LO 42.4: Describe the Societe Generale operational loss event and explain the

LO 42.4: Describe the Societe Generale operational loss event and explain the lessons learned from the event.
In January 2008, it was discovered that one of Societe Generales junior traders, Jerome Kerviel, was involved in rogue trading activities, which ultimately resulted in losses of 4.9 billion. The multinational bank was fined 4 million, and Mr. Kerviel was sentenced to three years in prison. The incident damaged the reputation of Societe Generale and required the bank to raise additional funds to meet capital needs.
Between July 2005 and January 2008, Kerviel established large, unauthorized positions in futures contracts and equity securities. To hide the size and riskiness of these unauthorized positions, he created fake transactions that offset the price movements of the actual positions. Kerviel created fake transactions with forward start dates and then used his knowledge of control personnel confirmation timing to cancel these trades right before any confirmations took place. Given the need to continuously replace fake trades with new ones, Kerviel created close to 1,000 fictitious trades before the fraud was finally discovered.
The operational risk world was galvanized by this event as it demonstrated the dangers of unmitigated operational risk. In 2008, many firms were developing operational risk frameworks and often focused on the delivery of new reporting, loss data tools, and adaptions to their scenario analysis programs. However, even though firms were developing internal risk systems, the amount of new regulatory requirements rapidly overcame their ability to keep up in practice. With the news of Mr. Kernels activities, many heads of operational risk found themselves asking the question Could that happen here?
Page 66
2018 Kaplan, Inc.
Topic 42 Cross Reference to GARP Assigned Reading – Girling, Chapter 8
IBM Algo FIRST provided an analysis based on press reviews. The highlights of alleged contributing factors to this operational loss event are summarized as follows: 1. Mr. Kerviel was involved in extensive unauthorized trading activities.
2. Mr. Kerviel was not sufficiently supervised.
3. Mr. Kerviel used his knowledge of middle and back office controls to ensure his fraud
went undetected.
4. Mr. Kerviel achieved password access to systems allowing him to manipulate trade data. A number of reasons were cited that explained how Kerviels unauthorized trading activity went undetected, including the incorrect handling of trade cancellations, the lack of proper supervision, and the inability of the banks trading system to consider gross positions.
Regarding trade cancellations, the banks system was not equipped to review trading information that was entered and later canceled. In addition, the system was not set up to flag any unusual levels of trade cancellations. Regarding the lack of supervision, oversight of Kerviels trading activity was weak, especially after his manager resigned in early 2007. Under the new manager, Kerviels unauthorized trading activity increased significantly. Regarding the size of Kerviels positions, the banks system was only set up to evaluate net positions instead of both net and gross positions. Thus, the abnormally large size of his trading positions went undetected. Flad the system properly monitored gross positions, it is likely that the large positions would have issued a warning sign given the level of riskiness associated with those notional amounts. Also, the large amount of trading commissions should have raised a red flag to management.
Additional reasons that contributed to the unauthorized positions going undetected included the inaction of Kernels trading assistant to report fraudulent activity, the violation of the banks vacation policy, the weak reporting system for collateral and cash accounts, and the lack of investigation into unexpected reported trading gains.
Kerviels trading assistant had immediate access to Kerviels trading activities. Because the fictitious trades and the manipulation of the banks trading system went unreported, it was believed that the trading assistant was acting in collusion with Kerviel. Regarding the banks vacation policy, the rule that forced traders to take two weeks of vacation in a row was ignored. Had this policy been enforced, another trader would have been responsible for Kerviels positions and likely would have uncovered the fraudulent activity of rolling fake transactions forward. Regarding collateral and cash reports, the fake transactions did not warrant any collateral or cash movements, so nothing balanced the collateral and cash needs of the actual trades that were being offset. If Societe Generales collateral and cash reports had been more robust, it would have detected unauthorized movements in the levels of these accounts for each individual trader. Regarding reported trading gains, Kerviel inflated trading gains above levels that could be reasonably accounted for given his actual authorized trades. This action should have prompted management to investigate the source of the reported trading gains.
2018 Kaplan, Inc.
Page 67
Topic 42 Cross Reference to GARP Assigned Reading – Girling, Chapter 8
Ultimately, the unauthorized trading positions were discovered by chance after one of Kennels fake trades was detected by control personnel during a routine monitoring of positions. Kennels inability to explain the fictitious transaction led to a rigorous investigation, revealing the depth of his fraudulent activities.
Lessons to be learned specific to this operational loss event include the following: Traders who perform a large amount of trade cancellations should be flagged and, as a result, have a sample of their cancellations reviewed by validating details with trading counterparties to ensure cancellations are associated with real trades.
Tighter controls should be applied to situations that involve a new or temporary
Banks must check for abnormally high gross-to-net-position ratios. High ratios suggest
a greater probability of unauthorized trading activities and/or basis risk measurement issues.
Control personnel should not assume the independence of a trading assistants actions.
Trading assistants often work under extreme pressure and, thus, are susceptible to bullying tactics given that job performance depends on them following direction from traders.
Mandatory vacation rules should be enforced. Requirements for collateral and cash reports must be monitored for individual traders. Profit and loss activity that is outside reasonable expectations must be investigated
by control personnel and management. Reported losses or gains can be compared to previous periods, forecasted values, or peer performance.
