Stress the Management Action, Not Just the Starting Portfolio

Worked illustration

Should a stress test assume management will respond?

It can examine a response, but the response must be an explicit scenario rather than an unexplained rescue assumption. Preserve the unmitigated result so the reader sees the starting vulnerability. Then describe the proposed action, decision trigger and implementation conditions. A lower modeled loss after an assumed sale does not demonstrate that approval, liquidity and settlement would all be available when the scenario requires them.

Basel distinguishes static stresses from dynamic simulations incorporating management actions. Its banking context does not govern this example. Changed holdings require new assumptions; an action-conditioned result cannot be described as a frozen-portfolio sensitivity.

Specify an action contract before calculating its benefit

Write down the information available at the trigger, who can approve the action, the eligible assets and the earliest execution and settlement times. Define size in currency or units and explain which mark determines it. A fixed amount sold after a drawdown is not the same quantity as that amount at the original price. Include restrictions, pledged assets and operational dependencies rather than treating every holding as immediately saleable.

Record transaction cost assumptions separately from the market shock. A sale at a discount to an already stressed mark creates an incremental expense; the gross proceeds are a balance-sheet conversion. Do not deduct the asset's entire sale value as another loss. If the scenario already embeds a liquidation adjustment, explain whether the action cost is genuinely additional or would duplicate the same deterioration under a different label.

Reconcile an illustrative action and residual portfolio

Start with invented wealth of USD 10 million: six million in equities, three million in bonds and one million in cash. A 20% equity decline reduces the equity sleeve to USD 4.8 million and total value to USD 8.8 million. The proposed action sells USD 1 million at that stressed mark with an assumed 2% execution cost, producing USD 980,000 cash and USD 20,000 incremental loss.

After the sale, equity is USD 3.8 million, bonds USD 3 million and cash USD 1.98 million, totaling USD 8.78 million. A further 10% equity decline gives USD 8.4 million ending value. Without action the same later decline gives USD 8.32 million. The action-conditioned advantage is USD 80,000: USD 100,000 of avoided second-stage loss minus USD 20,000 cost. These are synthetic assumptions, not an actual liquidation plan or execution record.

Test timing and feasibility rather than perfect hindsight

Specify whether the trigger can be observed before the action must occur. A response chosen using the eventual worst point uses future information and overstates what management could know. A policy can be predeclared yet still face delays from approvals, trading capacity or settlement. Run a delayed-action case and an unavailable-action case, keeping the market path fixed so the comparison isolates the dependence on execution timing.

If a response relies on a lender, buyer or counterparty, identify the evidence for their capacity under the same stress. A general relationship or normal-market practice is not unconditional stressed access. The FSB's liquidity-preparedness recommendations emphasize reliable resources and operational readiness for margin demands. That supports testing the response itself rather than treating a paper contingency plan as demonstrated liquidity or an executed offset.

Inspect feedback and the portfolio left behind

Selling liquid assets can leave a more concentrated or less liquid residual portfolio. It can also remove future income or collateral capacity. Recompute relevant exposures after the action instead of scaling the original stress result proportionally. The synthetic sale example removes equity sensitivity, but it does not model future income, tax or financing consequences. Those exclusions matter before interpreting its terminal advantage as a broader economic benefit.

Where multiple participants attempt similar actions, execution assumptions may be mutually inconsistent. A single portfolio's sale does not establish that all portfolios can realize the same price simultaneously. Identify a feedback mechanism before adding extra price impact; an arbitrary second haircut is not evidence of contagion. If market impact cannot be quantified credibly, preserve it as a stated limitation and show a separate clearly judgmental challenger rather than invented calibration.

Ask what must be true for the mitigation to work

Ask whether authority exists, whether the assets are free to transfer and what confirms the time and cost assumptions. Does the action reduce the intended vulnerability or simply move it to another ledger? Is the residual portfolio still able to meet subsequent obligations? Report the original loss, action cost, changed exposure and later outcome separately so the claimed mitigation can be reconstructed and challenged.

Flag missing baselines, premature settlement credit, proceeds counted as gains and hindsight responses. The workbench does not simulate this process; the example establishes neither executable capacity nor a production risk-control capability.

Review checkpoints

• Preserve the unmitigated result and state the response conditions.
• Reconcile sale proceeds, incremental costs and residual holdings.
• Test delayed or unavailable actions without using future information.

Source notes

1. Basel Committee: Stress testing, public consolidated guidance on static and dynamic approaches
2. Basel Committee: Supervisory and bank stress testing, range of practices, public publication summary
3. FSB: Final recommendations on liquidity preparedness for margin and collateral calls

Continue the analysis