Cannot believe this, more than 120 guys have visited this forum so far and only a handful of answers given ...

QuoteOriginally posted by: CreditGuyOk, you are trying to get around my question ... 100% correlation means all 5 bonds default simultaneously, but you are pricing this today, then in reality you do not know, maybe in reality correlation will be much lower and only will default, so your hedge will be imperfect after de fact... also, default correlations are not typically "simultaneous" in the way you are proposing ... I want to leave the domain of "artifical correlations" and come back tio the real world: how do you hedge recovery risk in 1st-to-default contract? Typically you will concentrate yourhedges on the widest names but you do not know which one will default and these very wide names can have a very different recovery rate... I want to ring up your desk, it's been a slow month for me (vacation and all) and I really need to kick out the jambs. I can understand why you may like to look at the problem this way, but it really has very little to do with anything a practitioner comes near. Assuming that correlation is 100% what is the value of FTD protection?? Who buy's FTD protection when they think correlation is 100%?? Couldn't you find someone to sell the risk to and still make a profit if you could convince them correlation is 100% and you bought it when correlation was assumed to be 30%?? As ASmith has said it depends on the term sheet wording, when everything defaults simultaneously how do you decide what is the first to default. If default correlation is 100% then assuming a world where credit spread is strictly a function of default probability and recovery rate then any difference in spreads should be attributable to difference in recovery rates. If it's worded that you get the credit that causes you the most loss, figure out what the relative recovery rate is assumed to be for each credit, factor in your exposure to each credit and hedge out the one that causes you the most loss. If you get the average loss then put on your hedges to the single names in proportion to your notional exposure. (ie. 20% per name in the 5 credit case).If you can't with certainty say that the one that causes you the most loss today will be the one that casuses you the most loss in the future, I can agree with that. However that's what you run into heding an FTD with a single name, it's called basis risk and you can't get rid of it. You don't happen to work at a certain large (but getting smaller german speaking credit player??) do you?? For the longest time they were of the belief that FTD protection sellers were not exposed to correlation risk. Seriously, I'm not kidding on this. Crazy, no??

Yes I work there, and you also work around here too, dude! Or you work for LEH maybe ?However, I want a much cleaner and more elegant solution to my question: how do you deal with recovery risk in a first-to-default basket? Pls leave aside the 100% correlation thing from now onwards...QuoteOriginally posted by: RowdyRoddyPiperQuoteOriginally posted by: CreditGuyOk, you are trying to get around my question ... 100% correlation means all 5 bonds default simultaneously, but you are pricing this today, then in reality you do not know, maybe in reality correlation will be much lower and only will default, so your hedge will be imperfect after de fact... also, default correlations are not typically "simultaneous" in the way you are proposing ... I want to leave the domain of "artifical correlations" and come back tio the real world: how do you hedge recovery risk in 1st-to-default contract? Typically you will concentrate yourhedges on the widest names but you do not know which one will default and these very wide names can have a very different recovery rate... I want to ring up your desk, it's been a slow month for me (vacation and all) and I really need to kick out the jambs. I can understand why you may like to look at the problem this way, but it really has very little to do with anything a practitioner comes near. Assuming that correlation is 100% what is the value of FTD protection?? Who buy's FTD protection when they think correlation is 100%?? Couldn't you find someone to sell the risk to and still make a profit if you could convince them correlation is 100% and you bought it when correlation was assumed to be 30%?? As ASmith has said it depends on the term sheet wording, when everything defaults simultaneously how do you decide what is the first to default. If default correlation is 100% then assuming a world where credit spread is strictly a function of default probability and recovery rate then any difference in spreads should be attributable to difference in recovery rates. If it's worded that you get the credit that causes you the most loss, figure out what the relative recovery rate is assumed to be for each credit, factor in your exposure to each credit and hedge out the one that causes you the most loss. If you get the average loss then put on your hedges to the single names in proportion to your notional exposure. (ie. 20% per name in the 5 credit case).If you can't with certainty say that the one that causes you the most loss today will be the one that casuses you the most loss in the future, I can agree with that. However that's what you run into heding an FTD with a single name, it's called basis risk and you can't get rid of it. You don't happen to work at a certain large (but getting smaller german speaking credit player??) do you?? For the longest time they were of the belief that FTD protection sellers were not exposed to correlation risk. Seriously, I'm not kidding on this. Crazy, no??

