Guide détaillé à venir
Nous préparons un guide éducatif complet pour le Portfolio Stress Testing. Revenez bientôt pour des explications étape par étape, des formules, des exemples concrets et des conseils d'experts.
Portfolio stress testing is a risk management technique that evaluates how a portfolio performs under extreme but plausible adverse market conditions — conditions that fall outside the normal distribution captured by standard statistical risk measures like VaR. While VaR and CVaR estimate risk based on recent market behavior, stress testing forces the portfolio through specific shock scenarios: either historical crises (the 2008 Global Financial Crisis, the COVID-19 crash, Black Monday 1987) or hypothetical shocks designed to expose specific vulnerabilities (a 200 basis point rate rise, equity markets falling 40%, credit spreads doubling). The Basel Committee on Banking Supervision requires banks to conduct stress testing as a mandatory complement to model-based VaR. Under Basel II Pillar 2 and ICAAP (Internal Capital Adequacy Assessment Process), institutions must identify, quantify, and plan for stress scenarios. The Dodd-Frank Act in the U.S. mandates annual Comprehensive Capital Analysis and Review (CCAR) and Dodd-Frank Act Stress Tests (DFAST) for large banks, requiring them to demonstrate adequate capital under severely adverse economic scenarios. Stress testing serves several distinct purposes. First, it reveals tail risks that statistical models miss — particularly concentrated positions that become illiquid in crises. Second, it provides a basis for communication with boards, regulators, and risk committees about specific worst-case outcomes in concrete terms. Third, it informs capital planning: institutions should hold enough capital to survive their defined worst-case stress scenario. Fourth, it can uncover unintended risk concentrations that look small in normal markets but become dominant in crises. Modern stress testing frameworks include: historical scenario analysis (re-running actual crisis events through the current portfolio); hypothetical scenario analysis (user-designed shocks based on economic reasoning); reverse stress testing (finding the scenario that would cause portfolio failure and asking how probable that scenario is); and macro-economic scenario analysis (translating GDP contraction, unemployment rise, and credit spread widening into portfolio P&L). Critically, stress tests complement rather than replace statistical risk measures. A comprehensive risk framework uses VaR for daily risk monitoring, CVaR for capital adequacy, and stress tests for tail-scenario preparedness and strategic planning.
Stress P&L = Σ (Position_i × Sensitivity_i × Scenario Shock_i) For multi-factor: ΔP = DV01 × Δrates + Delta × ΔEquity + CS01 × ΔCredit spreads + ...
- 1Define the stress scenario: select a historical event or construct a hypothetical scenario specifying shocks to all relevant market factors (equities, interest rates, credit spreads, FX, commodities, volatility).
- 2Map all portfolio positions to their relevant risk factors: which equities, rate tenors, credit issuers, and FX pairs does each position expose the portfolio to?
- 3Compute position-level sensitivities: equity delta, rate DV01 by tenor, credit CS01 by issuer/sector, FX delta, and vega for options.
- 4Apply scenario shocks to each factor: multiply sensitivity by scenario shock magnitude to get position-level stress P&L.
- 5Aggregate across positions to get total portfolio stress P&L, accounting for correlations specified in the scenario.
- 6Compare stress loss to available capital buffers: is the institution solvent under this scenario? Does it maintain required regulatory capital ratios?
- 7Identify concentrated risk exposures: which positions or risk factors drove the majority of stress loss? Use results to refine limits or reduce concentrations.
Cash and IG bonds provided only partial offset to equity/HY crash
Equity stress P&L = $10M × (−55%) = −$5,500,000. HY bond P&L = $5M × (−35%) = −$1,750,000. IG bond P&L = $2M × (+5%) = +$100,000. Cash unchanged = $0. Total stress P&L = −$7,150,000 (36% loss). In the actual 2008 GFC (peak-to-trough): S&P 500 fell 57%, HY spreads widened by 1,900 bps (−45–55% in price terms), while Treasuries rallied significantly. The scenario illustrates why portfolios with equity and HY correlation suffered catastrophically while Treasury-heavy portfolios were spared.
Linear approximation; actual loss slightly less due to convexity
Stress P&L = −DV01 × Rate shock = −$50,000 × 200 bps = −$10,000,000. A $50,000 DV01 portfolio would lose $10 million if rates rise 200 basis points instantly. In the 2022 rate hiking cycle, the U.S. 10-year Treasury yield rose from 1.5% to 5.0% (350 bps) — a scenario far exceeding the standard +200 bps stress. This historical realization illustrates why stress scenarios must be regularly reviewed and updated to reflect current macro conditions and risks.
