Parametric Var model

[mrpmkt]

I have a question on Marginal VaR decomposition. I am referring to example IV.2.15, The Risk factors are S&P500, FTSE etc. You have shown the marginal VaR decomposition into these risk factors. How can I further decompose the Marginal Var into the actual positions of the portfolio. E.g. if the portfolio were to hold, MSFT, INTC and BP.L as some of the positions.
Can I decompose Marginal VaR to the lowest level? Is it based on the weights these positions have in the risk factors?

Thanks

[coalexander]

Hi

You can perform the marginal VaR decomposition at any level, and drill down into individual securities, but if you want an exact hierarchy of VaR to aggregate up from the individual security level, through risk factors and eventually to entire portfolio, you need to define the risk factors as the actual portfolio of the securities that you hold.

For example, if you hold just 3 UK stocks with returns A, B and C, then for the exact marginal VaR decomposition you need your UK equity risk factor to be X = w_a*A+w_b*B+w_c*C, where w_a is the proportion of total capital that is invested in the stock with return A, and X is not s standard (FTSE-type) risk factor.

Hope this helps, Carol

[mrpmkt]

Thank you, this make sense.

[FelixTanudjaja]

HI, my name Felix
i came from Indonesia UPH business collage.
i want to ask about parametric, can i calculating VaR for fix income(bond market)??
is it right to used modified duration for sensitivity risk?
and how about forecasting volatility of yield using GARCH(1,1)? should i used return of yield(LN(yield t/yield t-1))?
can i used "proc; make garch variance series" from eview-6?
Thanks.

[coalexander]

Hi Felix

Parameteric VaR for cash flows is based on a PV01 sensitivity. This is fully explained in Section IV.2.3. Of course, I am happy to answer any specific questons you have about this section.

Interest rate volatility (GARCH or otherwise) is usually based on daily changes, but if interest rates are extremely high, some people might prefer to measure volatility of log changes, because when interest rates are high relative changes are more stable than absolute changes.

Best Wishes, Carol