Managing CDS Index Rebasings in Historical VaR and Expected Shortfall Scenarios

Summary

The document examines how twice-yearly rebasing of credit default swap index spreads can create artificial jumps in historical data. Those technical moves may enter the loss tail and distort historical value at risk or expected shortfall, including stressed estimates.

It considers leaving the jumps in place, additively adjusting earlier observations, or multiplicatively adjusting them. Keeping the raw series is simple but leaves risk measures exposed to artificial scenarios. Additive adjustments remove jumps while preserving past absolute changes, but can produce implausible historical spread levels that drift with repeated rebasing. Multiplicative adjustments also remove jumps, yet alter historical scenario sizes in ways the author considers unrealistic. The discussion offers no preferred solution or tested comparison, so it frames the modeling tradeoff rather than resolving how to construct representative risk scenarios.

Key ideas

• Periodic CDS index rebasing can introduce artificial spread jumps into historical scenarios.
• Artificial jumps can affect tail-based VaR and expected shortfall estimates.
• Leaving jumps in the data may contaminate risk measures if a technical move enters the loss tail.
• Additive adjustment preserves past absolute moves but can create unrealistic historical levels.
• Multiplicative adjustment removes jumps while changing the scale of past scenarios.

Tags

# Historical scenarios for VaR/ES of CDS indexes spread


# Historical scenarios for VaR/ES of CDS indexes spread












Historical series of CDS indexes spreads rebase twice per year. These events introduce two "technical" scenarios (even 10 bps, not marginal) which could affect historical market risk measures (e.g. VaR/ES). The question is: how to manage these technical jumps to avoid unintended impacts on historical market risk measures?

## Answer by Micio Geremia (score 0)

https://quant.stackexchange.com/a/84100

I attempt a list of possible solutions.

- do nothing and keep the jumps, hoping they will not fall in the P&L tail and inflate VaR/ES? simple solution but crossing fingers is not sound risk management, also because these jumps could, one day, fall in the tail of stressed var.

- additive rebasement of the past series? this would clear the jumps leaving past additive scenarios unchanged, but would also change the level of the past series, twice per year, accumulating unrealistic past levels

- multiplicative rebasement of the past series? this would clear the jumps but also change past scenarios in an unrealistic way.

- other?

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.