This article explains what stress testing is and how banks use it to manage credit risk.
Stress testing refers to a technique used by financial institutions such as banks to estimate the impact of worst-case scenarios for loans, investments and debts.
In simple words, stress testing is like putting a bank through a really tough test to see if it can handle difficult future scenarios such as recession, emerging climate risks such as floods, famines and hurricanes.
There are four types of risks mainly used in stress testing - credit risk, market risk, sovereign risk, funding cost and net interest income risk, climate risk.
It includes the following macroeconomic scenarios for stress testing.
- GDP
- Unemployment rate
- Inflation
- House prices
It includes the losses that could occur due to changes in market prices such as :
- Interest Rates
- Credit Spreads
- Exchange Rates
- Equities
- Commodities
- Counterparty Credit risk
It is related to the risk associated with investments in government bonds issued by different countries. In the context of sovereign risk, a "haircut" refers to a reduction in the value of financial instruments such as government bonds or securities that an investor holds. When conducting stress tests, financial institutions often apply haircuts to government bonds to simulate scenarios of economic distress or sovereign default.
In stress testing, funding cost risk is estimated by considering scenarios where funding costs rise significantly due to market disruptions, changes in interest rates etc. Higher funding costs can lead to reduced profitability and liquidity challenges.
Net interest income (NII) risk refers to the impact of changes in interest rates on a bank's net interest income which is the difference between the interest earned on assets (loans, investments) and the interest paid on liabilities (deposits, borrowings).
Climate risk refers to how changes in climate (extreme weather) such as floods, famines, hurricanes can affect the financial health of banks.
There are two types of climate risks involved for Stress Testing.
- Transition Climate Risk: It is the risk that banks will suffer losses as a result of the transition to a low-carbon economy. This includes shifts in technology, changes in regulations and the adoption of sustainable practices.
- Physical Climate Risk: It is the risk that banks will suffer losses as a result of extreme weather events or other climate-related disasters.
Stress Testing Scenarios
Following is the list of stress testing scenarios for credit risk management.
- 10% drop in the S&P 500 index.
- 4% increase in the unemployment rate.
- 3% increase in the inflation rate.
- 100 basis point increase in the interest rate.
- 5% decline in real estate prices.
- 20% reduction in global trade volume.
- 15% depreciation of the domestic currency.
- Major cyberattack affecting multiple industries.
- Sudden exit of a key borrower with a large exposure.
- Trade sanctions imposed on a key trading partner.
- 50% reduction in tourism due to travel restrictions.
- Large-scale supply chain disruption causing business closures.
- 20% decrease in corporate bond prices affecting fixed-income portfolios.
- Major social unrest leading to business shutdowns.
Orderly Scenario : It assumes that environmental policies related to climate change will be put in place early on and will slowly get stricter over time. It is based on the Network for Greening the Financial System (NGFS) net zero 2050 scenario which limits global warming to 1.5°C by achieving net zero carbon emissions around 2050. Both physical and transition risks are low due to the smooth transition and reduced global warming.
Disorderly Scenario : It assumes delay of introduction of new climate policy until 2030. Strong actions are needed to limit global warming below 2°C. It may result to higher transition risks and increased physical risks.
Hot House World Scenario : It assumes no new climate policies. Global emissions grow until 2080 which may cause about 3°C warming. While transition risks are minimal, the economy suffers from extreme physical risks.
Types of Stress Testing
- Scenario Analysis : Banks use scenario analysis as a tool to prepare for different possible future scenarios. By simulating these hypothetical scenarios, they can measure how each scenario might influence their financial well-being.
- Reverse Stress Testing : In reverse stress testing, banks start with a predetermined adverse outcome and work backward to identify the factors that could trigger such a situation. It helps banks in identifying vulnerabilities within their operations and risk management practices.
- Sensitivity Analysis : It refers to how changes in important factors or variables affect the bank's credit portfolio. For example, how investments in commercial real estate would react to increase in interest rates.
Different Approaches in Stress Testing
There are two common stress testing approaches which are bottom-up and top-down.
Bottom-Up Stress Test : It is a stress test performed by banks. They use their own models and means to perform the stress test. It is based on the banks' own customer data and external granular credit bureau data. It takes into account shocks at individual customer level and later we combine them to see the impact at overall level.
Top-Down Stress Test : It is a stress test performed by a central bank such as Fed in US. It is based on aggregated institution data. It is based on the methodology and assumptions developed by the central bank.
Hybrid Stress Test : It is a stress test that uses a combination of top-down and bottom-up approach. With this approach, the central bank gives banks a detailed set of rules on the results that their own models produce.
Statistical Techniques for Stress Testing
In stress testing, one must forecast the level of risk in a portfolio based on projected macroeconomic variables such as GDP, unemployment rate, inflation rate and house price index. The following are the statistical techniques commonly used to perform stress testing.
- Time Series Models: Credit and economic data can be analyzed using ARIMA or GARCH models.
- Random Forest Model: The main benefit of random forest is that it can handle non-linear relationships between dependent and independent variables. It can also perform out-of-sample testing.
- Copula Models: Copula models can be used to understand dependencies between different variables such as credit defaults and macroeconomic factors.
- Bayesian Probabilistic Model: Bayes' theorem is useful when making predictions conditional on extreme events.
- Monte Carlo Simulation: It refers to generating a large number of random scenarios based on probability distributions for key economic variables (e.g., GDP, interest rates). The credit portfolio is then simulated for each scenario to estimate losses and changes in credit metrics.
Deepanshu founded ListenData with a simple objective - Make analytics easy to understand and follow. He has over 10 years of experience in data science. During his tenure, he worked with global clients in various domains like Banking, Insurance, Private Equity, Telecom and HR.
Share Share Tweet