A limit is only a control if it binds

Graduate · enrolled learners

This lesson opens with The JPMorgan London Whale, 2012.

What happened

Traders in JPMorgan's Chief Investment Office, one of them nicknamed the London Whale for the size of his positions, built an enormous book of credit derivatives, the Synthetic Credit Portfolio, whose notional size tripled in the first quarter of 2012 from 51 billion to 157 billion dollars, until it could not be exited without moving the market against itself. In late January 2012, while the book was in breach of both the office's and the bank's value-at-risk limit, the bank temporarily raised the limit and then hurriedly adopted a new value-at-risk model, the number meant to measure how much could be lost, that immediately cut the book's reported risk by 50 percent, so the risk system said the position was safer at the very moment it was becoming more dangerous; in May the bank revoked the new model as inaccurate and restored the old one. From January 1 to April 30, 2012 the office's risk limits and advisories were breached more than 330 times, on all five of its risk metrics, and traders and risk staff repeatedly attacked the accuracy of the metrics rather than cutting the book. The trades lost at least 6.2 billion dollars in 2012, by the US Senate investigators' count, and the bank paid about 920 million dollars in penalties to US and UK regulators in 2013. The controls were not absent: there were limits, there was a value-at-risk measure, there were reports. They were overridden, gamed, and ignored, so a risk system that existed on paper reported that all was well while the position that would cost billions kept growing, because a limit nobody enforces and a measure that can be changed to fit the position are not controls at all.

The decision point

A risk limit is only a control if it actually binds, meaning that breaching it forces action to reduce the position rather than a note that is filed and forgotten, and a risk measure is only informative if it is honest, meaning it is not changed to report whatever the position needs it to say. The London Whale is the case: JPMorgan had limits and a value-at-risk model, but as the credit-derivatives book grew and breached them, the model was swapped for one showing half the risk and the limits were breached more than 330 times, so the risk system reported safety while the position that lost at least 6.2 billion dollars kept growing. This is the foundational lesson of risk systems: their existence proves nothing, because a limit that is waived when inconvenient and a measure that is tuned to permit the trade are the appearance of control without its substance. A control that does not bind is not a softer control; it is not a control, and it is more dangerous than none because it produces a false sense of safety that lets the position grow unchallenged. So the discipline is to judge a risk system not by whether limits and measures exist but by whether the limits actually force reductions when hit and whether the measures are fixed rather than adjustable to taste, to treat a model changed in the middle of a losing position as a red flag rather than a fix, and to remember that at JPMorgan the risk reports read reassuringly right up to the multi-billion-dollar loss, because a risk system is only worth what its limits bind and its numbers are honest, and the London Whale's bound nothing and told the position's story back to it.

Recorded loss

$6,200,000,000

Orientation and Year One are open: anyone can read them without an account. From Year Two onward the lessons are for enrolled learners, because progress through the later years only means anything if it is tracked against a record.

It is free. We do not sell the list and there is nothing to buy at the end of it.