HY spreads widen much more dramatically in credit crises
IG stress P&L = −$30,000 × 150 = −$4,500,000. HY stress P&L = −$15,000 × 600 = −$9,000,000. Total = −$13,500,000. Credit spread stress scenarios often use significantly larger shocks for HY than IG because HY spreads have historically widened 4–10× more than IG spreads in credit cycles. In 2008, IG spreads widened by ~400 bps while HY spreads widened by ~1,800 bps. These asymmetric spread moves must be captured in multi-bucket stress scenarios rather than applying a single uniform spread shock.
Reverse stress test identifies specific failure threshold scenarios
Reverse stress testing asks: 'What scenario would cause failure?' With $50M capital: if equities fall X%, loss = $200M × X. Setting loss = $50M: X = 25%. So a 25% equity decline alone exhausts all capital. If bonds also fall 15% (−$15M), equities need to fall only 17.5% to trigger capital exhaustion. This analysis tells management exactly how vulnerable the institution is and helps calibrate whether the capital buffer is adequate relative to historical equity drawdown magnitudes.
Bank CCAR/DFAST regulatory stress testing, representing an important application area for the Stress Testing Calc in professional and analytical contexts where accurate stress testing calculations directly support informed decision-making, strategic planning, and performance optimization
Investment portfolio tail risk identification for CIO reporting, representing an important application area for the Stress Testing Calc in professional and analytical contexts where accurate stress testing calculations directly support informed decision-making, strategic planning, and performance optimization
Insurance company Solvency II Own Risk and Solvency Assessment (ORSA), representing an important application area for the Stress Testing Calc in professional and analytical contexts where accurate stress testing calculations directly support informed decision-making, strategic planning, and performance optimization
Hedge fund due diligence — investor scenario analysis presentation, representing an important application area for the Stress Testing Calc in professional and analytical contexts where accurate stress testing calculations directly support informed decision-making, strategic planning, and performance optimization
Pension fund liability-relative stress testing (surplus risk), representing an important application area for the Stress Testing Calc in professional and analytical contexts where accurate stress testing calculations directly support informed decision-making, strategic planning, and performance optimization
{'case': 'Liquidity Stress Testing', 'explanation': 'Market risk stress testing (P&L impact) is separate from liquidity stress testing (cash flow timing). Banks must also stress-test their liquidity position: the Basel LCR (Liquidity Coverage Ratio) and NSFR (Net Stable Funding Ratio) require portfolios of liquid assets sufficient to survive 30-day and 1-year stress outflows respectively.'}
In the Stress Testing Calc, this scenario requires additional caution when interpreting stress testing results. The standard formula may not fully account for all factors present in this edge case, and supplementary analysis or expert consultation may be warranted. Professional best practice involves documenting assumptions, running sensitivity analyses, and cross-referencing results with alternative methods when stress testing calculations fall into non-standard territory.
In the Stress Testing Calc, this scenario requires additional caution when interpreting stress testing results. The standard formula may not fully account for all factors present in this edge case, and supplementary analysis or expert consultation may be warranted. Professional best practice involves documenting assumptions, running sensitivity analyses, and cross-referencing results with alternative methods when stress testing calculations fall into non-standard territory.
| Scenario | Equities | HY Spreads | IG Spreads | Rates | Duration |
|---|---|---|---|---|---|
| Black Monday (1987) | −22% in 1 day | Not applicable | Widened | Rallied | 1 day |
| Russian Default/LTCM (1998) | −20% | +600 bps | +100 bps | Rallied | 3 months |
| Dot-Com Crash (2000–02) | −49% S&P | +500 bps | +150 bps | Rallied | 2.5 years |
| GFC 2008–09 | −57% peak-trough | +1,900 bps | +400 bps | Mixed | 18 months |
| Euro Sovereign Crisis (2010–12) | −25% Euro | +300 bps | +200 bps | Mixed | 2 years |
| COVID-19 (Feb–Mar 2020) | −34% in 5 wks | +900 bps | +250 bps | Rallied | 5 weeks |
| 2022 Rate Shock | −20% S&P | +400 bps | +150 bps | +400 bps | 12 months |
What is the difference between sensitivity analysis and stress testing?
Sensitivity analysis examines the portfolio's response to small changes in a single risk factor (e.g., what is the P&L if equities move ±1%?). It is used for routine risk monitoring and hedging. Stress testing applies large, multi-factor shocks simultaneously — reflecting the correlated nature of financial crises where equities crash, credit spreads widen, liquidity dries up, and volatility spikes all at once. Stress testing is forward-looking for extreme scenarios rather than marginal changes. Both are essential: sensitivity analysis for daily risk management, stress testing for capital planning and tail risk identification.
