Irony 101 Or How The Fed Blew Up JPMorgan's 'Hedge' In 22 Tweets
Many pixels have been 'spilled' trying to comprehend what exactly JPMorgan were up to, where they are now, and what the response will likely end up becoming. Our note from last week appears, given the mainstream media's 'similar' notes after it, to have struck a nerve with many as both sensible and fitting with the facts (and is well worth a read) but we have been asked again and again for a simplification. So here is our attempt, in 22 simple tweets (or sentences less than 140 characters in length) to describe what the smartest people in the room did and in possibly the most incredible irony ever, how the Fed (and the Central Banks of the world) were likely responsible for it all going pear-shaped for Bruno and Ina.
Of course we do not know exactly what positions were undertaken and what the mandate was for the CIO Office but our assumption is that they are 'smart people', had no positional constraint (i.e. any asset class, any instrument), and more than likely were human emotionally.
1) Post QE2, JPM's CIO group needed to hedge tail-risk of bank debt portfolio.
Credit risk was rising rapidly and had reached levels not seen since the crisis...as the real chance of tail events was creeping up fast...
2) JPM traders/risk managers are not stupid - can manage curves/levels in 'normal' market but firm needs 'extreme' risk hedge.
Critically - these guys are not dummies - they don't simply buy/sell index protection or curves (as some have suggested) in ultra-massive quantities (since risk models would flash) unless there is an edge. More importantly, they can manage risk at desk levels on term structures and exposures (and even jump-to-default risk to some extent) but on the aggregate portfolio there is a lot of un-covered risk of an extreme event (which seems ever more present) occurring.
3) VaR risk model can 'comprehend' most of 'normal' market moves but not extreme tail risk (illiquidity, basis, correlation shocks).
VaR models (even the most sophisticated Monte Carlo engines) will have significant problems integrating the kind of extreme risks that need to be hedged to avoid catastrophic loss (such as basis risk (the risk that a hedge disconnects from its underlying), illiquidity (some models incorporate a market-impact but very few do it well or even close to reality), and most assume correlations to be relatively stable (not a dynamic variable - which is critical to the pricing of many credit instruments).
4) Goal: Find as low-cost a 'systemic risk' hedge as possible (with longer maturity than options allow).
So, likely at the behest of management, the CIO office set out to find cheap ways to manage this extreme risk. Equity protection was (and is) expensive as evidenced by skews and term structures but credit offered some room for value (and moving from options to tranches - not important what they are other than levered complex credit positions - allows a longer-maturity hedge to be placed - typically at lower cost than rolling options premia over time)
5) Hedge choice made:- Senior Tranche (Low cost, Low spread delta BUT hard to model risk in agg. book & very long correlation).
Senior (or the highest position in the capital structure) tranche positions provide attractive hedges for a systemic (or even cyclical) tail-risk. As seen below, courtesy of Morgan Stanley, the hedge is 'low' /'medium' cost and has high sensitivity to systemic risk. Implicit in this is the fact that it is very sensitive to the correlation of asset defaults.
6) Bought CDX Tranche Q3/4 2011. Provides systemic hedge (will payoff if things are VERY bad but relatively calm if things are OK).
See table above (perhaps was a more complex tranche term structure, cross-quality (IG-HY) or pairs-trade but the endgame was the same - to get long correlation, at a low cost, in a systemically careful - unsuspected - manner).
7) Tranche is sensitive to spread movements (low), volatility (medium - due to hedging gaps), and correlation (high).
These characteristics appear fantastic at first glance - not too sensitive to spread movements overall (ceteris paribus), volatility will cause some drama (as the position will need to be rebalanced), and while correlation is a big sensitivity it is directionally in our favor and has relationships in line with spreads that should help us.
BUT, sometimes the tranches do not always behave exactly as one would expect - due to the second and third derivative interactions of the model parameters...
8) Correlation is market's way of thinking about systemic risk (low correlation equals idiosyncratic risk dominates).
