Bearing Limit
"Bearing limit" refers to the amount of cost or uncertainty a person or organization (or any other component of a distributed system) can contain or can be expected to bear. (Uncertainty here may include risk, surprise, instability or variability.)
Between 1994 and 1996, a stand-off emerged between two speculators. Yasuo Hamanaka, Sumitomo's chief copper trader, was long on copper. Julian Robertson, of the Tiger Fund, was short on copper.
By mid 1995, Robertson un-owned about $1bn-worth of copper. Robertson was convinced copper was over-priced, but the price was still rising. Other speculators (notably Soros Fund Management) bailed out and took their losses. Robertson held on.
By spring 1994, Hamanaka probably owned about $10bn-worth of copper. When this was discovered, the price of copper fell 30%. Tiger made $300m in one day. Sumitomo lost billions. History refers to this incident as the Sumitomo copper scandal (and not the Tiger copper coup).
Alistair Blair reviews the story in his weekly column. "Mr Hamanaka ... couldn't squeeze any more copper trade tickets into the bottom right hand drawer of his desk, where he had been accumulating them for two years in an attempt to recover losses ... think of Nick Leeson times three". In contrast, "Mr Robertson ... was not betting his bank. He had conviction and self-belief, but he also had discipline. ... He could remain solvent for as long as it took." (Investor's Chronicle No Free Lunch, 19 November 2004)
Rogue traders such as Hamanaka and Leeson exceed their bearing limits, and may damage or destroy their respective organizations. Meanwhile, Robertson remained well inside his bearing limit.
Of course markets are irrational. Sometimes rogue traders get away with it; sometimes disciplined speculators lose out. But distributed risk management (not just in financial trading but in a wide range of domain) always demands attention to bearing limits.
We introduced the term "bearing limit" into our risk management practice following the Nick Leeson scandal and the fall of Barings Bank. (The pun is deliberate.) More recently, the term "risk-bearing limit" was used in a Sept 2001 paper Reflections on New Financial System in Japan (pdf). We have also started to apply the term to questions of knowledge and intelligence. (See weblog post on Bearing Limit and WMD).
Any given party has a bearing limit, which defines how much cost and risk it can bear. Above this limit, the party cannot be expected to contain the costs and risks allocated to it, and these may spill over the contractual boundaries to its partners. In the worst case, a party unable to bear its costs and risks goes into liquidation, and the remaining costs and risks then have to be picked up by another party. (Think about PFI and the possible demise of Jarvis.)
In some cases, the bearing limit can be determined fairly precisely. This is particularly true in cases that are covered by various forms of indemnity insurance, since the bearing limit can be taken to be equal to the level of insurance cover. In other cases, the bearing limit is itself a matter for negotiation.
Within a hierarchical organization, there is a bearing limit at each level of the management hierarchy. In other words, there is a maximum level of responsibility that can be delegated downwards. Above this limit, the responsibility remains with upper management. (For example, if a trading bank loses half a billion dollars, this cannot be blamed solely on a rogue trader with an authorization limit of 50 million dollars. To pretend otherwise is either foolish or corrupt.)
Between 1994 and 1996, a stand-off emerged between two speculators. Yasuo Hamanaka, Sumitomo's chief copper trader, was long on copper. Julian Robertson, of the Tiger Fund, was short on copper.
By mid 1995, Robertson un-owned about $1bn-worth of copper. Robertson was convinced copper was over-priced, but the price was still rising. Other speculators (notably Soros Fund Management) bailed out and took their losses. Robertson held on.
By spring 1994, Hamanaka probably owned about $10bn-worth of copper. When this was discovered, the price of copper fell 30%. Tiger made $300m in one day. Sumitomo lost billions. History refers to this incident as the Sumitomo copper scandal (and not the Tiger copper coup).
Alistair Blair reviews the story in his weekly column. "Mr Hamanaka ... couldn't squeeze any more copper trade tickets into the bottom right hand drawer of his desk, where he had been accumulating them for two years in an attempt to recover losses ... think of Nick Leeson times three". In contrast, "Mr Robertson ... was not betting his bank. He had conviction and self-belief, but he also had discipline. ... He could remain solvent for as long as it took." (Investor's Chronicle No Free Lunch, 19 November 2004)
Rogue traders such as Hamanaka and Leeson exceed their bearing limits, and may damage or destroy their respective organizations. Meanwhile, Robertson remained well inside his bearing limit.
Of course markets are irrational. Sometimes rogue traders get away with it; sometimes disciplined speculators lose out. But distributed risk management (not just in financial trading but in a wide range of domain) always demands attention to bearing limits.
We introduced the term "bearing limit" into our risk management practice following the Nick Leeson scandal and the fall of Barings Bank. (The pun is deliberate.) More recently, the term "risk-bearing limit" was used in a Sept 2001 paper Reflections on New Financial System in Japan (pdf). We have also started to apply the term to questions of knowledge and intelligence. (See weblog post on Bearing Limit and WMD).
Any given party has a bearing limit, which defines how much cost and risk it can bear. Above this limit, the party cannot be expected to contain the costs and risks allocated to it, and these may spill over the contractual boundaries to its partners. In the worst case, a party unable to bear its costs and risks goes into liquidation, and the remaining costs and risks then have to be picked up by another party. (Think about PFI and the possible demise of Jarvis.)
In some cases, the bearing limit can be determined fairly precisely. This is particularly true in cases that are covered by various forms of indemnity insurance, since the bearing limit can be taken to be equal to the level of insurance cover. In other cases, the bearing limit is itself a matter for negotiation.
Within a hierarchical organization, there is a bearing limit at each level of the management hierarchy. In other words, there is a maximum level of responsibility that can be delegated downwards. Above this limit, the responsibility remains with upper management. (For example, if a trading bank loses half a billion dollars, this cannot be blamed solely on a rogue trader with an authorization limit of 50 million dollars. To pretend otherwise is either foolish or corrupt.)
0 Comments:
Post a Comment
<< Home