CDS vs Corps recovery rates

External question — Quantitative Finance Stack Exchange Author: Dante Original post: https://quant.stackexchange.com/questions/85858 License: CC BY-SA 4.0 — https://creativecommons.org/licenses/by-sa/4.0/ Adaptation: HTML converted to plain text; contact email addresses removed. Doing some work where I'm calculating expected outgo (not market implied) on portfolio of CDS contracts. For simplicity I'm using in-house probabilities/recovery rate assumptions derived from corps but being told it's not appropriate due to recovery functioning differently between CDS and underlying physical bond. Wondering if anyone is aware of any analysis done on payouts for CDS (protection seller) versus underlying? (or where I could even source this) From my understanding CDS is decided via ISDA auction which may differ from what occurs in bond markets. I would like to think these are broadly equivalent however or else they wouldn't do a great job at hedging, but I don't have any quantitative argument to back it up. At the auction stage could you not just get the physical bond at which point you are exposed to the actual bankruptcy proceeds? Would appreciate any help. Thanks

External answer — Quantitative Finance Stack Exchange Author: Dimitri Vulis Original post: https://quant.stackexchange.com/a/85859 License: CC BY-SA 4.0 — https://creativecommons.org/licenses/by-sa/4.0/ Adaptation: HTML converted to plain text; contact email addresses removed. Let's consider how bonds and CDS trade when the issuer is very distressed before a default, and also after a default. The bonds trade not on yield or spread, but rather on price, which is is the view on the recovery. The probability of default is effectively 100%. As far as I know, few people consider the time value of money, i.e. how many years it might take the creditor to recover something in bankruptcy proceedings. Defaulted bonds can continue trading, with little volume, until the creditor gets something, which typically takes years. The CDS also trades on upfront, rather than spread. Since the probability of default is effectively 100%, the upfront fee is the view on the recovery determined in the CDS auction. CDS mostly stops trading after the auction, but can be very active between the default and the auction. These are two different markets with different participants and different supply and demand. Not surprisingly, they might diverge a little short term, but converge long term. The most extreme divergence I know was Delphi's 2005 default, when some people were chasing the defaulted bonds driving their price well above the expected auction outcome. I've seen three good papers that discuss actual CDS auction results and try to understand why they differ from bond prices sometimes: Feldhütter, P., Hotchkiss, E. S., & Karaşaka, O. (2014). The Impact of Creditor Control on Corporate Bond Pricing and Liquidity. - quantifies the difference in the bond price and an equivalent synthetic bond without control rights constructed using CDS. Helwege, J., Maurer, S., Sarkar, A., & Wang, Y. (2009). Credit Default Swap Auctions. - historical analysis of the difference between defaulted bond prices and CDS auction results. A little dated, since more auctions happened since 2009. Sultanum, B., Paulos, E., & Tobin, E. (2019). CDS Auctions: An Overview. - newer historical analysis of the difference between defaulted bond prices and CDS auction results. So, its not a realistic assumption that these 3 numbers: the physical recovery in bankruptcy proceedings, the market price of the distressed/default bond, and the recovery on CDS from upfront/auction - will be the same. Economically they should be close, but short-term they can diverge. Just making up some numbers as an example to show how they could be all over the place: maybe a few days before the CDS auction, the defaulted bonds are trading at 55, but the CDS trade at upfront fees implying that the auction result would be 50... or 60. And then the CDS auction happens and the result could be 45. All the while, the economic models looking at the reference entity's assets and liabilities predict that the physical recovery would be 35, so the defaulted bonds trade around there after the auction... but a couple of year later the bondholders are pleasantly surprised to recover 70. A better practice is to assume spreads between them, and to see what the impact might be of these spreads widening a lot (for a short time), and possibly even keeping reserves for them.