Nine highlights from our 2019 Analytics and Scenario Forum
- 31 July 2019
By Giuseppe Aloi, Scenario Programme Manager
On 3-4 July, more than 50 operational risk and scenario experts met in London for two days of discussion and networking. This year's agenda focused on:
- Enhancing scenario estimations using data driven approaches
- Maintaining and refreshing scenarios
- Operational resilience: how firms are reacting to the new challenge
- Machine learning for operational risk management
- Quantifying emerging risks
- Approaches in place to quantify cyber risk
I'd like to thank everyone who attended the event, especially those who presented. This year, we had guest speakers from AON, ICBCS Standard and McKinsey & Company. Read on for my nine key takeaways from the forum.
1. There's widespread interest in structural approaches to scenarios
Many forum participants are starting to adopt structural approaches, also referred to as a “risk drivers-based model”, to estimate scenario frequency and impact.
2. Risk language should suit 1LoD definitions and needs
Although regulatory reporting on Basel event types continues, many institutions have developed taxonomies that more precisely capture the risks they face in today’s business environment. Our recent work on developing a new operational risk taxonomy has shown that there is significant industry interest in an enhanced risk taxonomy.
We've now begun the second stage of our work on creating a taxonomy. In this phase, we will dig deeper into level 2 risks.
3. Increased responsibility for 1LOD to maintain scenarios
The 1LOD is ultimately responsible for the creation of scenarios. They agree storylines, assess the impact and sign off the scenario outcome. Because of this, many forum attendees feel that they should also own the maintenance of scenarios. This means the 1LOD is responsible for monitoring when a scenario no longer reflects the risk exposure of the firm, and then for updating it accordingly.
4. External information is vital for scenarios but there's still room for improvement
Using external information to inform the scenario process is a hugely important activity. Some firms rely on external parties to collate data, while others perform this activity in-house. To support this process, ORX Scenarios offers its subscribers custom Risk Intelligence Packs. These are detailed, bespoke packs about a specific risk, which draw together information from across the ORX service offering. We highlight the information most relevant to the institution who requests the pack. Contact us to find out more.
5. More awareness around resilience rather than being more resilient
Attendees showed widespread interest in the emerging regulatory focus on operational resiliency. Many said that resiliency was becoming an important topic internally and externally, but that there was still a general lack of clarity around what operational resilience means, and what is expected.
6. Machine learning enhances the level of challenges
There continues to be a huge interest in the use of advanced analytics techniques to support risk management activities across financial services, and the forum participants were no exception to this. We explored some great case studies showing how machine learning can enhance innovation and detection, including:
A demonstration of how machine learning has been used to reduce the levels of false positives in automated AML detection. In this application, machine learning was used to match job descriptions captured within customer accounts against recognised positions. This automatically increased data quality, leading to higher accuracy in detection rates allowing AML case investigators to focus on the most significant items.
7. Institutions look to adapt their economic capital frameworks
In a lively discussion, participants debated the relative confidence they had in approaches developed for capital models (typically loss driven) compared to those used for stress testing (more regression and scenario-driven). Different regulatory requirements resulted in a difference of views among the delegates.
8. Developments in cyber risk management
9. How to quantify emerging risks
In a complementary session, the challenges of quantifying emerging risks were discussed. Delegates argued that the industry response to emerging risks often happens after big events materialise, instead of reacting when the exposure increases.
About the Analytics and Scenario Forum
The ORX Analytics and Scenario Forum takes place each year, and gives participants the chance to talk about the biggest issues facing the industry today. It's a free event for ORX members and subscribers to the ORX Scenarios service. Our events are exclusive opportunities to share knowledge, experience and expertise with peers across a wide range of operational risk-related topics.