Charta is on Google Play: practise on the go. Get the app
Charta

Characterise the market-driven scenario (MDS) approach to building stress test scenarios for investment portfolios and judge its merits.

FRM® Part II · Risk Management and Investment Management · Market-Driven Scenarios: An Approach for Plausible Scenario Construction · FRM2.IM.15.a

The idea

The market-driven scenario (MDS) approach builds a stress scenario for an investment portfolio by shocking only a few key market variables, such as an equity index or a credit spread. Every other variable then moves by the amount that historical co-movement implies, namely its conditional expectation given the shocks. The aim is a scenario that is plausible and internally coherent: equities do not crash while credit spreads sit still unless the data say that has happened. Merits are consistency with observed market relationships, the ability to explore events that never occurred exactly as stated, and a transparent link between narrative and numbers. Limits are dependence on the estimated covariance, a roughly linear and normal framework, and the fact that plausible is not the same as probable. On the exam, contrast MDS with an ad hoc scenario that shocks one factor and ignores the rest, and with a pure historical replay.

A practice question

From the question bank, with the reason each wrong option is wrong.

Which is the main merit of the MDS approach compared with a scenario in which an analyst shocks a few variables and leaves all the others unchanged?

  1. It gives the exact probability that the scenario will occur within the horizon.Why not: Treats a plausibility measure as a forecast; MDS does not deliver an exact probability of occurrence.
  2. It removes the need for any judgment about which variables to shock or how large the shocks should be, as the covariance matrix decides it all.Why not: Ignores that the analyst still chooses the narrative, the shocked variables and the shock sizes.
  3. Unshocked variables move consistently with historical relationships, so the scenario is coherent. - correct
  4. It guarantees that the portfolio loss is the largest that could occur at a given confidence level.Why not: Confuses a plausible scenario with a worst-case loss bound or VaR-type guarantee.

The benefit is coherence: leaving the other variables at zero is itself an implausible assumption, while MDS sets them at conditional expectations. It does not produce probabilities or remove judgment.

A harder one

An analyst stresses a multi-asset portfolio with an equity fall of 30% and leaves credit spreads, rates and exchange rates unchanged. A risk manager notes that, historically, large equity falls coincided with spread widening and falling government yields. What is the weakness of the scenario and the MDS remedy?

  1. There is no weakness, because unshocked variables should be held at zero so that the loss is attributed only to the equity shock.Why not: Treats zero as a neutral assumption; it is in fact a strong and unlikely assumption about co-movement.
  2. The scenario ignores co-movement; MDS keeps the equity shock and sets the other variables at conditional expectations. - correct
  3. The equity shock is too large, and the remedy is to shock every variable by the same number of standard deviations.Why not: Confuses coherence with uniform shock sizes; equal z-scores everywhere ignores the signs and sizes of correlations.
  4. The scenario is too mild, and the remedy is to add the historical worst move of each other variable.Why not: Stacks worst-case moves independently, which ignores correlations and gives an implausible scenario.

MDS keeps the narrative variable fixed and derives the others from the covariance matrix, so spread widening and falling yields accompany the equity fall in the proportions seen historically.

The hardest tier

A fund's main concern is a sharp rise in yields accompanied by rising equity prices, a combination that has been rare in the recent sample. Which assessment of the MDS approach versus a historical replay is most accurate?

  1. MDS cannot be used, because it can only replay episodes that actually occurred.Why not: Confuses MDS with a historical replay; MDS lets the analyst choose hypothetical shocks.
  2. MDS is free of historical data once the shocks are chosen, so the distance becomes irrelevant.Why not: Overlooks that conditional expectations and distances are computed from the historical covariance matrix.
  3. MDS can build unseen scenarios from chosen shocks, but the other moves rely on historical covariance, so check the distance. - correct
  4. A historical replay is preferable, because it is the only approach that keeps the other variables coherent with each other and with observed market co-movement.Why not: Reverses the logic: MDS enforces coherence through covariance, whereas a replay is limited to past episodes.

Hypothetical shocks are a merit, as the other variables are still set from estimated relationships. Because that history has little such data, the scenario should be assessed with its Mahalanobis distance.

In the app: 6 practice questions on this outcome, spaced review that brings it back before you forget it.

Start free →

More in Market-Driven Scenarios: An Approach for Plausible Scenario Construction