THANK YOU FOR SUBSCRIBING
In Market Risk Management, advances are happening in three different layers
First, the organisational layer – Many agile organisation forms and working methods were introduced in risk management in the past years. There are usually now no separated IT and Risk management teams anymore but they work closely together and have the same targets. Colocation, sharing the same language (tech/business jargon), inter-area knowledge sharing, and removing bureaucratic legacy barriers help to reshape the development and operational processes transparent and efficient. Another idea is the broadening of skill profiles that for new small-sized innovations not very many people are needed for the whole process. The specialisation in small parts of the process because of special technology and special skillsets is hindering fast end-to-end delivery. Last but not least everyone in the risk management organisation including IT people shall share the same target vision and understand the contribution to the whole organisation.
The second change is towards efficiency: the avoidance of manual steps and manual operation steps in the risk management processes. In the market risk area, the daily process tends nowadays to be fully automatised with intelligent data quality mechanisms, inconsistent data is replaced by data with usual typical characteristics. As far as possible default content replace missing data fields not available and manual corrections are kept to a minimum. Also for meta data like new portfolio/strategy allocation and for consolidation tree maintenance, rule-based methodological support is supported.
The next step would be to introduce AI and remove the last risk analytical manual steps in the risk management processes e.g. automatise limit violation logistics, if a limit is broken, usually, analytical work kicks in, why the limit is broken? What new position or position change is causing the limit violation? Are events on the market or the development of a special risk factor responsible for the limit violation. This work has not been yet automatised because it needs a lot of contexts and a lot of meta logic and AI is currently not implemented in the calculation processes. Currently in the adaptive setup, for each new base case, a tool is on the fly implemented that helps in the analysis and reasoning.
Currently in the adaptive setup, for each new base case, a tool is on the fly implemented that helps in the analysis and reasoning
The third advance is towards the architecture of market risk management systems – there are no monumental tightly integrated risk solutions anymore. The main reason was the inflexibility to quickly integrate new products, react to market events/developments and fulfil ad-hoc regulatory requests. For instance, the EBA Stress test with ever-changing EBA stress test methodology is a nightmare for an inflexible system. Usually, the market risk management department helped himself by running the analytical calculations mostly outside of the system because the system was not supporting the exact parametrisation and they were not able to change the system timely to use the system for the EBA stress test.
The new architecture is one where modules are not integrated but only loosely coupled and inbound and outbound logistics are stateless, flexible and support multiple types. In such a framework it is much easier to implement new requirements and also to adapt the system to an ad-hoc request like the EBA Stress test and then work in this consistent environment. The caveat is that this requires intelligent and well defined ‘glue’, how the different modules are connected and what technology is used for this ‘glue’.