Algos, Twitter, or Laziness?
I am surprised by the market reaction to the NY Times article about JP Morgan’s whale trade. The headline JP Morgan trading loss may reach 9 billion seems devastating. I can understand why a computer would react to the headline maybe programs missed the subtlety of the word “may”. But why are so many people reacting so quickly and selling the stock when the details don’t really say anything new? I thought Twitter restricted you from writing more than 140 characters, but maybe it has had a side effect of restricting people to reading 140 characters?
The loss amount referenced is useless:
So let’s look at what was written about the loss.
In April, the bank generated an internal report that showed that the losses, assuming worst-case conditions, could reach $8 billion to $9 billion, according to a person who reviewed the report.
This is it. A report from April, where under worst case conditions, the loss could reach $8 to $9 billion. So this isn’t a report from June. It isn’t a current estimate of the losses. It is long before Mr. Dimon actually told politicians that JPM would have a “solid” profitable quarter. As arrogant as he may or may not be, it would seem a stretch to believe he would tell such a whopper to politicians on national TV.
This particular report seems to focus on the long credit trade and not the CIO office as a whole. There were more positions, and based on JPM’s comments, and the Fed Stress test, the CIO office did well during period of stress.
So how could the “worst case” scenario be $8 – $9 billion? How could it not? Have people lost their mind. Whether the trade was $10 billion of tranches or $100 billion of outright index, what do people think the worst case was?
Worst Worst Case vs Worst Case vs Base Case
For weeks now, people have had a decent estimate on the size of the trade. In spite of all the confusion of what exactly it was – tranches, curves, outright, etc., the magnitude is known and there are even some filing with the Fed that gave a decent sense of the potential size.
For simplicity, let’s assume JP Morgan had $150 billion of outright exposure to IG9 10 year (I think the trade is far more complex than that, but let’s just pick this as a big risk number).
On March 30th, this would have been marked at 112. As an outright index there is little way it could have been marked differently (tranches are another story).
The widest level the index got to was 172 on June 4th. That’s a 60 basis point move tightest level to widest level. A 50% move in spread! Scary!
On March 30th, at a Spread of 112, a $150 billion traded would have had a value of $2.45 billion. On June 4th, the trade would have had a $6.49 billion. A loss of just over $4 billion if they had $150 billion (which seems high) and assuming they did nothing (which is also unrealistic). 60 basis points seems like a lot, spread widening by 50% seems like a lot, and it is, but it translates into a “price” move of less than 3%.
The same mythical trade would have a current spread of 162. That has a value of $5.98 billion, but JPM would also have received a coupon payment of $0.3 billion. Yes, this hypothetical trade earns over $3 million of interest every day.
There were a lot of moving parts, but from what is known of the size of the trade, it would have taken gross incompetence on the traders unwinding it to lose more than $4 billion on it during the quarter. This doesn’t address the performance of other positions in the CIO office which may have added to losses, or may have helped.
But how did the NY Times and JPM come up with a “worst case” loss of $9 billion. Seems pretty simple since a position as big as $150 billion (which isn’t what I think their position was) would have exposure of over $1 billion per name. Assuming defaults in some of the worst names – MBI, RDN, RRD, JCP, S, and index spreads reaching post Lehman like levels of 300 bps, would get you there. But that is the magnitude of spreads movement you would need. It could have happened. It isn’t an unreasonable worst case, but it was meant to be a worst case, not a realistic estimate. And it was only on that position. Under the “worst” case for this trade other positions may have done well.
If this was $9 billion of IG9 mezz tranches, the worst case would be full loss on the tranches. It wasn’t the likely case and didn’t become the actual case, but it was an estimate of worst case.
Will only cockroaches and TBTF banks survive Nuclear War?
Banks should not fail because they made stupid mistakes other banks didn’t. You want to eliminate as much systematic risk as possible, but at some level, banks will fail. If we see a 10% decline in GDP, banks would struggle to survive that.
The worst case on any individual loan is 100% loss. That is the “worst” case. A more likely “worst” case is 60% loss because the bank believes that will be the recovery. On a portfolio, the “worst” case would be all loans default. But in a realistic “worst” case assessment, the bank won’t assume all loans default. Some industries won’t be correlated, some loans will perform better than others, etc, but the reality is “worst” case for a lending operation is always bad.
That is why credit gaps wider. It goes from investors worrying about how much spread income they will accrue over time, to fear of losing the full notional.
The “worst” case for JPM is losing $2.3 trillion if their entire balance sheet turned out to be worthless. It does no one any good to assume that. The Fed Stress tests are a good example of what we are trying to protect against. A bad, but possible scenario is something banks should be prepared for. They should have the capital to withstand that scenario and it looks like JPM does.
We see much the same thing in discussions about Spanish banks. It is unrealistic to assume that under the most dire scenario they will still have 6% capital. What we need to look at is a likely scenario, with real values used where they should have more than the 9%. Then a scenario that is another leg down, to severe, but not nuclear war levels, and that the banks can sustain that capital.
It may be fun to send out the headlines. Rant about how stupid Mr. Dimon is, but if you are doing so based on a report from April, with no clue of what market conditions were assumed for the “worst” case, you are doing investors a disservice.
Where are the winners?
Some hedge funds are turning in decent numbers for May. Those that participated in the whale trade, the IG9 10yr vs IG18 trade which we recommended back on April 9th, or the arb trade which we discussed even before the whale story broke, when all there was to see was an index trading absurdly rich. The funds have done well but nothing that would indicate they made $8 billion as a group. The JPM trade is complex, and was outright, and the best funds had on rel val trades, so it wouldn’t be exact, but CDS is a zero sum game, and someone by now should be bubbling with excitement if JPM really lost 9 billion, but feel free to avoid facts when the headlines are so juicy.
Disclosure and Exchanges
I do think it is ridiculous that we aren’t getting any more disclosure from JPM. Aren’t shareholders entitled to knowledge about what their bank has? Why are shareholders exposed to a $1 loss because of an inflammatory headline? If the details were known, the stock would either not have rebounded as much as it has, or we wouldn’t be selling off this morning. July 13th, the day JPM allegedly plans to come clean feels like a long way off, but I would bet against this NY Times story being remotely close to what actually happened, though it’s a shame we have to wait that long.
If this was an exchange traded product, we would also have a much better idea of what is going on. The DTCC info is interesting, but hard to use. The tranche data doesn’t differentiate between first loss tranches and super senior tranches – a very big difference in risk all rolled into one number.
It is also very hard to figure out if any individual institution took off trades. Whether an institution does an unwind, assignment, or new trade, can affect the gross and net positions differently. If someone is buying back tranches on a “no delta” basis, it may show up as increased risk as the “correlation” desk selling the tranche may initiate a “delta” hedge.
There is no reason for the markets to have so many CDS moments. Regulators – fix it. It’s not rocket science. It really isn’t. It’s complex, but relatively simple and just requires effort and the willingness to accept that some trades that people like to put on now, just don’t fit a more regulated market – but that is the cost to trying to make a system safer.