When I got the Senate Permanent Subcommittee on Investigations report on the London Whale last night, I did what any sensible human would do: I ctrl-F'ed for my name and the names of my friends and enemies, gloated briefly, and then set to work rationalizing not reading the rest of it. After all, it's ridiculous for the Senate to investigate a basically legitimate trade that, though it lost some money, did nothing to destabilize JPMorgan or the financial system as a whole. And we've heard all the important Whale stuff before, including in JPMorgan's own Whale autopsy, and even then it was old news.
But then I started skimming the executive summary and after underlining every sentence in the first ten pages I figured I'd have to give it a closer look. It's an amazing, horrifying read.
What was the Whale up to? I don't think you'll get a better explanation than this, from a January 2012 presentation by the Whale himself, Bruno Iksil (page 74):
Mr. Iksil’s presentation then proposed executing “the trades that make sense.” Specifically, it proposed:
“The trades that make sense:
- sell the forward spread and buy protection on the tightening move Use indices and add to existing position Go long risk on some belly tranches especially where defaults may realize Buy protection on HY and Xover in rallies and turn the position over to monetize volatility”
This proposal encompassed multiple, complex credit trading strategies, using jargon that even the relevant actors and regulators could not understand. Because the traders themselves declined the Subcommittee’s request for interviews and were outside of the Subcommittee’s subpoena authority, the Subcommittee asked other current and former CIO personnel to explain the proposal. Ina Drew, CIO head, told the Subcommittee that the presentation was unclear, and she could not explain exactly what it meant. Irvin Goldman, then the CIO’s Chief Risk Officer, told the Subcommittee that the presentation did not provide enough information to clarify its meaning. Peter Weiland, the CIO Market Risk Officer, offered the explanation that Mr. Iksil was basically describing a strategy of buying low and selling high.
I'm not too troubled by the Whale's proposal, but I'm very worried about his bosses' total inability to understand it. I'm gonna file away Weiland's answer as a perfect response to any congressional inquiry: "well, basically, it's a strategy of buying low and selling high."1
But while his answer is - always! - generally right, it was also specifically wrong: the Senate report is, in its most gripping moments, a history of JPMorgan's CIO getting behind the markets and then chasing to catch up. They sold low and bought high. The structured credit portfolio had always been intended to be a credit hedge that would pay off on significant corporate defaults, and in November 2011 it did, making $400 million on a pretty lucky high yield tranche trade when American Airlines defaulted. This was sort of an original sin for the SCP (page 55):
The American Airlines gain also appears to have colored how the CIO viewed the SCP thereafter, as a portfolio that could produce significant profits from relatively low cost default protection. In addition, it produced a favorable view within the CIO of the SCP’s complex trading strategy that involved combining investment grade and non-investment grade credit index trades, accumulating massive tranche positions, and sustaining a period of losses in anticipation of a large payoff.
A month later, JPMorgan asked the CIO to reduce its risk-weighted assets. Normally this would involve winding down trades, but that would risk missing out on the glory days that CIO had experienced the previous month. So this happened (page 63):
Mr. Iksil told the JPMorgan Chase Task Force investigation that, in early December 2011, Ms. Drew instructed him to “recreate” the American Airlines situation, because those were the kinds of trades they wanted at the CIO: the CIO “likes cheap options.” Thus, as he described it, he was told to maintain the SCP’s default protection in order to position the CIO to profit from future American Airlines-type defaults. Ms. Drew confirmed to the Subcommittee that she gave guidance to the traders to position the book for another gain like in late 2011. In short, Ms. Drew indicated her preference to avoid reducing the SCP book in a way that would reduce its default protection and the opportunity to profit from future corporate defaults.
When your boss comes to you and says "I like cheap options, so go find some, and recreate that lucky win we had last month": you're fucked. Fucked! Everything's fucked! Quit! Become a blogger! It's all over for you.
