Stress test

Statistics can tell you what usually happens. They are not built to imagine the day everything goes wrong at once.

A stress test fills that gap. The portfolio manager deliberately creates a scenario of extreme negative events in one or many risk factors, then works out what it would do to the portfolio. The scenario can be a past crisis played back, or a hypothetical crisis invented for the purpose.

The point is not prediction. It is preparation, so that when a real 'black swan' event hits, the manager already knows roughly how bad it could get and can act fast.

The workbook (section 17.4.3) pairs stress testing with a second, related tool: sensitivity analysis, which measures how the portfolio responds to ordinary market moves using beta for an equity portfolio, duration for a bond portfolio, and delta, gamma and vega for an options-based portfolio.

Stress testing goes further. Instead of an ordinary move, the manager creates a scenario of an extreme negative event — recreating a real past crisis, or inventing a hypothetical one — to see the resulting loss. The workbook gives no specific percentage or threshold that defines when a move counts as 'extreme'; that judgement is left to the manager. Its stated payoff is direct: a manager who has already stress-tested for a 'black swan' event finds it 'easy to rebalance during the crisis hour' when the real thing happens, rather than reacting from a standing start with no time to think.

Illustrative figures. A PMS strategy holds Rs 40,00,00,000, split Rs 28,00,00,000 in equity (portfolio beta 1.1) and Rs 12,00,00,000 in bonds (portfolio duration 6 years).

The manager recreates a past crisis scenario: a 30% equity market fall alongside a 150 basis point rise in bond yields. Applying the sensitivities: equity loss ≈ 28,00,00,000 × 1.1 × 30% = Rs 9,24,00,000; bond loss ≈ 12,00,00,000 × 6 × 1.5% = Rs 1,08,00,000. Total stressed loss ≈ Rs 10,32,00,000, about 26% of the portfolio.

That single number is what a stress test is for — not a daily probability like VaR, but 'if this specific bad day happens, here is what we lose', worked out before it happens rather than during it.

Chapter 17 (Risk), section 17.4.3 (Stress Testing and sensitivity analysis), distinguishes stress testing from sensitivity analysis and links it to being prepared for black-swan events. Expect a question distinguishing stress testing from Value at Risk, or asking which sensitivity measure applies to which asset class.

• Stress testing is scenario-based; VaR is statistical and probability-based — different tools in the same section 17.4 toolkit, not the same thing under two names.
• Sensitivity analysis measures response to ordinary moves; stress testing measures response to extreme ones, using the same underlying sensitivities at a more severe scale.
• A stress scenario can be a real past event replayed, or a hypothetical one invented — the workbook allows both; do not assume it must be historical.
• Beta is for equity, duration is for bonds, and delta/gamma/vega are for options — matching the wrong sensitivity measure to the wrong asset class is the easy error here.