monitoring arms can be utilised to estimate probability and severity. The absence of data can also be circumvented by models requiring only the use of firm-specific variables.
Climate change
Being one of the defining challenges of our times, climate change can also affect the performance of financial institutions via “physical” and “transition” risks. The former refers to extreme physical events attributed to climate change like floods and wildfires, which can disrupt businesses. The latter refers to how the transition to a low carbon economy could reduce the nominal value of carbon-intensive assets, affecting banks’ assets holding. In response to this, some central banks are already delving into the use of related measures within their stress test models. The Bank of England and the Banque de France will be running 30-year climate change scenarios through a bottom-up approach; that is, based on bank provided data. The European Banking Authority is also, for the first time, considering the inclusion of climate-change sensitivities in the upcoming 2021 EU-wide stress tests. The Federal Reserve, on the other hand, has not made much progress on this front.
Stress testing would involve the crafting of plausible scenarios before selecting the appropriate model to simulate any adherent financial risks. In the case of the Bank of England’s first climate change stress test, it put forward three scenarios:
- global temperature rise is limited to less than 2°C due to timely action
- slight inaction which still achieves a less than 2°C rise in global temperature albeit more in a more destabilising manner
- significant inaction leading to spikes in global temperatures and a concomitant spike in physical risks.
Sources
Bank of England. 2015. “The Bank Of England’s Approach To Stress Testing The UK Banking System.”
Bank of England. 2019. “Stress Testing The UK Banking System:2019 Guidance For Participating Banks And Building Societies.”
“Bank Of England To Test Banks’ Resilience To Cyber Attacks”. 2021. U.S.. https://www.reuters.com/article/us-boe-cyber-tests-idUSKCN1QM1H7.
“Bank Stress Test”. 2021. Investopedia. https://www.investopedia.com/terms/b/bank-stress-test.asp#:~:text=A%20bank%20stress%20test%20is%20an%20analysis%20to,capital%20to%20withstand%20an%20economic%20or%20financial%20crisis.
Board of Governors of the Federal Reserve System. 2009. “The Supervisory Capital Assessment Program: Design And Implementation”. https://www.federalreserve.gov/newsevents/pressreleases/files/bcreg20090424a1.pdf.
Chan-Lau, Jorge A. 2013. “Market-Based Structural Top-Down Stress Tests Of The Banking System”. SSRN Electronic Journal. doi:10.2139/ssrn.2250163.
Dent, Kieran, Ben Westwood, and Miguel Segoviano. 2016. “Stress Testing Of Banks: An Introduction.”
European Systemic Risk Board. 2020. “Macro-Financial Scenario For The 2020 EU-Wide Banking Sector Stress Test.”
Federal Reserve System. 2013. “Policy Statement On The Scenario Design Framework For Stress Testing.”
Financial Stability Institute. 2021. “Stress-Testing Banks — A Comparative Analysis.”
“Fintech (Financial Technology) Market To Hit $460 Billion By 2025 — Global Insights On Growth Drivers, Key Trends, Strategic Initiatives, Value Chain Analysis And Future Outlook: Adroit Market Research”. 2021. Globenewswire News Room. https://www.globenewswire.com/news-release/2020/01/29/1976561/0/en/Fintech-Financial-Technology-Market-to-hit-460-billion-by-2025-Global-Insights-on-Growth-Drivers-Key-Trends-Strategic-Initiatives-Value-Chain-Analysis-and-Future-Outlook-Adroit-Mar.html.
Goh, Joseph, Heedon Kang, Zhi Xing Koh, Jin Way Lim, Cheng Wei Ng, Galen Sher, and Chris Yao. 2020. “Cyber Risk Surveillance”. IMF Working Papers 20 (28). doi:10.5089/9781513526317.001.
Hawser, Anita. 2021. “Global Finance Magazine — Regulators Impose New Stress Tests Around Climate Change”. Global Finance Magazine. https://www.gfmag.com/magazine/november-2020/green-esg-stress-tests.
Lehmann, Alexander. 2021. “Climate Risks To European Banks: A New Era Of Stress Tests | Bruegel”. Bruegel.Org. https://www.bruegel.org/2020/02/climate-stress-test/.
“Stress Tests”. 2021. European Central Bank — Banking Supervision. https://www.bankingsupervision.europa.eu/banking/tasks/stresstests/html/index.en.html.
Vuleta, Branka. 2021. “How Much Data Is Created Every Day? [27 Powerful Stats]”. Seedscientific. https://seedscientific.com/how-much-data-is-created-every-day/.