- theblackcarpet
**Posts:**10**Joined:**

Hi,I'm implementing Value At risk for a bond portfolio ; I'm looking for a simple way to do this since I'm only using Excel. The portfolio is composed with Senior and Sub (also tier1/2) debt hedged against interet rate risk ; I think the most suitable VaR may be Monte Carlo's VaR.Any ideas ? Guidebook ? Thanks in advance,

superficially your problem sounds simple but I think its more complicated.You could do a quick and dirty solution using the duration of all your bonds. But I suspect that the risk in your portfolio is credit risk - this might be more quite difficult to handle even in a monte carlo

knowledge comes, wisdom lingers

QuoteOriginally posted by: daveangelYou could do a quick and dirty solution using the duration of all your bonds. But I suspect that the risk in your portfolio is credit risk - this might be more quite difficult to handle even in a monte carlothe two obvious elements are credit risk and interest-rate risk: for interest rate risk, he said he was hedged, but that may not mean 100% hedged.For credit risk, you can get a crash course by reading the section on credit risk in either the PRM Handbook or the FRM Handbook.Two main worries: default risk and "migration risk" (the credit rating of the issuer changes) and you can get the probabilities of those events from a ratings agency

QuoteOriginally posted by: ppauperQuoteOriginally posted by: daveangelYou could do a quick and dirty solution using the duration of all your bonds. But I suspect that the risk in your portfolio is credit risk - this might be more quite difficult to handle even in a monte carlothe two obvious elements are credit risk and interest-rate risk: for interest rate risk, he said he was hedged, but that may not mean 100% hedged.For credit risk, you can get a crash course by reading the section on credit risk in either the PRM Handbook or the FRM Handbook.Two main worries: default risk and "migration risk" (the credit rating of the issuer changes) and you can get the probabilities of those events from a ratings agencytrying to pick a fight Poops ?ratings agencies are as useful as chocolate teapots ....

knowledge comes, wisdom lingers

- theblackcarpet
**Posts:**10**Joined:**

Hi,Yes i agree that Rating agencies follow credit spreads instead of anticipating them. I just want to calculate the VaR concerning credit issues (i consider that my portfolio is "fully" hedged against interest rate risk).I don't understand why you guys are speaking about default risk ?

- theblackcarpet
**Posts:**10**Joined:**

What are the PRM and FRM handbook ?

Last edited by theblackcarpet on October 15th, 2009, 10:00 pm, edited 1 time in total.

- theblackcarpet
**Posts:**10**Joined:**

QuoteOriginally posted by: daveangelsuperficially your problem sounds simple but I think its more complicated.You could do a quick and dirty solution using the duration of all your bonds. But I suspect that the risk in your portfolio is credit risk - this might be more quite difficult to handle even in a monte carloI already calculated a credit sensitivity (and i agree with you that for fixed income bonds, it is closed the duration/modified duration) but the V@R(1,99) for example reflects what could be at most the loss in one day, in 99% of the cases on your bond ? i'm right ? With sensitivity/duration you can only say what would be the value of your bond if the credit spread moves from x%

QuoteOriginally posted by: theblackcarpetWhat are the PRM and FRM handbook ?there are 2 (more or less identical) qualifications for risk managers, offered by 2 risk management societies:prmia offers the prm (professional risk manager) qualificationgarp offers the frm (financial risk manager) handbookthe two handbooks are the core reading for the 2 qualifications, and will take you through an outline of credit risk management.One way to do credit risk is to look at the probability an issuer will migrate from one rating class to another, which will affect the price of its debt ("migration risk"), along with the probability that the issuer will default altogether.There may be other ways to do credit risk, but the above is fairly simple, once you have the matrix of probabilities, which you have to get from somewhere (either ask a rating agency or take daveangel's advice and ask a chocolate teapot)

Quote There may be other ways to do credit risk, but the above is fairly simple, once you have the matrix of probabilities, which you have to get from somewhere (either ask a rating agency or take daveangel's advice and ask a chocolate teapot) Credit migration as a risk analysis tool went out with the ark. just becuase its simple doesn't make it right. and you will find that it is your approach that is tantamount to asking a chocolate teapot.

Last edited by daveangel on October 15th, 2009, 10:00 pm, edited 1 time in total.

knowledge comes, wisdom lingers

- theblackcarpet
**Posts:**10**Joined:**

Thanks for explanation, but what about VaR ?I have read that my only risk factor (in my case) is credit spread so with Monte's Carlo methodology i have to "make the risk factor move" with for example 10k simulations, using a gaussian and i get the new bonds price. Does it speak to you ?thanks,

QuoteOriginally posted by: daveangelCredit migration as a risk analysis tool went out with the arkI see you're stipulating to the existence of Noah's ark

You don't say why you are computing VaR or what data you have.If you have a diversified bond portfolio with the interest rate risk hedged, the VaR will likely be determined by economy-wide credit shifts. There's plenty of historical data on this, with lots of models.If you have a less diversified portfolio, or if the interest rate risk is not fully-hedged, then specific factors may be more important. You will have to model these.It's hard to see what good a Monte Carlo simulation would do. The main purpose of Monte Carlo is to reduce dimensionality. You have a one-dimensional problem.

Yes the most suitable VaR is Monte Carlo or historical value at risk. The var-covar approach is not suited since bonds behave nonlinear with respect to interest rates. You could of course linear this, base don the duration and still use the var-covar VaR

you are not saying why you want to compute VaR for a bond portfolio ?are these Bonds in your trading book ?As mentioned in previous posts ,depending on your answer to the above questions ,you problem might be more complex !HS VaR will suffice IFF to cover the risks that fall within the VaR parameters;if you are computing VaR for Capital requirementsthen your problem is much more complex,read about IRC (Incremental Risk charge ),if the latter is the requirement ,then you need a complete framework tocover the folloing Risks :-default risk-transition risk-recovery risk-default and migration correlation risk-hedge roll-off risk-Basis risk.-liquidity risk-concentration risk.

GZIP: On