By Phil Read
On 18 April 2008 UBS issued a “Shareholder report on UBS’s Write-Downs”. In 2007 alone UBS announced losses of 18.7 B USD in relation to the US residential sector (in total it would amount to over 38 B USD – the second biggest loss in the history of banking). The report was an attempt to provide shareholders with a perspective on what had gone wrong. In November 2009 a Consolidated Securities Class Action Complaint was filed (423 pages) in the United States District Court Southern District of New York alleging fraud by officers of UBS.
Together these reports reveal a number of games that contributed substantially to this loss.
The games I would like to highlight are:
• Hands Off: the player establishes a reputation as someone not to mess with territorially (in this case the player was John Costas)
• Rubber-stamp: bringing in consultants to do a study of a project you have already decided you want to do (in this case Mercer Oliver Wyman)
• Pseudo Science: using highly specific and meaningless ratios to justify a decision (in this case I will highlight VaR, but the manipulation of AAA ratings would be another pertinent example)
• Premature Sales recognition: where players record a highly contingent sale as a way to increase revenues for the calendar year (and in this case of Investment Bankers also therefore their bonus)
• Public Challenge of Your Loyalty: challenging a person’s loyalty just because they have raised some legitimate concerns with the way the organisation is approaching an issue (in this case the Risk Management function)
John Costas joined Union Bank in March 1996 as US head of fixed income; was promoted in 2002 from COO of the Investment Banking arm of UBS to CEO of UBS Investment Banking. He was credited with increasing UBS corporate and investment banking revenues to 51% of UBS revenue in 2003 (from 22% in 1999). He was celebrated by the media (eg. Euroweek: “The Rise and Rise of UBS’s John Costas”, Financial News Daily: “Costas Deserves Place in Pantheon”, The Banker “John Costas is Enjoying Life”). With his celebrity status cemented by the media, Costas was now perfectly set to play the game of “Hands Off”.
When UBS launched DRCM (intended to be a UBS hedge fund run for outsiders) he demanded that control of it be given to him or he would leave to become Morgan Stanley’s CEO. UBS handed control over to him and his lieutenants – including 3.5 B$ capital, 80 top traders from the Investment Bank, tied Costas compensation to the percentage of assets under management, guaranteed a 1B$ bonus to him and his associates over 3 years, a 35% success fee and a guaranteed management fee of 3% of assets under management. DRCM was shifted from a traditional hedge fund to one which was a high risk investment vehicle for US and was allowed access to UBS’s balance sheet for funding purposes. It was therefore critical that UBS manage the risks that Costas trading strategies exposed the company to.
One of the Confidential Witnesses in the Class Action Complaint alleged that , senior management in Zurich “never got involved as long as [DRCM was] making money.” Due to Costas being allowed to play the game of “Hands Off” risks in DRCM were not properly supervised, and risks in the remaining Investment Banking business increased for many reasons, partly related to the loss of talent to DRCM, and partly because of other games.
Shortly after being appointed CEO of Investment Banking for UBS (1 July 2005) H Jenkins commissioned a study of the Fixed Income business by consultants (Mercer Oliver Wyman). The findings of the consultants were that there was a big competitive gap in this business. These were of course known facts. The consultants went on to recommend that UBS invest in “Subprime and Adjustable Rate Mortgage Products”. This was a classic example of a “Rubber Stamp”: Jenkins however wanted to have a consultant propose it. In addition the recommendations suggested that in order to grow the business one of the key factors would be a “streamlining of risk processes”. In fact not only were risk management processes streamlined – they did not fill the position of Risk Officer.
I would like to focus on two key numbers that were used to play the game of “Pseudo Science”. The first is market risk VaR, which is a calculation of the amount of money UBS could lose in a given day. The model underlying VaR is based on an evaluation of assets in the portfolio compared to historical market risks. If an asset was not considered a market risk, it would not be included.
