How Negative VaR-Equivalent Volatility Can Arise
Summary
The document asks whether the VaR-equivalent volatility measure used in PRIIPs key information documents can be negative and what such a value means. The answer says it can occur when the loss at the measure’s 2.5th-percentile threshold is actually a gain. In that case, the risk measure’s percentile input reflects an outcome above the initial capital rather than a loss.
A hypothetical product illustrates the point: most proceeds are used to fund a coupon, while a low-probability outcome wipes out the investor’s capital. The answer says this construction can nevertheless receive the lowest market-risk category under the described measure. This brief example highlights a limitation in interpreting the metric: a favorable percentile does not rule out severe tail loss elsewhere in the distribution. No derivation, broader regulatory context, or additional examples are provided.
Key ideas
- A negative VaR-equivalent volatility can arise when the measured loss percentile represents a gain.
- The answer illustrates this with a product that has a rare total-capital-loss outcome and a coupon funded by the proceeds.
- A favorable percentile-based VEV can coexist with a severe loss in another part of the distribution.
- The document provides a short explanation and example but no derivation or broader assessment of the measure.
Tags
Full text
# Negative VaR equivalent Volatility (VEV) and its meaning? # Negative VaR equivalent Volatility (VEV) and its meaning? Can a VaR equivalent Volatility (VEV) as defined by KID/PRIIPS law be negative and what does it mean if it has a negative value? ## Answer by James Spencer-Lavan (score 1) https://quant.stackexchange.com/a/40377 You can attain negative VEV if the "loss" measured at the 2.5% percentile is actually a gain. I build you a product where you lose all capital with 1% measured probability and pay you a coupon with the proceeds. Under VEV, it will attain a MRM of 1 (I.e. cash-like)
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.