The critical part of our story is comprehending what it takes to crush the most senior part of a capital structure of a portfolio of credits. A few defaults here and there will not affect the most senior or 'secured' aspects of the balance sheet of a portfolio (which is what we call idiosyncratic risk - CHK there, ALLY here, EK then, SHLD soon etc.) but what really worries a manager is if systemic risk rises rapidly due to some event (say a huge liquidity crunch or Greece leaving the Euro and a run on Italian/Spanish bank deposits) which could cause many firms to default simultaneously from a total lack of funding or credit availability (think 2007/8/9). The probability of this 'systemic' event is priced into these credit tranches using the term 'correlation' - low correlation in generally good times when systemic risk is low and idiosyncratic risk dominates (individual firm balance sheets etc.); high correlation in systemically bad times (when something systemic is sinking all boats).
As seen in the chart below - correlation had been stable from the start of Jan11 and was rising modestly as USA downgrade and European woes picked up - though not crazy enough to make pricing expensive...
9) Q3/4 2011: European/US Chaos reigns: Correlation high (hedge doing well) - CIO office reveling in glory of smartness.
See chart above for Sept/Oct movement in correlation - rising
10) Nov2011: Fed/ECB start coord. global easing program -> starts to crush correlation as systemic risk is 'supposedly' removed from system.
And here comes the critical aspect of our story!
The actions of the Fed/ECB/rest-of-world with massive and unprecedented easing efforts was perceived by the market as a tail-risk crushing event - i.e. they removed the systemic risk from the system once again.
Look at the chart above to see what happened to the short-term correlation (red arrow) - it was squelched to levels we had not seen before
11) JPM CIO office forced to sell IG9 protection to manage tranche position as correlation drops (think: delta rebalancing).
What this meant was very important. The tranche - which had been purchased as a hedge for JPM's aggregate (likely long) book required rebalancing as the 'models' used to price and risk manage such positions would have demanded some hedging of the hedge. This is similar to maintaining a delta-hedge on an option position as the market moves one way or another and volatility (a secondary parameter) changes. The trouble is - these systemic risk tranches are HIGHLY sensitive to this somewhat 'magical' measure.
12) Q1 2012: Correlation plummets massively: the 'tail-risk' hedge is needing huge amounts of index rebalancing to keep it 'stable'.
As correlation plummeted - i.e. the market pricing forced the inputs to the models to change which altered the risk sensitivities - so the tranche 'tail-risk' hedge itself needed to be hedged in increasing size. This is likely when Iksil began to be forced to sell IG9 (this is the index upon which the only liquid tranches are based) protection - an oddly bullish position given the bearish nature of the tranche hedge - as all sorts of wonderful second derivative interactions played havoc with his models. Given the size of the firm-wide tranche hedge, and the implicit leverage this tranche infers, this would have meant very heavy index protection selling (in what was not a hugely liquid index anyway.)
13) Mar/Apr 2012: JPM CIO corners IG9 index market as forced protection seller on tranche tail-risk hedge position.
This meant that the JPM CIO office began to sell more and more protection at the index level (dark blue line below) which forced the index to trade differently to its intrinsic or fair-value. These kinds of disconnect (red arrow) are often arb'd by sophisticated hedge funds - but this time the arbs were being frustrated (orange line) by a SIZE player dominating the market and soaking up their demand for protection (the funds would be buying protection on the index - the opposite of JPM CIO - while selling the underlying names protection).
14) Funds complain of richness of IG9 to intrinsics and how technicals are crushing their attempts to arb - the Whale Is Born.
See chart above (red arrow) as the whale began to dominate the market flow. This means that no matter what effort the hedge funds put into the arbitrage (to try and move the orange line higher - back to its more normal zero level) - they made no difference - and in fact were often hurt significantly
15) Momentum takes over and Iksil becomes self-aware - and potentially presses his position (unbalancing the hedge).