What historical scenarios are commonly used in bank stress testing?
Standard historical scenarios include: Black Monday (October 1987 — global equities −20–25% in one day); LTCM/Russian Default (1998 — emerging market debt collapse, liquidity crisis); Dot-com crash (2000–2002 — tech equities −75–80%); 9/11 (2001 — market closure, initial −10%); 2008 Global Financial Crisis (2007–2009 — equities −55%, credit spreads +1,900 bps, interbank markets frozen); European Sovereign Debt Crisis (2010–2012 — PIIGS spreads widened dramatically); COVID-19 crash (Feb–March 2020 — global equities −35% in 5 weeks); 2022 Rate Hiking (rates +400 bps, bond indices −15–20%). Each scenario captures a different risk driver and risk type.
What is reverse stress testing?
Reverse stress testing starts from a defined failure outcome (e.g., capital falls below regulatory minimum, or liquidity is exhausted) and works backward to identify the scenarios that would cause this outcome. Unlike standard stress testing (which applies scenarios and measures outcomes), reverse stress testing identifies the weakest points in the institution's risk profile. Questions like 'what combination of equity decline and credit spread widening would exhaust our capital buffer?' reveal vulnerabilities that standard scenario analysis might not highlight. Reverse stress testing is now required under Basel Pillar 2 and PRA (UK Prudential Regulation Authority) guidance.
How does the Fed CCAR stress test work?
The Federal Reserve's Comprehensive Capital Analysis and Review (CCAR) and related Dodd-Frank Act Stress Tests (DFAST) require large U.S. bank holding companies (>$100B assets) to demonstrate capital adequacy under three scenarios: baseline, adverse, and severely adverse. The severely adverse scenario is designed by the Fed and typically features: unemployment rising to 10%+, GDP declining 4–6%, equity markets falling 50%, real estate prices declining 25–35%, and credit spreads widening significantly. Banks must show they maintain minimum capital ratios (CET1 ≥ 4.5%, Tier 1 ≥ 6%, Total capital ≥ 8%) even under the severely adverse scenario after including planned capital actions.
What are the limitations of stress testing?
Stress testing has important limitations. Historical scenarios are backward-looking and may miss novel future risks not present in historical data. Hypothetical scenarios require judgment about what scenarios are 'severe but plausible' — too conservative and capital is wasted; too lenient and risks are undercapitalized. Scenario completeness is always a challenge: risk factors that are not shocked may create false comfort. Correlation assumptions embedded in scenarios (even implicitly) shape results significantly. Additionally, stress tests are typically conducted quarterly or annually, while market risk evolves continuously. Real crises are also often faster, deeper, and more correlated than scenario models assume.
What is a scenario probability and should I attach one to stress scenarios?
Attaching probabilities to stress scenarios is controversial. Some risk managers argue all scenarios should be presented without probabilities because their value lies in identifying concentration risks, not in computing expected losses. Others argue that probability-weighted scenario analysis provides more information. The Basel framework does not require probability assignments to stress scenarios — they are designed as tail events whose frequency is less important than their magnitude. When probabilities are assigned, extreme tail events like the 2008 GFC are typically given probabilities of 1-in-25 to 1-in-100-year events, though these estimates are highly uncertain given the limited historical sample of major financial crises.
How do stress tests interact with VaR and CVaR?
Stress tests, VaR, and CVaR are complementary risk measures used together in comprehensive risk frameworks. VaR monitors day-to-day risk against predefined limits (typically 1-day 99% VaR). CVaR provides the expected average loss in the tail, addressing VaR's limitation of ignoring loss severity. Stress tests reveal specific crisis-scenario outcomes that may fall far outside even the CVaR estimate — particularly for portfolios with concentrated exposures that become highly correlated in crises. Best practice: VaR and CVaR for daily monitoring, stress tests for quarterly capital planning and board reporting, and reverse stress tests for strategic risk identification.
Conseil Pro
Build a stress test heatmap showing loss by scenario (rows) and by risk type (equity, rates, credit, FX — columns). This reveals which risk types and which scenarios drive the most loss, guiding risk limit setting and hedging strategy.
Le saviez-vous?
The term 'stress test' in banking was first used systematically by U.S. regulators during the 2009 Supervisory Capital Assessment Program (SCAP), a public stress test of 19 major U.S. banks. The SCAP — which revealed a $74.6 billion capital shortfall — is widely credited with restoring confidence in the U.S. banking system during the peak of the GFC and accelerating the recovery.