So there Iksil is, mandated to chase "future American Airlines-type defaults" by buying lots of high yield index protection, while also required to reduce risk-weighted assets. And, obviously, not particularly desirous of showing a loss, since the SCP has been profitable every year since inception, and since his supervisors are the sort of people who go to their traders saying "I like cheap options" like it's some sort of insight. So he did "the trade that makes sense": put on tons of offsetting investment-grade long-credit trades (actually selling the forward spread, selling protection on long-dated IG and buying protection on short-dated, which is net long credit), which both provides some carry that he can use to buy HY and provides an inversely correlated portfolio to reduce risk-weighted assets.
This is December 2011, and the market then rallies. Here's a toy chart (white is investment grade, orange is high-yield, up is "rally", not to scale):
So the Whale is making money on his IG position but losing on his HY. Sadly (page 78):
Not only did the SCP’s short positions lose value as the economy improved, but the long credit protection the CIO purchased for investment grade companies did not increase in value as much as was needed to offset the losses. As Mr. Macris put it, the investment grade rally “lagged” the high yield rally.
So what did they do? They got long more IG credit. The result was that (page 85) "by the end of March 2012, Mr. Iksil had acquired so many long index instruments that the SCP – which had traditionally held a net short position to provide protection against credit risks for the bank – had flipped and held a net long position."
Thus while Doug Braunstein was spinning a story - a story that I believed - about how the structured credit portfolio was intended as a stress-loss hedge that would make money in tail downside scenarios, this (page 278) is what he was looking at internally:2
If credit spreads blow out by 50%, the portfolio: loses a billion dollars. Nice hedge guys.
The other result of this activity is that they kept buying high and selling low, in effect, selling more protection as the price for protection went down. This selling itself produced mark-to-market gains on their long credit book - if you keep selling lower and lower, you can mark what you've already sold lower and lower - but they were more than offset by losses on their short trades, so that in the first quarter of 2012 they lost money almost every day.
This snowballed, so that within a few weeks of March they added $40 billion of notional and blew through risk limits, and on March 23 Ina Drew "ordered the CIO traders to 'put phones down' and stop trading" (page 85). So they did, and, unable to defend their positions, those positions collapsed, with IG (where JPMorgan was long credit) widening dramatically over mid-2012 without a corresponding widening from HY (where JPMorgan was short).
That much I suppose we mostly knew though the Senate report provides a wealth of new details. What lessons can one take from this? I think you have to start with: what is hedging? What were they doing? The Subcommittee is a little frustrating in its rigidity on how hedging should work,3 but you can understand that after reading quotes like this (page 44):
CIO’s most senior quantitative analyst, Patrick Hagan, who joined the CIO in 2007 and spent about 75% of his time on SCP projects, told the Subcommittee that he was never asked at any time to analyze another portfolio of assets within the bank, as would be necessary to use the SCP as a hedge for those assets. In fact, he told the Subcommittee that he was never permitted to know any of the assets or positions held in other parts of the bank.
Nice hedge guys.4 And CIO didn't just stonewall the Subcommittee on the "so what are you hedging?" question. This is amazing (page 46):
To clarify the risk that the SCP was intended to address, at one point on April 2012, according to an internal bank email, Mr. Dimon asked the CIO for the correlation between the SCP and the portfolio the SCP was meant to hedge. Mr. Dimon told the Subcommittee that he did not recall if he received a response. Ms. Drew explained that, even though the request had been made by the CEO, so many events were unfolding at the time, that she did not recall if the correlation analysis was sent to him. The bank has been unable to produce that analysis, and the Subcommittee found no evidence this analysis was completed.
Let me tell you, I worked at a bank, and I am very familiar with the approach of "nod, smile, agree to provide that senior person that analysis they asked for, and then never think about it again." I was not aware it could be practiced on the CEO.
The problem, it seems to me, is not with the conceptualization of the CIO as a "stress-loss" or macro hedge, or with how it was initially set up, or with the fact that it was a little fuzzy and the correlations and "what are you hedging" response were a little opaque. The problem is that, while everyone talked a good game about tail risk hedging, no one was committed to it in their heart of hearts.