We suspect that at some point the daily rebalancing of the hedge's hedge and the constant and consistent rally in credit and equity markets became too much for Iksil who just became another momo monkey, perhaps leaving a little too much long index protection on the basis of the rally extending...i.e. over-hedging LONG his implicitly short credit/systemic hedge
15) European sovereign, China slowdown, and US growth risks spur deterioration in credit risk - meaning losses on IG9 index position.
Between his huge size and the velocity of the shifts in the index as things began to go wrong fundamentally, Iksil was in trouble. Not only that but 'correlation' began to pick up and so the hedge of the hedge needed to be unwound...
16) JPM CIO faces huge losses from small move in spreads since they have sold so much protection and tranche unbalanced.
He found himself the dominant long player in a market in which fundamentals, technicals (arbs), and his own models (correlation) were saying unwind/short - which starts the pain trade for Ina and Bruno and more than likely this is when the bells started to go off in risk manager's ears and Dimon got the call...
17) Funds arb the skew and press IG indices knowing JPM needs to unwind/hedge the index hedge of the 'tail-risk' hedge.
It was a poorly kept secret obviously and left a market smelling blood. JPM looks for any way to manage this position - i.e. find any liquid hedge to cover the overly-long index protection that Iksil had over-hedged the original tail-risk hedge with. There's not enough liquidity in the original instrument so any and every credit instrument gets hit...
18) All credit instruments blow wider (as JPM - or front-run by peers) look for any and every liquid hedge to manage over-hedging.
last week or so this has been occurring with the worst happening post Dimon's call...
19) IG9 skews normalize (today) relieving some pressure - seen this the last 2 days.
The arbitrage between the index and its intrinsic value has smashed back to almost zero (-3bps or so - see chart above) as the technical pressure from the whale is relieved and the arbitrageurs flow can 'correct' the index back to its idiosyncratic reality. This likely slows the pain trade as the arbs will unwind their position removing some pressure from JPM now.
20) HY18 massively cheap (and HYG underperforming) - suggests long HY position but scary technicals remain.
The on-the-run high yield credit index (meaning the most recent vintage HY spread index in credit derivative land) has seen its index spread (dotted line red arrow) explode relative to its intrinsic value (orange arrow). This is an example of the 'technical' flow that reaching for any and every liquid hedge has caused in the last few days - and leaves HY18 extremely cheap to its fair-value.
Today also saw HYG (the high-yield bond ETF drop dramatically) as another example - and most notably - perhaps JPM was the cause of the HYG collapse relative to NAV at the start of April (orange ovals below)...
21) Did JPM unwind some of tail-risk hedge? What further losses did they take on index hedge of tranches?
We can only hope so - and would suspect so given the recent gappiness but it is unclear what losses they faced on those positions and what model parameters they are now using to price the underlying tail-risk hedge that is likely up in value significantly in the last few days - this is where we suspect the truth is being hidden. At some point the market will once again force a rerack of the model parameters and we suspect the worst is not over here - especially if we see systemic risks continue to rise rapidly - more rapidly than the unwind can occur - key to this will be the number of defaults but the MtM volatility will pressure JPM's earnings without doubt - though it seems unlikely that JPM would not have managed to hedge a lot of this systemic over-exposure - though basis risks become even more complex.
22) Summary: JPM tail-risk hedge imploded thanks to Central Banks' Systemic Risk reduction - unintended consequence...
The key factor is that if systemic risk had remained in even a 'normal' range of possible regions based on history, then the JPM CIO office would have had no need to over-hedge their tail-risk hedge position, no greed-driven need to press the momentum, and no need for such an epic collapse as we are seeing now.
The point is - this was a trader/manager with a good idea (hedge tail risk) that was executed poorly (and with arrogance) but exaggerated by the unintended consequences of the Central-Banks-of-the-world's actions (and 'models behaving badly' as Derman would say).