Automation of stress tests and the use of models

Automating stress testing at banks

The level of automation is low across each stage of the stress tests cycle, in part due to the level of change seen over the past few years and relatively low levels of investment in technology and systems to support automation.

1000

Related content

Business people with digital tablet in modern lobby

Understanding the costs of stress testing

With requirements for stress testing increasing year on year, it is surprising to note the limited regular monitoring of costs. Only around 10 percent of banks reported regularly monitoring stress testing costs, while a further 50 percent of respondents have made an adhoc cost assessment. This lack of formal monitoring of costs could suggest that banks have simply been responding to regulatory demands, with little thought given to optimizing processes and embedding them into the business. This is also reflected in the level of investment which, over the past three years, has been modest compared with the annual delivery costs.

Our analysis also found that the level of automation is low across each stage of the stress tests cycle. This is partially due to the level of change seen over the past few years and relatively low levels of investment in technology and systems to support automation. While this may be expected in areas such as scoping, planning and governance, there are areas such as work flow management where the lack of an automated process or management tools may contribute to institutions having poor transparency of the entire scope of stress testing tasks and dependencies involved. This is having an indirect effect on cost assessments and hampers efforts to optimize processes and reduce timescales.

Future investment in automation

The good news is, over a third of banks expect to invest more than US$100m per annum on stress testing in the next three years. This investment is expected to focus on data frameworks that support stress testing and other risk management processes, and also in supporting a move towards greater automation. Increasing the level of automation in the stress testing processes will improve controls and reduce timescales. Investment is also likely to focus on improving data capture and reconciliation, model calculation and the aggregation of results with more than 60 percent of institutions seeking a high degree of automation in these processes. Automated solutions are also being sought for the population of regulatory templates.

To a lesser degree, but still improving from the current position, greater use of automated tools in process management and scenario generation is also expected to benefit firms. It is important to note that human judgment and expertise will continue to be needed for reviewing and challenging the test results.

A focus on models

The introduction of a new capital framework (Basel 4) and the implementation of accounting standards such as IFRS9 are likely to lead to a greater reliance on the use of increasingly complex models. In our survey, about half of respondents reported that they are already using challenger models. A further 37 percent intend to develop such models. This will require substantial model development resources, as well as appropriate resources within the independent model validation teams.

Use of challenger models to validate results

Recommendations

We see a clear intention from banks to invest in systems and tools to increase the level of automation across the stress testing process. It will be important that banks focus their automation efforts on key processes in order to improve timescales and support internal and external demand. We also believe that institutions must continue to increase the scope and quality of model validation standards to ensure robustness of existing models and to derive value from new models.

Connect with us

 

Request for proposal

 

Submit

KPMG's new digital platform

KPMG International has created a state of the art digital platform that enhances your experience, optimized to discover new and related content.