The UBS-Adoboli Scandal Shows the Problem of Negatively Skewed Risk is Still With Us
In my banking courses, I point to negatively skewed trading strategies as a key cause of the crash of 2008. The recent loss of more than $2 billion attributed to rogue trades by Kweku Adoboli at Swiss banking giant UBS shows that the problem of negatively skewed risk is still with us. It is an old story, but with a new twist.
The problem of negatively skewed risk arises from the way incentives change when you gamble with someone else’s money. People who gamble with their own money tend to prefer strategies that have a small chance of a life-changing win against losses that are frequent, but are capped at an affordable amount. Putting a dollar in a slot machine is an example. All you can lose on one pull of the handle is your dollar, but if you are lucky enough, you can win enough on that one pull to pay off your delinquent mortgage. That kind of risk is called positively-skewed because its probability distribution has a long tail on the plus side and is truncated on the minus side.
If you’re gambling with someone else’s money, you look for something different. You want a strategy that pays you a steady income with near certainty, even if you have to put a lot of money at risk. In exchange, you are willing to accept a small probability of a catastrophically large loss. If the strategy pays off, you keep the winnings, or at least an agreed share of them, depending on your arrangement with the source of the money. If the rare unlucky number comes up, you probably lose the opportunity to continue playing, but at least the money you lost wasn’t your own. That is a negatively skewed risk because the long tail is on the minus side.
In the world of finance, negatively skewed strategies are easy to find. Writing naked call options is a classic. There are many more complex variants, as well. One, practiced by many banks, is to game regulations in a way that keeps officially measured capital up to the required minimum while tangible common equity is allowed to melt away to a sliver. Bernie Madoff’s Ponzi scheme was still another negatively skewed strategy. It provided both its organizer and its investors with years of steady earnings before it went down in flames.
When you start with negatively skewed strategies and add a bonus-based compensation system, in which losing your job is the worst possible penalty, you have a recipe for systemic risk. The trader who initiates the risk, the trader’s supervisor, and the CEO at the top are all in line for their share of the winnings. The higher up the ladder you go, the more likely you are to keep your job and your accumulated bonuses even if things go wrong.
Although we don’t yet know exactly what happened in the case of Adoboli’s losses at UBS, informed speculation points to a negatively skewed risk scenario. Probably Adoboli was making trades that should have been hedged, but increasing the profit by neglecting the hedge. We can expect more details to come out, but that is all the old part of the story.
The new twist in the Adoboli episode is that it comes after UBS, badly burned in the last crash, implemented some of the safeguards that risk experts recommend. In particular, the bank reportedly has greater power than in the past to claw back past bonuses earned by rogue traders and their supervisors. In principle, clawbacks should alter the probability distribution of gains and losses in a way that makes negatively skewed strategies less attractive. Why didn’t it work?
Part of the explanation is that clawbacks are not always credible. As a practical matter, it is hard to extract two billion dollars worth of pain from a young trader who spends his bonuses on high living as fast as they are earned.
Another part may be that even when clawbacks change the objective shape of the risk distribution, the change may not be fully perceived. Studies of combat soldiers, drug dealers, and people who set out to harpoon whales in skin boats strongly suggest that young males are genetically programmed with tail-risk blindness. If that is the case, even the threat of prison may not be enough to discourage excessive risk taking.
Still another part of the story is that rogue traders do not operate in isolation. They, their immediate supervisors, and everyone right up to the CEO are united in the enterprise of gambling with other people’s money—that of shareholders, bondholders, and taxpayers. At higher levels of the hierarchy, the way to keep your job and avoid clawbacks is not so much to discourage the use of negatively skewed strategies at lower levels as to enact a plausible set of rules while maintaining the ability to deny that you knew they were being broken.
As Andrew Hill writes in todays Financial Times, even well-designed control structures can be “powerless in the face of an embedded corporate culture and a system of skewed incentives. . . Rules and processes are far easier to change than bad behavior and big bonuses.”
Perhaps, then, excessive risk taking simply cannot be controlled at large, complex financial institutions. If so, some observers, Martin Wolf being one, suggest that the only solution is to construct ring fences that protect the essential banking functions provided by giants like UBS from their more dangerous operations. That is the approach taken by the recent Vickers report in the UK and the Volcker rule in the United States.
We don’t yet know if the latest proposals for ringfencing will work, but on the whole, I am inclined to think they are worth trying. If we can’t stop people from gambling with other people’s money, I would at least like to hope that it is not my money, unless I choose to put it at risk. If ring fences don’t work, then we will have to try something else, like, perhaps, banning young males from participating in financial markets.
One Response to “The UBS-Adoboli Scandal Shows the Problem of Negatively Skewed Risk is Still With Us”
Enlightening Dr. Dolan. I appreciate the article. So negatively skewed means there are ever decreasing small chances of a catastrophic loss and zero (or near-zero) chance of a huge win? Contrasting that with the lottery, near zero loss probability ($1 ticket) coupled with infinitesimal (near impossible) chance of a huge win (i.e. sucker bet) shrugged off as "Hey you can't win if you don't play".