By Walter Kurtz, Sober Look
As discussed earlier, we’ve had a substantial divergence between investment grade (IG) and high yield (HY) CDX spreads that started this year. These two indices historically move in tandem. There has been speculation in the market place that the JPMorgan CIO’s activities may be behind this trend. Based on the recent article by Euromoney, that indeed seems to be the case. Euromoney indicates that not only was JPMorgan selling IG CDX protection, it was buying HY CDX protection causing the two to diverge. Effectively JPM was long investment grade credit and short HY credit.
Euromoney: – It is clear that JPMorgan’s CIO sold substantial amounts of investment-grade credit default swap index exposure in the first quarter of the year; market participants maintain that it also bought high-yield default swap protection, with total notional trade sizes running to tens of billions of dollars.
Those trades were unusually large for the credit derivatives market, despite their concentration in indices, which are more liquid than single-name default swaps. The JPMorgan activity helped to fuel the global rally in investment-grade credit spreads in the first quarter and contributed to a widening in the ratio between investment-grade and high-yield spreads, taking the latter to a multiple of roughly six times the former.
This is not necessarily a bad position to hold, except for it’s size. The thinking probably was that if the economy stumbles, HY companies will get hurt first and their spreads will widen much faster than those of IG companies (making the spread between the two to widen). But liquidity in CDS markets has not been great in recent years and these large trades began to move the market earlier in the year – initially helping JPMorgan’s position.
|IG vs HY CDX (Bloomberg)|
But now that the market participants got a rough idea of what JPMorgan’s exposures are, they started putting on the opposite trades in anticipation of JPMorgan unwinding this book (which may already be taking place). The spread between HY and IG CDX has narrowed substantially since JPM’s announcement last week, causing the firm further pain (remember, JPM needs the spread to widen). The vultures are circling…
Euromoney: – The exact details of the trades put on by the JPMorgan CIO have not been disclosed. JPMorgan is understandably unwilling to shed additional light on its holdings, while its market counterparties such as hedge funds have limited visibility on offsets to individual trades, along with a strong motive to talk their own books by speculating about potential eventual deal unwinds.
And now everyone is asking the same question. JPMorgan had fairly thorough VAR models that should have shown a potential for a large loss on these positions. Why was this risk not flagged?
If you look at the chart above (IG vs. HY spreads), it is clear that the historical relationship (on which VAR models are based) has been broken by these outsize trades. Once you have a sudden change in correlation, the model needs to be re-calibrated. For spread positions this large and correlation that starts out close to one (which masks risks of large long/short positions), even a small change in correlation will have an enormous impact on the perceived amount of risk.
|Illustration of how a spread position between two assets responds to changing correlation|
As the mass media begins to pore over the issue of JPMorgan’s VAR models, the reporters will surely miss this one critical point. VAR models generally can not capture liquidity risk. In particular it is difficult to model how outsize trades can impact correlation. Once positions become too large and liquidity declines, even the most effective VAR models break down. And once others learn about your concentrated positions, no risk model stands a chance.
Latest posts by Sober Look (see all)
Did you have a comment or question about this post, finance, economics or your love life? Feel free to use the discussion forum here to continue the discussion.*
*We take no responsibility for bad relationship advice.