Page 68
2018 Kaplan, Inc.
Topic 42 Cross Reference to GARP Assigned Reading – Girling, Chapter 8
K e y C o n c e p t s
LO 42.1 Operational risk departments look at events outside the firm to gain valuable insights and inputs into operational risk capital calculations. External events can also be useful in many areas of a firms operational risk framework, as they provide information useful for risk self-assessment activities. These events are key inputs in scenario analysis and can help in developing key risk indicators for monitoring the business environment. Additionally, external data is a required element in the advanced measurement approach (AMA) capital calculation.
Subscription databases include descriptions and analyses of operational risk events, which are derived from legal and regulatory sources and news articles. In addition to database systems, there are also consortium-based risk event services that provide a central data repository to member firms and can offer benchmarking services as well. ORX is a provider of this type of data.
LO 42.2 When comparing data in the FIRST and ORX databases, we see significant differences between them. The FIRST database has a significantly higher percentage of losses for Internal Fraud than does ORX. In contrast, ORX has a significantly higher percent of Execution, Delivery, and Process Management losses. This could be because not all Execution, Delivery, and Process Management events are reported by the press, implying the FIRST database is missing many events and has an unavoidable collection bias.
.Another difference between the two databases with respect to Execution, Delivery, and Process Management events is that ORX data is supplied directly from member banks. However, not all banks are ORX members, implying that ORX likely also suffers from collection bias. This is in contrast to the FIRST database that collects data on all firms, including a significant number of firms outside of Basel II compliance.
LO 42.3 ORX and FIRST databases must be viewed with caution, as there are several challenges with using external data. For example, external data derived from the media is subject to reporting bias because the press is far more likely to cover illegal and dramatic events. The use of benchmark data is also a concern, as there is a chance that comparisons are not accurate because of different interpretations of the underlying database definitions.
One of the best ways to use external data is not to spot exact events to be avoided but rather to determine the types of errors and control failings necessary to avoid similar losses. External data can still have a valuable role in operational risk management if staff acknowledges any limitations. Databases can provide valuable lessons about risk management and highlight trends in the industry.
2018 Kaplan, Inc.
Page 69
Topic 42 Cross Reference to GARP Assigned Reading – Girling, Chapter 8
LO 42.4 Jerome Kerviel, a junior trader at Societe Generale, participated in unauthorized trading activity and concealed this activity with fictitious offsetting transactions. The fraud resulted in 4.9 billion in losses and severely damaged the reputation of Societe Generale.
Page 70
2018 Kaplan, Inc.
Topic 42 Cross Reference to GARP Assigned Reading – Girling, Chapter 8
C o n c e p t C h e c k e r s
Which of the following reasons is least likely to be a motivation for firms to use external data? A. To provide inputs into operational risk calculations. B. To engage in risk self-assessment activities. C. To ignore any operational loss events outside of external loss databases. D. To use in the advanced measurement approach (AMA) capital calculation.
In the IBM Algo FIRST database, which event type accounts for the most risk events? A. Business Disruptions and Systems Failures. B. Execution, Delivery, and Process Management. C. Clients, Products, and Business Practices. D. Internal Fraud.
IBM Algo FIRST Which database is likely to suffer from selection bias for Execution, Delivery, and Process Management losses because not all events are reported in the press? I. II. Operational Riskdata eXchange Association (ORX) A. I only. B. II only. C. Both I and II. D. Neither I nor II.
Which of the following statements is least likely to be a limitation of using external databases? External databases: A. must be viewed with caution. B. suffer from collection biases. C. do not report all events. D. cannot be used in internal calculations.
Which of the following statements was not a contributing factor to Jerome Kernels activities at Societe Generale? Mr. Kerviel: A. engaged in extensive unauthorized activities. B. engaged in rogue trading despite being sufficiently supervised. C. had knowledge of controls to ensure his activities were not detected. D. gained password access to back office systems to manipulate data.
2018 Kaplan, Inc.
Page 71
Topic 42 Cross Reference to GARP Assigned Reading – Girling, Chapter 8
Co n c e pt Ch e c k e r A n sw e r s
1. C Operational risk departments look at events outside the firm to gain valuable insights and inputs into operational risk capital calculations. Firms should understand that external loss databases only include a sample of potential operational loss events.
2. C Forty six percent of all records in the FIRST database fall into the category of Clients,
Products, and Business Practices, more than any other category.
3. A Because not all Execution, Delivery, and Process Management events are reported by
the press, it is likely that the FIRST database is missing many events and, thus, has an unavoidable collection bias.
4. D The use of external databases is critical to firms operational risk management calculations, an
example of which is observing fat tail events at other firms.
5. B Mr. Kerviel was insufficiently supervised according to IBM Algo FIRST.
Page 72
2018 Kaplan, Inc.
2018 Kaplan, Inc.
Page 73
The following is a review of the Operational and Integrated Risk Management principles designed to address the learning objectives set forth by GARP. This topic is also covered in:
Ca pi t a l M o d e l i n g
Topic 43
E x a m F o c u s
This topic discusses approaches for modeling operational risk capital requirements. Collecting data for loss frequency and loss severity distributions is an important component of allocating operational risk capital among various bank business lines. The loss distribution approach (LDA) models losses with respect to both frequency and severity with the goal of determining the appropriate level of capital. For the exam, be able to compare the approaches for calculating operational risk capital charges and be able to describe the LDA for modeling capital. Approaches for calculating operational risk capital requirements will be covered again in Topics 44 and 38.
O p e r a t i o n a l R i s k C a p i t a l R e q u i r e m e n t s