However VaR was manipulated to enable UBS to acquire more of certain positions (eg. CDO’s) than would otherwise have happened. CDO’s have an inherent risk (in some of the UBS models this was estimated at 2-4% of its value). However according to the Financial Times article “Crooked to the Core” on April 20, 2008, traders at both DRCM and in the Investment Bank played games with this and therefore VaR was underreported. One particular game was to “insure” the 2-4% of the CDO’s by what are called AMPS trades, which then allowed them to hide the CDO’s from UBS risk controls. This allowed rapid build up of CDO’s without triggering the risk management systems in UBS. However of course, only 2-4% was insured, not the other 96-98%; and yet no risk found its way into the VaR.
There was of course a vested interest for the traders. One of the Confidential Witnesses in the Class Action Complaint stated that “CDO traders were motivated to create larger and larger CDO deals in order to increase their compensation”. “CDO traders were compensated solely based on the volume of CDOs they structured, and not on whether UBS was ultimately able to sell the bonds they structured.” Had risks been properly quantified, rather than manipulated using the pseudo science of VaR, this would have had the effect of capping their bonuses.
Premature Sales Recognition
One critical element of the accounting for trading activities is the notion of “mark to market”. The concept is simple (that you value your holdings based on a “mark” of what you would be able to sell them for in the market). It became clear in 2006/7 that there were significant problems with the mark to market accounting for CDO’s. John Niblo – a trader at DRCM – sought prices from Wall Street dealers for his 1B$ portfolio of assets backed by subprime mortgages. Based on the responses he markedthe portfolio down due to market deterioration and illiquidity. UBS Investment bank – holding similar securities – did not do the same. This became a crisis for UBS. Niblo ended up on administrative leave and was later terminated. The effect of this was that UBS was valuing its assets higher than the market – a similar game (in investment banking) to the game of premature sales recognition in industry. In other words there is money on your books that does not correspond to real value. It is a game (and of course the value of the assets as we have seen has a direct impact on compensation as well – so the incentive not to mark to market was high).
CW 13 explained that when the risk management team would “nix” a deal presented by FIRC traders based on UBS’s risk policies, the traders would either finalize the deal in complete disregard for risk management’s warnings or “go up the chain” of command for deal approval – the FIRC trader would go to the head of trading and then the head of trading would speak with the head of risk management to have the impediment to the deal removed.
This game – whereby the risk officer trying to do his/her job has their loyalty challenged (by being accused of being an “obstacle” to the company’s growth strategy etc.) – is one played in many organisations with people or groups that have some governance role that they are supposed to play on behalf of the board or investors.
As we saw in our previous article on the collapse of Lehman Brothers, games played a huge role in the loss of 38 B$ at UBS. It is a staggering loss, with huge economic and social consequences. It underlines the critical role all of us working in large corporations have to try to minimize the games played. The stakes as we have seen, are high.
In “Games at Work” we identified several critical elements to building an environment where games are less likely to flourish. One of those was to “build processes and structures that are less conducive to game playing”. Specificallys we recommended:
• Sufficient simplicity: Complexity is a breeding ground for games. The complexity of the products put through the Fixed Income group kept growing (up to an including the ridiculous CDO squared). In addition the organizational complexity of the duplication of some of the work at DCM and the Investment Bank created this breeding ground. Clearly the Board of UBS was overwhelmed by this complexity
• Collective design: the DRCM structure was proposed and advocated by McKinsey and agreed and sealed between a few senior players. Perhaps this decision could have benefited from more eyes looking at it, and more mouths commenting on it. Similarly the change in scope from hedge fund type structure was the decision of a few people.
• Visible accountability: Rewards and performance must reinforce accountability. In this case they rewarded reckless growth of risky asset classes. This must have been obvious to all who contributed to the bonus design. Allowing John Costas to call the shots also gave him the illusion of being beyond accountability.
• External orientation: UBS lost its focus on its customer base, and got focused on internal measures. As a consequence it saw a massive exodus of its customer base subsequent to the events recounted above. In fact customers withdrew 147.3 B$ in assets in 2009 from UBS.
• Boundary crossing: structures and processes need to be set up to promote cooperation and collaboration. Clearly risk management groups were systematically sidelined at great cost to the firm.
• http://online.wsj.com/article/SB119214581308956665.html http://securities.stanford.edu/1039/UBS_01/2008711_r01c_0711225.pdf