Hi,Has anyone read this paper by Hull&White. Want to share their thought, especially about needing taking CVA, DVA functions in the bank to decide how to choose the discounting curve. Thankshttp://www.rotman.utoronto.ca/~hull/Downloadab ... RvsOIS.pdf

I have read it... and it is tricky to implement it due to the systems required to talk to each other... but implement people must if they have not already...

I think it's nonsense on so many levels.

QuoteOriginally posted by: TinManI think it's nonsense on so many levels.Tinman - you can't leave us hanging with a statement like that and now argument as to why

Well I only had a cursory glance at it, but his argument seems to be that we value using risk neutral valuation using the risk free rate, so both collateralisedand non collateralised trades are valued using OIS, neglecting prob of default.You then make CVA/DVA adjustments to find your risky MTM.Now I'm sure JH has forgotten more about RN valuation than I know, but it relies on a risk free cash account which doesn't exist.He uses OIS as a proxy but I think he misses the point.The proper way IMO is to use the replication argument, and allow for the fact that CSA and non CSA cashflows accrue at different rates. Theoretically non complete since you need convexity adjustments but there's no way around that for now.I also think he's got the cart before the horse with CVA, that's a far more back of the envelope calculation than marking a swap. I would rather take the impact from different curves than use CDS or similar to calculate a CVA. The idea that I would use something as flimsy as a CVA calculation to adjust a MTM I got from transparent and mostly liquid market rates is not sensible to me.

I found myself feeling angry reading this paper. Not because of the material, but sadly for a lack of it. Where were the fancy Brownian Motion gymnastics, or at least a cursory hommage to Karatzas and Shreve? Why were there no fancy change of numeraire arginments to demonstrate the authors mathematical brilliance? How DARE they expect people to care when the paper could (on a base level) beunderstood by undergraduates? It's disgraceful - Hull should be stripped of authorship of the Hull books and forced to do REAL research for I Karatzas for the remainder of his (Hull that is) 's productive life! Rant over.

I appreciate the irony of EscapeArtist999, and I was a little disappointed by TinMan, since the paper by HW contains some non-senses, but they are not those mentioned by TinMan.I agree with HW that everything has to be discounted withthe risk-free rate, that is simply the way to give the value of money in time, not the rate of a risk-free deposit that does not exist any more (you are right TinMan on this point, but you are wrong when you say it is not correct to use the risk free rate to discount cash-flows). I disagree with them on some statements, honestly embarassaing when coming from researchers like HW, such as:"It is becoming standard market practice to use the OIS curve for discounting when collateralized portfolios are valued and the LIBOR/swap curve for discounting when non-collateralized portfolios are valued." (pag. 10). I do not know in Canada or in US, but my experience with the European financial market is showing something a little more sensible than such a supposedly standard market practice. This simply shows that either HW talk with very un-skilled traders or that they do not have any contacts with the real world."A natural question is whether it is necessary to calculate CVA and DVA at all. Can we adjust for credit risk by adjusting the discount rate? The appendix shows that this is possible in three special cases." (pag 14), this is true but it is just in very specific cases, such as a loan (very difficult to apply to general derivatives pay-offs). In any case one should assume also that the recovery implied in the Libor is the same as that used for the CVA (which is not usually the case) and that Libor is the rate corresponding to the bank's credit risk.I showed in some recent works that DVA cannot be replicated (see here): HW still spport the Brugard and Kjaer argument that I proved to be false under an economic perspective, altough correct under an abstract mathematical point of view; I also studied which is the relationship between CVA/DVA and funding spread (see here). For the OIS/risk free disocunting and the pricing of derivatives under collateral agreements see here.I think that once one starts thinking seriously on these issues, the HW's paper appears a very naive work.

QuoteOriginally posted by: ancast I agree with HW that everything has to be discounted withthe risk-free rate, that is simply the way to give the value of money in time, not the rate of a risk-free deposit that does not exist any more (you are right TinMan on this point, but you are wrong when you say it is not correct to use the risk free rate to discount cash-flows). I don't think I am. If you want to compare cashflows across periods you need to take account of how you fund those cashflows. $1m collateralised and paid to me today in collateral is worth more than one which is not. I'll discount that at the rate I have to fund it.

