Historical Stress VaR with Proxies and Current Portfolio Weights

Summary

The document discusses two practical choices in historical scenario analysis for stressed value at risk using a crisis period: what to do when a portfolio asset lacks price history for the selected scenario, and which portfolio exposures to apply. The response suggests representing an asset with a suitable proxy, such as proxy returns scaled by an estimated beta. It cautions that simply dropping an asset may be inappropriate, especially when it has meaningful portfolio weight, even if no proxy is a close match.

For exposure, the response recommends applying current portfolio weights rather than reconstructing the shares held during the historical crisis. It frames portfolio construction in terms of dollar allocations, so the stress calculation reflects today’s allocation to assets.

These are concise suggestions rather than a complete implementation guide. Proxy selection and beta estimation can materially affect results, and the document does not specify validation procedures or explain how to handle assets such as bonds beyond inviting that question.

Key ideas

• A proxy return scaled by an estimated beta can represent an asset with missing crisis-period data.
• Excluding an asset may distort stress results when its portfolio weight is material.
• Historical stress scenarios should generally be applied to current portfolio weights rather than past share counts.
• Proxy choice and beta estimation remain judgment calls that can affect the stressed VaR estimate.

Tags

# Historical Scenario analysis for stress testing


# Historical Scenario analysis for stress testing












I am doing historical scenario analysis in order to calculate stressed VAR for which I have taken 2007-2008 US crisis. I have two question in this regard:-

1) As we have to take prices for stocks with higher volatility during 2008 crisis, what if one of the stocks in my portfolio was not available or the prices were not available during that crisis? What we have to do in that case? Do we have to ignore that stock? Please suggest me for the other asset classes also like for bonds.

2) What market exposure I need to take for the respective asset class. I mean do I have to take calculated exposure during the 2008 crisis or the the today's market exposure?

P.S - I am using variance covariance matrix model for my calculation of VAR as well as stressed VAR.

## Answer by user18489 (score 2)

https://quant.stackexchange.com/a/24642

For stocks that do not have enough data during a historical period one approach would be to use a proxy (i.e. beta * proxy returns - in case that proxy returns = 0, then your proxy is cash). Depending on the weight of the asset in you portfolio, excluding the asset from the analysis may not be a good idea even if there isn't a perfect proxy for the particular asset.

Regarding the market exposure that you should use this is the current one (i.e. current weight of your assets rather than current # of shares). In portfolio construction you think of $$$ allocation rather than shares.

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.