Four steps to make your model risk management 50% more efficient

Developing, validating and monitoring multiple models is hard work. Regulators want high-quality, well-produced reports. Demand for testing is unpredictable, and with no shared understanding and patchy documentation, there’s a lot of back-and-forth. Maintaining the talent pool can be a challenge – but few banks want to add headcount. And of course, budgets are always under pressure.

The right solution might not be immediately apparent, but there are ways to move forward. Provided you access the right expertise, knowledge and support, parallel processing and automation tools can make processes for testing, documenting and reporting on models faster, smoother and more efficient – and at a reasonable cost. In this post, we set out a simple four-step process you can use to make it happen.

Why aren’t banks doing more to optimize their model risk management processes?

One reason is the way most quant and risk functions work today, which could make introducing automation toolkits seem harder than it really is.

We tend to assume that automation depends on scale, uniformity and repetition. And it’s true that models and documentation are often inconsistent across different divisions, with a mix of internal and vendor models, and no single library bringing everything together. There are also multiple data sources, and they’re constantly changing.

However, even small pools of similarity open up opportunities. And with some carefully planned tweaks to job specs, processes, formats and templates, those opportunities can make a big difference.

Four steps to re-engineer your risk model testing processes with automation

  1. Challenge the status quo. This first step is vital. It’s about a shift in attitude, from relying on highly skilled quants and non-standard documentation to accepting that tasks could be split between senior quants, junior quants and programmers. Once you cross that bridge, and accept that processes can indeed be re-engineered, the actual automation is relatively simple.
  2. Break down the task. Instead of trying to automate the whole of the model testing function, focus on the areas where automation can make the biggest difference. In our experience, building individual modules for testing, documentation and reporting has worked best – either as a suite of solutions, or to automate specific areas where clients wanted to make a change.
  3. Share key resources. If all members of the team can access the same documents, and you also instigate parallel processing, everything happens much faster. With a shared document repository and automatic updating of reports across multiple formats, the time gains can be dramatic – particularly compared to working manually in Excel.
  4. Don’t reinvent the wheel. Keep IT investment to a minimum, and work around current constraints rather than trying to break through them. There’s no need to change platforms, templates and IT governance arrangements without good reason. Any new tools should work in harmony with what you already use, using APIs to keep information in the formats your team already know.

The benefits of process re-engineering and automation

Once achieved, process re-engineering and automation can realize a host of benefits.

  • Validation cycles can be shortened by up to 60%; for complex models, that means from three months to five or six weeks.
  • Documentation can be more standardized, and reporting can be better, more transparent and more in tune with what regulators want.
  • By doing the same work with lower headcount, you can free up capacity for more models to be validated and reviewed, and deal with peaks in demand with a smaller team.
  • You also free up highly skilled and specialized quants to deal with a wider range of models, addressing a skills risk and potential bottleneck.
  • By cutting down on mundane tasks, risk teams can focus on improving Model Risk Management by spending more time engaging with model users and owners.

Overall, we believe that process re-engineering, parallel processing and automation of model testing is both desirable and easily achievable. With the right support, you can realize big time and cost savings without disrupting existing workflows or rethinking your whole IT setup. Check out this case study to see how we helped a leading US investment bank achieve 50% efficiency gains by taking a smart approach to automation.

If you’d like to learn more about how automation and process re-engineering could help you, click HERE to send us your query.


Anna Slodka-Turner
Global Head, Risk and Quant Solutions Posts
Mamta Mittal
Head of Model Governance Solutions Posts
Mostafa Mostafavi
Senior Quantitative Lead, Financial Services Posts

Latest Posts