Nassim Taleb, as far as I can tell, wakes up every morning thinking "I am going to lose a moderate amount of money today, and also annoy people." And that's a full day for him, that and the rocks. Every so often there's a crisis and he makes a lot of money and reminds everyone how good he is at pessimism. Not a lot of people can live that way, thank God. You want them as your tail-risk hedgers, though you may not want them around otherwise.
But instead of them you usually get Bruno Iksil and Ina Drew: people who are basically traders, get along with everyone, say sensible things, don't scowl or pick fights unnecessarily. Their idea of a full day is going to a meeting in the morning and saying "we are awesomely positioned for a crisis! tail risk! stress-loss hedge! black swan! whatever," and then looking at their P&L that afternoon and seeing a $400 million gain. It almost never works that way! Actually hedging tail risk is normally a miserable process of paying premium every day for protection that is unlikely to pay off. The positive-carry thing is just really, really unlikely to be a tail risk hedge. But it's more fun.
And so you have a toy tail-risk book - the report says it "beg[a]n as a small, experimental portfolio in 2006" - that happens to make money because, y'know: the black swan happened in 2007. Now it employs people whose distinguishing features are (1) they like making money, (2) they've done it, and (3) the product they're pitching to senior management is "hedging your tail risk." Why wouldn't you give those people lots of capital to play with? And as credit conditions improve, why wouldn't they morph their position from a money-losing, dour tail-risk hedge into a giant long-credit mumble-mumble-look-over-there-it's-a-tail-risk-hedge sort of book?
The moral of the story might be: if you're a bank, and someone is telling you that they're hedging your tail risk, and it's not Nassim Taleb, they're probably messing with you somehow.
Anyway, there's so much more - on the Volcker Rule, on JPMorgan's breaches of its risk limits, etc. A few quick summaries:
The mis-marking. The CIO mismarked its position! A lot! It eventually stopped after it got into a collateral dispute with Morgan Stanley, one of its biggest dealers, who were basically like "I notice we have no MTM problems with JPMorgan's investment bank but we do with the CIO, what gives?," and escalated their collateral call until JPMorgan put a stop to CIO's mismarking. Before that, though, you got lots of unpleasantness, including the junior trader who did this (page 96):
For five days, from March 12 to 16, 2012, Mr. Grout prepared a spreadsheet tracking the differences between the daily SCP values he was reporting and the values that would have been reported using midpoint prices. According to the spreadsheet, by March 16, 2012, the Synthetic Credit Portfolio had reported year-to-date losses of $161 million, but if midpoint prices had been used, those losses would have swelled by at least another $432 million to a total of $593 million.
CIO head Ina Drew told the Subcommittee that it was not until July 2012, after she had left the bank, that she became aware of this spreadsheet and said she had never before seen that type of “shadow P&L document.”
But now you can (page 111):5
Those nine-digit red numbers should give you a pretty good sense of the angst that must have been in the air chez CIO, but if not, try this IM from the Whale's boss after the Whale had, in a fit of conscience, provided "real" marks to some senior people (page 118):
Yeah, I don’t understand your logic, mate. I just don’t understand. I’ve told Achilles. He told me that he didn’t want to show the loss until we know what we are going to do tomorrow. But it doesn’t matter. I know that you have a problem; you want to be at peace with yourself. It’s ok, Bruno, ok, it’s alright. I know that you are in a hard position here.
Regulatory supervision. Among the most-quoted lines in the report so far is the OCC examiner who in May 2012 "referred to the SCP as a 'make believe voodoo magic "composite hedge"'" (page 49). Hahaha, true, but also: shut up OCC examiner. It's clear that no one at the OCC had any idea what was going on at JPMorgan, and never asked. Partly that's because JPMorgan hid stuff from their regulators; partly it's because the OCC was totally incurious and trusted JPMorgan (page 248):
The initial reactions of Ms. Williams and Mr. Brosnan, two of the OCC’s then-most senior officials, were to view JPMorgan Chase as an effective risk manager and to view the Synthetic Credit Portfolio as a hedge that would lower bank risk. The skepticism and demand for hard evidence that might be expected of bank regulators were absent.
