Why stress test matters
Risk models calibrated to normal periods can fail precisely when losses matter most. Stress testing makes extreme assumptions explicit and shows how positions interact when correlations, volatility, margin, and liquidity become unfavorable. It supports capital and liquidity planning, limit setting, contingency procedures, and board discussion. The exercise is valuable even when the scenario never occurs because it reveals concentrations, nonlinear exposure, and dependence on continuous financing.
How it is applied
Historical tests replay episodes such as a crash or rate shock on today’s positions. Hypothetical tests combine economically consistent shocks designed for current vulnerabilities. Sensitivity tests move one factor, while reverse stress tests search for conditions that breach a solvency, liquidity, or loss threshold. Instruments should be revalued, including options and collateral. Results can include market loss, margin calls, redemption needs, trading costs, days to liquidate, and second-order effects.
Portfolio example
A multi-asset fund tests a scenario with equities down 25%, credit spreads 300 basis points wider, government yields up 1%, volatility doubled, and market depth halved. The initial mark-to-market loss is 14%, but collateral calls and investor redemptions create an additional cash shortfall. The lesson is not that this exact scenario will occur. It is that the fund needs more liquidity or less exposure to avoid forced selling.
How to interpret it
A stress loss is conditional on the stated shocks and assumptions, not a forecast or confidence interval. Comparing results across time shows whether vulnerability is growing, provided methodology remains consistent. Users should inspect which positions drive loss, whether offsets remain tradable, and what management actions are realistic during the event. Passing one scenario does not establish safety; a portfolio can be resilient to a historical crisis yet exposed to a different regime.
Limitations and common misconceptions
Scenario selection is subjective and may fight the last war. Simplified shocks can be internally inconsistent, while complex models create false precision. Liquidity, market impact, counterparty failure, and feedback effects are difficult to estimate. Management actions assumed in the test may be unavailable when all investors act together. Stress tests should be diverse, refreshed, challenged independently, linked to decisions, and combined with ordinary monitoring rather than treated as a compliance exercise.
Sources and further reading
- Risk Management: An IntroductionCFA Institute
- Minimum Capital Requirements for Market RiskBank for International Settlements