Stress testing has been increasingly used in recent years as a fundamental tool in examining the vulnerabilities in banks balance sheets and their ability to withstand adverse economic scenarios or shocks, such as the one experienced in 2007/2008. Year-on-year stress tests continue to become increasingly challenging in order to enhance financial stability, the idea being that stress tests measure a banks resilience to ensure sufficient capitalisation to withstand such adverse scenarios occurring again in the future. Whilst many institutions have gained a grip on the general stress testing process, a host of regulatory changes and heightened levels of scrutiny have meant financial institutions face a burdensome task in balancing stress testing compliance, whilst ensuring they are adding value to their business through these stress tests.
Whilst European institutions focused predominantly on purely achieving compliance in the initial years of the PRA and EBA stress testing exercises, there is now an increasing demand to incorporate stress testing into business decision making and ensure stress testing is not just a regulatory tick boxing process. The first research piece therefore focused on gaining value out of stress testing, as well as its synergies and interaction with IFRS 9 and balancing the management of multiple regulatory requirements. This piece will explore three more prominent areas and challenges that surfaced during the extensive research The Center for Financial Professionals conducted with stress testing professionals throughout the UK and Europe. These include the move towards a more automated stress testing approach, ensuring sufficient data quality and the use of artificial intelligence and machine learning in stress testing.
View more at www.stress-testing-europe.com
During research with some of the industry’s leading practitioners, one of the most prominent areas that continually featured was banks’ striving to move towards a more automated stress testing process. There is an increasing desire to speed up the whole process and increase overall efficiency. Year-on-year or cycle-on-cycle the PRA, Bank of England and EBA stress tests become ever-more complex with more scenarios and therefore more variations in results produced. When you couple this with the desire for banks to increase the amount of scenarios that are stressed internally, it seems impossible to keep putting a significant amount of time and effort into every scenario, unless teams continually expand. Therefore the industry is looking at ways to go with either a slightly lighter touch on a larger number of scenarios or finding a more efficient way of getting an in-depth view across multiple scenarios.
It is certainly a challenge to do ‘more with less’ but one of the main difficulties banks are finding is automating to a degree of quality and speed. For example, banks are constantly striving to improve the lead time of internal and regulatory stress tests, this involves creating and constructing scenarios quicker and therefore ultimately increasing efficiency, but how do institutions go about doing this effectively? Similarly, one of the challenges of increasing efficiency is sometimes the lack of readily available data. For institutions with a global footprint, getting a uniformed set of definitions for data can become a near impossible task. Therefore institutions are constantly facing an uphill battle to aggregate data effectively, and bring together in a unified format. As a whole the industry is striving to move towards a phase where stress testing will be fully automated with less concentration on certain months annually. This concentration impacts head count and presents a new challenge in staffing and resources, and demonstrating value throughout the year.
A further focus and one that inevitably continues to appear as a key area is the data challenges involved in stress testing and ensuring sufficient data quality to obtain accurate and timely stress testing results. There was a unanimous agreement amongst the industry that the bar is constantly being raised on data quality, with the expectation being that data quality, accessibility and granularity is raised year-on-year. Additionally, alongside the heightened expectations on quality, control and useability of data for stress testing, there is inevitably an increase in the quantity of data involved in the process. This poses increasing challenges as to the way banks approach stress testing, reducing the ability to be flexible and meaning banks have a tough task in managing this big data to ensure accurate stress testing results. When you couple this with the constant need to achieve BCBS 239 compliance for stress testing activities, it is unsurprising that continually raising expectations mean data remains to be an ever-growing challenge for stress testing experts.
A third area amongst many that was brought up by various practitioners throughout the research was a futuristic look to stress testing in terms of how artificial intelligence (AI) and machine learning can be utilised in the process. As mentioned, with the constant need to improve efficiency and speed of the stress testing process, the near future is likely to see an increased use of AI and machine learning, both of which are becoming increasingly popular across a wide range of industries. The idea is that the industry will find ways to use AI or machine learning to undertake some of the tasks that are traditionally done manually. This will inevitably see a move to more objective rather than subjective outcomes, however the main benefit will be the speed that machines can navigate the process in comparison to manual use. One of the challenges of this is that it requires an inevitably large amount of data and information to determine the optimal ways and methods to use AI and machine learning, however there is no doubt that in the near future its benefits could be seen and utilised within the industry.
Overall there is much for stress testing professionals still to ponder. Just from a regulatory standpoint the industry is currently looking towards the results for this year’s PRA, Bank of England Stress Tests and the EBA 2018 stress testing exercise. As the industry progresses with stress testing, looking towards the future and moving towards a more automated approach, it faces ever-evolving challenges. The bar is consistently being raised on data quality and with increasing intensity and diversity of scenarios needing to be processed and results produced there is an increasing desire to do ‘more with less’. One of the areas being explored to increase automation and efficiency, is to utilise AI and Machine Learning solutions, but implications could be wider spread with this being implemented.
As stress testing continues to evolve and progress there are still many challenges outside the above discussed in this piece that need addressing. Join colleagues and peers to network and discuss these challenges at the Center for Financial Professionals’ 5th Annual Stress Testing Europe Summit, where senior regulators and stress testing professionals will come together to review stress testing requirements and evolving processes as the industry moving towards a more automated approach. Over two days on September 26-27 the Summit will bring together a diverse line-up of 20+ senior industry professionals from a range of institutions to give their insight on evolving stress testing requirements and best practices.
For full information visit http://www.stress-testing-europe.com
For further information or help registering, please get in touch with a member of the team on +44 (0) 207 164 6582