QuoteOriginally posted by: CreditGuyYes I work there, and you also work around here too, dude! Or you work for LEH maybe ?However, I want a much cleaner and more elegant solution to my question: how do you deal with recovery risk in a first-to-default basket? Pls leave aside the 100% correlation thing from now onwards...No dice on Lehman, nice try though. Anyway so now the question is stepping away from 100% correlation to recovery risk on the FTD?? Okay then let's stick with that. What's the question then?? Assume two of your 5 default simultaneously and you get stuck with the biggest loss of the two. You happened to hedge out what you thought the largest loss would be on a single name basis and were incorrect. Okay so now your hedge didn't perform and your boss is up your ass over it?? How would I avoid this situation?? I would not hedge. Seriously nothing is worse than a hedge that doesn't perform. People will ridicule you for it, mock you, potentially badmouth your family. Smart thinking wearing that bulky, hot kevlar vest, unfortunately they came after you with a BASEBALL BAT!By the way it appears your only way to frame your argument is to keep default probability 100% correlated, because it highlights the fact that while your default probability may not change in between names you still can have a hedge that doesn't perform because recovery expectations can change between names. I suppose that's what you're trying to get at. If you're really that concerned with recovery risk make the FTD payout binary and hedge out your worst credit with a binary single name?? Maybe I'll set up the NEGATIVE CARRY MASTER TRUST to deal with this problem?? FTD Obligations, Callable Short Credit Positions, continually reissuing bonds backed by the proceeds unexpectedly correlated defaults?? Sweet my friends, who want's to service this thing??

CreditGuyIf you assume that recovery rates are deterministic, you need to hedge according to the probability weighted average recovery rate, where the "probability" is the risk-neutral probability of first-default of each asset under whatever process you're assuming. A special case of this is given by your original question in which the probabilities of first-default were all 20% as default on one implied default on all. Another illuminating example would be where one asset was very close to default. In this case the hedge would be based almost entirely upon the assumed recovery rate for that asset.I still want to know what my prize is.

you win a stack of certs issued by the n.c.m.t. mentioned belowactually I was halfway thru posting exactly what you said but then I got distracted... but then I thought we were looking at the sensitivity to this contract due to an upshift of the corr from 100% to 110%.the only reason to revisit this thing is to raise this Altman thing again about correlated market-wide spreads and recovery rate... a bell starts to chime in this office every time somebody says 'ok so what... ead=pdxlgd'.

QuoteOriginally posted by: kryou win a stack of certs issued by the n.c.m.t. mentioned belowactually I was halfway thru posting exactly what you said but then I got distracted... but then I thought we were looking at the sensitivity to this contract due to an upshift of the corr from 100% to 110%.the only reason to revisit this thing is to raise this Altman thing again about correlated market-wide spreads and recovery rate... a bell starts to chime in this office every time somebody says 'ok so what... ead=pdxlgd'.We have anti-nerds too, people pay us to get the tax shield. Come to think of this does this exist??

"you win a stack of certs issued by the n.c.m.t. mentioned below"Seeing as you were going to post the same answer as I did, I'll go halves with you on the prize. You can have the first half (nc) and I'll have the rest. I'm thinking of spending my half funding a nerds sanctuary.

Why don't you just delta hedge expected loss? take the product of default prob and loss given default for the basket, calculate a dv01 of this product, and use that to calculate your hedge?

"Why don't you just delta hedge expected loss?"That's what I said ... and you can't share the prize 'cause I've already given half to kr.

So the question is to find an elegant way to hedge a FtD (correl<>100%) taking the recovery risk into account. Mm.. I would first like to see an ‘elegant’ way to hedge a FtD with correl <>100% assuming deterministic and homogeneous recovery rates! As far as I know, the hedge (deltas) within any model you choose will depend on the expected recovery rates. For example, for a 2 name FtD with R1=20% and R2=80%, delta1 will be higher and delta2 will be lowerthan if the two names had R1=R2=50% recovery (everything else equal).

QuoteThat's what I said ... and you can't share the prize 'cause I've already given half to kr.Damn that time zone thing. When are we going to one global time for the whole world?

DoubleV, you are cute, you have just won a candlelight dinner with me, but your answer ... maybe I will explain to you how certain things work... QuoteOriginally posted by: doubleVSo the question is to find an elegant way to hedge a FtD (correl<>100%) taking the recovery risk into account. Mm.. I would first like to see an ‘elegant’ way to hedge a FtD with correl <>100% assuming deterministic and homogeneous recovery rates! As far as I know, the hedge (deltas) within any model you choose will depend on the expected recovery rates. For example, for a 2 name FtD with R1=20% and R2=80%, delta1 will be higher and delta2 will be lowerthan if the two names had R1=R2=50% recovery (everything else equal).

let me, in a hackneyed fashion, say that this exercise is mental masterbation. First of all, the existence of 5 credits having perfect default correlation is practically nil. Secondly, the methods that one would arrive at to say that 5 credits are perfectly correlated are probably highly suspect. Thirdly, this is obviously a stupid interview question, in which the interviewers have some answer that they like in their pockets and are waiting for you to produce the same answer. If I were interviewed and they asked me this, I would answer it on the condition that they answer MY question about hedging counterparty risk, and, if they got the answer wrong.....I would make them feel uneasy with a few raised eyebrows, a few glances to my loudly ticking timepiece, and then I would say that I have an important meeting to get to...of course, I would smile and thank them for their time.