Do you have the right tools for the job?
A recent McKinsey report highlighted some of the challenges that organisations are facing as they manage their model portfolios.
In summary the article stated that greater workload and stress is being placed on model development and model validation teams due to:
- Model inventory sizes are increasing by over 30% for many banks
- Related to this is the expansion and scope of Model Risk Management (MRM) activities driven by regulations
- Organisations addressing operational costs due to rising interest rates and increased economic uncertainty
With current time-to-market life cycle for modelling often between 15 and 18 months, pain points exist throughout the process.
To address the model lifecycle challenges and drive greater efficiencies across lifecycle activities, McKinsey highlighted 4 efficiency levers summarised as follows:
1. Creating re-usable processes and assets, with greater automation
2. Better model inventory management, and parallelisation of activities across the lifecycle
3. Established standards and procedures for the model development process
4. Skills, training and collaboration
Powerful software tools that have been purpose built for model development and model management can help address these areas.
Our Modeller tool covers the following areas to enable compliance and efficiencies, primarily addressing points 1,3 and 4 above:
structured, repeatable model building processes
complete audit trails
storage, recovery and review of model versions
complete model documentation
ease of packaging and sharing with independent validation teams for model validation and review
Focus is our Model Risk Management solution that primarily addresses points 2 and 3:
centralised model inventory
standardised workflows for all model lifecycle stages
remedial action management
dashboards and reporting
resource and activity planning
documentation, standards and reports repository
Many organisations are driving modelling efficiencies through the adoption and use of Modeller and Focus.
To address these challenges and drive greater efficiencies across lifecycle activities, organizations are adopting and using tools such as Modeller and Focus. The benefits of these tools are significant, as users report that they make modelling easier, faster, less error-prone, and more straightforward to understand.
Modeller provides a system and user interface that enables scorecard developers to see and do exactly what they need to. The Modeller user interface provides ease of use throughout the modelling stages, from data processing through to model development, validation, and reporting. With its many capabilities, Modeller requires little training to take full advantage of (McKinsey lever 4).
Modeller is rigorously tested, used, and proven as a best-in-class tool for in-house model development teams. It drives process, control, transparency, and efficiency, enabling model development teams to increase in size and quality output with ease. Modeller also provides many built-in tools to make model building easier and faster. These include hold-out datasets, sample weights, grouping of special cases, auto handling of special cases in model builds, and standard reports. Field Reducer, Grouping, Reject Inference, Modelling algorithms, Explainability, and Reporting all comprise automation choices to enable the user to strike the right balance between automation and user oversight and control (McKinsey lever 1).
Automated yet controllable Field Reduction is another powerful module in the era of Big Data. It is fully parameterized so that users have full control over which fields are kept and which fields are dropped based on predictive power and correlations, with full manual override functionality.
Model development and validation teams can quickly and easily compare traditionally developed WoE logistic regression models with models developed using ML algorithms such as Random Forest and XGBoost. Simply swapping in or out a model algorithm on the same modelling canvas means comparisons are fair and informative, without the need for any duplication of effort regarding data processing or model reporting, etc. (McKinsey lever 1 and 3).
Modeller records a complete audit trail of the process during the development process of every model, including design definitions, characteristic groupings, selection/dropping of the variables, shortlisting of the variables, all versions of models stored with the ability to roll back/forward. The user has the ability to annotate the audit with reasons and justifications, making documentation easy, efficient, and comprehensive (McKinsey lever 3).
Focus and it's fully customisable dashboards provides a consolidated view of model health and remedial actions across the organisation. The Focus workflows capture all decisions/approvals for all model lifecycle stages and have been designed to contain all the steps detailed within the organization's standards and regulatory requirements. The remedial action management ensures all actions are logged, assigned, viewed, reported, and automatic reminders sent. Nothing gets missed or falls through the cracks (McKinsey lever 2).
In summary, Modeller and Focus offer organisations a direct solution to the challenges highlighted by McKinsey across the modelling lifecycle. By using the right tools for the job, organisations can streamline their modelling and model management processes with key efficiency levers. As McKinsey has pointed out, these efficiency levers can improve transparency and consistency, reduce the risk of errors and attrition, and ultimately enhance team health and achievements.
“Make sure that you always have the right tools for the job. It's no use trying to eat a steak with a teaspoon, and a straw.” - Anthony T. Hincks
If you’re interested in becoming part of our User community please get in touch.