Home / Wiki / S / Stress Testing

Stress Testing

Actuarial and Risk SciencesUpdated 16 Nov 2026

Stress testing is the assessment of how a bank, insurer, financial market infrastructure, investment portfolio or the financial system as a whole would fare under severe but plausible adverse conditions. A stress test applies a shock, such as a deep recession, a sharp rise in interest rates, a market crash or the default of large counterparties, and estimates the resulting losses, capital shortfalls or liquidity gaps.

Stress tests complement statistical measures such as value at risk (VaR), which describe normal market conditions and say little about rare events. Firms run them as part of risk management, and supervisors run or require them to check that institutions hold enough capital and liquidity. Since the 2008 financial crisis they have become a central tool of bank supervision and of the oversight of central counterparties.

01

Types of stress test

  • Sensitivity analysis. One risk factor, such as an interest rate or exchange rate, is moved by a set amount while others are held constant, as in a simple portfolio stress test.
  • Scenario analysis. Several factors move together according to a coherent narrative, such as a recession with falling house prices and rising unemployment.
  • Historical scenarios. Past episodes, such as the 1987 stock market crash or the events of 2008, are replayed on the current portfolio.
  • Hypothetical scenarios. Events that have not happened but are judged plausible are designed by the firm or its supervisor.
  • Reverse stress tests. Reverse stress testing starts from a failure outcome and searches for the scenarios that would cause it.

Tests may focus on solvency, measuring capital after losses, or on liquidity, measuring whether buffers and inflows cover outflows under stress, which is known as liquidity stress testing.

02

Supervisory stress tests of banks

Modern supervisory stress testing began in the United States in 2009 with the Supervisory Capital Assessment Program, which tested the largest banks during the crisis and was followed by public capital raising. The Dodd-Frank Act later required regular tests, and the Federal Reserve runs the Comprehensive Capital Analysis and Review (CCAR) and Dodd-Frank Act stress tests (DFAST), whose results feed a bank-specific stress capital buffer.

In the European Union, the European Banking Authority (EBA) coordinates EU-wide stress tests of large banks with national supervisors and the European Central Bank, and the Bank of England runs concurrent stress tests of major UK lenders. Supervisory tests usually set a baseline and an adverse scenario over several years, require banks to project losses, income and capital ratios, and compare results with hurdle rates such as a minimum common equity tier 1 (CET1) ratio.

03

Stress testing at central counterparties

For a central counterparty (CCP), stress testing is the main tool for sizing the resources that stand behind margin. Under Principle 4 of the PFMI, a CCP must hold financial resources sufficient to cover, in extreme but plausible market conditions, the default of the participant and its affiliates that would cause the largest credit exposure, a standard known as Cover 1. A CCP involved in activities with a more complex risk profile, or systemically important in several jurisdictions, must cover the two participants and affiliates with the largest exposures, known as Cover 2. Principle 7 applies a similar test to liquid resources. The PFMI expect credit stress tests to be run daily, with a full review of scenarios, models and parameters at least monthly.

The Cover 2 condition can be written as:

R≥L1+L2R \ge L_{1} + L_{2}

Here RR is the prefunded resources available beyond the defaulters' own margin, mainly the clearing fund and the CCP's own contribution, and L1L_{1} and L2L_{2} are the two largest stressed losses in excess of margin among participants and their affiliates. Suppose the three largest stressed losses are 120 million, 90 million and 60 million. Cover 1 requires 120 million; Cover 2 requires 120+90=210120 + 90 = 210 million. A default fund of 180 million plus 20 million of CCP capital gives 180+20=200180 + 20 = 200 million, which meets Cover 1 but falls 210−200=10210 - 200 = 10 million short of Cover 2, so the fund must be enlarged.

Authorities also run system-wide exercises: the European Securities and Markets Authority has carried out EU-wide CCP stress tests since the mid-2010s.

04

Limitations and criticism

Stress tests are only as good as their scenarios and models. Scenarios chosen in advance may miss the shock that actually arrives, and repeated exercises can encourage institutions to manage to the test. Models capture first-round losses better than second-round effects such as fire sales, funding runs and interaction between institutions, which system-wide tests under macroprudential policy try to address. Results depend on assumptions about correlations, which tend to rise in a crisis, and so carry significant model risk. Publishing results can support confidence, as in 2009, but can also create pressure to soften scenarios. Climate scenario analysis has extended the method to long horizons where historical data offer little guide.

Last updated 16 Nov 2026