There's lots more on apportioning the blame; if you want you can ponder who comes off worse, the JPMorgan executives who mostly didn't tell the OCC anything, or the OCC examiners who mostly didn't ask. You might consider this, though: the JPMorgan executives and traders are all named in the report, and most of them have been fired. (My guess is that, post report, a few more of them might be.) The OCC examiners are less publicly flogged, and as far as I can tell none have been fired, though at least one seems to have been demoted. Incentives!
Speaking of incentives. Ever wonder what a whale makes? Page 58:
Some of that was, of course, clawed back.
Lawsuits. The last, like, 50 pages of this report is weird in that it is just "here is the case for plaintiffs to make in suing JPMorgan for misleading the public about its CIO position." What an odd thing for a Senate Subcommittee to do, no? The rest of the report dips in and out of pretending that it is designed for legislative purposes - make better rules to make banks safer! - as opposed to just grandstanding and excoriating JPMorgan, but the tail end is exclusively - and devastatingly - "we would like some JPMorgan senior executives to be sued successfully and a lot, and ideally prosecuted." I wouldn't hold my breath on the second part - "too big to jail" don'tcha know - but I'm pretty sure they'll get their wish on the first.
JPMorgan Chase Whale Trades: A Case History of Derivatives Risks and Abuses [Senate Permanent Subcommittee on Investigations]
1."Well, it's the trades that make sense" is also pretty good.
2.Though page 281, to be fair, actually has the portfolio making tons of money in some tail scenarios including "Currency crisis" and "Many defaults," though also losing money in scenarios including "New Financial Crisis." Also, to be fair, even that chart in the text shows some convexity - the right-hand column has 50x the spread widening as the left-hand column, but only 20x the loss. So sure huge gap-risk hedge whatever. Wait no: the convexity is negative, a +50% widening is more than 5x as bad as a +10% widening. This gets worse on a gap down.
3.E.g. they make a big deal (page 48) out of JPMorgan's hedge accounting disclosure in its 10-K, raising their eyebrows about how JPMorgan didn't do accounting-style hedge identification for the SCP trades (which were, after all, not accounted for as hedges). That seems lame to me. And they discount the possibility that other banks do similar macro hedges by saying (page 49) "That the OCC was unable to identify any other bank engaging in this type of general, unanchored 'hedge' suggests that this approach is neither commonplace nor useful." Well, sure - except that, as you just got through saying, the OCC couldn't identify that JPMorgan was doing this, because the OCC were totally oblivious to all trading activity everywhere.
Also, tangentially related: the Subcommittee staff seems convinced that derivatives are "inherently higher risk, due to their synthetic nature which meant that no real economic asset lay behind the positions to stem any losses" (page 68), or again they refer to "the high risk nature of synthetic credit derivatives, [and] the lack of any underlying tangible assets to stem losses." This is just magical thinking - "derivatives are scary because there's nothing there!" - that can't be good for the regulatory results that will come out of this.
4.Almost as bad (pages 44-45):
Given the lack of precision on the assets to be hedged, JPMorgan Chase representatives have admitted to the Subcommittee, that calculating the size and nature of the hedge was “not that scientific” and “not linear.” According to Ms. Drew, it was a “guesstimate.” She told the Subcommittee that there was “broad judgment” about how big the hedge should be, and that she used her “partners” as “sounding boards” if she later wanted to deviate from what had been agreed to. According to the OCC, on April 16, 2012, JPMorgan Chase told the OCC that the SCP was expected to gain $1 billion to $1.5 billion in value to offset $5 to $8 billion in firm wide losses.
5.Also (page 110):
Data later compiled by the JPMorgan Controller’s office as part of a special assessment of the SCP marks during the first four months of the year indicates that the mismarking likely peaked in March. The data showed that, for 18 selected SCP marks as of March 31, 2012, with respect to 16 of those marks, the CIO had booked a value equal to the price at the extreme boundary of the bid-ask spread, had booked one mark almost at the extreme, and had even booked one mark outside of the bid-ask spread.