I like to discount everything at the risk free rate, by separating the return on collateral (given by the collateral rate) and funding costs, and expected losses for credit risk, everything disocunted at the risk free rate and added to the total value of the contract.I do not like effective rates including credit risk / funding effects and collateral effects. By using effective rates you cannot split the total value of a contract into the several contibutions to be allocated to different desks in the trading room.For more on that, please read the works whose links are in my post before.

QuoteOriginally posted by: ancastI appreciate the irony of EscapeArtist999... _ Glad somebody got it.

QuoteOriginally posted by: ancast I like to discount everything at the risk free rate, by separating the return on collateral (given by the collateral rate) and funding costs, and expected losses for credit risk, everything disocunted at the risk free rate and added to the total value of the contract.I do not like effective rates including credit risk / funding effects and collateral effects. By using effective rates you cannot split the total value of a contract into the several contibutions to be allocated to different desks in the trading room.For more on that, please read the works whose links are in my post before.Can you hedge all of those elements? If not, how do you justify using a 'risk free rate'?Only risk free cashflows should be discounted using a 'risk free rate'.In any event banks don't really use a 'risk free rate', if they did, they'd discount with Bunds.The reason for using Libor was that it was the assumed funding rate, plus there was no real alternative.On the point of decomposing the value to attribute to different desks, I don't see the issue.The problem is valuing the contract in the first place. Any bank should be able to value a book with different curves to isolate credit or funding elements.

QuoteOriginally posted by: TinManQuoteOriginally posted by: ancast I like to discount everything at the risk free rate, by separating the return on collateral (given by the collateral rate) and funding costs, and expected losses for credit risk, everything disocunted at the risk free rate and added to the total value of the contract.I do not like effective rates including credit risk / funding effects and collateral effects. By using effective rates you cannot split the total value of a contract into the several contibutions to be allocated to different desks in the trading room.For more on that, please read the works whose links are in my post before.Can you hedge all of those elements? If not, how do you justify using a 'risk free rate'?Only risk free cashflows should be discounted using a 'risk free rate'.In any event banks don't really use a 'risk free rate', if they did, they'd discount with Bunds.The reason for using Libor was that it was the assumed funding rate, plus there was no real alternative.On the point of decomposing the value to attribute to different desks, I don't see the issue.The problem is valuing the contract in the first place. Any bank should be able to value a book with different curves to isolate credit or funding elements.I can think of a couple of things I'd like to do with bunds, but discounting isn't one of them. That being said, I'm not a bank and don't plan on getting a banking license any time soon.

TinMan,best proxy of risk-free rate is ois. Interest rate is economically speaking the intertemporal rate of preference of current consumption to future consumption. In a keynsian economy this equals in equilibrium the marginal return of the capital invested in the production process.You can use this rate (that is simply "the rate", not the risk-free rate: "risk-free" is superflous and misleading) to discount or compound cash flows (ie: nove them to and fro in time). SO you can discount cash-flows related to the contract, to the expected losses (eg: credit triggerd) and other costs (eg: funding costs). In the end, summing up everything, the value of the contract will be the SAME as using effective rate curves, eg. the libor curve, if the creditwothiness of the obligor is the same as that implied by LIBOR.If you do not see which is the issue in disentangling the total value of the contract in the single components, that means you are working in a lower tier bank. In that case it is understandable why you cannot catch my thoughts, that are admittedly a little advance under a conceptual point of view.If you keep on referring to dynamic replication, I am afraid you won't go to far in the current market environment. Things are just a little more complex than the mathematical trick of the dynamic replication.HW are wrong not becasue they do not use rocket science math, but beacuse they do not understand what is actually happening in the market and its mechanincs in the current days.

QuoteOriginally posted by: ancast Interest rate is economically speaking the intertemporal rate of preference of current consumption to future consumption. In a keynsian economy this equals in equilibrium the marginal return of the capital invested in the production process.Seriously, that's your argument?QuoteOriginally posted by: ancast If you do not see which is the issue in disentangling the total value of the contract in the single components, that means you are working in a lower tier bank. In that case it is understandable why you cannot catch my thoughts, that are admittedly a little advance under a conceptual point of view..Yes, I'm sure that is the problem, your thoughts are too advanced for me.QuoteOriginally posted by: ancast If you keep on referring to dynamic replication, I am afraid you won't go to far in the current market environment. Things are just a little more complex than the mathematical trick of the dynamic replication. I didn't say they weren't, but thanks for the concern.

QuoteOriginally posted by: ancastFor more on that, please read the works whose links are in my post before.Hard to do since the links are not working